From 06afe168b20654ae55d880b2860455d814cb44dd Mon Sep 17 00:00:00 2001
From: "daewon.lee@interdigital.com" <daewon.lee@interdigital.com>
Date: Mon, 15 Dec 2025 11:57:21 -0500
Subject: [PATCH] Minor updates to align formatting with MS Wrod.

Including perl scripts that allow diff view.
---
 38202-j00.docx            |  Bin 1068306 -> 1069263 bytes
 38202.tex                 |   46 +-
 3gpp.sty                  |   20 +-
 scripts/latexdiff-fast.pl | 6031 +++++++++++++++++++++++++++++++++++++
 scripts/latexdiff-so.pl   | 5927 ++++++++++++++++++++++++++++++++++++
 scripts/latexdiff-vc.pl   |  913 ++++++
 scripts/latexdiff.pl      | 5470 +++++++++++++++++++++++++++++++++
 scripts/latexrevise.pl    |  584 ++++
 8 files changed, 18959 insertions(+), 32 deletions(-)
 create mode 100644 scripts/latexdiff-fast.pl
 create mode 100644 scripts/latexdiff-so.pl
 create mode 100644 scripts/latexdiff-vc.pl
 create mode 100644 scripts/latexdiff.pl
 create mode 100644 scripts/latexrevise.pl

diff --git a/38202-j00.docx b/38202-j00.docx
index be8ca8bdf709572a6da0756e31339e6b0c47ae18..4073387e810cc2b9e8279ea590b4188447f76129 100644
GIT binary patch
delta 59717
zcmbO<%klg|#|?AY>sNTI^mQI(V5oh-#URMQz)+rFl%k)KpIn-onpdJ%k(=`-_IA<g
z6IJ{CL+>s0-g&|&_2weW$F|?(95$qdUOSkY{cWAs+f|w@xt)<5915rBg}nQuvh%dB
zT#~)lJe9xZ3z(gZrcIo_^~fwK?pg2l&;I}6>-~E9A1}}IUuL*FS><c};t!kJ&+n8|
zS}0vvXR+_iU#l(WugG~BtAClJz;!b9{lq1E?c@JGw*O<Y|ASJwnQCI2MyJ8v8OPtH
z=1)rf=(>OYE@7k6OP^Q1R|}N7oBBM@<gI&o`9AMopEW-IvT3(06IGw+eQC<IbgPGZ
zCx7EAdgpAM75C^&W}<?|x4WH_Q=Zf(dNSM3?znRO*DE!}I)0}wyp?-Ay(-K!ck-{j
zrJ(Z8wPRaW^Q0~EOJ_}={C)9}{Ggz3i&b)hg1#@_ve56cFKd62(53eBK%-Kls@Xli
zgVldpb)VOGc;n=siSj>gDla{GQdROe|NeIum7=m76SXSNFI(_s`9yc7gW4-yrtggu
zimMlSdS2}5!k6NeKOdzBHnE&twe)Ic?WYN>nPJm>XMNDqx6^27EnGW)Lc=FRuFIYB
z`WF&+>6`v42ol<P|7gSS%z$UJkMCHxGPik~qh6cpLAA5`DdxPNroCU*v6Jl*i%_?a
z@2m}R={F~7UVV^yw0dJ&>&_jDKc=w!;APt6EmbC_TF<+6!Nv0F*LX|*w55e!I=Pgo
z<d14{p)=1Kv$WO*^J+29IE~wk_5n4mVXso}J*j-+CHZyXn#q>J>ah~OejnE<3QQ3>
zD%!QPiFM-c1I7zH=N&i6bo@}bOX+aMA0y}5`7WC)JkQokZsqqrF|Yaa7oMorgX^0D
zer-B-$8XLIrJq;o_x>>8TYN;4b&Knj8o{j&M^qT<S1KLm->t&0Bl~6fjvd?&4$W3<
zIC4+4l*xVKcaHsrwF+@Z{2tXTFgx;^Ir=%v(&Isy;<FrArF6F)JAGyGocMtD%e!~J
zDU?}WB=k(&^E#(W7qer~w4ZEG&dp@K=~Wi?wDeG2-sa;C+x|Yfx6rHpXMWXs{*~pO
z#;Vsjednq3wca*3#p9GISaBy_M9%eM_=^qs#<wnCKEuEub4B{nqv?BZND9vpW(k_D
zo8n&-^in=mI%VUEB^8Hko^+d(#VE|2Fh%A2l5*8O`}esWj%KmkZ1X?LHK2FhX)ZCT
zuwvB}w=>>ee6jlTHz_y41xKe{-CSR6E^q1i^l6pu>DfP~#~u~s%x(*Oa^#j}`MI~J
z@;*&U(7#%~d0YCvgO_cDerTPp&-h`->k;_UGkwmrwAXbp%52{_UcOdPy1Z}Rai%zS
z!KFXS0|X4`u`1oRoEXd7Y0X)Co%2GjPt1HJlj1b#J^K!1%j6ebGOzfX7`=W!>klFR
z`n~Fko)X#@Ok3lsnE%yYWy*LjdN-2u$^_eEQ%=k<>rDI_ykJgjtop_GN6!Ac(y~Of
z>Zx@um(!+P&yBtg{qKDQqNfNO>aWP$zoqqC@8h3>mtsHIeG|*xx@z@#mBn#quOADT
z61z9%-zF`NWk*!*Zu%1*x0d&({mE-h?dH2xcE1*_zx7B%|6`kl0Utx|4I$orwtZqU
zyO(PD#=aK07`}J;OdFFbmzymMDn89H`pr4VeUb@7so!La(^uGE=j*L7Qa@q#rKEcQ
z&fKImdqupTrSiYsJUf;D-0Ib~6`R-PuV&wLHqqmn^q+s<g0JT8P-H9&+T>kt9y?q8
zWZuVP1;zc#XFAs(E}W&oU*P&<YnFcUMY$WI4?j1lWIW<~DDJ;5TqEatd;IEO20Pz$
zO08#{dh#rfTl<rPg~mJ9hfX~sx%6nhr^(0T!eI)1FET`nZPdm0=o$ZiH?5Mn=y<`(
zP9xDtMT-@heAd`(6*#v|dr^to(Z4}#rOq~KnLh>IZ}*S4``?xtT0h0b_l)e-RLQQS
zZlO1hM~yF3#6D(ob6p^n|8m>Lyt7haKEg(3!Yg`Gqmq0Vu(m3TWqb_MW;&awR&dDh
ztf#K^<>31Z&+?qVXDnW$c{79MoXpv$wifXjt2I;#e>R+}))6w$7fohWFHsL<+H|08
z$%jR(J1XiYI<c_q(qIw#`?j9Z<i&QMg?A@NyLL|hRUaVsG2+>d@;CYXt|DS9Y!tj*
zep?o~PqLAz^)C+Ecf`+arp@2TKapH7<a_F`eX_sA{BohSJqO3ymwd}EOnQ-BcQ)!!
z-o^J5<(DZqdmTRS!L+`np7G<x$OlW_NKLHg*RTF3%k^i1!v%AfjsG2*`oDZzU%zj<
zuR!>eQ^m_RYyO()yhCl~;{Sioe5sh|^V^vJ@0rVQd_HQwWJ=08p`9|-cIr#<LtmCI
zbKK%-+WJ)2RC{-Jn#onA^S`>UO}T$f<ILKi7we7XRJ>FDe6404yp!=|>dA?_%;xzY
z>RBnYBz1vgo=97Wb@6-kh2ahcyBY-4cd<)_)GxK2pH>?5Dm^{t4zug$Eb*+J7cPeW
z{j&06;Mdcc|1Yn)BsR;K{aZ`|-~8;LIe#+`-TUVB?EKXkrg}|2KCl0*G*%Nh@5*`D
zBunwi?<J8&_Q^BDZZW_8|LNkdNwe=xxSHqde&n^H)ds10iI{KyymFuYxmCV7(&^kv
z|3geY9}@OHtzWRks`~xud6KP?PviKXmPDql`NXw6iLvTH>h4oB{PVs?|N0fXU|0V3
zz?Dv0Q#&nJ?OvV1lz6m;Z*PzOAAvWLm36at^%Wm+gdcnK>hU8ksf)=EuZf0Cu5b+Z
zWv=3@VAyr^PL}h6d+SdLn6YhbK3O@#{kV~Ds>Sz;o0nDM9?dva@4Rfc^{F`R^IUfp
z&N}w*Q1!P99RHqK?me+|E>}nUFUiX5ktgOw8GQ11bxBF!k`4Ew!Y{Yy^mH}7d;a_T
zmhCfttegAYC*0mr{PEN&{a1W06{QvTcW$h)(&{<GDaxc7+I!IVf9J3GsMlO88E;(G
zxjDzMVphnU&l7_dhl_t)6#3;<{n0Cjs<l)r(swLeqmgpL*10Y9<eWa{v!`6Qo&1y<
zG|A;NXQ9mpsTnHQR&Kbxbz3i!{kF_@PSXRXN12!^CLeA*65$*AD|wb^xaH>m&Ech7
zEPD#myyjG@6?K&^%U)EptYDf_(V{NjoPX+5mbDpqJKmg=An<$Xli7ap>wK0Jw;rp1
zT5u??&veC~RNptx^BHq}g}=9bT7G`r=9ZLqEP4m6H7@<u=9fsHcBWzXlwH9YCjUh$
zLlVyEbW0e^y_=-@=Mvi@cD0Mg7g<_Xmr2ysM_Tz^dSlzUj6<*M9QUQU)xV@FuNiTx
zv<sSaUlnZ0=KO2ZYAP=k6LabR&CR0Pk$cbk)|<FHw@u}qaL07QBm?yut8S$zTrG2X
zx7PCJSDr;8ENdRWI<Ul+Z7rXVh~t%|TN_VGm4BM|H}l;A=~rvIPd~pKG1I7#@7S?~
z;?LZxPDBV<Xg*K7xK)o?bVo|;V~3-{DhVO>Edl&2Y#Hi1Bfgpy2AL?SEZI@%&(w9g
zWW|egVf&Q&iHeEgJ%^sTEWNOaZL50RV~Zn~uWUPFwXNlGw}jO%MyuB8+nVl8P`S)p
zFEfq*Qlw9Ik+8kq)L)%teD+&Ubg!15=*<2j`1wRGp}S{=w05vu{CHPm(VhjLmh8=u
zdp2A3iiu~}qTRd)-`TrfT+le>_4DVxS~`c;>aWV1@`Pb)?A?|18=q;-KliE7Q7n+p
zPD#D(zzSi9nkh$rcYk;heQ}mq=`(MmGN=FiY1&_Uy|WW%xw!i|1{84qylkOx!0J(`
z>y5q%f3_IqK3>nV!m;#y+L^7|A38k+)%?9?{E}NL+<Ne_B@^pCE%ocFg>J1aO!rTQ
zryHd;$_4)y@Ugj_@RpG!qW*l()~Mp8x57+cd!%?8l<Amc9M~u=<nB3bn?{Rn5}TTS
zj#7BUvDe#~h1ZtpMZfZpzue=a8RWA00P~CvH%VFFfbPBD=dzk_%Y1LY__d_0^psa0
zTYG&kmA-v*y{aW*3)_|d|HW2w{tGIdS?;s*s(IWt(clHnGn-B4nHCton18(9Zq;Us
zHigy9(e=E3%eG8T)ya05RA|{{bShHp*^Ky>%`bU9yF2;ScpvSUd|S!s=K8e$<J%Wp
zC|;!(@wjNcY)G`Iw_lIF<(ZEiD;RzjY0C4&>s4!Y?+eyTi!^CX>{-MyfkAoN(O?cv
zfehn24o|yhf7H>=i}1}8Sz@x)K%w<~{qeP`?(>2q0=I3{$?z3%lPOrW!zW76{bS1N
zq}$j1HO=-E{N->v5_K@}@af&G$F@x}S#zBy;^!CsRsY^7J!g5W#GN6wu;EYzm(hgN
z=?hN?E^OfbEV=Ye&9ZAdO)5f9B+uIU;)d?&EixIq9_-v8**WRz*WgSa@#(8oPfv|2
z<({QdA9{^fc+s@=JdVrNSF)yE@d^E}UD-1A;Op4By(cVU7ph&zIJWC)tJK<#oP0OE
zT_2D19pY&?;Td~>|C^Vc?^#^0zI-%qYQ#s@ZMxGkrW<(VPk()7$?lA;reF2W@8;#-
zR%=$HR(wQk&f7<=%M#B@Z25od3$Oj5Px39&5mPiemSslOw|_pIWh3S~g-u9|@44O9
z{G(?k#&18NnqXCaCb+^f?%iUJ#d}UwG3Q@AXZCvK*6+HpYY!$SCqy6D{Wz`q4s-eH
z>w=|Qn4eV0weC}!yE-FkPtu7Er(Sooep&YGMwQ#=E<J85ozffq7n>As2E7${*s=Cu
z=-=~A&E7oDFLl~>2%oB79zH+ObkEsHHT{O3Z$iuRKF-?fW8=CyQYSnyv;5@D+Te5_
z=^&km?^?YKwi8tO?z(K|ak`=T%cA(N#scy57iL}R`<lNPe15F)VfKn%9j;}X8&s#x
zy;XBR>$AAJ#K9(;RT|%V|J~dowSs+lvZukmqjK{P?>ce+kaM1az`E49-St}=PqeF8
z?05LD#Npv0)o&u$Y4gY;#hSl6h4G}2kZ8+OHI1Yfe-2EESN(RoQpo>>SJ_v_P3<b#
zN|rg1%T(loKYUzkeS3ZS@uN(8f7aCL^*ojcvo-rDVtGbdEPB@0tc7AP+;e=*_`}ja
zw&?EJRC?z0r(%=*qMO1yYBVlCD$=Qs)0>_w%M-auasIc(mY=%~cC{@vWxxLAZfR50
z?7%-K&PdhvZF>Gja=BO1$w^$oS<6?{S#N#h`e^O+%g;HE8VHB2QSjz?V>ff>5$7k3
zYmEwnT&~XAcJsgTeG@DHKRU{aU7ki?gKtC>?NhsEJA2c%lGCvn{5tZDs;n2Z`;|AS
zSk<36Cv-n`wf+q@M*eFBQ+dya3$}8E8TxiUw|?Dwn`v{~lB`{|I}EOsuPly!BhEcd
z_0?6q(_)sr&u`@X+P6n$2GjC`RS(`bOkcZ0&G$s(yZ?Xdw@GSTxO*q}&V+Z<o;ns>
zIkoPW)Vb%TwL8nL!=)F>f0@mH%z6pG{hfWz%PLpzcdozo;oITevASC-?-^Y;tK8S4
zeE2rw+QR2rJEqMvlDYI<hVQKW>I*U3|1?Ub?)#J2xV!JUYvRT0`D-+VyyV&C!jnUo
zZ-z`!bL>buTO0iP|GWubLf6JGw^f{+$rY=<A|=%@f3M2MrO{@`&0h*%-~P^C^W?0#
z|MLAORi9WhufItvsNTVF)`yQX#Wv3S`}<*X_pbl1mp^>_|L^ts!<&lOd4-QFzYuOU
zirC7%m%+4S<NP4&*qa)LN{Tx_Cx)^AyppnPL$iLS$?D{n%#yFG+n-xI?svHGfO(J4
zh51E}n;DOvTJ$cr@965%XE`w;@0UHXx**1F&$l-J_`#~lLcXC#)pGvT)bHZIb+AWj
za@6^IGedgq_U!-r<@^8Gp8YRnn`Ez^@Ko^HaDi3rjit^%eKQ|NH(Ouj>-Xh8-Yn6~
zR-SmnB*fQn-@%@16Xu#+ZCR;YF(p4KxoEBNwO*&j$!X>)UmBXPZ<#Vt;JJ1(1IxJ^
zKZSa$WIV%6o!87@Uw?Da$~ALNGEdE}XG=d7CacXRcI~Kami?ckX~&O??TO$P=2K^3
z;|sR$nG~<!$5A>Z-D8d2maZ*MpD+KKls9XwiuAtc&r8>9y$zqFV|GHMYV(!kd2@F$
zYN~B1)Xn|PDSlx80eR~UH+53)q_BQo@1`O*`J}Vi>BI9)&5jo+emKK<J1QXNsd`_J
zM`HbL6VY=r4ILj>O*$ZN;iN8kNopIf!u}aA@&#W#G+ukWZDKc*s8jT<2|@ug*Th+e
zUs)`s8fJK~Pl$<oefifq-ppM4^-ClhZEybl(=|`%+}3~hc5Tf0a@(M&(l1nX%?hbs
zw^LRxn0LG7+srGo>^k(rTXcO@x(|yN^Ihp(W3t9`TYYTgP5o8VbfO;>No<w=7bbP{
z;C=50`OD|6(btaGzi3!;az$xC&7`&4^F;FJsQe6BnSS+Sli%7{kE=UkC+yn1Pkdjt
zZiJoYtX_d1byLk79Bl$z8y~Ow{B=@J!LBIFgBzO<=5sYm#|ylWxW&eE@4&)oDkUe~
zLY}Q$<lfb~<M5%XdYup!$0t5!o_m^FH|eghIevK3aX#+*Tw-~1^sX9i;ni7E^`^b9
zZW|j*qNkDX{e@o&t(q=#C`3K)apT;1esS?~rWrNtGrl&QjAjlLJ$s5hD<yaei)Zwr
z$K8yZ4Hli-7tb)``CW})3KJ|@dTaz0M649<FtyCmm6TbhW4-Kx8dv?I6Mb%P8$O(6
zNHp)W;}E#_EbdXK!@?VmiMAp^&kFWvg>PGax-Bd5cFg>;_X|EO&*y2qYw$}UV8#3g
z3X!iE3b_?dE$b6q&D1d|fj=cS;{K&euA=q7mNo3ysIV|OLwjE0&#xXF3ir-BZ9A*S
zZ}9G4OA7NN_pWK53Ku^8by1)?y#8K_%GqM3^Vw6G)i#`aJuAmNcIu6t{p*{i#d>@+
zsF+#btzsdl5n`pWrroJhfwM_9BrTl9^P$Db7jBpGxw&>Pc;vX?!)e<eeJOnB9=Y!+
zk@IJfzRD;4dgDa-6_YZ#rq*tpbU=ERZA9L!muy$>%So5JMp?D#Nq+T7Y3Wa7W9Y7b
z{rGuN$luq$7WVUM$Z5K~yQ8yt^S<{oN|xuYsT2p?4Ak>nwf4r(?HN|zKdyC1F`s0|
z<MB_5z3zvoFQc|U6MyKt6J@(J*0;5HsYwf|hsC)0iB`z*CrBrFSDZL!9<^n=@x7&G
zw(R`YE;brU%dQl4wBP9zZc}IOb>H^!>DL7l>&0__SS<@H6iw~pTQ*hfyHojv8(b!3
z8;><RJmysu_Z6&>xAgVrk?POsKK<Hpu};UHgomd#{#?{(YH4q?NZNJ3Y{r|0j|vAY
zI-Wf|*}J$uXMOjttd12oCap-?yvbY2d)A?qUfr|*S*35}b$N5_BKxr;lVv>RZ@!y;
zo!NcnCpVFL^J)5AvyLS0J?c4~PbQl2__;(I4ht#IClz{Y{~byAc;efEJ#soln^wMR
zi<;SeY|nzvU3VS@9pe*=ZakV7{lGEd<1NX<>MQ)Xh5K`scYn*8p&_}c&3E;tJ<%6k
zN&{@?@9Eb0_odfxqw|qwg-2`OPn_JXBX_e+{LEe{l~cR4B&XE}&1|2R+s8NGQX;rK
z?~!!V5xp%H2>~`IFFcx<TUqH^@<i18aVmFTPlTg;=k}7w>((VT*)MJ!72|K}TP_uC
zv#D0EVtVlfJw2{XS}*5BMlMoby(q^^=EtPFaR;_Hb?+{-FXf3fdUaFo@sw2?g(qc9
zTclHXsN~gh&6?A4es?@O>b;Y`rT)Hpw~4jtS^FoQpx}BLEA^s|#a;b5>v`@y4cgMP
za+{Lpja4<W-PctkgFb$8o*7&4i8mq3oX?(T)wAeSKNhb3ocuG|PKwXi7cIPX^*qPx
zE;X*Q?!)i+xS|^l<z1*1(9j9Dp2r~}cQa37Pj=a+Sc4}j!J?+CH%(|?o?M@M@Cd_1
zS&4nJ25;2X@|)Z`_dMN?g{MD9nWt{=<h`l4L(VitNvagjQp)bSChPQV+JS=WZ5R6%
zdb`;QbJ<^9mB|~-#M3kR#XP6Q?QZ>VI5r8}+=xEvzRj3z|2@B~lh0H|v#%;k=lVW2
zxIXvI-(?pzacj!4>DO<)P{L~D&m!DkpA&sXJ84lu)Goa+|LG5MPhMiXW4xa?*k)%Q
z$K+Sr+fN4GlzI2N!Pl1OB=e<<{B;%Ls=?}?-=FLkua>kFSvX6ik8{?KOfl7d^(ImG
zZLuu-@5xyOKKO9p((31W1-n!vfBl-Gv%E;Cjo<Gnd@OSDB~b>+hzRkg_r5)=Uw>0>
z^Rci4{>Gtk6IV+obk*xuE>yB!y@~H%onqhf_Qydpy_wgm^834e&`5qSYW>($RMPe0
z^_7!r_MV-ivUq3B1d%P8B@7=Owi+HYQcqsWuy+=dPQfmn1FL5(`8a9YUIyXsUX6cr
zQnqSr{Oiip)ny)HqM~@uv{qPI@!WN$yf^h4ik~$;YH`Je<+t8@^!3`Utu+d{)e6&l
z)?Li<HV*ML;9tdZTXIq?ujal*iiPTm(yQ+sJM#7Y3)N@?V{mt}W#d}&;_1n^doR4o
za?bS&>6*29HE-Lyj5A_$&g^GDELGxL+Hh})u)W(AtN-#04FCVXsXbYA$79-q^#@Ga
zPi?6eoOwe$tUE0t_@-{^CXba>XHK0ub%0}nixyJ{lfa^g<%Y9QpDE1Yt6l3{?|HU#
z-=cjfHO?$z<p&yX%v{*)kyN;2|K0n~_Pb4X`r-HU<FjYKPuKr^_;b~cw*Tob{vZ9i
zzu|vNy64`gmdGEMzuc8py&&QmKE?b+mSC;+gV5$b%;ELxH`&a)x35~cYmsKwq&>Sz
zCLioxwLGJ3=dKs7{11!5Lgd=#pIPz#fc4R@%a?o1SmG(D>Z-Sd--+>GTuLC{Ke1&m
zQ!Lx}w?wKuIsH5S;jJzaTY2@<7L_9NF2CviAS3Qy*cA7-<ojX)wctd#eH@dQ8&vkR
zw3nE9&wqHBKhumqz5d6{qe|ZU+ctYYe5V<EZrSST&tIC<RXEmv^L@QmK>We&%`*$+
zIU`FZx^C7uvv_4e|NZvE2Uj+7a()!q!@$;dS>ZF&;z{N$nGr4-w?3RqnY6Gw!8Kl)
zVVjBIq^E!EPPG`Qcb{Bu5s>|2mJ(ar-ry^Gc}8CzudA5p`XG?uQhB~i{mZlb6=6T}
z*#EWWf4&*Ik3;?9ZgY-(H?O}q`&ItruI$w;hf_0uJxDLu@hFw?-U;!GIc4E3k&N~?
zd^7Ir#q0e)X1ZX$&*`7r7p}5imUeqm#g%})b#+~vI@&#4PHt8`dQbQN<>|_w^`A#G
z=gXdN``#hcm3^X-zi=9JU9bG(CU%MXP3t-)t@CDQXgkEulWAe_BF62Bq(FqjGN(k=
zC&~J!ZPOmknf&+>3+ElCf=}-&+ulC0KKYsF>ZX1f^G_=uEW1)aTiMNAB=CF5^BsbV
zdnPC_JnDB46?gc2?4-%OhWnqoJ~9++HVT+_QpCnp#CnVLrjldHbt~KSXKzc9jr)_g
zB)UHS*NPRZ!hbgKA1#+-4Dj3dN9>WzY&U^>3|}V5J0(;l&QO--jp(QpRGZ8pkv9F}
z(g2sq%UKrWs=1sL-oqDv_CC{}6v+)LjS)=sI^q|8Fzt6a+QObZ>*dUC;u-D7BgCC%
zwejg6X_TF7$iU~*<<haNaAKMvL*}eAUfkB=^A4@A7c^*6PT9NcrqWrbA6y$xdh0!8
zT>m0sSK<x<hibO>zcuDCy`FRX#IE?@?@Bd2C$e8G{cZBsh<~$wNA-bV!x*MH!9wQ_
zKCb^c#ZSfal54Dyq4R|}hwg~|3bUuBo!oVzj*)BPEAB;^S(6ecn&04_*;-jB?Z_zR
zn}5JdM&a{5fm!tuiP<steJ{CQsaQWpUw~1&;)>X&eHR%{{%SjV&2dv+MWg!5+smb*
zr|F(`(9T)2x2S#L;fL3@ZJo**BO0b6EYLIg2jeTT9rqQbeSg&1FxIeM<+)+Gu53#8
z^64^r&A8(m_A&OGe{2q1@VDaQ=Unqwzn1gs_CHNx+}FHsTJcL`>G~NL8fzG|d(Z7N
zv2(26Zl<mDDQOSm^~=v(zvv#5&PdEnQ(0fq`A$&n$=BR33_HX#cI=t{en}h4$~&2B
zmKUWjt-K|1-Q?<(l;U!x*>`rne0EHVL-Fz2y>_avpPn$_ZVYnDVU^{vypZ+r$xGdY
zbvm(jcASx%9&2_pY(6CFI9c&zefN$Cu?dUA^^N~HmAWMgaNH>rX8QTG`Tul&`@etk
zW4|ofChzW%Xv24Q^0CJ!BGPR5SkoUDZtW`K@-%5%eaA6B=zYc+YpZ>i6?AU~soef%
z;c>b=G}^5@FDURY-y)@=PED^r|BvNo{Pq0xnWL`t-Oc(xe=ona|EK3YecFmx&iwj+
zGyI&cd31cKy~(xVkM_Fv(>EJbJDM5Dc<or7HSrH``vNl;=ZUr4o5JODI4=h#MDK65
zyfQWQrX+9l*(<d(GnT%#KWV#RtzB<Iba->g);5N3FZ}|}ct=iI7bP`?=?mYV>xp;&
zKAoRmfBuh@+^wk$4x&b1d{@4)S4jPT{{O%FmU{WcDyI&J>Yq}N+|zL5-m@<+mnc=O
z%eU|l%AX!`?a-Y55RTB2brrsKl6Bc)hcpg`vh>7Q_a83LujAGK_jTEpYafmGzH5|<
zo8fXZ<#Has|5U!E%8Q+kUUCulKEWk>=x&_Q*&>+)&*csrZ`+wd6H-pJolm~^zJZB-
z!lk;7(7P@55!q8XJ0?t@@p4IsV&cyZ&mYPQ8Y>LD_c>Mz1nDcLCR{)0dst56hKcN5
z!^%H3>e|j*b0<CR+OR84Yo*(R$iwMc-aeJ*-M=kYSi7;C;Z2y@#gds@%i5U3dvbgq
z-MDAF`$Su=rd(1__-VagoEy1=AJ4hDv^jQJ;=#>JZ%+wNxm{ntGrc|BCGgr7J@fRb
z;U^BeAJjgkw&&2HI}D#=Zd<W#R%?B+-Em4W!(7$A!yhI*ZTq4ksdb^BvBcu*#mR}U
z=ZRl<v9gZ+qVUcm{{CM}J!ZYA&|cqt^ReU#J73|+T`S@_^A*<EGuLqci=Y0HY3K9(
zzu(mBr#v+AYb`M9{1h{zrM~-v-Mpi&GKH04CjKF!bI-G$dvg5OT*f~?YDAUS-_ogS
zU_K}qAathPqVvIhQGv_L^|Pj>?+8=g@A<jp<i#G-br*ZgzCHi8HnSmkzEu2V>qDtR
zwg;{~4NY~qaYCC{XpP~$8DE$gc3E^zbYI{7S~s>VM|S_|0JjVNvZ=+-4BmRyUtq7g
zt;y;eXqNkAd77hQ{mlEzkJJm?yPtmXx<s{(^x+_xBWDT{ja#C2UKTVEF%7>Wb9Qa;
zPLYnrwJ%Oc?^B%mePW67G=F9DnNrIqSubMoc(KKE*Wu6l#*adtO!b?zPew}So$CG9
z{vukr%G*v}ooanXVA>tQLqSZ3Hy=6heBzDz(j<vF?|h!GW{~LCn6TxHSZ3ek$QPW)
zHQbq*I6qz}(N%0Y{zKbq|BIBsy8_ITuM-&F%;PBJy`8$`k$nAolMhlgj*=M+7cVq^
zP2W;7<?c?dkFDKD-#iW4AedtiaqXr0T7eqDI>(J)IcEIQxHi46*MR97Lu~6K#_R=w
z_qjeY|ET|T`u26!$FJ?<ugNVd?vT=0B=jQS@ApCvffu{lo(N3evBtiyhpCA3Y?3o)
z6z8Tem5lbT>Dq<Yx>#~sOtSMsoYsbNdd0n*wjep_Z+S%LH*e3hmasIpj8u_@sVjVp
zKbX2cxtX5l$GmA(_p)2+zU*A*w(g0LR0_!ZkkaPe$i1!J#m=AS%?gPHLO0eMI81lE
zuwgMLvx(C8t>&}JOcEJBd6_f!Z0vsPIBl^{5VuD846~&@IbK$(8huMwXBF2Uapt_e
z;79Hae$&&{$IF;&_bpu{{7<Dc^Yaq3t8Z6-ef8pZ=zF2**F<amCzW<DH#;75<-pQO
zQGflqub53<TpILf+OqmZFTJL%{;)zwb3)3LrI(^2{_;gh>^vee=imP8u4<8cYk#+^
zNl6Q(M=sgWm#TEfty@FqK>vQd#d{t^uF5+(O~cG=)|I8RrafF0B6ID%Nufo`%D&(G
zs#p3=YrE_@L%gZmd5KJR2-EGaIlb+EqW*zeaZ64~y)HM4T%295H&JiWoqB`Qk5+^x
z1t-@!Y@PI4IB!|~lV2YVyX9_Y3YHvrIg3kd#itw_=BoxZGdQnbQ{8HPqv61GHs*KN
z-m39^Pc?G(F^E;WY%p^VgHoqjyhT`+#HNhXV*YH$6@(;r2Cc7NSR(Xs+QXjtb}!<0
z_%ZtXcKyhXxsshGF(<WqTiOYmt)Y$eI#yh!%$w@cl{eJBSjMkk>z7t|$XTH4>5qBm
zr-p3f7ZvznyM*85N%0I{R{PTp>q^2!U2J0VvbR0yv=Mb~Y0A)AP?kP<=|72*^<R$b
z_;htEbiXo{`Lfk>U&^X!r#Ek_UGQ=7-R5Kc7fs^Mu}Zg>#dz5<*UsJa>g2p-Idfw=
zOY6-l=Gxc(S(?63<+0}v%}5>Y#IMiwPc!{FyD{l%fUbM-jumAc51sUQV?X-Ne|qSI
zzqazqv*k03ZZE8Bwfl7{?2+8ug`zurb)Ac1-ODOIonU);abtvJ`;-rYd2-*_OniA&
z&bqvtzL;Uw^LL%o_A+kq3HiP(B9W)>$(w%$>!Rj8u9wv7-?_8;;nmRi%Ex!#_KAsj
zo5#$_-JBe?L)zd+Fvo)p_CXUGu5Fcha_!D4PTOfe^WUyjnzNhTg?oeU_0Z$j52wsJ
z_cOD1qq&=#`1wEAdzq}BZE;`vx@wyKFPn?&_rB;qZ+%7H)TU0(L8&{gGf=wf^sVOP
zH;lIqZ{~3=tBm<mUu{@x|K8jEjllieR|TS7w?CS-Df^WTXaDaD0?MLC-iD=^)aLAo
zz9;SE68M$(^rOs}?flCR<!(RR(#ki(=@hTgFZFfC66qdlHtU?uZkBPMR9SO!o?4~F
z?~R;We$DM|`1X1E{4)M+^VtOoLl(qMypvonW%NyE+n$qgKiroyKbNX6@9bIJ!~D4G
zW%Q;MMXFDECPph8+WC3jeH(e>^`=bCbq60`lS(<=%X&*fZD;k#?u>>N=@)M8vX*DR
z6LNOCh@|5Si6f`)^4?t#k?!gG`A3tqllhf{5_|aCP9(<P@a>)d`ed+4?73NAm-Fl@
zH90MOb}jpbV~754lm5qZ$agyDnV9-V?RO^lzFJj!dRaxM{NxGV(^$5pNXJYy`u(8O
z-#mHItofS^PCoz0kR^9W=Vh_K&s_H%pWQQg+$T4@tWc5qSnR*!i}5Yzj{>vS{+#!u
zcx`sbH^o^2hv&7Vw(5l^7X*ieN?tu?(!T!w!QDzVot)nPCdu0rXz$#p93Gw-5;Eb*
zv3j?*gtxtEuB@JUD(eMYwx3FVU9qp*!`ghtt8W{83SOTK?F(IdW4)n}s`<u?%X4bo
z6Q{XqXlHEus3W!Su@{$r*Fn7xXC@x$TC^_LrhLf~rB%6)cTRjCS-p7Xci#^_ahf0R
z=vA&vp7rJVOV>r0JRH~G37k=$CA(8f&MR76_w=Kz`a{cJ&EwjBO;_dGdj*wA-l75v
z=Cz&GYjODaY{jiejRj_m)eDpE9^T;5y8Oq-%m8V}Sbl~+SGlvhZdOTH38tM)TFySh
zHsR8qOOp-rDxIooKS-3M_FJC#a$4rw|L2nJ=f(H6{a)pH{<?+B{9hTnHyFQTZL;J4
zU~2lIWUh9_vt5(w8)vRvlq)5^+i+3jD%Vv_oHO2>N_v<6@(Fj#k10?7wORVTXx(Dg
zoAxnWBBc4Q;X^G$%cLIDPLZy-4*QZNPA8bzPF{)e4y`mYPxUn~$ts%1S?2p{)jWZ8
z(?{*w6(yvXs=FAJt=#Np%OKbz8=Db0Nzc__Qg{T{2ClhXY8vZq)~o8Dbq{y?I(_|w
zuZz-G7}$Gcv)<BAo%xqdG%c{9NyToilE}@8HvW$jdEBqah!|wu{@s?-9dNts1H;^t
zZ(|xPZzR05;M^xXVVB1Ah9&*q)OW_7?S8oTRqyxb+nsEKSA3j!Ov&*3<ZjvcJY)5Q
z4=-*j&6-g5y<4MsU8mc+pSKF?lg_TRZG5eC_aM`$ZF{l<uLM=wM1I(ny=vddtFLDH
z`mIu*_3Blki{BHW-~(IE2_9UKRq1^#<_7B?K8uGjCo@AHgm@|*ZI#xIZEZ1rxq*Lv
zbTm(|CNEdn^%o4Y<w~mV|KVT>{UmzyLq*N|<((%Fl(%XeJbS5bhTD(nMa-LzJ-=3e
z!IdL9{YOx(>!u5NAFg~X{rm5!UH9^)+TZMo{_<VVBBUQE%)Nh{bE);cZ5tLyuHu~{
z(jWfVWal%nd9O_QyH5tUMl-y3(7isXG2nf!^6~a#?$*LJvv^eIr+j#G*-yAt{_xQz
z=E}^XNiQzTo}HT7s(ztv){EVa|DHeRh!AFVd|dxB(bH6SL$br}6XB;)R)-y0vijt%
z?H@9ee2jB%=rk-Z{M&h}IW>uE-Mzl&YZu(Ec=AO5<P%Nr$rj7Yrr(q6GqDYrdwun;
z0xe#JuR1K!uXWCKS1{RfypC&9-~WEn8wYdS{^PFYizXG`3_Yq^!~XifKK2ZqMJk-z
z+fA>qnw~GusQ)e9_?AsBJn~PY;qhnd8Opm|3wEtxp1&{SCY!a~RBhMw8(+*y?_%ow
zCEJ#~_m#EigU(Y9XWceST6*hqsr&nCgruir&&unFJ>s>oyZy_h8D|cq-m@|1@t^zA
zHtq07HdT@9(`9a4<hdSbx~{qQ=D&Zx<|_WH>ucD4>BSc3KYwHE6VIj{_j>dGy~Qi9
z+_n>)`{(a*@>t3GQ)<6MGhbep*0S5L_89Hx{`JN@ZuyRfGh<~%)BZ%qmYOf!Fi-oi
zu}X!`rj43MSxgMdI{M<Hb>-ae8Eik8w}~}JpO<0cvp{cKVU9p{jd@NrLRL1Wf3yXD
z9de$Qr2kW?ck&~p+8r8?RqDTXto~rw%D!THJImQL{T&m%pY!bG^Wjr*pAtGxMbIpQ
ze|dxUjn^G4&t138ahR}%(VL;I=KpuWJCS;aWaj;`^PN9YmX+mIQQ*PzX4?#|-b^~2
zRcyQFMpurdod37uFO>vt>~VN}cmIRK_g?K`JYn1TT*~H1zrbPlULDaMjwx#A>!+oD
zKk;f+)?Jofj^xV{&!#MRY!rTba?B4sC&s0xmTq6VwD9BgREzHJ>A!p{mc0L-z0>nh
z*_z8Hvpcn}xE9P5YSYWMvyuN~u#Cn16~j88?3qHcoeWu<gAI&Sd-~XJd+2WQ;SAQ{
zUvl)rm0v}x3Z|)q`Sq%6nlWr<+;GZgXHa1MDN)(Uo7UY{i(kGeHlfILGoO!{(RB{B
z-9CHLjtBLe?Av$AqIqfW$xkMiMb#_|XFguUUp1w_Ng-{2(A1d^7cS!Aw+bq~IBCy;
z=NqF+oK0O<$*jFEIq7(y@50rQaRt(D3NELTJowG_Y@T)8(xyygx%0u#|KBat;9hvs
zQR(%IgyQ-Q%&+}7U2nLuy)aklbSd9zKS@cYuQ%h@Y(FTlwD42diNziq8{Aw&3Ke6y
z%wI46q!i`Ay(9Jfv|J9&3yhck2<2?{{CwQTt7ygAgB%aei<|hYJbJkLfV;zy>n%T{
zzi{j^X+Fp>$FbNx+Rf%ujB$u<fQejw>sn8}DuMVz(%+9AtzT>~_r#nJr{711>26H#
z{?fip@U~w_S|jh9cg+3n6>_#MccR6A@A8&E&U(ssPsS`2nS_Z4nPeufP8XXx%O|>c
zO(@&Vj|aQvIx#WMIJlr<dd4cAsShoW=K1aG{;#+0^a3mPEu|MPs#-p{`E$P7`>F}C
zlFJJ(IWAhYXpwhJuDZW)y-MnpeNt1{TW08O=JXG?_GNkV^0v+FM~~TAc_d?|2r{)L
z&u3)2{y9W`F7G{dHeVBw=U1Lw`FgjganDtY^)sFqID0n!ob@8N&0ng^V$P(}#QXmq
zGR-h^xaVul{#JCZQ=gWQrM9c6=N6w{?bro<?NQ6@eqBE@;a!=?!P=_x8zbu-<&~v7
zZ@+MCie)oRy{awu*F8@Exa#er*BlE2i#{%Pw)4Cfd1NxvS}Pwl|4g>)x~zRqURbR$
z_*2FDO}5*kLHOrs^Bw#4ADP4v{UrJS>3T(R?g-g=hnMcH{w(9)y<FS(ie6}L;Gch-
zsgq((w$-d`+<$SwN1K|S31_D)G;KQXR6oyieb<f#=LMRR@{@lwM1Cunx9rxm=3Qmq
z^8>b&?TufMX}9V{*2As;EX1GJa!Ji`{~N~sb5gHEGrLCgwoNy>lp+ldf16_vJ@=I9
zy|S3K=PqTx-(nYYK-RA``{=>gO>19mIv2|9aCPn*-qgQ^EmBI3f+4fQdXr|Vg}-li
z{;9g6{@|*#-lVs@SsPBhuvqK5m~FK}&Fs3)u&bQAt_nXmDqHHI7%`n={ZT3PEB+U)
zLIP$sNd5?#*{Hu(`Ge|B1Gi?s<F8dkg86&W48tEMO|8l*U-e~I+*b~Ej-2`{pUT%9
zPN_fq?snKFXI}YD>Xy^|KQvmsJ2;u=kYx$agJt5e`PTJqFTCzf%+u<6d0t7Lb%Sog
zWe$=1+^w8>i}~hNcI@9Px*)Pm{kz!w9(5bN#MPNo!c1e+E=)Q)v0FtqK+CX!qcMQ7
zhkM-}=95$ZI$W%rw3h$Lq+?ktRSm4}Op>~*dUwTwwR;a1cuo$#aAy~jbX31WWU0gR
zlJMzEzfZCDDc^eSVg0&mE=Nm6ColK?+IXeOtu);H^}_7@3wN)|rvCG=4&LAs7rByk
z=lR&Zhn_jKZ#`BLk?qsI_1eP?Y-;Ul)homu<)d!SUiVA3CG7MyRq6NX9069YpSfgi
zfACv;ZhLS3R-2QtNw)$*BQJJN$loz1HRJciuOB?E7pc03@-(j&?06|%zwPQB37N&a
zn|wrUUBee1-gCM4Jb%}%@aIj7zm+K6vYDp;YS!b!c1q2Ll}aA3%Vy;Bdh(qAZD#ni
zdXWmdjjK_}JZm@e2>V~V?tD$&XnJOD{J-N>ncS)jjSH>z9Vs=t{(9mvjkpb_|B_kc
z!am=3nzTRr_okH}r<~lcq;t_pBHFoL%2aHV>y(#^UcFZHva6eWYOP%8O{uuD`0|Ao
z4$}{QzUD7)>@KwLrMg(yk2KcS)3;LD9({2L&0UuuW4$U_V$l^1VM(d0k!$xpE%JX~
z-<$B`Y1S{@KdRmGjZeZ{OJcrU{n8p^X!7LgO4p2@1AH|b4%Z&;xgepeWU{Mt<zbHA
z3H5PHHMkCQU6nW!;9K?eH2Z(?J*)g*{Qf7ClXd?9?~}*=_ja4t_Z;l|e(AK+jrq@l
zn6!5{=O2pxdh@V*&Wt+#4GiaHx`hl3ZH%fSj8dIVRF5)v<=JljB#^{&!}MxEtlX3X
zvofS&SN6{05?QTXX!<X?i)j<5yb{+pG5<>$A5}u@!&MgNxS49qYpLOq{iu>3V&iFI
zbJ6h6#cRIp`<H*KxO%^N!KDi~=Dqo$c75fw_9B@}k3D10#Z}p?T%5n;T!(t-;mbD)
zZ#-s;o4@GvTcea?wI*qI+~0TnUXk!D?!kv8wi9ClxSU>>7xZcG+Y<M=L!ke#_`}MM
z{f+SlXNU)1`kP-Lo6xb?qW6S*^Ho{bPg}nx&RWtY6n`XxX){X`OO37Xs@$G}>jy8f
zbEHex_HGhCDDphUm9e*yr7v;jS^?XAHy96{+Zr10exPmh5~ew~<!{S1&b)Ek)cEW)
z)#YoRM(j*}y3KmpPMLX!jD5eEMm{-px?CmH$@ki-2lmH&6=!C!d)B`>FBmoVbe`<O
zE@jVc=|!Th*Fh=m_FUyHZA<Q2JBbDsGToJHT=}@P<=f)??7j=9miL_%<T7jPI+U)n
zM=5jHM&&KX>UWp7yz0|q+jah`@neICTWP)-_W!oMRa(2S__|(D;3wPDk6+z5XP5lJ
zVn?_B@tUtIPv1(lDp5Ha>NCB5^`)6NW|sz~s~<SE;dP2__S&S4WjhYc^KP8Ys`+N_
zmm1TP_xEnxY%xc$W6^KX6G6*a-fyu@6)T&zUN&`C#)Q2Us#7H|b!Q&FYjQX+?VI%}
z5nt!NjZ+u<Y|``9bDz02+kK|xfxxs-nbe}4yz8t_y_m9;ZD0DQr|-=)R`1eZclTkv
z@}jou-LL9aq#Y1cpSNa3(9F5VR95iHrdBQP+kSHUoA*xZXWV%yn*Hvp<&602q0anF
z^FGCBect?3l{J6UD(1${>-TRYRZYvUwhNyr$EOo;^ekU&lGWD)ks3da$GvI)l*>9x
z=SVc97#vj%vu9w>`zX4NtMPiHt-~LVn12RyXV(8;VIEUA)jH#8$)ZWJ!D69Tjw~>*
zW9muNczf%{n)?TTiN#0hSG3r_X?w)~_|r|E)rZ+Mno4izylYz>Xj8Jm@4VB4nuVwJ
z&TNwK=T-f@wOwt7{oNXC*Uz~ppLp%b3GiKi@~)>>w9(xe5qH)qbDq!k_RZ(H_|5kE
z<BM#`lI7Pg)Zbgj&T&5Z?Yo<9#%^cLTwRtgUa{x!#^Qh-`A>E^{{Q+jbK8Z>C%I+L
zK3H|yLX*?@)J~^{-Jh1^987<iJuAu6M*8GrruSww@;_cLU7XDQPxEd853}h6n>x?4
zB~c${ux?7d$rs@LJMQJX$%P@_pO-i^xj!+i;km{)xiBU3x*c10{r%$!U7_;Mkq3@F
zP7ZJh{8U=oVt03fP1)o~=O391q8scyvM1j#<e1yIe)k?;bLKYrcQY<baWIMvxpAjG
zJTO#d`YQ8R<pOahRn_8tW^0sIt&MSSPna919d7?4<n^t)#-?lun{KXITlg-iqA2<C
zgXC+kT^>)-Hx5$$u}Zfz?PI;H9n*}yQs=Eqr(N1F$T7XPsZ%iN7Q1FW%kouk$||AP
zUxQ^CukF@S>1TM){Mu-?Iz#xoDJQ3u<nr%juwivb=3J0Haqc>0y@+?LqPJIY?metI
z<?x-TQzy5JhnU{}pxd9$e?wuzKHpOu8$aLvVw5>+bw}2Qa>f7BpNur61Qz8Ts2BEM
z7-V_-bV<aH`AyS3C*_<Dc(Ck6spkykk47igF>k9mv7##UfvubV<V1~4(gjP~WL7V^
zx28&<zC2eXmvKqgE01-yPDb(!OTKKfcY7GO^vk<VW?IetqTb(2+n2AtrQjg3>_x_o
z(39W(R&}x3hMMk=liaZ83ExULg;UEuwLQIF-`uYF{pZ&ni^$zy8Xlf~d@;D}&Bf-=
z&%V9z4DFxim2mFy#mmemzvo7r+OB-N)mQ5H<I9Wr4Ua57F3}Vzb!hjob37`)%-^&5
zREH$aSLvE2TyT`5DwT7x72A)F7e@<ZkGm{jap~$`Q79yFqdtGH#JWm{o4@$%@>>or
znY3h9+=0jSW%hOUzH?=Aoj>ww>=As~SA1`SdrC#gwF>RIRv|W*@82t|m5*ylU%e^Z
zto^OlGiig5^Zdg?Dhqqh`MNA^{HfELE;`Y!_rm&*H{-u%oLQKBy6`@K#F=PCz4ND>
zFKF|xXxcaV>W&MYZt{}+mxQ+SN}BIsDP+z3boZ<m-^%1KyX$`^9MWuOxE6c*gX@vg
z>FcF}J>Fbdtrj^=P~)4P?eqPm>t`)|*|$(y*`em!s$8yW*X_2J8{FP=or~L=w|*A;
z7Tx=tZ`0S`YTGa^<ZxN(y2)4i53p|Wd^c^?$-bpUbJsd7)hf%BO#L}+;cmNg%ikCL
zv6+4-GbMAT(22N)&rKTt6zgva1csg1wT;K@YTl3blEXr($5XFQI~e}!kMk<c_)}N+
z==<(6+3{)Lme}gfC5&-OJ71}42|Mrjv~P-oc@Ogr{@t}MlXRrrB3JL+w0Nb=wdZG7
z@?H)8>AL3k!aQNm;@rCrGmRF!%1Q`${rOdS!8@6KM_LsBYrb*Y>Xjc`di7!$N4@{j
zm(_;WeTNP_*V<B}AN2P{$Xs33=pW9DxIa`}*!g^a_j87pM4_dpzjccLSnN~7_fKNm
z`$e8x62i_J$N9)xt)7u}yikXYCnhWT(ehIlc`C(sT8U0(k)5@+N#mSOP`T&M!)xje
zFPFYh>AsDZH>a{PE86zJiA%f8gIddLyh}gSALP=$r^x@^@%?4tFCk|RrOK`S_TPY)
zJ?+7tQpxm-Hchv3x4O5=O%hpl%O`w_Kz%QZN8`t&YqcH;3EjIQw-rh^C%rrP^7u{z
zw&$v|PqDvHHqFfJc&)qM;Nu(C1xbR5_1-UAIK>~<+H`kYyqdpCIHPZ2nXl--4)rx0
z&Z1v+=hizfJfXL(y>5AXiJAApRqgEy7xc1~n3t|Q5Rf{Zr)F<U-=yCwD*6_=?C|G)
zvpV{bbGf3<vL`*w>jYE71DC$DTNJzgb(_SAA2Sjh;yU6h_Lt3VKH2>BGt;E^TZ%3V
zlum!$p|I`7hT5A>FN3o;I_fRtFS`Bj>9)NaGneOayg0w<YkhoZUtoOdha`2g-!H}H
zU1RRhiDWu#+PtfKg7muiXLqd(pX3r5DR}x6?;b_<!g@i8BI!CV?_cX2bzB1jeqS&A
z9BSkG{q()UyJnonr*B}cSFH=5c0h&YQ_~UA4;#H@=BQfLCG6%3=eaPQslQI|4rhVb
z%-ZAWQC7Ru9?7^CePF9^|M=_uuDJJqe{rl@)nhqrt+wNAqt6mEY%(QPUj5p&PIHr6
zpTh2ayHnX#|4g%*R1s=@L1(tQYFJR|Y5T}WvMa^@?aJL6FiUQK>`@1G7hmz(vkzD8
zzYu!W-M#j)_l?YpJ1UZ{c&6D;v8sPNP5-rH<;r5AvuTHtr|F3=ej(I#;B0-G3&*ob
z%}G56Hs{V=y|wc2BGrXI?k}48JN5RbH_tY|-gmzt#cYcHo%lQLy{{uSdHLN+<~dZ_
zwOr$L*Q&*zn}2gy8@RB~<-fBiG1$zcM&Maj+euY5-p}C+q`rKXwYKs(Xux#T`qqcu
ztM?XMU*q-FEUoJP%f9&&_g<@LdsOJ$xv2h%^_i-mJAvJO&L_iF4+mAY1Sg!Fw%Mw6
zZ`a{pvMtV+POfpQnv&x7!bLNIH(GqB)T&1jsndd2Kewzd@e5PocqJEZUgECv+`;<%
zMAO7$DhtZlFC1~zj-L4UX5*w*ak-_%BI&P|RLUKmbjZ_cf4}gPOQ#n8$a(K*&dd;F
zf5R)M-ln={f#JQtCxJmcg}%o)_gn9MckD?;Zt|u#52S*`yS7&_i}4uCCfX%k&dXfu
zaIIu*@uvBQey<a(iM%dzV_UN&lY*hgzexgzyE<PM2|dl0*SfRd+nkkq?=2R;o?o%N
z=<mu|>|D${ZvSLsKVa4z_FpbmFf?3eu}@@6oUawX*!p_LHqFr4$@x3>yx#EIGod^?
zZTptwrz?0^G$tl7DQvNcebnG0X6q6;#Yai#S;g(QlU~lvjyG7$H~on#$7h?lM^*%7
z%r*2$tKy7Y7wuaVpm@pR?zF8-XNz@iSQEm$ZRtd>8$54BTbbO>>|UG|{jl)eR;I)Y
zCt3b%ewZMB=KKZ^`}*TwlP$RCT0G4%+>}+b=Jo2fwbgPrT@3d9Qvba619&do^~(P{
zoB@Xmrk=C79i4K}C81vQ;F1$D6<+gOEte!U+}2=OoxS^K*}~lHqV?;p)-Cv*@8ZwT
zbcJ>M`3I_DKCyz{i;`}K{P+D^I7fT)wq~}}7}vzO8<jHqs*A<XzptOhS(hcoG-<`X
zKesQp{x%K|3fup;@d)?TDy7NF@$R4dCUu`Lzsi4jdzH?Mcehq+-N`-u;jhs0XwBn`
zbh|%V+-uKOnDOE1Dv?_ytS5pbwdZ*0X}n8+!hZ1j-#7OU3Y_0?cX#KhjRj73zeIPe
z<h6OQ<fh4~qA%gQG$!&FOQ#FgDA)hLz1uva-FI2fwuZ?i)0$daPb;aWNZTIR`)Nh_
z`^$?wo+sZwm+<KRjp>hB9L%fs-+Wy!^WDlNLH5u0&HM@R@qKDuc2#c`&wYAi#_Hjc
znegIB_tp3p%VwXO_IXLs{xU_+`_>M9lRZ^l%AejJRvaOCx<AsUb-!KH`~)7xiG3-L
z>WdYnr;A3a?o8ZzdQ;r8o~!z&i(XDyQ?q=9phc+2`bqJt|M8r?R(ko($Ml**i)=20
z%s;YK?MMBlJDiEUTkIAdlq#6_<e*z`zmn&=yiXzPE{AsOX$oCEs8rR)?k{<uyg<)*
zUrU-}sBYQ%yq^{7H9d)!be;G9k&65Cn%8-6PGw5{DwU-SIqdVFPQ2oA?9?o+t9^a<
zFY#El*!gaMBl0RDT7m0*0{4;ocXAe5G(Mea=Wq7kVVi=+%1y5Ee;$9ivB)Pg?&{t{
zInUe#=JDrx{3}{~w!r1+#GSj&^1DbM4X`t~wZiy5qu$%AeDhT^%dDG!^VoW|$bHfm
zSS97`7m!~(d2)SP#j=bU))SbHyZc!$kTIWoculC;tOxvmbbRx#C3+gq;kkOjwqo98
zMjrK>qNZnEmh#*-o0HX<GiS@gc@nudlxLLB;JO}qa(k(;n`vKTr9ZEu`S!BK+Dyw^
zWS4Aky?eb@dTXe?a;C+kV~;W>h+PrAU7Ay}>9^sHL(AJ1&%dlBRBz?{K5jv@QfB|=
ztNEMnJcyptb3tQk>pmrMyPgf9RZ(U$beZC>x462ql-*KPyR+i!nvZgS8t1R_opyVR
z;8%xo*&DNS|1ax(@w!W|`0GsD*=D(`3fw%NBrfY+Qt6hUI3-Wo(ZcD^F2R;dTJ8#&
zGcuJU+aAwS^8egxJV_(#bGu=EbW7s2-bYrmGzv{i?Q&liuU_R@9&)%madP|L)oP(%
zZM^Rut$S#`aMqo<tJMMxL)|A=c=?{2%{nV<*X{PKODUnk7b{PvtMIyCO22Zh?D;Lv
zZz9KoQ|F~wFmGhxcMN_lTNL+l-|U&I_^#L7D=$Cu&3xB;wxo+2bv`R7vW5ofM%KGp
z3a39#vp8I>tn$g=vD-IS?-tqm^Hs$>P2VOmDp>ce_l{97oD*0wBX+x%<tp{5P2cyt
zeiq_0FYjjK)x|O!lDI^!mU<W&$tL&wiJbAcVmrrVA*-jH@;b~{zv}XHUa)hyHgh?9
z`VpJA2PC}|?u$kTuFMtR)5hfd)$Psi-!{AJf7PaYHfzL0zD$^Z_+9CT9}6`cRZCV}
z{HyBv&xF%|JMU2eSvlv<8COLPEzW7uGW(&ewukZCKfa6w%PT6%-;~AlwAn{ZEQyYL
zoBmxrqO<jn=b_>lo293n4)k%~c(;Dd%zMm-Uh~8<IyOK4=J>=$R;lE);nmRQ=MSHC
zK7GB(acaFqX=DkHP<wgG!#BAS2UTv*VF|8Mw2Ilj{oc|yR#I7<$L8JMo4@An&IF+t
z<F)dFbs_<IKec}*#KhE}k$=70W1s8h6thz<OSay$y;3v5uQ%|0d#P?o?AgStOzYQd
zulvBRmyzSLFgN&G?7Cp97hLRRY3+>W^Y)7FY)NzM=<&?1fBdTA<E7|s6J5_BlX*K5
zlGl5jF;3$&KB^yl^WcT8uNb{%1kO#K;<H$1*1eYgiIZI)iktj<t(%uup1is*O1t^}
zO|dWHGV;3;st#Y%k)8Z@-m_ilqN^|eDqX4@^tN$Q@Xgd~LI!&RZaj!dkFEY0lYTco
zeQoxS7t_<`Zrzpj_Sm}mImc=pPkBeM?AW$uk#Okhbt%Wc+unZf$$rQ$&~k!ptIKOI
z!I^vApTC*1VDiUGp@OWZ!X@7|qh?=7;Qk?X)qhpTwiPQF&v)wF)y~)caqF?icUCz8
z<Ksb7J<|3)J#VP0YI(zE7FXU@Q5n}4iZP`M5`miLO|Jf*Y*y+={_c6e$WbpBZ*xN1
zy5#%@8UDF{_KDt1l-fN(cGI%XbsEJ_FX#G1O|A1`xf{PI<D>Bp4JVmDkDceZe6aCY
zvnXNz^5?5REbMr2+@Pv{?i7WY41ZiD=Sm!|5jy_#-Bd^22cIVMhNLD>j^;bR|ER3<
zzO$U$(wVB6iq9_J8sDJuH~Vs!^^r5js_Ol=Ti=~1y!W44&nhKZm0PK2&0}vq;0l}n
zbN&4Pe0Jg<`co$DS-VQ?tWfIdZB;f_?PlJ*k>)YpPqZf3rCzcNl4bheC(N+^)8g9~
zw9W)?KNr;+zq&{3f%KA4m77LA2Oea-sJN!Jd`al5x-)CDM4dmohClW8wY$^i8q8!E
z@HDNoen#r)eNPvsyx6!Tw|iD+dXd`OSkC?Lvz}P*_p)z0D*Q*4bMy31>NOR0rj@5!
zEdL#||Ns1d8}r%qX&?VMTWRX+PkyN%FKBV_va8#l)}Es~9{!O#6W<`eX3fg7N@vqA
zhpVLQL&7KNzkk@j_DtTHJ<k{(eYEEE<7$#S`=tHqmQPRXA3fL{z2$kEn&?8#dlNgG
z=C_#i@!fpAbD^fZ$c)cw28JU2jT|dgJPwq9xLNDpDE%<mXnW<)g19|XoH82Q6wCR{
zfBye1fA*&E#a}V~&tI;~d-1xNfj@)2f0d|Q3TyT}_1oL-eOQ~ZYpvy`ecQOtg?(7}
zed%tqUh^{+_jaqU-y>Ro;mMU}Z&s;zp1HVqhQZABDw`_B{vGhY9(GV(&yeZrwkdM?
z7Z=t>t$OP{>($)aD3ePM&qrIC*<XBAZv9E*;C@ZN9T)d4zI&$VH-o5~=xM%7?^dOB
zEm~Uq^y2)jzcYXAuoBsMC@;TYo4Z-a`&;j>?0TcIEWINpb!q#V1wwNV2$WB#PiX!x
z+pTkRnUSum`_WEuKeZ?9`vMCe{@wqd@xJWCT`$%C-K{JM>y~(PLDK5+#^`_XFaL2h
zv3wDz{8n;TH1>{J--GSh&u$#5U)%k~*j%PVdf%L#iS>63*0H}ecs^H;txfuk`tz@E
z_E<}w-m@frebud18+A1I$*ibtUn+Yi;+t-L{KPeE=~`VGw|8xw82vX#Xa3pd%^PdX
zro~H}%NXX?hAXcvmOFPYSmD>Ff^%7k^RpICtq2L^<}g*C+qQ)}$|`BmO{>xiMk}?h
z6rNlCXjW``T&V1dbnl$oeY0#+A~W}CSO+cs6*I~AcHxq%KkuZwpS>Aa9N_NT>?|Ap
zh4-n(gZf?lZ?rB>vkA5O*5w#jDjWLsq4aZuHx4HYzv;}i^<IB}o@Mp4zB$sn%&r_i
z$GhL!_H*6$%xR0%e&z`j$VwHPy+8l#Z{wM2x$Kz|+e&#}OxtGlHEHUW4EDax^9+_r
z*^yi&-rr|Wy|5^=Vx9J)!U*m6W<9l$<~GIpJ`swO3cl^A_i2&8{77?Nf$*{EUmo4h
z%j~(f^uKw)x2>M*R{lI8YWIE3ajkRr?yO!s$8`O+_d?RL-itk*j?D_|-zJsE|D0jl
z#5X*Kll*6S_T4yUb3FRHjFhBIzU_O~^3CElZc(*&FWxK8nC?6K`Gt38Y33$dtleDZ
zzCEz9?15a<{!MRB%+oik?<#PO3=Y|Pv3GU#_q(@dT$8{5{n(V&-TwX;Wg1fD)H9z=
z&emn=I~KXK+V#b&`5}|D+`?vk&R2A^Zr;W$_ak=m&!1YId;1ow(_iiQxg;!hlgZft
z1^2I2OHTZ*NHCgfTY7wg%He>kZhyiy+IW_I;N17tb(_Tgx^{zqS+3Sef7aGpmHMAI
zzwlJ*MbZc6Rk>-0uiu)+{J%Zk(wh0)xdvM+=5zZV6*4|rJm>5CTP4BYMejV^e{#L@
z%5;6X=Br=M`8D64b~(Xy3;SH@xs$JLiEPYX`uW+zXVK+_GCN%l=8DKJo$$2q$eg`;
z1@2EPRy~Y3dt(1PZ8n>Cdt0swzS~^(MsY=b&dG1z13PlIT;HD6v0mWq>yN%4zW+61
z_Al$(JGGHR?D4l<p~qCV&lNuRamRC!+iO>!EMC4?*0XHZnmcZKXJ<;E<9z;Q?*+?;
zGIlpssF&8>7HIz5*)e;;sw9qnwOMohCTK?Wn?JmBW8qdE$p?x5Pt3mW-Sse|@5+X0
zrI%SR^S`=Xf39(kr<uF^p_$g(rd^CbzG||>w{w<3rEey#+B~r`PxuJ;hvL}_)&?Ed
zKB*e(`$wky=A>&is`vQ5zj<R}ZGG-s->trU?{j>IuctbT%uW<K9a4I>L3-`x&5z%_
zu{cm7eXftK{ok>qnQh9_vsQeqP}%-|?nD=ZIPYmT@wRCnzvKzltF30-?b^en((~To
z{~3q+S8_`{=lC4l8u^7`Us9Ha>mSYqvNm1+@69R@w76vbY5&fGAIekhc1JTlkv`)R
zuCva5uW0M{C;294C(kg7b+4)FYDux&-*2Q?E1-5UN6jYfWzocPO}-4F@Qc4+ubXm)
zkwNK5X8^;3C#R1po$bod@~OY{vvy94>l`(wInC1FLHjr~f*x+x_c`QRA%FhSqo<P=
zDMhUeUH9&!-vy7gR^F#&Jx)(naMY1VU6#<YgV)Bj=*i{?n*|Yje!kiD-u=W;{q&t2
zPm*MlN)E2j>U!+;s^jyHb4?$=a=AU#k?6k@$Fo_dZ<C_a#6p*<_dic7-AGHSPZ7VK
zXms~W^PxxO`$Zql;Z9`B&sCULdUt}4<kr8|6W6TVe8z6VnXvekXO;<l;q302zv0M?
zmP(d-Mj@sf1(`ye<@4>Q6#doRGx>|>v)JY)(FSS%72VA1GiGP+oY3;T#Bw&*chQga
zClxQJ9>3&ze>1<z<;bSrDJh>`UOL0L$aQ!9Q^vQVA3tXoPcnYV8?o^72iC5&p0jL@
z^q=OLvG#o7$1hx8eY4jW?(#^@ad<bK)hXTLbnrvXKU})qMcs_wWh!mhZz=jLDw%mc
zuEi#niMdwnsilFTQ$`oVYv%XbJIz$O-mP*=n0H|(<MW6okw4$MBpUE5S!?fq99mZ{
zTBvd7PW`QS)oXqgEP1%wgn9qn#`_2K%eR-`PyS!^^0ue2f@9}#_P6&hv42n(Fcbf-
zIKgP$-)W27`_)%(;{EnE>kjjh*mr%C61<Y%OD5d$zV-jk|DXH+Z8oh9yFc}ExH7N)
zx0%oVraj+Q{^EpV#X8f!#^>i&zPnNR?*HcRx!!q7Hy7=9`cePz&AuIP_Bg10>0HOL
zF8BO4Thp+e&l8XLaZcshZ8B|vThP|yf7yTApZmG-v#PqWqmY6gCsRwv5ue#Y*QdVT
z7q%|uX58mvlTX}X{4rgp_|8R{r+HI$2cN!MAG7Mg$FOPXGx{_<`0xGhuPZqJIVn+Q
z`!UP-z=Z-UCoke+eQMm@R9XKhf8z)7Q)#BW*G~VeIy<j=s?){SO__RUFY+D=w+>C^
z>i%wc#*<5JT2hbaL5>5TMYnZ(bBXC~keb}IT*kY1$%7}4Rm@c_o-xhey58)W>ajgX
zG>VgJOdzcC!&7$~>Ki^&<x`&K^XNg$#%WUbUjJdO<-3r(AeC#2fpv03eZlqaV1*?W
z!ku?wt)g$ZESkE-+dL$r`uaKBkUxvAh|N3u;E9Z)kXqGCCIRV$*wXXOzxC5D#_hW0
z@kDq>lJL(K`D^xKVt3M5@_uaV%zWa#$*+^u>T>U)OpE%MA1|N2sL73A)G+JvRQqFb
zpHB2#`_upSt!C5b_N3;m4<naYG?vuc|NnC{{^vopubsOjmt1U>>*K!rI8Uxw^7H?w
zoPXt6H>Nx7uKl5~_=?|4PLUUzqYhR|+cxrZmQ2}aQF6{IN9vALv8;oxX!Q2w62IhS
zy7%@6Ffc0|`|0uP>uNQ=aQ}^yQ&_71ls#Fz@Wlk1^h3d^$8tmLZLOtLdHv4?#eWQW
zTwix-{{myl=Lc#(vx#kg+kC)v+xgQH2j<o@n76Scdt3<iDBxTpcJ}Ls&Usf<&aW2v
zz4zSwvkUqhbtaz|2@##R%v^f)4I%4`uWEFZdKql|pJl~p-I||XSi7O|$A!S!h@X}<
ze)s&OP3Lt#-1#lneE&<Msc-b>EVn$Y)wF5S)y+$)>hCHY_^NX2TCcoyY5uOloNo!w
z^FNvG%2F@6`pEVnOMR#MBlhzKQ-90foY<9kl&9j~W9>8bvXduBuF!k4>{?fJf6?v^
zfuaX9r_71rF0OZE`(SXuzWkSuUbmg<zh|qjSj0a!Pb%hAmOJBi%tg76^JrM9uk^;J
zZ*JMPo!<EPQeAre3#K1WT9oc7zqL8MRc!eJKf7spT4Mc=dOdf3y3V^M^HrtB{~ABf
zW1mC&PD=8BemT)Pb-JIAh-#9T`4)){<(tf(P4`=IB66mi=(_p$G_T#y*Ymic<&}53
zZCY>aY4;nxOJ<u-o@pPj@xAw|pzN3><zXqgg%j^>u>86-=;X@ebD!%&EYCL{);k^e
zc+LS0`Nwk{o_ZX&Wc>Gj^9*5)PYsew-)}oEw$*#3)5_O(-X><dR-3eSZ9Z${ZdtHX
z(8ESE<+F3*zS|4mU%4Ra|0@0Mx1EK_;)!9;lQXW%Y}X5Dzg8x)K5!2k*9Q&n^I_Az
z1*-ioJ|T3~*};<Qsz9rj<f?gm+v+zt7*ASseR<d6zK)$D|Kt5Hv(7);TvM-Evg2#j
zeCABc1#cELyqS}BT5r8WrjCNNS?w{)=Z6%9X9pCi?Y^5XWw}S}YTX~6@8UNUt+wxs
z{;=CPT4#sWCY#COA4BtwtxAr*d!0dTdxS@9=eO8izfQv?`&T%gcJ0gbt_f6K<sNvi
zUcA)BR5v(bh0t;3ri8-AyN^>o*FMZQ{B>;prxew$Lh<{>;%)KMFJHZWGHutb4bjh;
z#Qr^MFg&wf|6if>IfWJ$0iVSepV%#2KHp*2LP-zt`&t!E<>5Ydj3u$nRz32aFTLUx
z+LUgNURLt9<$lvqn;)}wX8(2#zv6u?+3Vn>9YXcRQRhUHyH`7FF};dn2%Z@_bzK+Z
zt@&O77Y!LqS(M5`T(lUgmSk>Aope!=BR?$hhvmT^)k<kD*Dsrx=T&dtxv%zvXxsXu
zjcwg}>rZcW_!Mf|#moIJ^~s(lebG7x(K18Rw(5G@=(F!D^ey8nd1k&od~ETVLr?G8
z2%P+8THYd8FTZd7bK_^S=j4_-KF~k%?Tqv@ch;K4@21*+p80OK{;xGZRxmCQzxv`-
z-?j%IXJj0=XnyItJUgNCwt&7^uT{ss-|Jc!H}mI3ElfMs*_g!~)wS&Q%h;RJIsG%v
zeOuf!#j<GLp1W5#wr%<GP5VUnowbL4FP|BIV(+AcX|{YauMBtv>T69n%$qK@XxoLx
zO|i|WIAOBwLS0JEca3>_{{D-2kt$QO{tL4^pAKjIzsL9Go^5`#_RVI6nDsRuGd4^-
z<~6;<yEdchNqd>qQ;U{kQl3&vf97j;t37!zgROm*%2zgF2j;xO?Q!j!e@-#a6*N@O
zf3WH5+Dj}JVG}*-UznXgCi=2|PE&vAiiKGPGb3-`(#osl%N4QSZWohNzenoCi<v<`
zvqCSrtWtUvxh41G;WZrcztzp}E~=S+VaD{s;_J6+{(e>eQh)00&Ze*75)=OGRnJ`M
zoHa95zgz76&x4mQ#T%x@Yx6%ePd%yHdhzu9O&3#bI`nE9dUj4%t6o?k`+Q&Q{ZsY(
zzgsDN{gt>YqQIok@`>geiC8ZC+U&KFZ!EdPo5XLpo|7*=Cb{{<s!jjzTWtILEWyGz
z<S=7Lt6a&$-M(&rzh1t3>;4g^*<va8W@i1XlrxtXV?M{c=gscxOGH1c->Ds|<)gK&
zM?-5(Kij_8>m?iczHgeVcH#A%3m(NW=dYPOcdakYU<tiybmi_!my9X<E}58UF$e_u
z$V@%H#x!|*hnYaa<tK-&8?0)puY9q$n%ZC`JMI3yPUU+(lTz2~t9^d@L{r0~{m;*W
z)OruC$!{;5w#Ygt_y2^*=KY`E=P!$|$$sj;-%YU8PMPCz)Un5gF-b0U1z$5KUCsaW
zMz2EZ-=ZAn`s$~y$F?o?y*BMgVOhKEBmaB4|IY3E`1ZPZ+q&*AN7`)vyzc&8fBbCR
z4c>g$73cFBFK**&FOTp^{keMU4^g{3_Z?|bIS)UpU#tK3%c&r<$7rKeiI|>~*{YY9
z)p}MwtAAG6?{@xP>uld0IlsPS{d>Cb<MQW>$`k%S+i|=;|C@z+eWBFf^6v7)AEm#|
z7-B!(c>c5NW6_Ds6ZKr@?)?=Ke_QZz3y;wCVD+XCy+_~F^SB87+xK^CbK-Ho{cE3n
z|8rjd&53&pUFI~ekm5Bjm12B0<Jp{R8%}7Z#wSU9ymj|XX6@`oubR5e!m<zVSNN!B
zNwSK12Ul&}%)Vl6_*>O#?Rn<)ng_SZc|Hzn3eW$eqQ7#N$Lzuj+se-@s^7jp>7_77
zkkO{2^4<AobWXNK9PK;N8M-p9eZ{ten^$OF^-*Zs>AmUlcJ8Y=yE<A0R9D?xbLH&s
zmhyd7c5{mLgU%a1(XH>(KGpN@Oz`|BIaL>tH#+_<(vf^&Ke>KQ?9dhbm^Qm(!{T~|
zKkZ-peP5jT+HWGea$P3F<B5H`mp;}?oH2j6%7Xu{E<?F?;p=C=9=G4#&MV;&p;gJo
zbi{$B_<mN(f@u;b?xa5TJL9;0_nd?^lcY9Ij&qi0yuG#KwGDUe6~^Z`9v02?y|nzq
z5&nyNt>acbO){E(r%-o#@FHGu;g=Ps`EFa^DyTP{nD=SL9RJnZ&$v3c2CO~p`fX-m
z;I6|<=ghcXEWdDhm6*Yq#*Ux6y*Egihb=4co>Z{)kL%ULN0zxPQ%h2rqw>&hPo7f2
z*E=Q+rKj0~IB!qPn^cmmmjCsSVvDK2$jex(yFn)&w{v7$Ctdp#G+}YllFmQ1+HqWh
zQ&X8%>6(7dtdBnJ-6&EeCRSRyc={!WKeHruS9$6+g`O7Hkeiwo9mh4F`QOiDC&OlX
zUs^wH^Pv^JJSnmJKE0V(H+g&b-X~KHnmR1r1$^=nceH+fH`I9J>Y0ZxHhNF|rPH+g
z+sY~S*7J8adUH*lmc_W6dtKvEp2PeXV>Kf}rdE|*^S?dUN3EZ?zWv6L$99(+HhVg-
zvRyf<)T>peC;yYte))Qix;}r^A3K+{m;7ko|7#D&{<}7;n`OluTw?4_+da&AvZFQJ
zE@l0l|DQAW{yb4Ab=6lzzi8f>?1;bTZ+<&A^Za(t&T~a3GrNCIZ+Nb}-TaR4^?yx{
z^YthCUXyX1U>YeAyQEH-#X_^bA?8`m1&7G%p9?#mMp_BIs!e(4{7mlV4puWUyLw^2
zb<f|<-j*u4%*VpE=+lXlXDxQlf6%_7ty@MaeXCoTT;$H$uso~R0s7TzlcKL`|M@+!
z-`y_n%yHhsPmeGyN$g|T7vjZkXwy0=Zt<ib^*;i4rz$mlsp;#AWuLue>(4oPpXym|
z80)NR=F6I1^sJaK{8Y;4xaVf`lc%4|PTHYczfEOt=n3(!PB~YvlrY_0D6*&jvWF?N
zcTVJi%bh*uc9-%a8bv-BeB5yC_qU!U+1(|!K8h>V4$Zi*a0+98u*ck+5p^B!ybdb8
zml8U1W_7sX&5Bp1`BzGLo%1X|MZB79To`+-ew(V{g^BauCVp#G_b{=zwOnlFwbj`f
zJM>I+%eMsgdVIZTQ^IU~Nmjb)n|<5`KIItCt<rWy54QjNWEkTceqVIA5Obe<kKdV1
zG3Tec-zz^HTR1^!=8sLY^jv2jpK;TVLt8mab;oP>`i(DS@4f#1TK^@}Riz7e9@tgo
z>n$n|uV$*>J%2;bu3cFxzePmcR(e{c^)5IxZd;yx_pUSFR$lp+b97=%bGKxh{G#2T
zc#3AN+IiO_=I#dBkcYSDRSB{+E?dQP=U~&e=Ylg!=i4op`nDo;>x}d(7R(&&3*1YE
zcgQTwovd}rSJl&1cG1O4k{h}W_yQXWc5r&J70a=%)1GkA=zskbztg4Dt7p#1+H>{e
z3~!G+leY2y5)ZahX3>58aIfvoP3!%dI_6njcaSdpCj0i(g2%kR%)4Hj&akn6zxCWY
zFP`buS3>i)rB$!WJG4#B!K!F;XqLlSW}}odJvNTVzKA*=`w_w^dH&pDy_Ji#RNQR)
zrFP0~<z2a^^z;!6@8>%?ZR=l|K0M5je7;-!c<&dDLbv4f89&u7%1@krseAu|ozoIm
z1g#aCZ#lK9h*x33l}{!!OOK^HPY{{+L(wVVl<5YJ4{K`kKP6pnNa~eh@ZQOG<%r0U
z+f5=n*sNY2GCp9%(z0si7rsJ4R;wGkuUj(v2N>3>dhy>_C7$0{s5#HkbJMm7^%AD9
z1xtFPu1x#L6Qy=6qAK8WCWF<nSuanjtUEtFq_0nQ`Q$Ih@^@Cw`(__vzEZe1+WE#s
zE6@DDWum|RcqO)^v5WJrw%VfK>u9~e)Bm548K;rP#E;YGSvnqR$*HoQEaLb>aaXDO
z_N_9;NfNu*pLT_*${4Smy-htJG}eD=<hS|{cVvyXHomzQ<D4SBEjO|6TVUDh+6`)7
zR)qiXobYzVJRiw3Kda)WKWGhjpy;z?S5uFmZ=dR$h29KHI--tD@~rQx>s)^!sp*4r
zNw)L;mIUS(yRKNhW+|(l=j}Aj&-UuXm2$Q3`Z}*URAiTZGSA5ps$kMDJR@;+?YT`q
z*2R`F*9Xt=IyQ&(ua&~pYPXg1ZrSg2QFqzDMJI>d<fV48lfkjJsh+8hDak8t{mxuE
zQDI{HB5~ms`9;yvRT}Fu6RuCw(z2i1-ej7vgU{A++1ocZQzq#|D;mD4)R^_|)s1VZ
zx4!qSoZMsd_u72*E0cv?7!)}kcL}7IG;s-=@2zlH<v5|<b)nDB2QL`EK4cW+4tmfj
zDxvJS%2r|V423@ecV~TSSiz-nLTmQjlT+B2tUS>YWGOZG`?QYNwLkTqzuW)VKKS;U
z)yp(CYu-w$t>0KMyR^{zpxNt!)8U*9zrG%u(8e?UhS9wUOHn=vt6J5h7cWAZ7JBcR
z@XUnm>5SCtmRFo?e)iY*`*W8ynalS|818>uH?6<@qLxWx|70eOl2p%ETP9r3vYH*S
zoGsL{JdSOBZ;a|w)9+Ic9X9)<6S|D8GG+bgg!S`Q-QL(4!*%ww;UwjBUX>Du6WdIJ
zr=CpfpHnJ1M>R<y;*wpz{G#ge#SfA<$o7;h+#<w$+MAjAda+<Whpr$?{oD<TJJzk%
zI`K}vYvsZ~_cvaV7N1xqFJI50*6*)6!$#U<z7B8A?+5ylY12L{#O(aIlX-8jk$phN
z&sk54RwV}Urkvg^(;zGLGdebL-|@)IHx9jzVmkGI#7yF<-L*F4>dLnfk%d<mq-G`M
zzy7S^?tEsy|HI!6YpfQUt-X3j_h(Uk+nS#}cc=Yz+kE}xR*sxfhQGd>o$qcn@jcHH
ze4r$TT}^1l`4;WpQCmC>i`fI&UUDd<T?(wy`YWkmTQtQgr@uG$@V1!MTh(&zT6mpM
zlVDgWkeG9*>_y&g-X+_f?FwM8*p+DZ;`sapM?Yt7zt$pWDqjC-@7uc@-<KFGeVr1l
zysDnnRNUl-g_@(l)q~S+DE^+rdFi+jbA#rV(Cv=v{}yG*ytA%&c10p)@~ufvOF{KV
zlk3(E&kMc6f4U1;vt8O&{Kolf|AN)KT+X{`?OZckvT{;saJbU7Ucadu-tFA!6!*Ql
zY=yv$WxA==(Q3zDa6jd4KW3vCo3;7LbKm|;vFcCiJ%Y;>n-BiZ?q9M}<?QmhP5K8G
zPfJXalz%Z(*ky<H*WHh$%`OSfcUfO#v%<?HeDy7+$5;B;H*~gpgl<eccS3f4xLDdH
z{gp>}7^X!Wid0`?X`Qcf)c(HF-p${QE`Lhj{r)7s?Fvgl_A8IORk)X|@A<jF#5U`v
zLr4C^1jU(EKcniuNL0U3zw+Zo^8U5_{kGqfZETw5e)E;t_@}+SYrI|9mX*OJ+^}fr
zw)a-jO-nu;;TLSWWTdlu51Z7oHiH!>-pzUbX;xkm%cBUfW7BxNdrxOugiMe6`m%I?
zaU++mH*?U(7ZKhyk}+5P8J0@TTYRg{$A9I-je5-+%c?z>%vdcup<W%-h_`5%{(l~O
zkM6eeupWmwpPN_vPyX`yf6b0f_9@4M1<&$wezm<?o^_;oo4MY!YCEI5;UcX%NzpBz
zADxJNJZagc1Hn~J9M$WN{tLf2*T2e%BYW<fYfK?2mdircU%NALddMrgcFu+OrB*DP
zRlVD>OW}x-uUxiJMn&e1m=pEZ3!?mWk5)Fz_-0kNZC)Pp=xqDCJgMJ5C4YB*53`u~
z^2lD9$gFLPm~N=vI(Tbq;DSDZ6?Ug;H@;%Hdz5#X+_|1S%_kq7Teiv?^7m$UiZf}J
zDl&<F+j}f;)7%QhS$WxVQ?g#om#a*T^s>1*Ye{Zt?f0FPioH3Jy80})`?3<e>zBzX
zPB)D%6L!~knN_59=*H2Y+ow`BuIo+oy4zIqG2@x%p1%dRC#x@voaW0pNmItYdf|(P
z|Gx{QI{0F`Ul`@;ouBOjalu@p)lW?V_?`Q=g`HldG;yY2K<>(!{<o|AnpX<1m~GVe
z=cC`@r%UzzUHW@UO6;%E^;?rXZ(N^JZ+A>H;HfCr;_C|DJF8w#RCL_-?1au2jc-yh
zD|RaVHPl%D@YlCv`*!^_>D)DGiNQ>TWVfv+_w8=?yEW^T0z>MmVz=dwU1oA!4m8a=
zA+j}Q-{py6OsanDL9zEtQXcrRo7qm53Tpj1bE3WV|3yAqkL`H*VDsJeQxCETTrW<!
zcIt^yef`FY?Fk0%>zAD=OG>#p?ZuTFF%|4@uYS3f&8yI}p)G~!wO^oRl}w--*G{%=
zDa)kJN2@NKaGYQ8+NT`vs+OI_%15@yZr{tryJPj;{L7CWd`y)Nuhq{?p4IGqxAV?~
zcizt%eHAa<;i|7Jac+CwCh#coK$!kD-kCLTGW9D$+d1lYE@>1ls#Chaxc&RXnE}zy
zwQEw+zUkjL?ws54g`?P3_%f^3l-t|?mdxX9mQ|R2>cz${Cz8*rH21OOUJh5f^{%~A
zao)*#GeeWFyQ*HSdzKKL)?_u^w7Ye^VB<Mo`)^aHe9-a9nVETJ8tby^{Dh+7RR2YX
z&bduncI}mdS7Wp9%=*rWejhDv)lQsPpm})Ky#}fDWmS+K!|Ep-yk0rst9I{h*X4Fz
zndR%-9<uq!g5T#nUfeKl(A=7)lK-Tw%eQ)eruQZx^U~c5a+78(-|##0gf4&Z;%j~q
zn~Z+ysZP9mqV_wh{0W1c<6)_vCe#+V^;>pIeBxViE&O16>dkdm>g|^Q4PMuneM#(_
z`unO+aT1?=cX}S}o<65}`cwV>_bU&t>h(Q*#K>~lXEh(L%0B{$x$bMD`g+oCiOf!8
zz2^7!M8E>Ygs!Gp-Q_7!b;i?omD_$%Rpm*K(6W%bx=6vxKRY`2RZ>u>Vx8T)rZp!$
z70n!t`PA-rUh>Jkukrm_b6x$!t@jhYo7`XI=n$&4y`*f)v5w?lH(SnM?#Ognxpm{>
z$(xGb>LfUGX=hpb&(C!{;k3@j^y7;qGFF-|&qh9NSwF9Tk)Th^Wx2FVi=%wEuKQed
zOO?%9O(}@A<0scarV|gfC;Rn%<;b2~e(Fg4x%GK9->!Z=#&32i<=cb%_PQ0nm)GC_
zZTLZ=QiREU)y*@-2~X#p*`BphFQD_xq9(&!E$fp<e^+N7UGlj7#Wh(E83kbmtzC(s
zmi#9US#7!7#knrnn*T^ajJULswt=y-|G6DY*Tt-u|6yU{l<RY36t_KAa6aarF;gu0
zq3gP9tpUyg9#i#8J^P=%?0R&ld-gmdCYG5D^+^ZjOH}TP+5ctZ2NA}zpO=(vI<vJ{
z)m8kkQXn|IbyH6!|KM_)aX?CS9%t_=Uta%tY!1`jy=*(<S3h&2`iU9)iWW}$v))(7
z-PZe^!o3?#YCc&r#TKicU)+95XF}pmqq|yGOWnn%-FdUJMyS#(xMh{kk`HGpckz4k
zDF&4K*X!I~?b)m4ZD$|^$yPzr<}l@St~{pYz3!FT*O<gmZQY&SA^h`?Ombded7#0l
zFyY^c(yI%0KAOHS_JWzAkLKq%>*ASDJNNuGQ2Bh!_l?V5pS`^klRZ5{<!9I%rQZ>K
zc6t5uYG?b|i;O0g=Y$_)JvU*!u$O91S$N6he=NuA{~X+B?%r@F{>tRXjs_n}qTkrt
zaiuc-mG2WfW%nyvNO;4GojcuhzXV4q#NDs!$`TBHS~ExH;k%See1T0m8rvqW`4Sw`
zu%LR@F=2^YY|~z?&wPEs@`PIIg^3Gtf0a4jK4Or3>)LG*_c{67E+&_`ufMq~sd?tt
z`DXS0uVQpvCwAAXO9p&juYYgC8o?j_w}rFYuD8a!`~O|rIB8`hOIDUud}Xm+kHw~#
z<vQ2>!#s_BW+#-m{aWw8E&I{mdE&y?il*h?nRh2OG&v%8?nG_(REfekd*N=eD+jLi
z-ZheSS+-0u`F`}9o*8=jDldA(pT1rw_3@FWn}UUZ&ccW@HGc%|*6)&5m0w&^Z+G^P
z+{P_kOuR4kEJ8N4F=T1;>|B)iA}LK@SL%M<&rP%L`^zmq_^mW*<&l6KDW6dD7u8b)
z`rLH;TAD0U=G&B6PK&nAJ~D0p=I#gEMRw=@&;0s$rr4psy}y1<7p|?S694~>WAV=?
zdukKvH7`$Hzvj<O`~Qdchuf{MKa@J{Q2%O!KX2Aq=qYu^*#GK2nqnNAo8<StLCD<f
zcbkYpUcr4yiJuk0hY~*i4eh&SyW!Tn#k2J$bk4rK*1=u2q)2wkJNaApS9F&O{MRXL
z$bP<CJvVEC(7B$r^|L!3>HXRulA|f6@w%YW;X8Bdhg8#8HPOp2j}_X^UiWt^&z}0~
zzosT@wCBg9?i0J!7uiv7dp7)xQ;t)rB->}-9YJRN`!>~agiEKsTz~ztzR&+N=KP->
zVhY}DKb2l~xwqp~>K0X19lqrr%VchRn(+8Z)@e80sn%Dg$$pr2p>an@;FPbQa^x2!
zy%h{zr9UnA?9!L1JiUA72q%RnezFPi@oVckRi7<+*WD*0VA08v*Q<9j=&sdjb?UjV
zuwaKm>k2;QD(_3Hc7OY}o&Su{`9q?U&mK3}5nsGpCoOf|p{sk(^L{<+ez|V)v%Nee
zsdKK_onX~+b+S-cI_q6XsHj_zPec00slV8ZCTtck-uE-4=z{RQ$E`D#-ebDjxOCz@
zm&LCI#ktxy)!$qHqOSb4mU&rK<EwA;|M4;$F;lR5$TQ35!)u1<c^lWC*L?o*%T&b=
z72$7u_-^b}zIS({@Nawm^l3a-i^Xo|iikb0`IBL6#?4f}Es?8y@65bEMw*A&zRuWZ
zCD@)h#VJ@kRi2ZNc~+wJvhC^1CmvSfTfDb8A}UeTFldRjZ)kIUnmE_)176lk&l-mP
zYufv@;d^znpPNCd3(M!czagh}mbJ<~n{@iK(BcjiR~P%KLi?gyFIPLh{(jSHNz0Ni
zD~o<VDdTi=<9H!=CFR#l5&ybvCr_|Xe&FtRtNiGtV+;|$$_#Yoe!ls3<q4id{#P1a
zYs;KICMk9VU3T2znY!%#-^luk_u8imo@&o-&s@;bc;4ne=jT6`lPt5Qew{PBEAv6S
z|LddQ+h*itmjum|Ub-#&wT1A1q44L~`rlk_Wm4u^R>qZSx1Twg(z~Ba!=t;RW^=<M
zmTg+Rl|1q6d*r5`YuI<>Z^Qmwa`8di*G#{B)M@AZyk)StR$Kn&wTn&a`Il`z`-eaN
zp1)Gx<Am<CtZb9sie*dRNxq$M_)(&?o})>*`x%qfeEo{T2TBf2*udU4rBJ(SddoY3
z+46gfj|avv{Fqs5D{Fs8m+vT3q(k!m$(r8_LU}Iy3)yr>xzyF8A(<h|rE}Basx!K)
z#b)qbF7TC2Y5Es4!KZmw-n9eH4fVdp=4U+b%lU3`Ti&Jm)pzQGW3GFDUYV+|(U&dh
zz2k3L)piYC<<9-{G#t-w5It!Z!*l4X_LSR6yEdIzQYEurrbp2;=s`fcThHd5Wry+^
zyRYrn(*KnB^Tqi;$9cDNZ@(|^{W)Ug^=JEigO01*?+RfmF|B*8BY&)*@_5Ts=9zkH
z>u(s<>r8tiF!^QP$8yzGnx+YHzM8uq<`&4`Wm249cY5bxp<u1-kJ>eNid<tf))l;e
ze_cE&GTtU(;<~x5-4g^0e{H#$rTzc&I-zavHkUt46V{t;x7SPT;kDWdIRmDj`F+)Y
zzNhuOcbRqm+kJGm!Jl2V(hW)9pT^sr4}X33_=L;#6aL5eOx}OQ)jGyTxB1A|E1e28
z$-d0<czf&qB9C<Y$Mu}Kx6tMEES|1K5109^UZoLu?%o**+xI4c#renF&Si3B-ivH1
zH@#Txae3?gnWtv%cu^Z|GTBFbbp-p7l^HtXSL@W7mM*%Lzl5{HGUK-Or|$iw9xlNN
z^)nyk|GQx<=GS-T?~kMFZ@gKydDaaUF~O^hrmt5S-6?#){N##Q1z%fO=El&a!WZUk
zmvg@N>3I_CWA>zZf75;)-5&PuOZJiepKbY{zWk6_SD|;yZ^ni_KJykNbyhfNO-l~x
zbYIi=uj2jsipaQ<NsXJ=pO#d0cPd(O&?L9)>QxQP`r|FhJ`B_9HfW~RCMq4-Yds@N
zQtv@l8{>wM_!nlp(+ilU^rVPw5)0_`>`)h;G1of#<aCa*v(;Nq>pm?pQnGyA&RtSu
z=*)0lO|Hbf?Zsrptngm1q87GKNhhmh9Btmb+-0|_!_IP%a$CluZBsw89d!w2d@^(A
z>1VxRvCLOZbm~v9xy7K6=f6_uux`G8r^d8mUam?0k+0s%^u|nbzFq9&y`*iodgUUi
ze+EzI<T48Ud$`_C|4_cU|G}JgnTa+(zj1%xp?*iO{LUe5`Ay;P{{K9}q|^6G!*cqb
z6MKK{sbiK=e3tZ2^})TT-)8Q9c`M$);`7bTC!79pv^AUlxx3@m-g>4Jp>j1x)m{F5
zyjJ;qS;>MPk!NBYPdpWu>9MtiZZwMD^)u>x(28BV@;u&Fh4xKg*7Ug_w_otMM(c5p
zh0C0dpIZObqEW=@XpGJj|GaxT?%vx1J~yOIUS;NXDRGmV%F+#SGu9;XN^+!ax-cUk
zwDfGr(t{t}^b<dwce(TFx9&ED`j*2RE7x|0)}>twshhT{<^|81*Yca@%WJJ#BdnYy
zb!R_c`17W<mv-hI4)L9JYh_i+7TZ@j78XyQGE1$Svj4Azm#cJ;;hu;SjJ_PVEhFrw
zb+%<Cp3h!i9n1dZf8@tY6}5NcyJQ_tFVUJ5nPP7>S=Vav+)Yx?=f4eX>c8c2hMBou
zU=7pCuP>D4V^!p@o_sF;Y2qs`yHz(@6P_&I{ABUMAKbjvN4)ZvHs(eLPrMywHeLMT
zF{90U7az{iD^pata_UXAN6wolu^ZbjKiu#<y_oy9*7ftxw&@mEZ+tZApzr((5r2*!
zU*0-@&5|tZ#3d!ST6Vk5T6s6^ocl?|n$`-Z@JRRir44h8-mmpcx~rZaq4w&1x6Geg
z$yTbq`>sari&@uIr0Rb?ME0mxdbmO23%$ohD<?f=k?Qx$Ocw3hF8`D#AZgXE=9IKA
zrv)!HX-OI1m5u)U?Mm0r&6D#IRi5ux(!KTN^@)mGW(;ehlnge#Q9XV+Vshl#z;o_1
zvZmZx{HprgN0a*aT~nSePHJ=1GP}FVYTd-dbMq6qVw_%Tv9H;=Yq@lgv@PGdNBy6o
zT%W~wZTK1S(LAQ@pV^a0J(2J~wFYl*Z{NQ?<m+_lIWy%JsXe-P=x69(#k*TOw4-MC
zusvU+TBzW3-MLYI>x!2Zi@y75v)0Xum>;KG@RjRN$k~8WjXC>Srqs{s%1KHxe0qCP
zuk>?=UEbACCM~u1*w@Qi=e{^g=fmy`O!j<%o~w5z{SM!A^3`FRgs8IWbk;kM<~(uR
z$Ew<>kp6xm<A!5fV|Ro<TzO+b#p`MNMX&Yo#%=qh`ifO^kyMhsOZWEw0r{cxcsH6m
zEIpZeZy~#rJ;$E;wUXJdb1n-ms`sAT#HjeoDp<UxaMFu~Cg=B0o1Kuo_`^(#6NZh~
z{N)!Lo_Ob;X7N9#U#RAa*Tc=8F=sbh9Pg?5V)Jvw+(YTBW-ep9GqL~Jq34@*Yt&eF
zvV@lwOxCYan^gNOYUTwcv27*6tLiUA{D?VqJo0>zPR+R!ic;M3ck9)#oq2V0*4)Tr
z^#KdIRJi7+F8>v!S<|-dsYKt@O#dFA=`&_cyf?i-=b04u{Ows!r?j4z$lA4V;>Hh^
zIyH9d8t$pqWWVOxYy0V9#E(6>?X3&gVpqzxxwY*)a_IS;zkffi+OKd$_dv~|BEO0H
zHPhZU-SgY2>Ntycu1^WqgzRWF5AoKi%1;{37Jr*wFU>u_TDwN?=99N9iJH%5ccd-4
z^p~yhXJBTs<XP_90{8Q0?Fh0KeZlwH;*_z+-?m(>D;?gOeFIGVO>MH4em|w>-qF*e
z=4RiP`&@3@%cDY}CpNup$?eNt5f?fuCFQ1J&guusmsREkU;UKwWU*n+Z6ltv!hJ7K
zn()d?KC4_G?pnVq@rB1K?e7l`nf2tJzP#kx$Ntx$?WZr?Z;M&qz4+^@Z}Iju1#hFT
zYkpJh7j<ozQM5H@CF^RT*UwAtcsAEG@rFOnQs7lB<n69_$(LrDdXA+bDs$#ys|_>$
zPnocjInF)n%v*7}<5Q+xotC&O+~Dc-BdQ0>ZPq;MIAg2yWsYGMbA8(-c^%a#i=eJM
zEq8Yb<h_l0{(SKr##HS?M*l9JJR7FTXwp@Z!_Ry{_0P*Wl}~SJ#`c@5O#4w;sb|x8
zj&-TM<<bRFEEU%rrak==!<NCzd}r@7t?jbRPA`T2KY1{J`YZpwD!mhbC*Iz$^xI|r
zJ~gq|6Xm_HALaGD=EW$xXXb>)`hTA;-r3l#_L%)U_wGMkd*_R8SMpl5v20Uvgsoid
zmL=v<%wMfuT$z}|@0Xw9vQMuxkzLQcN4P%u|K7y^e_J;{jaT@;{EqHx3#E0ef9j_>
zu6j8k%l34j;*$E*|9c;6)Ey7}vybClhojKU-BA;?d&T@^(j|F#lM?vYu8G}pm*uU$
z`L*4sC9pwPnx}n-y|#Jk12bQJxqjD&dDc;`(sL%C^xC=Ez0~P~!Ic!QGuMk+#dE(H
z2IO8})}|5X?NcUje$lLz@}@TvcW^{4GHQvJyw3Yw;%ULZ&i(f#^X+$Q&bfTXXi0kV
zdG5st6+d+H9Lv~#*kn4~dMV@6EqS_d(^7Sh`r{|3>8V8@(|NU0^yLiKT;6Mj-Bmnu
zlEq5TEb>YV+{pD+Z?@9)(~G7#T5a%I5)@%lAm@4SALH&BdtT0%nRB3Mz3OzarpMOF
zN{@C)Z8Y+(?96^;b@5x-vQuAj{MC#m7v>1XOBzp({UDq^<%jd!+Aor73~ldHrXH*l
zO8=k!RKd6YqCP{es=U>*JN|rHs&Pq{=Q<T_-#T^2NKIenczRatHMU^OTW?qJuU?td
z&$nFRYfVAyhEqy<EB<cOwR<P^^5<cT6`O3|FNu=;aM*K2=+b}3YHrQ&e|`Gu?Pr&R
zk}Gr{9c~WbdvRdt&AKfYuKc{mzT07qr^fVswLDL&w(!ik^6`5;yTYBPk#`eMcxQ4=
z=Tm<ioR{1dyz%$-PqV_ka<<o-O<5r?_C&5?kx9^#KR(Z+r!z#W&Pq;?vUfflq^lt1
z)PL<^WmLN#qmBsc#*?1H%$H1eTz|1tso`*HSE|QX9m73Kv)!{-yxjM|*Zko51J+9p
zEt7}~>bk#zr`qGgr5`iiMb*!2toAs)Ai1K~s3Kd%<9CyIlTicv{KP8F$qKjUb@g-~
zv}G%oaOv;}kb1<QdH2t>bg$>y0*ZW6C)7Ly4+!Kl`bW>ap8HMra)@+Q$od<~2`lE_
zpSaSg`9#U7Gp-#E6R+Rh{qX6#(k92G@9zsY{oMCq`+|G3lv1ofeddWNY<~5xV#1Bv
z_B=iRHud*(rNEazK3}#jeE;m%<ke@cdkK8H_UeDtto47YQ-!s@|I^v?eQN5CPgBn_
zCr>`RJz!^{+BfM`+vj(0EN_&zTPuD){?oIli*Ic{RK%P;e6_~-fAOj6E!y|x{wRE&
zr1|^c>>cO-C&VYk*Ht~a|2RNLUfizV<;ddt^+hTl4PGo}n6YJ1;O`Stx9$|0J4>6}
ztC8Qj+jiZd51V&C`ryiy_xFXq*yOV%TRT@?6n~VL^DR;H#hnNH(r?6Z{nVQB>E!Lz
z{qlMDpPkzNDR*CGz?&ny`cs!DJY0YMluf{%{s6at&=poz+?Qr?X)Qe9yz_RHU9scQ
z)}~YT?<^(i)xOnuJlv#wQ#U|jZta&Up<IzeKPtua-al(JkjqX?z2;y3|4aJp8+*<y
z>%YAFXVXo!`n5G_v!49R+?Bx3%D6lI-|?EmJ~wlAMV<XS)#&f1)Ti}=>#HQCwe#*A
zI)7?<=n0-{Z+p+$Z%<j4|98?&oByZcyL>%o_}n?4KJl$^*~jX7gV=NFr;b#gK9gW@
z-{7%C%|t=TUWs{G5k_+1bN)^coq1ztj)GBZ%stu4#drK$Ev-zw^Vj9e*R&?S>f9)n
zEHu+KS5;rTXCB{sU5_LG_y7O7pG$y8!lm9QbZ_MYAw^}KKj)SlsPX@too17IC!B3@
z3s1tU8*Cw|W+fVjv-)_J)Hm8q<dONXGi=q_%#hT1|FX8vi^%>ot)kG*`bbuc$9X@0
zqx!>Xe4QU0{FX4TQ-9y``trBu+JZrIc4+-ta_7mTC$W<KmLk$=S-cbOM6ch)cx@h!
z+)tUs(9y=S<5h>3UsBu5{{LU*!N?!qFZV1s!>D!r`P6SKciOq%C<?t=e{RZ$C%;Wr
z9@EcEoxb>daai)4N{I;~%^6M4mKg5rQAnAipm9gw#>)4Bhg9DxFS2}`E!DWoIYlUR
zr?XaJ3ZJx!?eZVf=h@W$G7IpLv;TD}ZQqx8t;;T}j{g6yocvnjs!+9Q(?Qway8RB*
zFZXh|{9JkBa^(52%TMn%*%p3}-TAWq*pHKs%ylb%*7(L8F7}PqVwCnhI!}?iT;|TJ
zuXF#M_Imhab!eVXS8J1GhXZHx(ekI~o?Fg2|5@MPX1S!>Hnrx9yF4Gg)2OSNv?JZ9
zL*-DxMN3INyQLx;ty0I1Jh8m><W~y!Y7gf;ug0`#s;b2gG_T(3WLg_C>B<q2lJw`j
ziS?-r!e25PT<iqwrmDuYzUo~UuQmJ4gT>`GKQ4J_PI|X@cU)S_t92cdCF(OhPY15Z
z-LPUu^T&-B(lwONbY3Z%Zk)f4JKJgL&&>H1=l&nsw57**XYBot%~RAJx#gLca!!v)
znIdr~^B%MA>#N6)vj&@-S)X)a8vhlQBVFa|a>d)2>UnDZWR%<gYsx?Vx$)oM4FR7{
z#(qB_v|i=u(&>uZxxI^K<lAd|ulcj;z;v4kjzvCK{|7G7k~A>#?UqQpm6)t2(R9UU
zz46A<7jMj8`B>$=)ULJ6;oT@2&N$s|<=5O)ZpM{d0wvie6I3sUR;Q*$tcdo#BjUyO
z<gV77d()Y&F!j~1o2-{HW%^mqoTbbk_B>ix`!`<V!|dt(ZzOWcu8Pg)f8#&<M$}y4
zuYIahdbk#^dl<3m<mHnmm(0zy+_iGKcu;!5q=3Yi+D@&N5{lYe5B?R@@;Mo~C}N9(
z;ljt4uB?`va_wiDV&&5Cx~&WcOFy1!uoEbVjjK6k(eh;d-ptEtC8qV;l!7-*zq^US
z)|YJsbIMcw?^(=Ou56J`N$d~VzRGN0%#IUl`wOSHFECxa@!tb$&DAoZG1vY^qzk55
zw^%5h$zgBEOFCPYV6F6E&1tpuB61Tl_rE)ADAw!Antb8fgvF8yk54}FwO`g9cl%?C
zMS!8*u0y_i=Dd5vn=7|obn~(n-}+~v>OKelC;tC>nEm>`QrT3{2-fsdHHVvLJW=ZS
zk=mKgfBE6n;JA+obH1}biZPWBHM{n`hRGyq+QHE8y+{8gC}(rrJ-g;#PNdFap-U^G
zB!XqrgOqn(l1R6{RHavJq`2s%cJ5Pe$Wp-x?VIAZ>`XXa_c~w~|Mdcsitxz#MN2GO
zd^fACln4(~+uS|**2L4-I<C5G*}c@+$$HPd#lQc)efMOZ!tcpW{XX}iKCV|gwZwFa
zTs&{#k>g8ccqU8Ddzuy^6~wu<WYWDG2H{JO_O;CMTK&Fs%bE_p={HR(uCM5rw@_)-
z@>;2C!|AIle$1@x=d9Zi&9UH;+_c3(ES&oFUtiukY!%{O%C_rj(9f!=5jPBVe5Xoy
zq*X@#lv?K^zamByv=ms6^>$d%M_bT3LCe@$(KW~N<{C$o$c6k9@|oTv_^(j<+=RRg
zwo8Racy{`C+}wXe*x|bAio-u9&iL`<s^i4g9TS3PZ~FHqrS5$6*4nds3c2TM-#UDB
zvFE|L^*fjrYcIVs-`{0T;O%Aeu4MJKxt!scyF62`@Aje&ri;Ai^`1WZ{7a!W@Q>C>
zmsp<zpS}xu={O__T%NSA+BJ2i?<3K0cFBK66LyxR%9MrfY%2NR>K5t6?eMK@llYzA
zuWx469e@6}=BZ)K`sX(MGopST?0NC{@OM*VLGILV2E0u5^&6jEZmCH!cyPHZ+s<>w
z%zz&8$ted`XIoAB5WoKO(ibx}{C&6kgLj<fcBQ{3Jr7+!m=pH!@8vV%a&BAP=1l+Z
zKfCR-EW^J?Wv0B>PFpS(Z+<t;_MFS-M7!#QbNxGdv;Vzhyt_!v;Bo_R^<KX8&nst0
zKU*)<7hDrAB<pdrh+}#EPsQ{N#o6}kbG~-R?2Iu8{&JE3UG1|?cYXH%xOQrOo#^4a
zG43DSg_&-vdh|Ga?D@9$|A~3RtA96_^f&y!6tA^inDzB~1BL&xychZBP5alyo4?kk
z<^B{^d86iQ^Ec@qiP3J`J174~`lU0o(|>IC60HyraDG#BWp(PF+F+^Zkk$34JGNAA
z>Ac%!aa3vFGxqTLYLWX_IBi}wrQY~#lx>&x5AWbi9fM<QJs)p$?mnU}pRZqJY5dLc
zrlQ)#yN}*`T?zKN?)qi+OU+|>8xDyI9|EsQeYrR7)7A&8o=pEb|4grFy~_tq^K8(X
zRIYoU&8jXQ`oAZ;B<%am-36ODr~B5=k5jE)bG5cYu3+jP*rL<~@S4;!cW>|GbqKfl
zR=F=}mC=*h=|BA!T9l&=)fxIqFblTq&#!-*aKrF){_3B$|0i#smHa+nQQrAq+nnQH
znYd-WS-i|onW@-cS>?~Q7tOcdFOmNI^xM*oqbh5Qc<aT3{MR^W2ks0^-sf`v)1qTF
zDeYZvd0BJUTPjWI;rV#{k+g9cqjX!=#hi<MEwLTXr*a+%nBa3LLd7j|LZ{HRq~*JB
z&&c<hF>l&^9g#Z=<0miv#PZGCLv2RLtdtX97Gz(XaocUuRh64(uN~i_{C*?j-j;6@
zmWh|dtXaKi=8dD5wd=({@pWXpJJeY8eW^CbyMOl8_b)x!(Yy3_xX3%-@}AW_<{f>v
zMH%<qGETIXnffl&{dkL3@VZMD##0qFSKlgs`Zo6d(VVnt->09k|ERx{!DFdM7r%hg
zO7}%~9~Mm%;e1f@ytBFZ+Ovyqtq=b{CTYp!7xg;gFsHlDl#hNlE55z&te=zdbwPHc
z*<{cE6K!9w_?GF#%%&x)yzA;efm<>sRorHjU16HXBYm+o(b>u%-~R4diT0Qo@iDfI
zr<9Ltb_FN0O|D;OZSExe`1ovtU)Nk$e7khfL+I4R%omSdCb>?Z{r;YX(6Z^<l&2gp
z=vqJD;P9&j9>;X-EV>Lk`3}h_HhMm)FP$)%kIVhKV@turhA*K4D{P9&WlM}3`I<CD
zEIiI1_20uI`@~~LoK?q04>zTCBEcW6R>ZP@<alwVt<UN5;};5k8lpBf;_^Q)l?43!
zb4tRBLrF6^r;KmUn$1@nFS1y1s5^3fS;_A|{qrS9g{ev{cN90jHGQq1!DK3UW8u!@
z7wa7rrg<u@_}+Y{B*11DACpi!qtVx(xS4B|MHpr-eUNo@lAmA#w?dQVuDYM48*XHB
z1k6!&{T8HSU-MX~@7?my6)F=F-`Gmq?o(mqyzpq3_iBc|L<I-;rcV}8t%o911j2Ys
zR@|`c+hON_#bQ2F;i+<-!cX4IlV&#FUvpDUGPbdPfr|<kx9H3^Rv#hG@5dFTHcst+
z=*Sro8GPCG(edx*i%*x&W0O%^^Qur;z#!9VTG_`)HPz#c+1n4SZGZHWBTLSE8cW;3
z-7KYVKbdjAGM>y@+7$IausTO`LE<c<*{&-y)|a^+-F)js+7r>qhDjZTFXh6SB2N8E
z`r3Z@gudzIh4sc?lNL;Vaeexd+8f~-yYCe4TUxDn&g`_ceQ}@-U!d}qIEl4t=F&_I
zKIh%%&x!rD(f`~#zjM!eF68WbT9Q3?)t66Qaxbq<bZL2h@bUBr|5Z7>V&Au&Dpy)@
zi+%3?l&1IdR)jxE%iGKqEx{VL{Dh%%riAD0zO8SKw_j99pUN|<zrN<ddDE`_wg0cJ
zs9;&+_ChjhtAK_7ip_U*SbCZqEZi^HHoBj?xn{ELDGB$L4ZkirrnrAly4E&Dmg7Q8
z(JPs?mO@ecb;SJbSM*CAdbz)5dreHvnfoT+UoP&8|7Y{Ra`y%K(<gq*Z{MxDGxSUJ
zn%4&{?%&ayzWY=hZ${vBd(Wj>_1dfl%GTeX``6Oyg~7`RlR0y1j>idgs;LOHyW1yA
zE)S97J}hdM$Rf1v{_8|z(;T1Fxtn$;DGQz!$=<xo$y375DoSO3lv0w27vJt8pI2%3
z`X2q6<o@x5Mww^X8}&a1qAi=Vf@Hgrn_Oc0E^hnW)_(TqfwY{5#})k}61uj}+gSgB
zd7?lJgRPwYA3cGidXZ&+Tjp<i*1<EQ@13~$il9@1Ex*r->Hmpq{IL42*ecOF%}1Jo
zd(ZdzX#a^2ax;+j4_q(YBiM2JtXtn5_U{^U%t!ThfBxGd&sf+K5%V!2Lx6$rMBJsy
zmN}c}Xm2l-2|3qj(JOo-RQuDFB0eD@nKGyP<sHkM4izm;=zsK2W1mK&m?2L@fUh~%
zk|L%H`&e7s4_R_=DN0wCZcKZ-HE3dj+?#dt);b8wXl%^h-1%_tHKs24e|vt_?fbjr
z|6<;A7ymqL{`J7e{<iJ5OP9|%*Ian;FaG`u`!D;?PJI)9^X}Pi*Kaz1yQKbk_n*G6
zDs1v^|Nlv^&)ji;kIuU*S&J1*k3aq3Y@f=Mx}e45tBqiX3y<QB-GA;fo%p=WG;7&K
zgXCkbjR$1HZ->p9-FC4wZ{?=0J#9OEMB4uw)ul}F^jylLeKLu`W5y*Xsa@jr>mB~w
zOy^`};yh`^a{7a0oA1o+R?|N{%l!A_)1j~vlQp;mSA_*B#ayTtVEH}gi`xnF-&0I)
zI_o6OS*Y}6{+nAXO1J%3&DdP4R(<fQ?4!fI8ajy^47hCD6c)L>b3LBIoNZt#ZEgEH
z@uR?z){Ls>5z-Mg8@A@Y4X;dV&drW6W#xOE#N4z;^WeVljW0V_>D!*+|0ULMb#0e&
zh~Hi(zhg%7Vw?B;{O~4oWBtF%t#uk-lV|*W>nZ=lAXiq)_{D9%6s{R}cF5XT3P-)!
zl{CXxZ}xU&r)9+&|FdRwPgxisbC==lT{fxzFEj7?hM%>yRMs`$%-<a!{>rIYB-&Qb
zCx7qquSHwq@}~c<F5I3yt7v<it!=4j=eEm&pMUqC@87(C-M0DDwy&#N)c#?1{r!9E
zuDR7{TYmn&C!Qzk>NUpa26yMH+qIwE`}_FGsokN@ej*HKgr80DxRok%{YUd-$>X~-
zUwb~=Keujg_MHtIk8iA~`FwqYRi9gO;U>kNR3@?XmuL4VU-6l)nZ8s(ee?aSoog4%
zKMXLiuJ3+W_VvSuKRfoEm3lDs&7nOXZ|4}-*IVp5dOv`Bm*s!azkj#?x3X~h_dEZ8
z?7Nlb`(xLe&0b^rX>!ui+3%<Ph+IFtyY#Be|7BsHom2{Cj!cyaWjbu)Sdo8*XK^LJ
z^2X~+Dqo+>xbo4&iD^II)auJm9%(v$uH3lpY3t4DPrpt67!vlX@oJRj(?h;bAI*5h
ze%?PyRI_w%O?|-r>*;a4(@ocA@&0R{`!brFO-)CsYonHARse(G#x170E4v#{m$#h`
zjhdx*wCdphPoWOJjm`dC(XTGPdB0U?L8ZC!to!=*Yo<RlJ-WB&@X>#%8}-;<fBhAn
zc#o;J`0dh~uDEx<pC14DE%M{8$eQB%#S<#c<L|p>J;;ikx4B=wUU~kylvh$GBZD&A
zXFM`o;u$8lF<a9r?QV9LUXyNcmW|GmoF~Nx<}_rivMkYBsHeZw_t4yfe{y;YO*b3Q
z_f%h6lpe@_{#MN5dzV{gEWD97Yx#8jyQ`;6y{gS#QF`XmmM)Fhe~+K}p9{PA)#F}U
z>X~0t^6k!E@7{fP)~emBpMPCjpZ9*B*7nP*kKbG!v!n4^YDDKJy<#@~IF@UYALg~c
zy~Eg+EYi85s%Ao?31jQo&W(2%z3c1zTONM@_jT8mSvPdcp3Zn&xv~1?&Rx&Vni6de
zU&`F4y7GGb15@RPKX%-`eEILEZQB%{Ycy9(fB9ke(Ybwl`Booe=SbpPKZ#8uq@LBd
zUS#7dmmT{y`S)L`zY==8RdJd@!&Hadw_=(5vvx<`nkHHOwA)jHG2*QXhqF?J(y8=4
zv9Eq#TT`^s?Tx?JtQ=GMP|eAA-zh|tuf1`rYUkXk1(NBCC)nOcac=+8)#kP1MuqVk
z-MV1ArrW=amh<itaQND_P5h3o+4k2<i#)d0)Cbky*>mO4+AZ80e^q|l^7?h%Y|}^g
zRvdqRJ66f#ug>-E+ebeYhu3+0=kQ+l@$dT0smr>x*1z3-HRR8_*L8un7uf$<_2k?A
zN!vHCsoAV%7Uq0^U3mHV?Wc@tVp1PIzvKPdcJ@WKe^;wrx2qd(S+nNjbpO-0{k<N~
zNZ%4NNB+RWY~~j~>le@Z?EU`78GrM;w?8lM%63e%evqjl%j6e(TPH2vG|@fS;QWb`
zVvkZUY>mj7v1;D^G;_zNs^`u0OMOLu7BxR#m%*|(!rt5T-G27?^Plc09{4wL?ZHPM
ze<p4W`+Wb5-P(O;<i5;W@lxm6D{jl}Z+<LldTsTzaDt~WyOGzWKBGN<>sz<QnwHou
zi@N>$>lK6dAG?0FTCA3O*6VXQ`nHtcs`yPt{O=3y#&E?NlvK!_wRxbu@VD0s>kU`J
zoZM$T$l0~@w1L9UeaSD4o+;hmtYd$6Q|zfX&9<+WT`x60WXRd#qxF0H*X6Ug1e=xx
z`}VwX+g82c<({~(mgAegu^y3_u;NlheRI$u0imalPX(K-GTQab_0x8@LZh~K!X{Eb
z&bBkHQ?TWB^Qt_^HSP3)<V@4%w!dNBYAb|)veoKNF!jmejg*=kaICQ{%-{Khb6DWB
z<8%MYrB2()ytn7emSy66Q)?Ek^yT_-^z!PmC&~hW`*uY6GG*KmtcjTIzWu$A+~UG7
zcOKTKY*}YI$6<nsk3^&43i*=SId&Q4?N;3FZ7X>sHm=Hu+)~hH5F?OkBr=hi$8Dn4
z^lL{hF12>DH|14PE}T#na&{LV(^U6&_d4xLvLx>K%f8M^(_lT<J1H(zy~_O1m*5;5
zo=YM#K3%--JHxS5z*9w_GxqI-g{1;&yS&p9f(1_2zh2-h`{dG+Mf3TTLW~M8T)BCA
zml?PL?8MPEqy70Kp}01_B4r<YSr^XqhL-E+L2YP>!vdwUMe^?Kj8ix|7l$A8V%{J5
zd6Kv5o_P;XF&wu^lJ(m(z0_4!#CJh+M~AbZL3*HjaEEWe<<|=mO9ePAmaTlL_^foo
zR=z_5iS=DeLZ?4FSSldw<F?FY&UMGXr4zgZU+s%dlR3CTJD|GPsw05miC*BF<FiYc
zjvjhE;fUj#H4&Et7C5cE!yWae=1J@MNX=<&pSbROOkeisZwYTtu)+jSt53U@eY+7b
zcU_~>A)C~&r>EMlnaa2sKHPCCHn3}g!QQJ<&x_x5F82=BuGi}={<Tt6ENMzW+}l(h
zL%zxgKe<&$!}R8`s|KX{ZoRt7|7?%m*M%ZNfq|Y&k{G81N=2@5-F!u!Va?s79dcb|
zb&Zp*EauKSqLL!G@utkm*n4S?wP~9c_yxSSf91!g=og{o+gqsqR+U3}<+Mj;x3u?%
zL`->h^S+LJ-aL8fW%uek_U61|cXDG8)6^@PtHKg-s#fxw;-rV|^M6gBwkLbC*Y1iv
z(QCxZ?a$fk9r~K|ue<X9RP{UMG81;KY5Bcu=Zl#i{_fm4Ys#fJML$!{P0Y_;_vL$$
z!16D<Wz&D1shz*mzT%mk_=L2}y3ZdKM^w1>$Naf9Z%K#Xj2Zjp&6PVnLDsGQ%$+pR
zZ5QlU^Lr*UDK%MHEsr{L$L;Fe&}BJxpBnA2YrM>hJNL`==;n4)qw^P5wH{zwdX2?R
zr^SZ#T%m>kgNd7CvUdw}N;I^T?T}M8Nq?<sy~WUlsoeOdN5;=%(a!y@hi{%-|4wzH
z!-<L>N7bB%iH})kJ09&z$`ES%bH-Opqh46DZQ;M8p^u**n!Un$Y1B6%m54<qNmrjO
zU{P5ZR1r6!r@#MZ0Bb}xW6G;JTg^|Ny!C$W%^mya_1@hx$;C%NVQpXK{0m!MRx<op
zP_>|S&Nrn>nTc<vRNnAbF=kshA<<PuTc1&Z^~vewKPx63KK*dv@wYnd+2x;KJuwi`
zT;^1gQ{Of>iqBX?#im26Xhvs1q3R9egtKRU=SA|Joh+^zGq*_om*qK63GSi^%)-H%
zN;}>!EpBW1a%gMl4zs+NtXDqKjC+<BEPc!wz97b0qIk8io;r^k3x}7+YMYZYSPZk@
z&Io>R@=d+CN({W6nq2((PvZCVve#w1&uzb7FZDs{th}8@$hj}tTpM&_%fyb}f5ZO4
zw=jhJ@A}ETuN)0eey?(v6@JEI-^unZC#St%qTGCS;!MY<r>FP7OZ@&L?f2_9ioZ`y
zt_-YuUUK;V{QqnE?A_B{CvCgpB7asV!=qZ_e#8R9gMkj(O`OXO!`kMXi|JjD&{(c<
zb4ASS9YSLDo(+czHsl^J4AJWsk_fmlk*n@!V!XQKl^O5C18zRoJ-a_&`1#_s(fl{x
zmXxiU{^prR&xRG#p0_u-zxuDncuMqP<&tMU0{i0sUEjXZ@X<Ghgv1#Mmo^@(W4pcM
z7lSkFlW$3XP8erT-<=+?o9A+`%VM)X(?4z(%4o7Yc2&!pWvNBIz@hi=zP0)0x19O2
z?vQ6$d1g<9&g7HK@w`93SULZCmKpah^7@IbGXfb4u3syexUY;Ss-wv)W!kjehj(r|
z7VzM#wc^y3wsD2`j^!0x?q7OGY~A-uZmYK%ZJi(1UoL%j{k9uhj%+bH$+zSM-<#9!
zmz!%p|9R+p@8i-o_0>OIG>>xDH=CVo-#YQtvk1@L%)8=SoC6jxy#928@p<E#{yJIC
zeLf0hCMq&-KeOe>$5eAYo367<MgBHp)}F8TcgV`l&Zs!czF%&heCce*v*-2a?0tRA
z<n|f)zPw#er|;SFWdHwf@lTa&_y7NP|EXU@Wb@Hm)7tWNTeKOCo-T=cKHcb~qsY7Z
z)@98FYdKx!Pc6v3871Mg<I!ZU>uvpg=IJ~0t8d*5+0%3Skiu7iU>D8@mkVbcjX0I<
zvEkD_@6wA|Y!c_Da{Zdod+u(b<*^N$lKu&>J2q+kt6G~HdV8rUW6Y`Ag3Za|TGcxa
z{PFes`RSuy?B(>z|DQr9Osx+n{<T8x?7ui|rI`AfFE>S`gSL8?{WGo*Oj)uz_bcn!
z6?Z?_Oh|m|JC*nG*Xi=&tqt$>ORMIKzh1j1zOJUTQRCu@-JNz%)0W@;68)HUy8qq#
z|Ih!=p7W5^@A$EGJMO$Ry`jH1x20@hTC!OIhm+f?qf2J`C<-m>`d4?gy6xxbuwy@e
z^5}NX{eAuG>bLcu7p1?nwBiX9E-R9_q^JMzby57&Hirw`NB-~lIl0gIPR)0>{ho=j
zd+&%iX;vKB{qNM&vWNO_r}a9S@8$i%9ro&P_F<!}75uA9Kl;APE}gSz{hRH=U*t<!
zcN$D?YHn_-xORI5XF|g};n3TQKL|93v`MC}3f1*IBDm@1Szm?rW%UnE928&KVp4hY
z(pjFrZyvh!{Vg-R?s7M`@X7g_h_!W<GuGD|E>!&bBB)o@;CR2>bAGQUe}DV0j^<Eg
z=6N2p`J1?!Z{B}y5e}_@gjtKs{v7gI{efFRLteSzr+?9wPi9wxW_;*w{rRrC_5bs)
zPU3TKoSAoR_4(7gYcGG8{_e&1&-L-*uTSkRU+mK(J!QRf=ATyUO7q*ZtarO@FlCZq
z6AXQw5zNZO(s1an=7zFU%^Zqf3&fw6-ww>3Vy<AfU{A)v#geBOpMIX@xOKmQvF(zY
zf0JdO-j|!<q3U#fb*k#hTLQ|$DwlE=ZhJ46b^rgmhWU97A<H6n9F}_jhO6faLra%J
zy_e*va?5DONm_zJ?-(bgTTIax3kcqq9B`Pk#dFHb8Pin%dVW0nw`lX+vsXpliQWGB
zsczx_$IqWWzI4OF>(}etMy30@+-sJrR7ra>u?d_yWOci|{(1cQx>f3@&+@S63oiMe
zQ7!3VxOC>ab+eX!<>tGj!Ln(!UiG&;FO^lXw~f}<e|zYm9B6HEDJ0?Sj;E8A4rl#+
z{r2yl%FQQsF6ChmN{;<_b&t>eQ%<k_KleZTs5Qa+{>AKq13y*xd9VI;l4$<e@LBtS
z{$Hil9%To9{`B`+!o#VaIBm`4-A5MH)WrPRELzI?E9CW#Eo(C;8-F<dbn!iw+4)l~
zKQHI2-sG|<CiCUH`Ze<6o9FDio3wmUCwIBlq^f%N5RJ$kpOXY-KCy@MxpVCH6Pu~(
zDyUWwzs_WK&eNpE%v0q&mnTTDsBk-8-z{@r>}IO@gKewYZ#4Yc@ap0IStnoHXy;s+
z9v$oQ`SLZc#<Qh@yX$X<iEsRUT^Gc>6D5A}+jHJh!QFN1Hr5A-*BAbK{VMXj+TzRS
zPuJ^x%5^F=j=FQd-rju4`RTvA@5QZFe4Xd}>V3Mr-k!Bl6AbOkx9OkA`1dPlLScPz
zc>TwxbuSNXWR+x|P+atK)rnJvVe8-Ne7F>Ps)sXZr&mDi+~aMl8CU(5VYhC*79lL;
z9`WJtHCH`VgQH68SN|vyu?{&_UvywCUxbD)Yve4?7k4g(%02(~%=72!pA}{Q@`LTd
z*DhJ{toUhCd3AI{`b=N>>1qs1_ea!ZeD*v&yGJ_g`{~_G?}9(?=du3&J=Jc7{bIf#
z!OO2CzeY{Gv+oAG(#}6S1W%p{<Tc0>Jo#yNj&GE(mHmwUB1uJV53~PXllpqnJ*u+)
z;=A;<WsbAm&Wf~0-j|VfKQP7kM1VlK$u7H=Oj|zFeECBg8s;s(ojXk;X38x68z(nj
zX_HIM@jf}xcj-6Ii^gnzv*MR*6ga>AsnJ>05EJgMSzF#Da<;JYI%fH?^Y|>Va8;~%
zdC*X=vFhTZ-$#>H=dg$ur35Y!Jb7@c{mCTZ{Cdgx@kLy`etr8+Cg|~e4c@a$H*mwt
z&6Qo#WgEAyG}3GAoYcNj=;8WU1-JLIKm4QHA5UU`EFhQ?I{9bCrw0?3Z8^(0_uj2c
z^;^G>KHt1Vir0mQc|mfL_?eKuKhp07J=rx=<km!~>0$=cJP*zE?fATk_0HMP?f)b0
z<n4dC#^aucPW{_^eP_&-Sa}YI=g<AW;YZLSfB&?cj!Si6_bR?_J*uiZBmSCl>^iyG
z6PNryF;Ba8`Lq%pmXCf3?~Z2p1YCLgivL-Roa`)xg+;BLTRX+JuRq9G^SPqd{>QZ2
z4WTPntQQGt(wX8g-#o)Zf#Ik0g`n@NCx*<;dv<ZjBgV%{y*aaH)_+kEUFOg=UAd&J
z@$clxulD{~X!=}0D`?rzxRtlX|6NeJ@N1v8fZmK(pQiR76Pf1YFvlvo_Uw#P)Ax6*
z?7Xqx=FsEB4|4l1Mrt!m<CeTNal>V4JG&h+3st5zZr*)WNvzM@A;u$P`-Zu47e8`y
zdwLnnO-tPU_r~{|Jh`jo;_ehZu2<k=I+RoNX4!#{9~Ye{O0>Rv@^Pb2W>4_ssYQHd
z_dW_8IuV~<DYyQ)eSZ4vc(L!@xwEw|cx>`p{;6X1oJ$iJR`-41D)QKndC4~ZFRoEf
zAFTbgd}hF}{x>H}3sV>PZCbzV_SK&`4lasJ$FA+-nBS`==-+f>8JA}&Z`ZfTm*-q}
z)~7$^Nw%IRd*AjjPkUg(rw!FXPiuT%DtoBz|66{-ap(Jusw>^TR|N>MG)VP-&+-VB
zdw8|lR${UN?;5SUUsM-%C+v}1&n}=Blp5l*X~VtO3rz$xn36jdUDl{L-70(kHRGgK
zqX}k_mdw$f%gYM>eJe4)|BU^!{W|fVjQ#VkuWqlu={#-w=9#wkFO@e~wORds_}8cI
zcF@Q56FVONb~HJ<OX8dMmXfWfgZc8OXDs@EVqT26xT4{q-JUKYGnIQgeHugi_iiqz
zY~V3uFtdI%)l`FFVa@YJG4l>I+8!^j)L7Ihut9n9HKv*DOHOT{uh$(S{Klhmnt_M0
zfLr+r1t$rY6ZL9wSA{vQKMP1a6MdvgnN$6S+oGnU6GMM2Qs<m|<eg})a^XRbSyn6b
z+FQT!iFPpYC2ZY&<YYmOE1#$v3+sV33wQHAQ#rra*;Im8`H-N>@tYULCHt@Oh5q04
zZAQbC@a|JrEJIWT856cRznRnTe00ixai-#JE4VqPnrh^&nf|V#zCDXUxhIKnX=qox
za_G8C<)@Zc9}z!ywR{mr!IT1aKb=ao>xbseob4i_ZV`}OW8TT`sO6d@{3+wQ?BRgd
zRd-Z$EL~k9WEa_1CGMTw;;i72w6)%fwQs%VZ9UeW2NCsYNvB`=+))XZ3Ys4Ff5(iM
zn_PG8cy-uIy!Q&<l#dgp=+)2Dm3trdy7%K!8I`{*)&88p`zCcS*<T{cbI+;0#Cm1e
z>I+pnZ-yo*9T70Nwr|7p$6ARCb3ScY9p`aZN~9rRUvc2NAWqKsQ)!mm0)0zoHa|9>
zbM9!*{>@WW0)+ykvzN?#7XGs1>1EqT+|OTi2x_=AIY}R_O|6;tGe$&N{7}#P`eH#n
zmBV48_P3{AFu1a7QOte?&j6_#Ik)BZZ@BmTkjo+!Us;~cz=9fIyXhT=If~dh3#$?f
z=Davpvu^j9msJsVTKw6C+wZI^iSNC`d0S$C_TC?RVrpj2{1&M1H=}un{QZwkof8aF
z%^W{yUp{>6<E!~!!p)3VZ(6j%UGkCQv8vE|x8glsrY7;xl5x(<CuVgxNpT#Q*}HBX
z!=k&&iJCr_uPrlt{;-wrxq4McA4{OqBp0sL!guy6R0(eN_;+A#pX#-$t#S*KIRmE{
zXgbZF{NZ}r1^KDn{SB8J&s6c>Z!S@pki_PqvBXy`Z}&^{o?rI4I}R_-QJJ!kC6Q_E
zxk|bIP4%aH{#Lx{UE0>_;c_BJ>20vb4q^5~HfP+LB-Im7q!uwv|9K_Q@vh;QrTeG8
z<JdLJ_O8C%^W0ohsl;gckDph_a&DbsEa$!Y%o}a)FAm#{{3EYlySsGTp2;3vDsMRa
ze|&gedFm6#tHbVQ8<%n^bg_qfmFsE#;hY<!R`a9St7DT!J)eY0>`AuP7pYT{{xvhm
z#~8d~UJ#r#Mc{;z*z3ni?<9^k1us<nsA;&%FZ`FlvZ$`J@%tuK?Y6G&kDPh8yM*~f
z?0eVQ{;$|d`{l20JzBGQ>g=a~^y78q?*F;^rA43do{A4=OHA3)8jI4P6OR?1O*~-s
zlI88PCkD@2_?O2mPY|tF5j4D3Atj$`H|N#Ha1)!#E`wL6UeEM7^t0gARJLUb{fi8)
zuCd+7U8m<cZ5ykDOs%4P#@uTLJWCgx&^rI+l#`d$g<JOLO`JrQ|6ay)wlRz&NJj8t
zBKJ>+*Ol+i9<%7kNGbMM$kf3UKIMH(z=EwuMB=s7?+fuP4qAV3Uwj2~Y<<@Mp9b%i
z@H9v~kZ3cT7o*R%V~WAM9Xbz=Fxa`-s4{<$Xkp;6WhrMnF|mk;d5+GZ=m*R0o-s;(
zF7R)8zmqaYhxfCc{Z9U?Lsgu3T+P0g-Fo{bF-KRvb)VN1#wJglma{+G9tS>Kd_k_-
zb+e<^{=f)Pf%#5{erkM~dZ;HzgmHEKidg;EU*0Ufu({su&U>-HnR{>OT$?W*zqe(X
zx7dT}o32k=-~4;|{nSuqG5*g_|GvrFxwH0O&NSw2eqYZRO|xxCk>E<(Y+R9g(B{Xs
zUgfsFWEt~o6WeduFST0M5jgjF#OnFo(M_D*mg=$#ndD!}dsvugyNYniMVjtiabeqm
z=TGac#iz=?y_9pX{c7yRwwXN4DWwiO#c%1UI=xH25xl6^HSttOsO-I{y`kH_b_+aV
zP+urH)0pMFqFXKBVZ$?w_e~^?Bvn6swVVF&)eTFv!&CQsIb$<T+_~9;cg>y|Wv5Fe
zCLce3ZoRQAPoe7E37hA=Y&DX%{(eI{q2J+*>PD>&!}`{i&l^8$hc4+932e$STgc8Q
zJja{udzoppe%-$KJ9+#<J3pVv6@PxteXDw_;h`zcA~&v-?%b3iUehh3mD0lL>T#TH
z_o1CC(^jTzcwDl9|9PL>xuq6U4_wKA$yCw0jkRI=dx-{T9%h+&#s_LG%B!|0PMQ?J
zCZ*N9NOyn9u?*w-)atob2Y+R(Q@)X*u`<ZWY}f9vTvPXAc83N1&%`PoANTHl7S>*^
zr|iu0TjBFfYf+np6DD6cW}dfwdAs3}nh8Ix{KGGq9JsSe<>g0>&yH#Rdmp(~2^r`t
zox@Tcykpihjn9tHJQLd{H5A<vby?MXbK|i;E1S=FUc2}!?^~qqYQ6g1Ij>r6Y}bar
zV&BZa|CiL6=}n1TLMK;-w%wlE>M|jiH%)N?$JW5hqKU^e7tG+6PqZtPXxrGZVa6?&
z8QNQP8TQnu-*$Q^!}?(j<2&UulNbAbG!&cia{FGREk8Lve_7OP(8{$S_Vm1j*Ifqu
z4F@K>&v<*`d2;{ds%I_-J(ieFeO!NAcWt7=>*W6HGgCBJn;h5JTiczgd2=y&y~e|Y
zDJ?eJZpO}ZI`s3#<Eh7T1=uCL*<Guu=Kg%D;qE-k;PIn%0%>#p{=5Bu?%w!2`x{HP
zRYu)8H&6cRtn+#Hn#qNRyF=SvRF&pm+cZD>QS!veNm9Xj&Jz6ZY@Sb#GN}EO6aGr1
zr~Zfp*COSf*n=L2IQ*RM2tNIK@We*jvO^*#gYuGPLNcNqDt`K<HK$EKC-IQuzysa`
zd`>smm18fq339fon3)}#y6`NEa;#O@&x9q~Hx?uWwS^t!&bk^_Y_N%~(U5)3-RIw4
zuG4+F&VYgAhtJu~<{g19jt$JSStgeS7f8&r+WNeH`Hh25&n)6^@a2w@OO{fe{bcuJ
z$Ld`VUKa-F&A0eqVdpGUtl}!;%6ILv#-Z<5)Jk79MaYZH6WgT3!7%aH=4Pv!^n9r+
z;XHl!%*D9RMQnL+!m8-#+drAV_k=Ob$hFr^`)^j?axkav`?Jfh%FOEb7*7@Zr*L6i
zY1~4gW@huB8TDU|e3CajmbA85J#CqBhnsVIMu~`_PwC4|>oQCZc_d9L3$grs@df9q
z*^2@qFNkrhzp$<0!zbD14}UI6D%*6s8M7<CS$uI_MzfJxrogQqt>JBEvb}sUhjQ(t
z-g23p)9jwn80x>F)O9f^LA!}pIJ+-ae4=+;q{gHwsCkAKd%a4R^XhHxYnMGRuqkD`
zm?3`Qm&Uy%3s`Nqrf{cSRezUpAuh}0*85d+Z2JTsb#^rR<yf;ne7*Rh+@{zGshqsd
z0yk84^S=4On<VvyZSqI|hl~3r^xk_tYmY@mH^Z!{kB?Q(_;I)>xI-#g-gKYCg~=`r
z9Y=quuh%)nxH9<Uy*p{Q>SMDsR~)dMyzz}k+sDAi7rR#k%~9yd6^L7yTNuBjM1A$*
zEz4*2OI2@q+r*bB!!#i(RXA4sR?U@)q75GnnpAjmnpn4HtyXRSDbRTLAd|dlUcZ!b
z)Vh}J<8pRcHftSw6k2Q2&wPEpa+-;i_&lrT6&p`0YRKo#7TI#@5UWh2Yj1r&d+v74
zN5wO2`{ov$d8`pOp?JeQ_k$l-`o6lrx)Ef?i{7c6eT}})jve5;&k^{z$@US)O~pmW
zHh+CocU|~h{`>vE^@<Xh<=y5@;hs_CD(G};Vtr~N|8avNyN!33J~J@lm0%VsS+j;;
zxnJ&vchHICoaCf%aI!w<+B$J={VL%f2Ah;`Su82!Sa3FRpSJUCPUYPyKYkc&(!ODE
z$cg33vm145Z)Nc*Uss&4y1q(e`=`VwO!+emiaO2rz485V>eal%4V4cTnbjPrj=DXU
zQ_O8+FZ=Dht$c_7r3-v~cRaDy-a#hf&~ZieRL(ia<}7QB`^*_}==f#ZKKEN^RBR5^
zFAOcaFJkfSV+sFFy%Rw?Nt!KkO}n3;PIlW_$X>mmxTT`Xu1{Ivwt)ZnrsqeN$HXt%
z+gl+O^e`cY=XK<Km9V&`^v3u$!?mYZY8~gjrnus~YIpCZwO1Fiawx2N;^eWW>DN#1
z=pWbFzgqv;9(nVKl8eNxoaoSu=n%7|)@`EPCiT|O7wpU5X0CZ_f>EwZn%_R7Ei)JF
z6TVRsa9=P-$2Vcdse6mw{%2-j`2YXS+}PsV7SkO5b7_Cx^gF?PA7fc=kZaR{yj7PZ
zr_Y@?cf01%Wx~0)CwT{1-TnM}`hq7jT)LH-Jp@wkmxQXT>2j%@W7Xq#`0(sg(#Kz-
zPBRxq+zvl^y?(y<>6d3y1Mm6myR-I=+|15JKejzzwDINH`r5xgA0`F<FWvg<^SU*4
z&%d9%`#N0S=<h4_?Uftv?cVVDW>)0ey>a{Mrs^)6y!-3B*FW#peEgvp@;~x@#J}H#
z+G73DTdW<jZ{JP%w(?e(-?Y^^mUH*ig`PY2&$|3~-L0Ch{`a|d{JUR2vHSDa=T~Q+
z?lxY-p?)PdcJ24I+)@AQOJo1vdiH-df9(JK(mU(d#}_Kb+OGLmqY=M<*U#il|L$Ea
zwTr(qr91WS;Xe;I#+-in=5p*yyA6N7zYbsY@A%;-yMAdD{a<(f+t<}6r(dlJ{`aX&
z^w-V)&sVeR<bS<9`sMQT1MAo6M?U^Dy?%cFKG}>q^SzlhN3PZ1X}kUM?%eVY-eV`j
z&(zh;&{50JkK3Q*lIYy}`&QQLRezG7ZaZ~$e*VT!*A;FanQ|hq@x;BPSFzg5+6;DD
z^Xz-h!FN?lrh;KhPeFs3+!l@PR^HrZLK*tk#b5tlCvP4*eRJqe3!_u%2R`1goF??N
zbwQR&z07vuSwD{45=atVnaw+)w^`?!i$Lnx@5<I)R~~FMc>QBy>w`&G<E1**?DGlx
z+_I{;$)m`&GNLvzBy?&_+WU|9FNvMA7n6FsB;~*Qq(6oA-#e}xUKBH}`~K<h{7Jtz
z#q9h%vnFcF)7q=6b(x()l&&9eY&rHSq-U$m^vXXRwq^h88-H3_yWefuwg1}wU%EAS
zIz?A*aQ*dGt0Ff0@u^8Yx35U;`F*Nxh5dfZ7kwK7IIPyqO}yx7aX2K!<TRVWyQ{g)
zAHL0-RdCoP>yx3dpMc-_2^<+M!8}`9FRF<g<(jD|dER*O+EYpYmPK&noVq)UIfv1A
zwxU$Bwaf9`x7&7!ewVA~U^#lXqqp05PLfipLq)@*zF7i`ZNkG6Wu=1ePE*xb=C)vm
zvk*r>bKt_KEWs%*4%Z3=G8_w;k2}g3w{U05y<gIQ&QK%IVViiQ;!@Agii?e^_dT?3
zc$8PM!Xf2r%|?y&;g^4&S^4himwi@;Pp+9$AE|cr+oy&1jQ9I?m3RKLsh@fFl=Xxo
zkI#1+8m~xjcDpct=9L3~=CFG_I@@WRcT2I8cj^?caLK)sb`)-P5NdCAReva=?YFt+
zgh5C6cHTp`s~v=xb7i*_dT&w+ch0*$Pcm1bbMfIeqf@4vzWh#nJ^4trf`Ce)vYCU>
z{4>7seoIdlm7i@+T6H^^<>(a-l|qjCEeh8cIe8bKvJi4sb2`!>reatib7Z5#wi7#f
zPR!#!WayFPuq}Ln;-V0}c8)E!_dm!w2$}CQ;ko(ta{Jd*qxq_rZ+^Wy{i&N!{e|@D
z@9p>a=B<A6yUuR*x@X;=)P8PzdA&S6{8#;?lW&&&-FS9=*w*<mBD;R|?h1M>uW$VQ
zu})_FQJ>O9-9P_a)49{R&N=C4z`eV%Yt_21KKu3dr_i3Am#&KQXtmC~n`6h5RFhNa
z+jAzQk2z>n@cUAxqZS*NEGX5@eLZ3J$yK2yzLEVCV!J<{bu4dT?f5uv!wVIe?mJym
z_+IYZvnyBN-o4Fgb2uVb7z)RA7y2-5(iV8UaEqa9W?%h>35he(-J*4mHM=D`^spal
z+b}gs)LG@;wY=5QJ?|SG1RnR^C`g*a5uuRshttrHH%L>?NII)rYsdT<QjHSfj2D^$
zdbo6-hb1N-V|r_G)Pk+gJzv>KH^{eZVgXN*%!IeDVw_GJBtC7qIxqhRpTaR|IW2*k
zTbc9E&Wh7j+x)ctZ^=)M?e(?OV-6pADSR;chEHS{bJUaI{2D`hamCfy5fhcJOMUs7
zc*tnU#KP%Iuk+{g?NZh`!2i9)RsG?XQx-*w*lhn@T6g=|V#c+$584~I>D`*)vo+B9
z&j;hhynhO6ZmfA~bUL>F|J)5<*RA{bn59WR@5r%R{Ff(etvJZ<?%Pp6bBc<8s>7}e
zzZ=Rue0mS?Z@*qE!=_uYb;X6GlOC<}c0WsQ5L5F5rT5P~#mYY?s2@x!sChHRieFLe
zPr;pK(~FO0r0tWx8MuV4zxC#tbswKBWSUrO9j*J!JpS72>@)e3uU`LovnbR~B`>a+
z;oO4e#g}wgJv;s47w8;*B-6U4e(^@TjmLfOTFuBldHGdc_4lcF*DU^6vekhdoYs93
z+P1z}*`8Z^dyc^UTg+Ax8Lt=cz7$!lWnWRbrIF9h>9U)}#-9S*8@ArQwXyG5zc7c@
z4nb*YS)SJ!J^#PyotIdYb-_x?I{4t<BUe6_8r++xT<*jzrTZ>+j$)U=(`iOB^~_rq
zG&eq0=wo$PJNxvR->(-k&3tSr)wAPtdDORV;c@!5`yJC#CrfePl=_~tYh5#CzC~ZM
ziH!4K-d|b#_4D(bZGXOfe)fA`f9`#0|J8DjE+=i$%Aa)dYw6CIYuy<&b7KB~+qp~O
z>m#Z6`pl0QYl^NEC`=Qbc5Ux7?$7d#c9B~3iq|}kZL3?8Ic>uwjmhhZs%t+zTVQ>N
zsiw$3qC{9v^5jK@LwmsqI9T|c=(L(YuZ^1h`@E!+or3eK7yYxm#Aat{mE%#6+$p(f
z)46MTt8c5nZ*=%@viHV~A5SKz|2(Mf?%N^gJ2kC&-uBy4!FRosHI}(OnEY_#-;2)O
zuFEv))106A9BB}%F)ENbVyL}jL0jL_kF4`~PJB~Zy)tlRSq7`3x9R2AQru40&gXsC
zu9KG7?!8y@_CxOnKfYi6Y=6R@nTh#G(UU1HT66fT_IHYlFTa<?-j&)iagp=2wGK0s
zPF<SkcztWcJ(&{sh=z!FoZWthR41?8ctv(?f5G}Z>6ZHK-y0&>ZdqkGuQ}1hGkdzb
zbjQ;d8$O<qIQL`1x$5lgS9Pa-cTnhf>Xwspr09eYgTl1MYDYQ*6(wXpE$Y9!@<Clv
z7E6+)N+W}!`1QOShn^WpCz&TXa3@Jx@I9Kbd5Uo3@k53ksSextBUkchURfZ1GsksX
zv<^SWH_Dw)+cwl&<#5XU-xsm($Dij#Rkl05@A>}I5$~K?JMYZ8)4wm(*8kmh;B@)>
z5{DNJRee&HBCA7!gbTh`y?R@<pz7Hqp5XIFXLGZ%lkcpaXU=|CP;9}x>O)EF5BE%{
zv`d+DXZ3Bihi`2c%xm1ncgsa>!nLnYs}r|0R`u~ooAq(orD)y0mRG+z+V*{;L%{jo
z8xgD1#f)r4t`uKsiSWo=Rg!li`POc5%H_CHe4<*x@#fL5@|!Mb_{zp4H;7#^21VFr
zC0|zl9pTm-#}~|FylI!hdCEAg>p{)JXj_OES3Yms0P$hfzo&2Y>-|r)74UHgUpP~^
z*o9-Uk)S}m>hgPS>{)3&g?jU5waR)MDxYFawT*oiWehH89E`uWxR$?Ck~~{^f6|TH
zueLc>RP$$~Em+%ZHf8mhC%0rDwq9hLWoeb86u#sH`&zCoTW`O*cBWcEz@%8&%xsqB
z5?!{lGNy{hn555qRy=+2jUVfG;b7$pXLJ~%E)*9oJgZrsnsi~yABhW1X>5ImE&K|j
zURcS?FdttqlksL;il8RT^K~4KC(5#)IiC1+Ggsy89f7cR-NwiFub<gJ>q*kTtuZR$
zlkaQu9Wd&*^H||k$@+`Cq@Zz{<<=__<vX{1%R9@-W~tcm#Fjr|jqYb~zS1f7k0=rH
zD@t<hd&zKU@2+}K-4HCC$FaHO$+VdZj73!b9WY|Qbj<Rks;aZx>Z-IYSEohJovOH_
zF`e&-uT*l!w0B`YtP5X%K2x0_U{kDYrFYI_s``=#GmqEb_gK}TwW&dEkAEqsa<wt$
zZxH(v6Ou7gw@W77`JS~XbJT}32MPs_dqkYm-sg7c<HpHZ()CU{70-5@keU>f(>(iL
z-gz62Nc-KrPi9J=dbL5d;Lq!yNrBg<RnA=V%wYD<FHJ4thn}3_tGx4SW~HFcxx$Y0
zXET({a}%=PU;8pUa#_Wy>UD398L#ID)exYFQ-9c$v07WIUEtoj+iDNj+NwXD>3ofY
z^}LLi^z_haYc@MBma0GBz3rCq@&_{?KXT()+Qwp%w|Cyp<B}}r?{x4zpSkgnwNT%=
z4-DsJOjV~Xd7z}6nc?#`KvvPvl<B+<^8!|T@$j+*U6Fb1XVUu^bt}>ySac-oc(!%d
zoVa||(1Pte&xvn}r+27v@K!A=(B5Qi^I0I}?;Epe>2*QiXxv{hz5d1d$v4;R@&5AX
z-{TUqmFZ^w(@y{VTCz!N^{F}Q?9^N9UW?`Io4-crlm5gz=hH7=tqE)t5^#O6a50;p
zsTa2r)2Hdd`HwbQsJDN)a5dla(rzw=iWkl63$#LKDT*w}I^fKld*4_3I@{(ytlO49
zSlHhfo!ITfKYKQ(+^-*UQmq=c^&c(-1}7y=Oqs@hKitlqe=XAj)_U>qHjbIbo$S85
znE82S`paxuIZp9NndtM`d?`yh=<Mz~t!qxUZ_`9p{=4FAC*{AF_iz2ZL1X&UN8dk%
zW($3(nfBqo-#^EHGq*4;)B2NSx8IaI{70#cj{K?JyH{UL*?7yLbkF6g;M&|u^SSkZ
z%~qUAD=YpYl)t{?$+kIb_J!E9nS?F)y?9$_5SOAQ-;CtXPNntJ1kF6Y3nv;FS6tZg
z8QjKs@ekC-nWOEtde?`x*p{Q$gzl(83!LSPe;ev7lAUhmwDQb^|G#XerD>Wkns)2_
zRI}y3gV}c3>*aAt`utI?*HmvxssAcb=YRa_+TY&~D7^URRuVHu&?kuT^10*nzw>k!
z9gUBxS?-Zl^!1C-%zr5b_J*cRcjb+e8B8)mGFFF#aZcUY|K;xSqYYvu_60J>44?WW
zzw3}?KEB{Ls4Nl;&zx$wV|}i|>9w0Q1^suHs&Bfxt6pGJU757!ck?M%6TKh&IA8y6
z+5EIT#mO%5yE>nqY+nuPVK_ZulX$k`VPoo5h2Te9U-UeU$a-jVa<i#Y#GK0vhxb@+
zJnm~Pb|FmTaMO-cZ%xniu4Zks)X}lxT;5|eTg>@cOjYS^g*|)Od5kYybDX`QNx>@P
z$h(;r+1h?`>_}|pj@_`ICz<PrYutUQ_wRPp@8bO|C&6;~PDfvFBHzr}+-;UwUOJ-w
zQ?K5;6EjcRH}#i_{;$d4<^!)Wr=%F8vG{kpb2AR{L_gGG+mbK?)O6_7Qc2!9{X)q1
zi>o&Dx~PgB|FFmnRO1{}J#GAYR%hD20FA7aziGSVQg-uxRsiLINr`zPPd2FjxMf=3
zf3v<iSvzE3-S?1BGt|F4dwct5yW`Whv)tsiZn6Jc_4D1_Jl)87KItcapKiFu`KRv6
zvCh;f#woV@{at1~+mp9x-rD{{MlV=rHvhFV?Mk@yUB~{%-{$iQ;+OV5&O8xx>B0Ad
z@Bb_0vPjRKct)1@&%cB_2i~(kYBSg>8Xoufm{vr6%BgubUcK3I;Jv$spR8>5etl=P
z$4&S9X9=BNBQPc9Ua+La=`{<M{S7(B{qeU2L)3>wfeT%o5`4F0I&;_UZ#DYy_v7gx
z?P8sb19zo<)PDGV+)+$%_OhAgOS=u7w_al|D){hwZ{ma1e}2DW@?F2W$9m?Oyh;7t
zKc6i7cXH<cm!@m#V~qYSco@Oi{91ap_%lzYy~e^i>+kQ|v2}a4RP8IX?|avmznu%N
zS|7}<KD20Z!8GsajJJ6bv$s#|c<Hn}7tuIc_1kdqF&_)ZmN|-27dV>B`IhnqhCNmM
z`ZW7e+uGD^71y?0y*IPDyst}o<0c2SV)yJv(}I*$>vPX%-n>;<TK(MCfaBgyL0R$i
zmK~uH%!iow6yN!M!e;U<sr*9>UYg72*l6~ioAD}4-qFyM>Aj3m3GY5Vtt~gCRA$_|
z)BI*ne>bCUMLMYb%Ulw6r#?~b_zDHH+9orx;~R=N6>hidobo8GPLDCZc;?{L6H8M{
zA8(YtnfKm?qaIw`&xIEE=GRr?&rbe6{p6?9hY#0mWa9j*qhcbia`x}`T4l$~V=dC>
zE4N)@IbAHHuyE@k*T)+_=5d}aje2n<ep}`47DyYb<sE1DyDiHSc225~<SmY_g7lH)
zCR}icPOqLNwNLI&MdFsm7CWZAkfNl_z4adP%;pPKlcKf@KUgapzP;rhbF;!i>5Chu
zU5H_6VfT2k!+dJb<Cp?}r(!Qpt-xxd%kLoN0H^FEr-nV<T)NsNikqa)zdvKm$E53#
z@xbDY`-`rm=>1C*cx0Me*iwE^5Z$@Z=V=_DpU!EMbzlAYH{BJgf)oNy=2NcLgGzz#
zaXml(8U4JfBK~>(Q>C4={=`k3Fyq)EX+}flYnFLkbGvl3d%o8^^xmM-cx~r~DHnGg
z+xE58_x8%1E|&8Bx3cG|PUSl6EzPf#)*&=a>&b_lGA^&>Qm4PYaR9Y2w|~FVV*V=U
z$~DU)#SNu=OV{rDove2HuHdY|`B5{)>TlSj-^zL|!y;|*+~LLsfmvSle?89?_ifq!
z<2|S~ZTY^@q2OHaje>KM-3y-fAFw=rRZc_MHS4cyPK1~7!@X>!(5_%{8smkg)!kgW
zYMO^6cqL!=G3r`?)5EOUt6qlQli9JjkmsD_gtxg>!Cgh(-DaCqmYQw7$5bC1J@IYX
zx#kI9H{bR+R~fwR@bm*8`Gx1tFJ<v`wpe~9dK24H*JG~RHea;=`c3w9vCx7AodT|Z
zKTTO18^75)RKi%<_Uq1AM`<P9<gZtmSY^e;=C&v==GC0?-(amkTiU(u^U&^ZWc;(}
zxxJB&hgY2s%sj00ymjHd33AD1^~$T_Jigq#6sDcQDm;JjqhrQ%1Ou1v;=QABIdWR1
zH+Sya9lKP&$VtdFx8%)#_iV|rfMZrEb7rb}wRqSWTrPat@yKNQq1~5LD&2PYm8S{T
z7O7}la7$QL$*?8ii_-l&Pc>Zs{XVu$YOlnFrkQMg%@)s;#63PLetwcvz>_95;jMH1
zG=+(rzI!I-Oe~uFFv(%xh1_NAim%hRPdxLb@6RuWv|pEeo8oHqCdYqToe~%~o&W2q
z3w$4+RMq`2w3=P>`I`JXp;hNNxi%^;Sn{~dZ=q)N{k>+vxoeBeFIR@Ym{WDQ@2Mqk
zz@<rwop<J*yk@t4rBkchrt+0<w@wP1mAreNGoM$z6{zM&57EeZ#p&yND0uCru0LWA
z<L;Vq<iB6ZnSbSG#Z(hpXT|mVm{euGq%Sq4nH0NcEK}-8)l&UhZWDfM&ZJ`+Pv(Xu
zEOE{hj#+u=*ZTR3Rz3@R@J(1g`uEyA<-P1I2R--=F0C!dWPcEN+KuzE^`yn;BG`iE
zOl#ARN%j7%KdPZQJ?vFg+V=}bO6_?MaQIZ7D7H+HSnL#?`gn%me^K3N-ejJPgHxD~
zC(0O4i=M|D*<#CfJmD1M&3iSBk7Z_BJ<V6}?pk!gi{D|}?J1yEe~$9ZPrf(#;+c;B
zzbuoNer9^Gd2Y3l_Vf5@yt#J!eb;sIe|35H^HuRqk9GF`^=X?w<<0a-UvS=EV;;-0
zY0Kxmo7f_F=Hj_Ozpq4B>zt2lusbgh&dYyo?Rn1Y)42C9y|wQAOxKmIOV{u(nqlC_
zee~pnmkf~LiqxO|IV63-`NnPT3q1q1rr0n3t$Km&myKPnk#X3<D!0sLGZ)d-6)P_{
z-2EH2-R?ef(}MH<^)d5TH$<@2%9++apRs|dct7KX9lNB}-tJ`FEqwj234`*5=SB=s
z7y1Ghx<=g!$tpkmW3J^T#|5mld@}5XJv%qXh~Ar(*OR4dnZN#}So7oL`wUMMk2Kl!
zcjveTPWzp2BcQXnBmM2oP0<tIUVGj=;j8g&k6$&ATvi`>FLY;p>CA+wH9MlRKODK^
z6CQeQo&Mz8uhz%k`}gU4Lj<F-ocY$U8yd^@Z1vlr$70+oa6?FI`Mn9rrF~3iD%MS$
z=<m#SH*HRKTJV(Jr<v0JbbMQ}YggXi$oDH3udUnx8J|1Q!<-wtMRbeCC&zwqd*fRN
zdX6vWU9Muj^TRKRrj)7me>N?5d~>dD_3fVbjE)A6m**aH^*GO%W|Q^n+zizi<@j?2
zJ{?c;UPDvInTityd=f`jTo78q`N%{;*`8xd!m-B38;)HM`QLe^Sl#`x#2E{P+qb_w
z3)^(%m+*EI?UX)th1<EuK$&XW#4|sZ<+STNmfoCf|L(n9_V>iV`dy*>Plq#?{QP}-
ziKzY6ojT7xM{h28zKO5@sdHs{`PNxFVbaI?%BSV8TCe+l+2;FfQ*WNh>x+Bz_Uoi6
z`_I4Mw`1o&#$Iur&?9g2LKku?Gw-?`{63PoU2DpecQ0Fx``&dHoiyv!`3>gHtC|A%
zvLDE}JQq|cSv5ty-}PLnpU>=qdPotyMn1T5)n-Hf&;y&^1ZFzV2+-c>ywp)TtRwx%
zwfi4qGNbuJSuV0Ui#Bfg7?XZY>|YksiJhm8`4*(_ziJz@k)0)R#iJ?hS}Rmc6s9Gv
zP|Et%w^PjJ+S~NSj0>*HAL&|t;9U;GuH47~=7K*9xulH3_`1xswk$GG+IptGcFD|%
z=`EYiY)PMVtf6d<v*>}3Kkjl)ojJ*L<Jn{4%O@?o<MRB+4ReM_j*SK(g5r^r!)2bv
z?K@DR#;eFY;jP>%iIz3<Z@NjXx-sSL2h|_9mL9py{rJz14K8PltBOSpbPRR=IDR|c
z`%fft>5lt}ifcr|tQY^k@>^}qxsQse_Lp4hPlSBcX=mT`|AJFbnn%=|^;sKBEx1$c
zPyCa=y1w;Xaq^vuaT2)-zaL*#+cF_2qP*qVg-Hf8(krX?+D`bZ@O%H_oGD&ho9e&3
zv<}xfJ@r4|vsZFmGnBITe|*$>;@<>54wcM1)qe8~n^%}CHoclttjzW$O5f?*;bP9k
z&m$6MEp4s;ze_REG|Xge@U0axhl@EIa^yF$WTvh7kkVl=t?1caMedaU-=`n=c>g~~
zQ#4b3RqWIId8NNkd;eX({k`$uhi~nDHpTdcr8-D7n)FM#h7_wA$bU*u^(Zb+I;PYd
zW509j@9b^<5n?{K=anlqn%M6!nI5AURk8AH7_{M5&wNQH<Mo2Gws*XrEco_i=0&zk
zHg>r|+JS$cKRsZxob{YkY-#%btGZ{uHz;^~zC7)jgxF`v{q<Yg(u|7bGffzlWHC-y
zyPeNB=KE!a9*0@&JzpAfdbrk}FrK#hj;YA~naZ*(Q4^jWD4fygGc)y#{^=u8v!rzA
zok`N^d#JmyAt%57a<@$6sng;XK^>ZR>nt+0H{HEdBe1D1P1<u-{;9GbLKgpiFa3FS
z_vY@?t5delI5X}0<)@d<&Mtr7y|<>OGsi)p(X3y}eTLL5<0ox1<7PbLyz*>XLEAjN
zGkvY8Mxg;Q`a0KEpWT^qXPtO6Q?4ASb`Czqow?I|b1R2S;X7O3`iIatj1O*ednbL^
zeX^)D^4zM`!S`Z(?56%)H0RH@Wyd&6%i5gY891(JFY4HEa^t_##<f$tJYB1IK1x6O
zaB%?tBHos&jMeVNOMRbRJD7i=&G?*apuY0;%Zp37zuGET*e%>=xzcdT#=aTd#)ky>
zqb^7vDZRon>+&LxwI9Rk)0WH$4!qJ|z~?LbT{uHo;JOY&lt5kJ!l_-Esei-ObN0UO
zX4LJ-2DNM#ZM>u-y-6pJr_V8lc~k9)O-*7vhVRWTRK5;bT6(Zd`L^ylP%*Q8;*qa?
ze|||ks<JHoyjOj5(Z5tqvEPy`qIz1FcK-IbyjDPMm04Z7eZebVyLI&+_%_@SN?bXA
zp6<(SH<%=QL_XRaD(lv@^!vPLU)1?iNsq0ZEjC>^$T4}&Pw(prhh`YKR`VvVFOu%q
z{{2eF<gTAhy^<fdt`TEg`yunxtH1|tp>9);-IGlU6_q@9J^bs=N87etRTO$`lKWMo
zVDcQn<|XDlzFZ-OGq$q3Z!3$vyQ{v6pQq11Jl`*iCv}bUv6oprE+>+9EMK`gneFWL
zzgog`mYV49*;zN|vd!TuU)-a%8*z4R5&V%Q=kPYOYTxChVMc2@zo-}Hh}IgeU`^yz
zU~kgd_^{~032&YW2P2p_?>)nn!@K=f@HHmURR)!JYu==lUGEM~-52I$J>#qKoeX`y
z`n9LaehBSg$k4IbH%nbQJ@)U0Kc7Ar#IKq3Gq=cg|9^|QzSp`Z-&YHNFEsb+yj{Pp
zAFi=1wbn{uVtQqg`jzuLqtfNpkD+p@Y#V!arJBuK8-4KA9JU)-u>p(PZe*>@^t)6w
zcfIoMSBxcg*$WSQcD1aLm&-n9@8cSE@&6U^oRxR$*&e^OHGMp-VC}W7W?8YapHBSj
zfAMPiIq4wRX^BsIg9RrRE!<o-``4L@=-;mdIzKA!`?d7*v6cffuBMn)&swv=qvEpN
z_HS`_4^{CCG;NSmS`#gJx$od&vAzilE-sfBe5tQ8YyZ2SOXoFd9E{muKARy`RCmM6
ze5PHYegVP;Oq1p6jokh#GYM_r-T8Y<)ROGhYv~Hf3EWz;9(*c0mHZ;!`Y*MWI;nqd
zDf1RZquA`D#{W}8Vv3&sXG`I|Yg8++rLIZZa~U|F?%>#=^Zwk1TRuP6^-Z$5`8Cz{
z@9)D`-8QnjoasMN#4%$6yAJocfOqFzjz~_OaeFRf@he`du;)7UDS6+ca_sY7ZF_2R
zG*n6Bu4&EAZF~M-ZlCViV%)UxONx-9X4;C03vyQ8dDak+eg5T%qJN80eo7QfR(Y)Q
z`6%n=Rxz8}qDkikwy%DZZ_%^ua>&!<1-*smT)fO{&u_n}yzGR?L;ty@>F2j5bNo3j
zDATMd?_3z;&|=XqJoTL5<45(c>n1<c3Z3xDaN=*XQWrtzpqq8R_vg)im;Cyvc*+yM
zX?`IbibW4Eb{0<*3e`8PG_X3T#JoB2%%movbuHgh7*A;~TWOckUG*l#qT`(K*Q4s)
z%eMq^Jlii5l+?tkbT8&&?#<%uLefXp%%3jjBQF2%w9M=Mc6WTjmd;L!wSHvsvwrU?
zBlCs7?CUQ6vS)tu&?Wc7%Qok{)|l-3%XT@KY!=%-ee0b%<H{3ie;)U<9Jp5W@4`XZ
zGZwb)%p#k9zkYl-Uf(=S<*ZArbFK24kjq=@>ULL3tyvJ0S9>|__x0DOeT(yU-kB3x
zxp&u(GT&c)@9jSBoD*^&cG~u*i{`}cwz|dgwEmZMylVIS<)Jsl^Yc$Uv(9_|`RW1x
z70dn4tm0q0{@t6?e^YnH_=??sv}8_9{k~OR`&PDoTW?>pt7vMtsg%U>*QIH8yS0LI
zgk8+9mM?#;U-je5=_7M%ua<@EKU5%FRO+$I|FV^C<niy9&#zls@hh{|GH!p+9=l(^
zV(vP#eeIvW&v4Bvd2_LaU2exOs4Y{ynX_-M{yTlK&-u&4^Z&ivSoEsCZqEHbk^dj3
z)%|_C@6^uwf9$6xZethSyy0XWGwTNFv%9uVHa@Fg&mn9*`PyZ_yxCtEMHG32OC&rc
ze!tabWRkkWGvSxh3GHu}&#x<c`0t?)*YlRvGm>|A{_A~iHA89IlP^1p>-YbOa9yV?
zx`b0l;;8larlr3({QvcIUsv!kkyI{kcjFa3+I?kt%S8GaCukgIIVC+w`JjsTRo*0)
z4JKy=*+hjoIc)0huTGCKXjytix~}~T%hKDAE>ueC3U0V`j%^2<Mzhny;`1NUoy{Kb
zxg-Z{U|2HW;KT711!h^?J*xyW_Hrx<?e&<Qx<s3Qd#oGRu_d=tmlXJPO`GU;yt9Yj
z;_-}EN2hQyP5r0c*WmCm?9)`X;uW8+?>TYR)p+mPqX*{f(+ya%*Q%b0bMnoWXorZV
zZi9^?+lsv%A787#`cwL4;p6*pR{SwOx5M(w**`>z21Ujjs+RTsZd^DiVb!Y@jE&r~
z?ry2mKCiQ$u+(JljK~u&7c<Q9NPNF~hyBJY|E}KR^OdT4Ao4T!oySFu|1(}x3vbF4
zn0s7=v;KYNRDs&2-@W3kX#z3z^+}&MHYBjzP_p@$nkX4~nqkHIyu&Ber%ioo5@|SR
z5C0nbjqLlxjTuy)>bM9*$v#WE!>%)Ph05&q&6Z5Z#I<HP-W2t9_E9%u@{Zx}ILoLl
z@#@#i#7C8mw@VIiOFo$(Zy&zGo>jZj|Fpe(^{eQz)P3{PHr#5GH&gs+X?R}p$*%gz
z!Xoyvn!+EpUg#?MX|;=Kt`&3I751AkuQcthmlfCD@{F3{8DzCgb=U8M;R}L~zB*(Y
zn8e2#;<Mvc%B#$$*UgLG?(92zYqHs|`%MNFF~9S2ufC5+U9<h|`5mwPFFm<_?(3cx
zh88(hml>CD(7BRu^g^rTuC)jM$z3tC{ZzHZXuXn`ThFRF)4oPO`?<H~eb=vCi9GW_
zk-y<z!$WJns5ILb{b!%dctM2K`>FAqO_QZAXx8s2XZay;=*YQ}N82XfR;y)j<v7eD
z>n3!p?)N%_<!t86OZEwT>iYX-^Xk|g^%fiNSOzUMKDd2d{o|c{PhE~k+UMN8uWv85
z@O4IWWl!<Jm)C5A-`c&|SGQAN`BbtV&s9E!9erM&+w!t{dIT=1EaiDEJxx$`i^myW
zL7q;PlN_!~>jMnAZ0@g43*vOz+V)q@&_3zyHH#uC_ij<WgEyJNJC;O9R=tw=mvTbt
z?!p2NuC#^+oXqtdQGRZ@6GBZH)}}Q&Ty5Ag&wbaYg4vZz5<GUzb&ulnd}wp&fyKs$
zjN+4uB79nsZXAd&b>LaO-ARu@XXDAZ%{IpA301S%nvMUNIoAtTY-QQjqb{5$C~SCk
zdZDvm^!5OoRe#0T?w=|b`{%XyYsb%pw-1$-cg*uy=lgi|u_Nyu?Gs;;FlDw@md8bz
zM;bf)jBl2)n@wly&SbW%6KPOBJ^Q!U%YL1$^;@_7Rppf}==oz+uCpJUXq7i*3a&ja
z!CC+Qrbc2x)9&8%*7{}QG4&~*GZ_vtY)~wDa(Yo>iz)NOb-Q%cr6a?5pHC7!eR9`<
zU&_@Fsts)&VtkxfPP?`5jrcC-eO%3N`NLT<im$tQ{Wj)xKXdYNGiz$w@k&siT}OP$
z+w^Ik%UN^pGM!&D-MR8d%));mizdkG{+4;WGW+JzzZJ6%tm58xzvavG`Xe^WBb?OJ
zRjRBC7JkU)>snAFyTW1lsRr4H3b(yYmo7h*ci!LDv&g!W`}j7O^6RxMR;elHgc@df
zt=S-WigmK?^49pf>IVx&Z*EmL`1M+W$M*fjwQD|fnO>`k{=Y`MX2P-k+txjAz2EeA
zm+{-ortUjm!gOM9F1#!C_FZ6Vb6}Wq^4d?c5|nnFtO(qjH1B@*_jQ@|EKgUz{~!K4
zR8-!^-TeA)NI}c0mKYSsIr-lu*?RB%$7TY@-?RRbWtn@+*=KK#_n}+d>f1ImMjhz5
z%k;K)VTN~vr-;X|t@HOe?_INIH+Ll8F3*WFDw7`lv5+eIC#5WJCpRhLcA%cpw7!zf
z=`W;wIk^{=7+?AyaW+3DR#<2`V_&3~`>%;ye;@yNKHu)mUxNoy5z7zVPfuCW9+YNR
z@6^;{a&XafmlorL*EYFFG~K9@73})f!+yfZM>#)rj-VEcPHFpPS+S~4<@T~I4UcUl
zv==mcbnINxE_`^8#3gsRqVtYn>;iihrp{aZraO>%$sQ|a%ef8@u1c=F-}^;h-~5H!
zrgNT$^HhG%c~Yi4S9bZ6KMk9EmVGEVThXLi^0S^zbw>TYG;@Y?8+zs~vY6zaAl1y(
zy!cBod;ft|YAR&~hKU_QVi#`kX>m!vjGp&v>0jU6^)cVMe}4*#zOdu&rVVr2UpO9f
z3fs2F_YK!)P37Ay%Z&P^9+|XRYzk=!3|sAdX(}84VXjTT;~2_3W6s`APdgp*d)gM?
z?pzDT+dJal)yE2n&#RxGA*EXuteDkpre_ns`IAbca81;X;0&pQrqNygtQ*cTFgr!p
z>1ee@8wpsZZhz#WdwJC(k==_dE^2N_^XC0?YbKk*jkuN#Z%*--20!DS#*rcr*S7i_
zhi0@+VTbEn6V}4JK^$6(E-Q$!>!dBYW_)zXbk?=;%>ACp{Pmm9mUdR1ziDQDanb7x
ze)W_0R_}Vr6gKmNub|L%%gsE6Y+myO{wGd0>YJbRf#XtmGtZSzvH{^1Ctgi6f2k-R
z{(O1Pn&hh-+h$F2spa_^&&J$0Ke10<q5Rjkl56HM{^#8vUe>(2!dU*6^XWN<R`8pq
zeXzL1o1A7@{Ck_+tV^c%>W^*usIpvt$F8-OelMC1$5@=0_F7~0hr-oy4RQt(SM7f;
z5h8l{-_pic@6|LpC2LBji0^bXWsrZ-E3{#6{fr|Q*|g&1au3y<6I>N9yHd4sqtX1$
zAzT$rtep=`d@5QVPg0vztF+25@86X8qvB6F+F}gvCdJh34Bm1~LhHKI*V|(C)Bji1
zb`;6naXh&A>e{P@%^dcDqHBwu^?%X#YJH-8!*{CSz7qzF?~VLepIWc#-}>~``ul19
zn|FTu?h>5#^^g4fU*=BQE4*Yp=eus7sj_{7={p7gR`HKksodvo>^dxcVq@+8h>X=s
z({|eml)Ux7{94l1`8TW8U+KfYLw)P^E?ic>fU9&$=(?3V&mTR#<(*8q(Z>1R*EjsB
z-ulRO!M3G)Le@XrU9~sxYn0Es^!&(eac7QZo1EHmZy#6Xu?_Ddyz-^AQtnngD@fyw
zo6EH~``rIUNxS0z{fjf+yGPFU(u|}ZiK$ijanoham-$rxnyMEYE1T+F*^~M7oq8AZ
zid3%fPm&z1#v1?K5C4##4r+ik8(wcWyw2Eec%7-;@H%t5;dPdF!|SZ=hS%BJ4X?Ae
z8(!yVH@wc-Zg`!m-S9ehyWw@7cEjtu+YPVtm4D{^?iDsWT9AREa{9(o{3`W&w(A$|
zGhfhUd-eJe<FZMcCe2OnXwjMC*~Hp3ZR*;;@2jO$N<X*Xv<RL#Q!4oHLh0q_|H_zr
z_<Mifn(*~@+YhgMZNC2ekMcKfLU%`gIi2_U<JX({`=5P&_2z0`j@jw7tNFLCr+<I@
z?oEFFzt8sj_V2#4^|8nFw{aJRX5W2U_3_u6`ZAgA)5XPiAI;sJS9bsF?K0;FzV>To
zPy8oi7q@2i!^?Hm`(9kXo%Uwk?oji4@f^p_+$@~=@9WLF8!>XzE#k|&7vBGBo;<VY
z=KVV*-QRwk&ph_Q{ay78q2EXA?&ckP*(dJ2TS@BI*>~G+zw=wY_WzqKmV5KWe*Rf+
z_OWuWRc!h3oAvTH-o@JM$fx}g{B)}2?FIc?cl%hj#LeY+W7cmT7O(rI;Qjj-_m1oO
zd0#9!w=CXY^ZNa54L`CNH)rgxxh-|Iu$Oz*&yu-$KXYHtXfNI6eyr&8^Nz|*2a5jt
zdUI4RTKp$?nSEsTcA4qh?{B-~=_a0<XKc3p>^nzge(if}_T{>Mtxq}H+_d)B(dEmt
zx7|Nx@^{07zrO<8cCSCl6jj@Buk^aJu}saez@^u_Pj8nu|9c^3eTPiY+wdiG*Q>oc
z^?&k&)ApCoe>-gwS5f_^aI=<u$JhK5OaI>c{yMyT`t7oa39*yyMP9vG#Bh7z?VidE
zv7_-Os|#*Vx&QXV`8>;ee&JhQzpgK<EZr}CyY!v;<98No-{-G8mZsvQSAM%}>#RSo
zlCGM$e82nj_`=A^?l!$X#j^@?GM1l?zx}l=G>vb%>;Crb-gP??%<rwg6+g>&w)(Rt
zd9E!NlIBY%Ek4sTed_G1uU5ruU+g<&zxeOFH`e<*ui=?`)co=0g_$q!-aeb}y6*Cc
z+}rh2B0juXb#~r2t35KZ+`s4d>g-+^+rzHqz2ryb=fwS%f?rI3aDSX4v1#wz{ZE!(
zo}0e@$JT(48!{ig`p_Zge!Nb*zGmOr%s<83U&rMw)ZBAdVDYkx^Jjf!ocUx!PX4UF
zSxc<rE>2Qs?0B`p==8tkb$>qARo|ZWbKy<7kffV$J=T1$xBuq+-TI~H<}T-A_a%P}
zN=~~p{`j}$!b`pO*Z+6R9?SdqKR!0k>r(Z79{Js8`@a24IdtCHy}AE->|L|do}EgT
z-I*1a|8KglbL_8(*ndas<uiphSZMyax-IfTg2JWty3$?0x5O-%VHa@1GN5R;Ovc|_
zY5tg7dw=fRe`ji=^6j$v^>*>w;!hmAm1-W;xAE<?*KLm)6RK{VW{rCE?$fgO$9?U$
z?dK8R`}=GAv`do{Z2~V%dK)v*Po~2n);DA0!q1CY_e$8^egE^;+q9oo<@K*s+Eqn8
zyLrTUk?~C3!gU3|x0heO^E&Nnj@VmK!{V#Ieox%g`S<I?>x+YD{`j>n*KYg1`pBAF
zYfmiPJ*nOILEYEa7fMzuO+0*{r1aOOr8#e3ZQCtrzeV|}82eA(lo(l~ZApigoP1I`
zFD>!?@6_m!I{T9PJInpcf`6R*+Nk=&&32`hW`cF(^AxY6?s292x94sCW4&`(U)Y=}
z@9r;n5HILp=28A{X49$l-*203=PhhAy1D;ZecEK1PpfY4d*8ND{$Ad+C+Yt;%`P=J
zf9qiW;>77a>c%hI{^dO?{#ZEW#(&rEci-QtzjRdOvg*o*k0!r&=Q2>Q|Nh?l)1#Gd
z11dMoIkW%I4~Mj055CRtYVs~$uw_9%$KT_(ERP&te7vTUW36^aLD<O#!R$$m9T!B8
zf1h}#(#G8@vOc;;BGkJ><JhakjQd<z!e?Zu*3NJY5jA%637N5YR`@50A8|@u=RF*?
z)mdD4whMkfpb?w6Pjc;Do}|jBib+eKRP=2POkTyg(fxEqW5^NKJ@dU>rz$*}xlK@c
zj^r9^$Bvo5S+Yg1-8-@8(D%+=MKi1~Ms&Ux?qoN05jrn;{$x;n_6cS`wkZqMiu_E4
zuDCc&X_h)<`F%!y3`eeFIhWAFZ$*hao6L@@o}Q5(Gf6kWMKk{KZ;fE}6Xxr3`X*eD
z+1&7#@w!do?8i|bT6QMCsGo7Yt+nN6T8Q5arsUcKm$rS0U(tD|bJa1CicP_1nDZAM
z`OW!6C|%%e)|uA}S`U5i^r>gDQTMc6yw|g`ap&ZnGZrcoH`j`uKa~)FbaR-o(OEUt
zGp}<Zn?5aUIHx*g+XENrCA01Xw$>_1PJ3kBk!W#=OH+&OtAj_X>@tT{J<4qpZe0mj
z=42?mpr$40jFYwKsSUx6--R|l*{ryz*EE3ZN&l3?A`+h8Is}DTpZTS3I=b4j-b>41
zy=}2^)dSVzl3gE;CF-j&y%haot>kgkkhSH(A77SFdKWgiT)2~{QWXFAugXl-d;S}P
z*C<AwsSx*^C*tjAbSY3n-ifK%N^9mfF6BF29OY3qO6T4z5z%vw^Uhq-9HlCHS}bjX
zYhJ|aB|=&cEobeNy1Z<v-~};{`;iUbl<Rv1cuu)XMPH4uo0?$!Y30r%`@*joRr-|X
zzTwH;dNtxc$4Rl%R|1zD-eMV@-BU5wG)}j`*x6A*YQvW1wK`i^C1(moESqRO=jsU!
zjnxM_rik5knxc00OrV=Huj2fes}q=9XL>i?2^5hqxG$oa*<u~K<fELIW1ybc2boV#
zEc(^!Yo`mHa+lncx8nLKjr9s3>$28sZRLqO8`b%EThSes4EvTFJ(UkuaLk#x*!0{E
zjnICkDwU_H=PH)pTOlb|p{?Z;J%KIUO-qBTR6w!!_)J+v(Ql`;PR;ac5}2QuGv|Hq
z3Kuq>?~BzP6I`O-Xqt$gmke6F^3KtJzQxmQ?(~b*PZG-M@CcQD;k;#ul1Nd=^a!S<
z3#R(8U0$X+v8`&Y=}o390=X)mXRy0%=2CWDX~25zgXGVM#~aHpif)(srRC(*`abxi
zRJBU%OfAQ|7fqMyM-*-4*<p8e>#D+fck6Y9>fY7|ZKthVo+JCq+{tEL=GN(nZd{E;
zdKvas6RZ!^bk{3AjG5W>t7mhxtZ(mbmbOKl+k}dAIm>VIEPusZbfsVun}(&<vn;+!
z2gV!G9;TH?rX)@=^a?)7D|j}!HS=;%LVa!rr{t|U8+9JPIy~W`u|{t4$yW<%dQ=x5
z_E1eZQnNbc*w4O|Qx5&r3KqWq!B4YZSfa3teMRT7%;wL5k29R>Bb=Ky+^F+e@T!1s
zrS_TT1li^=e~(WQb`zHxG)FXl7I4y%$@HD%`Dm`1Y(e|AlzHh(CTu;nSN-%3-T-Hv
zr$;Qx6;&Qa<g4=Y6<zTX+#tSqOP1Q}Et8DodIDCRS}-^2sO7rMwUhR>)okwCbi-2R
zq0PM@q2A4o-jC&i*XAC%c&1)TtIW{d(}C0b+$O1N!{}LZGuTCp3-dCypPAmvJZk(-
z?50<WjI^$6)tu;Cr_NllPbqwI%1}wS&;HuXj>k=3LvEc{6b%1kHbE_ktLfy8{#z%e
z9uDddd2Lv}=;DswuA;fEpP7sPF1cmw@$^BKXzFPTv#_QRxvYHEHODIRRkzNmcQl#3
za+%iQBWG$gY_c_2y%J2{9S?u<d!uR4##q)p!P%DYu7(_ySg>Vvm-n2>!byd5UGwHV
z)Y-L2iv4@ms#%414y&8;7+)$&I@TWYM0=zBTBiBC{V$wytPfDt6Ma^?Y2C3csf%s5
z<lb6w_&ne1OJCc5D;`PW*3-0n_x?&q_iV}fBQJP%1{c3Gdcvoud3f%%BXhFS`mXL+
zr>^oqE?hUHHFu$y)#Iw&(}EtYX`8@x{r1)+Eb{N&J57w;i%%{{y&BchBdGkRXG`zQ
z$#b4gi8`ym8UAOhDdTtd*Q;!54sa}GtS}7{l-d4zmXg&OvsY2)pFG)`EEDbdcFN(u
z_ijnPTTouV(w3X;_3Y_F?q94vv>3m)>hMZ9DcDq-KljL)%UxEUQw{D<npqhZo32<;
z=G}4FG-2-2qyJ`2XsMpNuBCd%tmH|FnZZ&)oI8S#M;^OWHYbPA>)gjnSuGt4CUmV-
zW3fM9B;vdAd-<Ud(J3;j(@I;zCyUJdkkxgF^NWYyO0MUpme%KX`meA^Wvx5*-h}_g
zXUkKph4T+(hPwMOidydY?aI1z&P8VJrh`=u6I0xlO<}Fbu<u)yC1a|Y5_IKIN8%3d
zR~q||$Cf`ou{V)x&PJKfRvd~S{3a`GKFlFnV!*b0iOM&<7L`9&uBV1+>a_Yi6nwAl
z-MgEmWrt+%`8Mto=Ke?PMeOvS&HT3G+^3krHSF`U`*vSx(Wp6haI*b@^M)LTQs)J2
z9(CR4mMr-8MAiR?$%Js7#~P2P2L3pyrD&MUH6_$SuR_;$n^1nlv5!}z&lTuC?loMl
zyg^^!^r0S~wL%u}zfLJS`(eQ}({+Y5xgyuH<0r2^w3>BV!5Vdy^Di7PNPe$pt6lJi
z;j=<OXm_Q+N}gjPF2V)nm())dOnp3A>_+!b9j0?K9GyE}zcUt12)NbgaEFoS=@*_Q
zvjp0A$2Xqjn$M@A5<X$ufz)G?UIIFy6RaIGI<wCzs2-L*P-MY8znRC2dyYJdF7viy
z30`{qg4#OY7B94!-Y6xsW~-oEPr%2H`p20oT#s!nUXgmtva(>xp+h<i6Ix!cd7xo-
zXolzs<#TV;Gn(hOwk4!hNh_Y3$ye3!bc3E-&+$lK4>t*&w>IIm6IRGhb4n3uQsh_K
zEPAd(d8uMVQDDGc-bz_kRyCi-6t4MIy|dhF<TPJxl5rDBo?)h}_2-P~q~K$Br$$t*
zno-Ymn&;f2#iD%k4ZFj_EMBWDIi{Yt=&Dq>+r~|tzh6g8UVPw=rZ4-MY^@!-`}}jY
z-tcqGwEQYHcj+SzrFo}Y!awhsHF2A`r2W+XxXPTb7o`n1M=eNVdvCI$!(Z>1hR8el
zBN;Mwn$p7M7XAGp?9yk<Pc15J6Pj$eV_9!x#hu4p^)s26mnld$o}VKWl-#$h$c*ol
ziQ(Z7#?y{y)HY30KJqYFcuLIqFOP#OKlFw_4oOeEwWGGe{tK&<(*gg<Tut-ms|7LF
z?9-a)d}yYYqM4KL5|NW<19Ybzw!L~u<Dl#T6;0JPxz1BMZP_zJ{I)#4x?;_7%Xp(p
zkJQVjEIW6jY}2*+2djD~G-&?|dC9dud_vY!(=C^u%s8-ZZj7a4m*DI4vR9h@$0u>y
zJiImYR9JtqC)cTm%fmD?T3;U%IdkUbdKX8D`4QfMRWkKLl`1t>wy8!QJEGEJ9<4Dp
zUbu1l8K1~Ih09o6ANqNUh2%NCOIamW5N^(`xp8WWW9ES~GKX#Nt*jSen!jGjT&m!@
zfwtl0xo1KytG!6%+-Tg9CiHv4nzaW-Q_Vt@l2`S9+_8P)+dG@L<y0QXuTJ&8Gt)8U
z$lo6^p0XePv?mw1oHw1l(zIB2%R#B2$rqDt3(lPRa9|_X^Zc8Ijpxs#o@cQ;J9qEe
z!nXO$&wr>CFME`e#&`9$@t$Q<B<jnRjY6+ld_O-^y5^YW<fO-Pg=$X^E^p*0Y`^#5
zDC4cYKkI5E?N$fHEw%fjwVro>lzduCTCHu#+11~l9@Z9r+V%S3<o9Oke|GJC+5Szl
zXzPBv#^=x36Sn7tt^SnQE^NSAZM(Nj{`Tyb-(P(&``fL}X0a<L|KGn2VLv-utp6B%
zssH_!aj*2Ow+~AnTvm;d?K<~0b-U#K($|-N<ki=f&f1#U_FMfz-R3_(KO1~}InlO!
z`s3W)^Ws+5UH4i4((n4V*`MWWDmQM+En9rM^z7@+{(pbk+sXXcAHf^`bnTsQwVUtC
z#l746??n9kwY&PaKYkd^`Q`d1Nju;An0gU+AD<Z(3H6ud&o0WDw$AQfg^cj>jq2A=
zUY=1S6TIB;tHm<^<X?qn<{a~p+`aeyw!^D#*HqPh`>?s^{+HFE_1E?4YmV%fvNA}l
zivImJZR0Kd=%5Ly`)};MJ0-tNX5V##+RYEn*ByCK@paYC_un6H&gKr&yJr`_ux4B9
zGXF!hW|c-=;#;Qp2W0D51wB~(`|H=o$0hbzzsbJ(_SovO+dOA)@T<Ssuhye|>+2)k
zn{t0=9XrW*yr!z?(AO_}+Y77K&%W}=x%%<!*xfcs%UAv~_*y^7OD#qAyry*N^RBCZ
z?o{TTPo3(r;Mchz!H@ghH`~=3zdHM$5i|`keVUtqz;=~8{IX2b3m@~#O^?6FuRDFu
zWB!BFTkiAgZeR6;pOJC<f(QH;nWj75<CmM>|CnEQde1%nAJaE@2(V61xyP?DedB%p
z_uE%L<-cah;?&S_cYCY6zzfFu>k<-*cR#bqoEDPe7puLptte-sPq>L(&MxbM)2+uf
zAH6?)bkZ?l{=W%6+N|F#62HZAt)9ECw&eQhJ|^bg<(eYVnhrLRFa3E=`ziia6V39t
zEaJ*%#<gY7#yc`Hw{tcr6d&0lu%}hluj%TZQisR(d{x{wIhE&A1uDPXR_0irTmR%q
z{+(jCMf2}C&WQKxyW_RaM)DHdh1+Wj?`;bDCAQQ3&g{!uR_coDycNE%!{?3Tdd9cj
zrM0Z(;mP^`<dZJ1yb*U;p>%@r|4sMZQw7hOZ`L_^?}bhHpVi0w%G*{8{|*t0FPy(#
zL+8Qt-R4X8ah89!y6XM>iHzJ*dy%&80>{@n)^q*&lZ6(uRxLc_+g~ju`$cT`Cf%?f
z9lC8NV;%`wEeYcdN$=X;SkTFTI5_=f*0cxL=ij^8zTJ4A<2NQy%xvGKB*4l%y<Ss5
zdHN$&0p026H3hy;e{U)vH~pi!fZld%Gl36m)Ay?iC{JIcDWJQZUrpc$Bg;CDv*z1R
zX$fdD)?YRdX?s8M2Y=W)2c?a3*DU4ER6A&*a%JHrw#FTw-)z37vrYMS?zH;;$!0g(
z)%4bWxAgBjU)a^n9=e!0LcKT6%|ua<>1^+Ws@*+xVc{Q@oX#2RM&z34Ear<T3EWp3
z{6X#cu|v0Jv8|c@AiyAV`sXysw|nx7dSvf6Ch2ju@I9=byRqwM!ItPd26gARH3U07
zi)c)5s%)!HyyJ81#S)R4AhYUdex`$lR~9%_w`V$iRDOL>>wDJ02{wvTzr-@HUzBWd
zk>AfR;r2Gq*}NGq=0CcBef8Obs}Zm2!ekb060rGx>3Tp~{PpaR+<<K!L6gKPxT=o4
z&HvBz{pt7F{u#{Q6zVG_Qf3@*<PvzX&AE<!#m!Q`qOirYN)s;bo?TG<jN#(V-;U;9
zRf^jW?AiF#{PBXD3wt+C%9ZT8*l)XfXJFO}iIsKT-!E)E#wA|$qap0Dct+!VHrr6+
z=j#t|-Z5FhuKcNL^R<jEOcEEwGvgB;D*Sd2ou|u|V%5KG>(pD#8o^HX{>P>qD~Y@K
z`A^TYmOnF>D2L4XR`tfFN9~QWeO|s#U244E>eU)Qf`lf2wpc1U_vy_)FRBCoYlJM{
z_CJ5lQIp=xu!$z~Cq6$G7{KRi|6l!7J$R^f`xRXQX^!cgUIKj6|C$QeZI{;&NMPKq
zZ7a~k#I?d(rLXfS14HeD>D%oE{8`-3GB8X&a8hRab#DOy8Sn%-69dC3ZUzP+24ryb
z!t@{q0Tngyj3g5S!x3%<20>&64680PFqG#PrRb;TrA&`?5RjAsPX)0sFszhdU=To2
z+kI{NMF#;JrbD-;^E(RIFm>FW?&~Pv#dK0_`b<XwN2WOr)9*M6I5M}hGfj7N6)>6}
z<RqZSv_o=wlaqiW(^2>7M?rMNwCNnq0*=h*uChT?)j133GqLZQz8)lcNSgy9s_G)3
z&y+G_dYFrVBXgi07ew@oi-10}kqpoD4H7(J(?whbw3s$tnC{>zki<Oc82@yE!~9~?
zcY{P<o|*pKRlu3~;aUFa@u&Eurn|cd2*|)kq`lC?!SdYnVmAR>ro;EAZ*mjxV?OwR
zf4bsbe%a{??g9eS?K}jem_Iz^pB`|JUu61H2LXZU^&p{}kNBrEKH?XdzQ|obpGoV<
z^vmu7_RKX;!J?AW^*jUwrW?2mNHPC@#y`E`F~7+4e2`EfNJ#nxSd|E<wAE*ddO7`}
zhkzZ^r9aaZJq7G|nHdEj@lsq;nKM1XK|pMJlc#_blfJ_Ad7c7lGN96!nTdfxiUmEE
z43wuo@D#9N%2b;!=Ov&z{gju06my=Y0K}msUIL=i>wN^In6_$7UkZ|U2FZUkgv$T+
z5)hRE6*box85k0n(SwQAWV(*GfGvx#xd6lTkA?yw(;K}7Ojsl>VcdWE0$iX}sl}pT
z4H1(AaRsKoFc#pOzS3KOhe_Rby1$Qr0Vt_UG3nV)pY0=H&vMHFqFQ9SzpVh*^gliV
L0&EYQK<N(vM;D(d

delta 58664
zcmX@V&~ef%#|?AY>lrQvTRb_)z>vF?i$RcqfuTIVC`CUdKe;qFHLpakA~)wvY;^u(
zkFq-c3I78E<cylP<lfW2x3OHK>rG}(JWsSSHw(9k#wL}*SxP%qQf~@B(cQ0-UwN|s
z#{5a~lZv~R=!vLa-jouZdoJm=NMz}%S26pno`;D3ygdKC8spBTp7pQXAG)^NKecLU
z%X{+k;fqC|v+vtov{E@c=R$Fl#uD+l)w9a})%-mF|I6h!ce?MqY-y8Z6iTeJ>%QB4
z&f@f=&zm=2k-gkGd7fX5dFQ2!8FP1Ce4-=8S37;)*+oaJg)QbKhZhR%H1S<%dzmwT
z?v~&Q#qVbYeilk+n>ukp$+Nz>D)o!YxuyOr?mUtHYvwjTj~k5wJpm7{dv^A#EuE9N
z{u+m;wO~`++Q3QWi$jbpQq8>%S*ezqdp)vJ%{BMReq|CF{y}4zqK%B($s;Epq#Iey
zwm7*xvxcc9S^YD6-GlV+4wHkujQZs3ZZC8X$o5NAs_I|1V4km1Kii`4piVuj%PRFY
z`mg4nd7XUy{O4as=U4JdnPiL2x_tgctk{ysNb}4?yZ2PBXAnEKb#n{bf#VuJ$Lnm|
z+fS~$^5EePm7?9fjN2zK&@u1+*|uyO=dG%p5@IZ~OZHE=t8$`x-(02hrW4dA&A6U%
zsr9Pa>3Kd5&EmZ8kIYhi_KEl4autO&f&nqN3O3biWM;7T8kZ;D^*ElN5vDpfM$6-P
zyuq~mj>fC!bOb*v>sM0jNVWgK9~Qr8YSm2hI+>k8Zs`%%PtMsm`CPBsa*G@e7FE~G
z&ZKn-j!({g<jK5r+|YNP@SgcPCk^Cm&j#FAzi1K{@Oi%5+rvzo(qGuvHiadAIIS?n
zZmqF-vO1ee{rnxz<=qmeSTr+Ku{`rWz{EM3;U9NbPyG$my~*DL>$|!W8^rHTQBWz<
zJIE>Ha?kc(@59CtIqS~Pj5!_#Rof1BX0_jK)igQ0dy{TysUu^5Q5hfK^77h?4=wy2
zr^L;;zsu{xvMa)=rGe|*+s~PtnAWvfZN1R9=W4t87rcMIov&Ww>3-?C#(S)N=5X%y
zdhkuBkRk7|#_OO3t{*?%-1AAHYv03`+Q|of_*(@TEON|SUWw1WAsJ|=-m&Cb^@({;
zmRzzwWv3Bz(c@!F<%#E+Z+3VXE!6bQ%X}VGQ(x-(%bdkKt@7v8whfYp1480*Zk$=V
zB6rbN=jHnT->(}wG}wMyQ$KI+?Vl%3&RTFKJM!Mq``bdpQ{&#3bx+z;cWQ3Y?XKBP
zI=8sL#^%4Bs9Z6_eNjN_Pk!Fxmm5Sp{Oulosr~luxM)&Fe?`5hi^z=EtZ`4m9bI()
zS!;BZ$g?WVwwxIIZKX9|?RCx*zCN+?0<$bP^S0@8S*g#M^M~p1`}es&k{(U0|8BB`
zQ`&SV)09^E7pEq#R{1q|*UkA-8=SV~MsD@pa({8E>)w<30sM#lOju^@TvqwEqRxd!
ze0kNUnP~=}>$7v#-Aomh)R?6)%jog9d8r#rCha|;Ab<J!>)boHuN{46Bbc&K`|B=t
zgTQ-}b{)D`S;Xwcu9xI>KzmQClbqz8$vHmto;LgZXNs|%v-OqgnR7YyoKf$+S0`M3
zAG#cMJ94k`$+nZ9e+BCng`G38`@L~SCrfUnT2@=4bowP<;rLCr9)35Uzv}XRQ{EMm
zFXc+@vU9cfC~FQsXg+`EY~M37%e}q&*K_O-dn+2Zsk+SdN0UbPt}U)Vj!s^!$~F7z
z!|SW;95&QzzijFYId(m`r|8Ap$7w%yhlLuYPY|p3UeswXwVA7sB~YASaAVMiT{?eu
zM^EE@VUTUbEi$QSvBDvzHFgmk>9<7#t(qTwcVa7D$hCQkO~}t*FHb*Tzh{x>rX}7x
z4{u$O+<I>hTXn$9_|k(HcE)X23W(jnvwL^eS*!K4)WmYs0_t1C7aZ$y;O27IxsVj#
z&2%<Vt>BR06qeKHdajf&nEmi%^)oL$6=~iJFB^mI&Hc<;#X8~0)5gy7_&X0+cYdgx
zdqLKn^HkIOQ>zprJG!?&*L3@Iv(|^x>&t7|tP9tDS~E}dDm?N3Yv16?w7%eu^|$<X
z7Y1$xejTB?%qBdUla`0p=Py+Gpju{>@aakV%XG_z`ig()w*TGtEtWO<(%j(W?Ht49
zwJ86m>1)p4o^}0y3|u9dc<mQ+^8NfK^V#O|0neMJ3+wsyt^Q45sq-+bl$>;9|Kfx0
zKYoVaKYgr4JL{?H<(Dmg&Is;%X2$;i_so|c|CpUgm-`vn@W|(*_IKAqHzw4ZpOC6v
zaY28|OK;!y1>UApr}FBqz8#%rvUSm!R}R^)@?VLZ*A2FaPPOq8OW{+S>r>dP|LfQj
z3$FvoOA?QTaK2AnA(<!AwqWjxs05>eR+h!ZK36A3FJaVr`!nE*=fUW+GlKg=Kc_v<
zvb=sV_3xL1FC6|B-TJ?2&6d_@XC~M_52)8V_s=BF**5&hjIY{<p8Guf(`meD#v~Qa
zW@7`s0@;Ng&(hai2nyfuI^HfSzNq{4=C-vb8>(b`Y~AFm#OLmB+BqlgXP%kw3(XzM
zI__=!Vb;CBvNkBjmz>c!(@-co>2hjorEh?Vt6+-36W9Hxwz%efl3w;Rw_?}pp1@3}
z&7w2wf3CPaJEJM?=p>eyW6$e4migG&pZz61!{qSlcYj_z{=;Q`F;Vics>5W~ix<x5
z`APj?xY|{DC00--zPB;EN2>kx=^Y|xv<s#?-udh5KAm-D@~ytsHs^%(Q^j51HB8!A
zUpPnB=s^AFNpVjO)iga~6jz>a%6+r>qfq}cy^w=z?s(P<^e$R&Y^9;KYkvOy!!E|(
zVy)*-(W@^y=k8r@*VMY?>7zS*+15Y1yd-8O3oz-1_9~ve$^6!S>s8et+bydzH_iJ}
zHzQ=uXTcQ*uU6W;7mKl-8CAg@bja{Ae}b!`^E*K!o$O~NGH=54HkD4(&r|g@kq$oY
z+}`M!G?nZ3&P&@J>mM99J2UkLQyjBcz>(Q9d@0AW0$%oKhQ&NR^N-)rn1Q3^U8iR8
zv(6(cuXQf;e${y@a@7$}sr*mR>XUK}ret=Vdd9N%hQ8|k64ty--5J^@dhNny7S|Sj
z(trIZ(vIs*_N*P(56m|%o2zP4e7vHw&#~}-&P&F}XPP8`XtVlv@ubzySK?=R=CirO
z@NwP@pAY&a6Ydyx-V?uj^5*8FAI^(yeXMf!-sKFZ4~u4O|FEx^`|yeGX|5m2Boenr
zNIUK*daSem@UDJYJ-vtj-`t!uRp<XXUlVudwyDJv?ua>`v{An`>zYd2t#Yr|TYp~v
z%EQE^l>g!CjFftws6Nk-K%*y9XQb4(TJM>${f9t#K?~dLN1v7|hb`a3Sa^dYH?NGZ
z>&$5-_2+kVmR)o0YtoOrZsGosfs^OO?-@@I&SCJG^tm@WSzVRc%lW|BXekCg=P57J
zlkHPH6$_)Q4n0%ZdSMgWTJ^ZcB1aBiF<WA}ZTZZT{1%!vTg5*#tIZVrv3Ntx1i$2r
zll8uTc7(iQ|8?{K7WcTAe;zU&Z`IG)a!le`&dH&tC~%8c=>h8r;k27Jhs}Ktm}Gh_
zQgrm2Gw<Dc@n)6w*0Wx0YLm31tj_NJsMAz;Us(N^&l2GTr%6xOb=N$}o_N(x^{c9+
z<--5$mjfgUS1$Ic6IjpUcu@6`|Dy(_*DJ~;6n|#cr*1tMe~M%AA)e)D7HS1FdJ2Y%
zcoqDRTPoao@UNyL>pczi>q`sSoRk{hKe_!&Vsk+ML2c*5r>;d*GcaB=PF{L7W~G_w
z*$2@RkG`8G`if0v_p-t%DKF((H#nX-@nom(6Ry)0f2$Ib)=${Cu=k7I*%vGmlxDHX
z9da=_aiE>Od+*o!xt8W;neX>cTCG{}WNB4RqM+lhRdds~)=B6`+J4v-pCtKf`>Mse
zQhQhL``op5mJ*8q`)ixm52~iwp0@~@?z?NzqOUoREiGmqQaK%QtwZH;kx1%+$Z0V%
z=C)rxY3kcOl}(NJ)sDxrl~gOjbmjW<0v4t((A#tF={}=X+tjAck*KXN+F%{Y!X$Fa
zHbKH9ro?30J2t(vNRxvP97A3AG!`EEqsjJ2<adw#hL1n{FPdoAZJ2gP!{cFIqQj~Q
z?%k^HazPTF8!KI}%+~0+^4KJ^Fxyk+=;Yq6+<4n1H!L3f+Tr-<Yt!Ob{rO@Nx4kay
zxTv4-^|HOg-<z)IIFGqbbJJn^P=Ba`%P8VhI_CjK)&oabTY}2$f?vl-HK{$Dc_gOf
zjk@+0j`uNNVr~dIoP70btM;<)?%cAcVheZ8HQ2OFB38J#^*oPbzJE~hnH7=J|G4)a
z*Il|gxNh$W3ujj0j4dvEp3fDm7QA_`XO92lPT5vTriqhx?W@0ND0H8*YuV+W`C-ZR
zC&h1Uu}<9I5?FORX1UtsqjA;2??3H{{+#greB1dFw;Rt^=I*&r8u$L&H~SokhwgXs
zoqw>2sdi4voOSKtpAAy|u2Y}PJJa1>JpWt%(K8d{x1L~4uyTH;X(Ov&9@ye|FJ*7z
z`-|tU-Cnu%J8y9A^vAnxZ(HSczOOuH`(aC!Xyf8~S%Kr98Gg<3SvM(@Gb{XHZm7G?
zl5o@G*NmeAcwJ2jnh)mni0$5_d%carJ%{)Ft9mh+m;?#W<%ZAZdhA>AlVi1D&N(Ai
z!PG624S2mTyghn$dxq&}U!^zq3@kUUf8CSm+r|2N-csQe>|JLjzRsAH$Yrf6zkHoM
zYeAt?-lB>(*7Zj?jP3iF_IKXWn3OR~>Gt6ptImF3YqsC=AVc|q#zi--A4|WSwLosp
z!P80KRr{97Tyw4C<lpJ25OaFZ-)(J5?=L_2GUv~14WZ5#Oczy^I9ncfyuEWazuRSE
z=Mtf~%qFE+=NB_iZGKW0?f-}|_N1PD!p8||zADF~r)hd0pHfrvw*K6=&o>J<>gJiv
zp4+w9;@I9cUmMpx@m*rK1-|C26r1Co6KlwSE$w4VZuutOxPU3gGw%IZq+C>~vH7@f
zKvPg&la<7#kYKyt=bxQ9cld=?=c0!PO>3)Xueg>ba^HBf+Wf_D4BLAhJ#}2K6*#`t
zU3EWfZN;RD*H6D#DjBY85&gZbUa0H1?J~Wu9D9>@`^=fr5am4oul=XsS(}fXH<Fq(
zLrvKJ=wzAY+DUgeYHLStzq9k(gKvU&f*Lk|(s1%_Qn2=}nf`WP!tGDCcMgY^+1z&(
z5xR5KWZ}H`Ys&JN&y=P5n!h=`bmKnnxwjMc@5zXq6gYQNPUmITzO~EVpE^BHv&iaQ
z{h_~n6+6YRFX6Pj^0)rac{OI&U9XQ<OJBOX!ItCx#f@wCw!N<OW&fV{I#2mibK&2G
zzxyQq@|(Z!$h;@<zT@itpC9s~`EwmNoS7Ot)&H@_|7GmDd3Ra2r!V{R+u&T?{XZMI
z&X(SmpAq=%w*0dluPfq2#ESpkny0k!OW}t?Q}?UE^**7wlUg*&cA0Pe_ny;p{?%W+
zGdO4Z1kbSwRMnnc`C9d4bX#h%(`3uSuWSA^Pg-Lc_5YX3&gL_(erjY1F?em~l$XyG
zTzz!EyxqIN@AZGL?Eks{|F8TzVecl?J*t>e!=!OIcDDt0gJaJhkvAnTD-JcZmArVo
zitppW6t#*xz0583tCA&cTg}V-&Tnt3Yg%xTy<*k^`=?E5jK?o6d*|DCbhGa>9lcfM
z!9}t+#JTOit)2cjVXvp>vSlCD<aX9n@ZCDpBQ-fn_uj0KUb{W}e}3uzKlk|k58V$e
z(;lhJP@VCbN%$L&`G0ZtWg8YuTsHA_{<$A(c2C(l)8pQotX;Qi8%5XF+f{E37GB2p
zA@H9Tr`mO^<k;0B7MW$f&is}YtXp?+mhLlTOL)|?_v*rnJ<8g%?oHKtrubr8$|`UD
zg9nd&)l1h3lhv*g%lC-BvhV*<o%_~NACgpkZ^#Gp$f(sxsP0dgCvaQi{+$hfE@-_I
zoi+d3^^(guFUzif|Fh?9>e9b%lYiDT&UDHbtJ2v}9{WI6`1g*Y{Lj^@6Y3l7i#}|K
zJhkJb*0cH29yV%|cjwGnZl9gg_t5FV%;d5y3wE5k|JiUsYwqD#o0t8Pk5g3-)_<65
z>M0*}L&TxZ@Z^1^OAr0O@Np|Y7u9Mp`=YpXg;C_*GQE}Vx?VcTO|q&Y=5hJ6|4LoD
zRsUz&gfy12l<$wFt2`{ik5#|Ewdv&bx+h$7mYv+Q>gALC$GuHeIpW{WFWLLD`AdoG
zjSH?0$6epQdgy<pGIrbBQ|IbDH;WvdB$jxk>__yg6z+Qe#{FxnHtkwlv(@eN1=hu3
zjvpPj-P;kk=i94e?OC>~dewchez%6qz0I~V`gZlBi2Ce<pByi)S$d#6dk-VeCtF`3
z%XPcr^tEHpUC>(Zlau(tHL>9B)z7-l-x-^){}A((+Et>f{W;*O@=>RKM-Nr$gt!Ji
z^4aFGr=_(^H;YOA!L;LI-0|FE>-Jt=nfk(RgTR*?{D1%4m73AA^rT9y`=#F}8Sc9^
z+_XwJ@OY6sZ*H}AeS_wfhDiPyO;;zIoLp0L)5TCVa>`d$D?3R8Zsn}Ezd0K!Z@FD$
zZ#-j{^hBD`X(Q_)UWXO8i>5!;zTL-oOC@m94Q|h9g&I?b8^<oPGaj0K$Gb*3Kx1n_
z1b^!R8NUltPJJv{x;k=C?mE5Q+XH@F=5|h$c9Un}>NS^8IjyUZ$58Kb)TH9-DuzWm
zw``9Jg!gaR5_xofoQS}6GnFkoce~Rg>edMgcO1?<(RcGH=dri<b8;O1cw7v!)Y}qM
zYkK5duX*PZRzDGzy`P#m?+Nh4Zk+wjE9dU<bjQ<XODiPyoY~JHv`0`QBwAxlyVFwj
z?nR-0^Y&=0mQgRi?eVMBF>!WCy$y#*-A>p06VF|U{BuHo&T;dm-l*>0uQxQlnA|iu
zdTM7xq{yyaccW!)Wj4Q>e*9SO%xTjWa=Y0INq!Wb!g%OW)}I)+tNYjO`?8VosB-Ix
zwe6|fwtb)LtaAM3uAFaL>Zej8H-}pP7xR5>w@<7iNBQO9h9~O_AGF^;!Lsp_?}LTM
z>Sb?wihsTNA>iuNb!R+9r%$}>;`pGtf~Ua0;>5jtllS@Ad$`|L@yKoKs0eWJIasv9
z{mw?=wtP0J{x=U!e|B7~aPvp0&)!E`S~4t`U&)<!+J0e&ct%cfFMH#$Yz1*&jyicu
z-}@3KayJF1KW}z(ox*bQk@ott(|7Gz_e8%xWATgn>&(5n268MC)faVYiVMF?oLhQv
z9<N|%^fZ%uYvxRPmKyuiq_n^NUsQUbd<Q6wj~y|W^pwB(a{6;-ch(AJ&+OacuCdQ5
zlKZ>$bS>Vo%0#z+RCsX1SLt~1!k^}B-R@;<&weIM4E8zuB5Uf_j5q;a$Ix#(rfzIK
zy7xd*wa$lng$)%?dp-z<<Z}!6zgg1#FK32Ex#(Q;wWq%2g-+iR_x10S+_aC&#g82k
z6Xb8|FyH5XKPK(t3BEla?re5kl$23)JxKfYg|4%XKh7lj+}+i=o<r#Nl}C+^AAOgJ
z^}jiB+IZ381yQ>eMEp3_p;|J9^>hHwV%Ou>+xl|zJ}SKU5msE^?Ot}S_5L%NH{#;1
zu9JDst;x9<XyR#@%2#{xo6dvmgGY0h&yQ-;llYp^U&xg8rp08|G8OMR26KO8cHQ4}
z-29!&m8FKiH~syTd63EW^Yl+=-Y@&7QF8Buw~Xn(6X#BSxE<4Fo4ji6DWm+*eSNAD
zo0PKuJaV3yTl7gZAxpYlo~=Ig^J1fX7Ows`IcJQW6!pI>S(g@Wubi`Dm&&bbp54(J
zcbG-CIevFsu;I?zA8rReobWsR;dYq%ckZ@1Ia!nPLd`yL2xNDw^$Xl=dia|0aM{IQ
z4j0Sz*?k5n<vaA@{P(o{qh_I-mxnf`_=StcUcEhi!d^9lJ7o!5QeQL$_P5j@^1ooR
z#bDuy0~Vcmj8e8nJJQ!HR5$B?yY@rbTJo^{eb20$Vky1XSFg|6QabnX&TnV$t>@Cc
zJ>^m2hfm?%KGPIF_a;1=DBE4><|E>J+F;46+o~+l+rG$sKaqb}#DAUCp(Wow8CYuU
zK3prmA+tqsuQJ=ugZ%A*MukTIr~RwgH@SXdvgm~+1uiFV_;1$OS+5*Xm~!0p$8&3E
z^-h`ln|oeWi(5EP+kNcSB%S4jM*Z@0PUXj{Rs_H5+_bC1YdXV{^p9;zO3cdZkA9nB
z`i}KaNw9Fp(kg~!Ki+FHYjtf_d;Q?O_^XOK3$3X;O+G|@`jOJmZ)CRahUD56CltE&
z`ObNAyWaPiisY0}fk8KxUvNC)eDzsn+(VH#U5E9{n7rrcX>rWbd)>4qf8Bv2dqWiK
z#W#grxnVPbA>gBwi?i^n*n=LR3R1Kn*`0H)YiD@ii%pf<U!EO~etG>-<DQp|K}NgW
zR_pFveR5G{fa;Ai6W(m%|KauPifp@1*xt^SulXNN%AHX!X{{+4AeEH!+V`$&&hBYZ
zvF}!#^~}0hRJQZj1>Lzf8z%q!o)FQ#&Vl`byVbX3rfWZ~ON~D@|M>Un)$z+y)662~
z=yoqUWM}m~=*+SsM=QnV_I~;uZ*jpuIqr$yMQt@dCr>Nc4I5Jhwcob~+gNhG`q_Kt
zPtKe6HCg|k+1}n~Qds})W}Vs$-S$8M7h|0>%N{7`3n%9t;`BRt)?yahy*-~)KK)75
zyV$<#Yu}HA+B$&-rwJ|dR7@D;lNSnl@a<-kRocfc+E7q=OZXRSNVn9H=E$~EY45hY
zLwScf@*LM`JT9=3ywK9{>S?oQcx*@2#Z_H8Mcbk(KOQM=$lPw&lQzA6zo_#b>%I1W
z7av?9a)9mdin{RMTYIz2*lco@RAwCDjWglPuK#*iGyN^w;houh5ykPMx(P4THE(p=
zf7+~eMnI$>W6Jay|8~D=aW+m{qv|AOu(*Nm09%9p5!MDJqkhgsAN+ZD9R6OlYRTsC
zV|$X`t;yYfW2&{PlX+tAqOLiYg6p;VK4d4Kb>mOu&TbG?p226-_3F_=VW!4cNleVs
z&Z?i(v?_f1q^USuh}AcS!E-Ci*$0Kuf3({cY|YrN$g)%<;^9*XmjmmQFKd6|jyQfr
z=Y+XkzsBt62Tvb(aet}7&iR*uC!M|PAl3KL?)0U-%>UFl&)VC3?ej^VbKqf9TApxU
z{pl!w<Erf6$KxcAho?V&Cc_@DbMrvK3^8XXTl;Py`3R|2$Cpc&K8Uz~>CWahsX!rj
ziK^-stcn+dPS4&j{q?J^RkKbkc(2`+QXAi`)V(D(mes5yOJt?U&aJz+-sGH#5?Lv3
zYQ~=uEwb};P32=Z!?jnU`IWRketzWGs@l!2BU8Wbx|CyJby@%0l>3{nNi;1uttFk9
z`o`Pe%%w^~Zf&b`Lg+amqovx6S%Pa4YQMFn&QUX4e=zCes+uCf6Yu@);`7Dt9=>~=
zvsUW(1-@+EToZZ82W+<vt@dQPdXux;{+e&0cU{}@k85rxZJfAx8=LfvU53-D?agLi
z^z*x6D%<q6{t!=iY;)9hrF~5b-u_yTBdd(&S6)-qJ3UE_`_@B--4%z#oXnyhN5v&g
zwAAp`Pwkufc=MO@Yx^f=b8VjWzVCixq3XpFlePu>5>o8S8#g}cxRA<m`ONucl2Kpw
zd}_OMXw|};0D)~XQ?yS^@4t4vU*7J2>TEk*RmIbfj#_!;eq35Fk`lScf;}#}Mw2m8
zEIspJ&=XUwK*jA}Qj(n}-k%ZpB+{$UcPmTm6W+y}6jz6B)d@CgIoftrRX3{k(3TZD
zwlb;y=VxH}|Nl*`XZ{n9Yd7K*Qx4qaogQ*W{Oon69c5piugdEUF-=Y|kO-I=(iuB5
z#cXn9*4dK{`rj^oIC$OiNB#7k_5Vaad&@{0?{t~q_-tL8te4RAgd?}_cYlw2qQBXr
zV&6ZhbNlvx_<#E7?b-F!ziamHIGgwH?N2La^G&HXdv~ARw_m{ZYOu-SJq_ibX7Bf?
zZ}orn;^(^MhugRR(0$r|`*!`m$B*yY|5x-5pSB{FTfE*&T`^n5qVMhLFontM?dsR0
zC4N(~SvHfQUFe<oMU}c9^+dVlCqFGR;)`Lg$yT)Se)U!9%pv*Z_v0lNEj+PLHGaYB
z$}iD<>n7EUPt`xYX=$!r<;CP3v!0%{-u_~e_UTQ{%RCos-*NtD;_8{Z^rcyIp7Pz8
zHs5VStbMy4=l-+l*H61zbne?9k#f=K)IK+Hj{0whj#oPxu1^WvC1%lZuj#+mn%n#T
zT)rzmBmS`SZXt#hPHp>Yt0I}EoNM|1(0dB!|6l9>d;d`MTEr9canVziZgGLzyXV-=
z4RJaDmDl~2?aBw<f0j6{SX6pc`qSwT(VGo^^33Zvl$xji_`~n+_(P}b_p2&TPG@lb
zc7S`b#_OD@4fWOLvR>+L?OjVd*M%)Vd@^{3%EcVbmXjefmIo`e@HscGTDEf2!`H_b
z9*|p7B9RojUGW-MDwEO^)if*3sU8t_i9-JvG8E#@EV`e>5bv?KWlF2373*|9hl9&r
zzUa%`n<lyH?IfM+EmzHgJx!U;zp@XGOgdF<x5=J8)7Clcu#$JZ`o$7fx7&F~Ph6X`
z#B<)#&3`v{=iHd~Nby+3);Ejh3+^^BJ$FJ+_l*&6vE1~$=(vfwhYe%o;sjkIFZ=k#
zDe})}Sub(>0~dF(?U~}-=OSlji7dJ=6?l%Zgu(ONVwLBLr*&U?8ig#7XP(+Tk0<*3
zo(4DntJWR!PN>#Zty<@N$))~yw0rECz7q~VGiS{RarqmyqhZ58mJhl=_OE-y+Oqz`
zTlN3eirmW!wOS^7Zi(TVTp(z7WV1!T^12gU&pKBvI(;eZPsHBK&GPftCnwxv+Seub
z(4nB^xJC4{_m5SY>Zd#S%-^>na^AzJ4`(lZG4-D=bN0pqsk0|UOhv>NT#(N7omzi<
z=d~|-TnlUULNy;N`tCn-_|55BxirfP6Soj<cE3ZG5l`>iEfe;fR1!G-Vv}V40?)jA
z6MVflr^Mz}GS7Pd)ZFFr6_vwE>#ulhI8-6~ku__AK!cuNlE}sE$5r0bWuuS3H<}Xo
zOXEqPSvue4IV`mc%-0585J|n<mekDI`t<aldPUWb%^w~rcJDi8krXMtjbUHE$IJ!W
zo2Ffoom_cb>qMK|({(E>XGJe9(iC8hzL+>CF<R@rQl^XXTNPd9cexWxlGu((2fQ%x
zjLNKJ)?MRy>B-cvN&6;M36-k<yjQa%AX9i_={mi;D;bxUr1O4IGt_8&7@73ov&S1P
z!wVnp&0b&6l%Vs>f8vIS)b0{VUBSbL_CFMM(Dc20k}GSu`3tS2)FaN9ds_tTj>j){
z>6jcQc9qv~!bU!}1N$Byzwf&-_~ZHdEm9Sx`xfwAY0h|Y;Fork70Z!itJzD>Yb@Cu
zaVBmZi}XRZX4P3v2^SV<Jb5V2kX#@+<5!?jXUmuD7QqdS8TE}#stLR%ucQ_0SmNs{
zKkrJY`Fwr#slz3Ffy}NUOtP!~>+x|i#a^Ev^GIm=)#DQb9rPAP=4d*Ku42gz{AhG3
zo|kjgwml0ks%lg&jcqz*tEy4+Lr(i*o@S)*@1@pTpH6v~p{sS7r^Z*X^3R#KnGRv4
z7ejV0$@D+CbE0^*Gs~&^-fTX$2~6TUT<p%v-S9MIFiMr;xsWDs<xR#Rh7LW6{3~xW
z4lz8LTE-@kBz}9ZjynUB*hb@AFUgxzO1uJOyu+{Vsz2h)xqHEn*c<$!r>oDGG1u;E
zT_pTZX=~=^B}P~8uKxPzMSJLbq3PE|YyBsca%*kW3)(DVTCj3S*xt(uTkGFG$(TB2
z$I_J<yCyx=vRUQAr6QTS(nKt5Z?l-2o+0b<|Ib&QP+7hG{l5tJWZy??K9%r^Sqi>e
z<SO#)!SVI0GiH9)3EiIJEt1`*R=RRo>JBZ@_N&&VGT$OsJ>HicziRT*hb2OK#Rrc@
zFOkU(VY>Y_r?=ft)IZQGaLFa9)#ckF>lbHl*PEy}>5jqaBn7=&5%K|RpZsc>{ZfAC
z&p(#kcIKCrUNl@b71arNT4=+3)u5(-mA6*;2F6DHuWWbb-u9DucRHnQM#3(yWr;?W
zj4ndH`wG@vF-W;MO?MtoU!00j!P2<jPA^p+tvw`ZUwdJH!DOa+vz~msZc+d6p-%J`
zQ|Ge!3Jw=BS=OIQyi6<a+~mr6fBWTO@A;D>=X?-oNwVDcyjXRrtP?9s9s5gW-k&mQ
z!3WP9EzrJea#}&^=Z-tqa!v{>vRQ>PEK``pwc^grw3YTOZ?*p_a!>wabYf9Qa_%Jk
znX8_DTJu&Q<4;2Qf{)H7z2U_Nk40PSsm^cQ=NI`k#lJqd%x~QhtJiza$=m<AvoT=O
zN0mP_bi#IS`1kpFDBr(l5gW5sOj~Rj6LR}_f#URE>wauIU6lA_xv2HZujMm~ZZE8B
zwfl7|?2+8ug`zurb)Ac1-ODOIov3<wabbjH+msK+C31XhCceBXXI&mnpUg1pxm?Gz
zy>eT8LcVW{NaX2z@}~Zw!Mdn<k0tf`ckZlzXv!?~NY86opqMY?OLsr9*7%MY`@X*W
z?#|t{R9f%cwwuPfcfL#f(PSyGkkRyTcpb&ZmHqBYi`=O{<=+}->=&GsC28xk?B$&K
z(_ZY;ug%EGl;yeG_BwLki_dRDWL>V+-&(w8{evgL{Bv}ycJ4O%(x+;By#9u*%HDLz
zt~Z}r9`h$HtFxM7b6<I1x%S!euDqkBsn@SmP7>x<Jj>4YdH3Q|x92Lkh=ktXpfj6U
zv7LkA&Q^Bwh!+hn=iO0QT>nU|xti%@i~5@7mD3)382R71ep!V-UF%w7*xt>e7wUff
zc>M3+w_tvyzE96hUMp_2`zBev&%-#P-uA4mm8^yNWsQ@Mk6WCwFjzNZpK7;OIFqmZ
zPE&!$ho@Zm+t|8SZ}zlDQZ|Y!;<nuhSiqY&>G$+hxvv6Wd7`}KF7v!|C}N!xdUm>q
zq~i;bBd7oB-d%7b#l-dVk4e%_+pipy*u&Sx@WKC<a*52b^JkxE&RUoDH7bzpuGOrk
zozKqIo9}Kr^nab~Kb}Lr;hblT9<|5Fw;NaW7)@EcQ8(_<dCwE7-b~S(c-O6(=KSHv
z@!N{l6XZ7=M4o#rcjZod<fZ3*fAi%Yp4-1T$zm3Z&vD_49rLZWnb|AKceP*0Is08@
zes*lL5$na4tZynhj3uJ;Sa^cNLM5+mywbk@-oZG*eMjnBRO>7G<aSK#xG^bw^_5lf
z9!g2Nv)6xe+xx3gMQKtv$0F&aYeH@AGskr_a<^W(Bp1$k>&e|G&Y@hNvy=lraRjn*
zz3n@slNfOALe8T|lbYjGMdp<>)ION0{7~fW1hbEKCfRkZ*w*<xDlhG6Z&jb!LF3A$
z2Z}e_EbYp?<h@+Dv))TLc29|W;yuR}U-w0cuNS38+UyQow5nRGWUaKvntP5OD$~6b
ze%x3P;c(pHqh4U@rU}luOf?H*8n_q)Yv*zKe|)a~L7}Rq@~HBMFANVNo?KjG9l@`8
ziba_9&yF6m;QoV0m+QpV);Z1D#q?d_^1ACg^5Q<5@?8{`&k2uMC#P3`(W_Ey)yl@@
zuVZF=t<#pAvB7)N-m;CgUuP|iS`syF%{@KcXNlAN_8n05_w>u*2$xUX<F`2Sk7%A5
zC%=BiREe(=DWy#fNqkHJ4EZ|Fat0pDo<*%XHTj}O_O6A&ldt(aSh(V~mGzvr?eh0p
z1y)T>5m3Cc%qpw$lMBaHFK+Gp1@%+SS_LXwm*$+hpy_${Pg^WkP<@H+fx3TOnf6s}
zCI8}XX{^rc3==-S)cQ<HC97w+m8-@k<;r=-4omc{u*@i4mKiO*N!RJ@wL{D$yRXJ^
z8JJ7z$f~z*X9^O3$~582#eYiXt7AHUXjUEj`?WXV?1GiGOSXBO`THd4^z_+hI3;ZA
zf8`4L7Hu#%u&-OAIk6#%@7pnt#Iq|c8)w_xJ<OEwy=VPLlUp5&+PWT|<hvdl{MC25
zyW^&lo~(R3cGXEt6yAA6#K7-tpK3<Fi<()?4K|xUMt<UR+$_(Di3G*$z7XK;!FT+?
z!(+NK(p@wBCe}Y_Sf=%}E2+X}<Fm=_X%BWgO{sq{FE)R<;J>9C&Pp5HWlCGM-b3t@
z!#TOM_i=KYDzt8t*NaShnOtY-`JBJdDYNPrzueh?Tm{$4m^+i%&&SSiKV)-8-Cs3`
zsfATFui0}U&pLDGJ4`)B3BD&E7|!vR>ybZnw28SPb5pdytG;KaHXYjhyW*3=9@mfC
zpL2BlozqZM-##J8mwQI|i6?gTNzvQoRAU9q_Ii}_n4aB|cvZh7@mzeRnDmt$kBsIq
zKmVyZlVQGP=kzOX>uMh?dD-}C4_9E@_N|O>QxpyMY_6DjT=+#|PF2C%^AVHZIGEe^
zAKz|Qs~5X9RDt9EA*r_a3>u#`Je}@T<nO$5O3#q#Hm}XPsMGbz&hKZ=3>KWCP#F64
z#f$D~xe2{1V|J?K=E<@OUA6965n}m4p(T>#x0Cy3$qWBBFei!l&3hG~asI?Lle+?j
z2MuaYEfc)B+@actJN$H(Vz6Q9vF|ftbF!yQxF1uP;BFk_epN-zze!eS^}UzEI*)J6
z49t2f?Xh8v?!UPF=C<b4`mKBWK1pZQ>bJ9;KB3(*MQXzO58PLNU1-vI`CVyAug^LE
zy0@iXJl!|7d=m4ntav@GQsi23>Wvw5j$290`t9bxtbG3#|8)DiYsLEyUz>3|q5F``
zHyzdnkLiKV<$@f6>?h_QlKGe`c6QNv2Py9U$S!A}ZWF&xr@wsM#9<V+ovo#wU$Y`N
za*mDR?7a`(F|0Ly()s^H<CBg}r#8<%$@co=lmyOdiNe!1(Y_B}nH#8Va5UW1&Jp#y
zyI?1aQ}HZ2<!YuA=dLbl>Eb&1`GwQbu$0puE4Fc+XX-fckBRHqL)VJBE_MIVAeJjr
z58S)b|1q88X(B`L!S4;99vCKm=ijhx(!%<+j+VU-)f9ZXZr#}6{qQzNtCsildm>Wb
zKV1*C=>9zY)me@CF}2$#%By_qHk+&Q+o#6NsNc)9B=U3W+~xaU*@b8w|0k@cnz7=_
zw7V)>@3jf83|YLiT2rW>cWPGG6HzArwb{Zl%eN*d-d$-QnpW;Qb*KBfMYct43$B;m
zev-|+xt`x=UHY}AlToV}52nms8Jy4FZg}(37UBG6Gegg~n+cPIr+d#?>doI2EN~=;
zZMo5Q&d@Es9`393q9%#|u(=g&9umCd)WuhG?PfPVozvdM%5+zdNpaF~AE`SD{K<06
z)46|q%J7$!Tla;*HIk#$=;r2`2j=k~5Po^C>q^6w?S+Q*oU8A!PvbP)Cb@d^{yRIG
z1(p_ma*J@AD3B1^zT#2PE}=Ygf8Ea+jJgNgV#B{wEOKD?_^kaPZ_>G_(;AUZrd)0e
z_4^hia($JuW3XdV+NX5p{sQ+0OC@{a9_8L>zp3?T;`3PxUo|fJSn+n7+Rhh_HJsKP
zwsxy0?werwaQb~yAG`XbH&0(ix3wN;HGA-sO|JigO(n<9ay>pTv3>tky$W`&nXMv|
zF!3Oh%;eRIoKt7{L>I3ql{)nDV3(f%gJT*j32fz?Z@C0#uHZSZSvLL2?jP^}EW5$`
zxM7=~)h-ow8@UgapVqP8oLYEAQdVV^me#H#S1-S|aSY1&I<@>0gW|H)IrWDw8+r&c
zR{t{VPp<m4puzC%={^B7W}bD=SMEEqS$=tE+oL(N=j_@UuN!bhOnv#sIWnr61~1w7
zp6xj6B-s90Qt$0<`#7e{Y$tYSzFshQQE<c}A)#}Z<XnYxgCE~=Tl;9?q<5YFt{N`c
ztsBZMf4|T|`$h7Ri95RGRBs&e48GP|QGc=i$=9TqM|x&{4?8EFmY4bD`RSa}yeAn~
z6OAgpjCbT|7O1#iPxL%-J^MkqjuNBall=V$zy8&o=5nLnqCT>L^DldmVf)h>?;Ts-
z>(6_f`LASc=dG!GPW=DSQnqh>s`I`eaZ4$qdF}I+Qw&oxS|2Q)kbLR2&AEpR8BI@4
zOIRh;_sP~QU%Bn+g0>ZR-*sNoInQBo)}8YQ@0u|A=Z|*xJNCX`aX(8j?h)G+i-t!V
zM0b7^+1Ye+>*EjmK856MxLml`ZrO^L>3jFL{PVrJ@zDd_n_;Dq&%KnFX8CQij;xhg
z*mlC7W$G^9#SzO+hOVk!vx24c^s>01pvvZ5smWVR4qabSA8jVE@9PQ|<wD83uWmd$
z!YklmR6m<x{hJ-WE9Nf}6$zf%u=DR?qwV!uroELv*=8;J;=w{cC&RA?R$OPkqBm{j
z@}lKF>t~1ZbJQQm6Yb47p0@eLdy5L`ZI7J4sT}{rX%}Fud_%=*&fyIV>jLwtZyUOV
zXWf6lbkU4|I$Z$_r<tzR|5y~eE?~w3zO4^t9{;i3oXO<WhbQ+QIJdW)*S>L7ICZ~I
zxa5}Pk8Mtqo@p=(Oq)1C%R3>e;OznLb#K_a&Q0d4T+(0?=FQu9QE1AQ7mP2X4hBEv
zb#s`tr|H?XZEF{na!e{Mjy#u88mQ^H^Ds|Xw(h|!-;kJa!%C|LIZxe;D+%?bdu!Hy
zVzZ678#J|Zi!tlzqnYa2F=ec)uitnhx$fGGhgrRn>#ju|x_aQrg%tj&ePs)$zAXxU
zudpR*o8HYEvlSgOMbBmzyiqdioyB!+-_;J~?VXv<p{YwPCa)GNn6~h(`}M==vTf63
zLJfJ=?aIAup0oO$#GIrJO+F&~_vA`c+17K<yIT@?T}FOd-1oL;FB$U!N_NR_sqvrk
zq(QxRu2=PP2_CKHg3}91Ggs6(zJ6qX%{yl0*~W!d`;L@09FIKJx=rBwFYcI=r`Nqb
zt!B0LTd7l*^{2bsIlET;+SNHp>wH{D5$lm^NB6>plos}ueM~RrxOwb}HrVX+C|e`u
zqk<oQeaO{kw;m+8pW=0%lBM~gR{81ct>#QKUmn&M<jIWpujj5m_vpMs(@G_)`yZ-)
zIjxhB`ZQNNFkg8Cv-}TU@iQy79~A#;<{g-S*C{Yxaj#!s%52epW-}hX|MKhk_kZUr
zeC01*|7%$wqPHOM-%t1Vzq{>^9_ZfjX_LUT-gwW1*zX_eeJfU5mH+7A6|QFFdmtzo
z$fVZh?#A!EY|U(|f;k5^&z>oo{w#4rUPjw1x1eoWE|)XB#V&^|h>Pi5{hc>hK;ZbY
zeg%t&p18$FUxda^>{;aQvR=aW)8(k7FZUKm?9HfHl>DQ)w5FRswsOxKXXO=FN|vcx
z?>kw_@pxmC;TMV8=z4};xyAXJbC(z=eR;}jFVpjqV{I;9?PI>f2FWj_jHj*Je&kZ9
zL9u>8g^;I5nJnwVtJW6T0`a%jR65+a^RRoC{jvRx@dsy2556=dJ$P4(^5d-;MZH&L
zT|aF-d((f36?goR45rO2O)NFGzNg;u6&Np2TE~<wUE8}!{Gf<*`pV$?ujRRNjmZ&d
z$+v~$a~Km}-#WEkIpJ_*5aT)7$60*`e5U5io|$%fQgF)g_0BOx#z&6E@cDaotNos(
zQ<Q3DJxOcfJ@wQN^A}~isCC8LWfXW8s+|1hvu@IHEfatKg5Uglnu{K>?6PHuS}q*}
zN=OfaGFKdCpWe{CSfeKEz0K45L+6^jcP;L2d{=z6OrAYx_nMol4iwlYsy~RRo_73d
z+oalpQ=I{k`!`$`WiL@V8tOBBwbX<NfzYDl<((%tR0o-EU!2msrNv|(_r*!9nmJA%
zYiuLy>J!roEtCaZzk5$u>c@F6)6Gstrr{2YUCBj-d!d?l`7d>6I{)tw{9&-0xn902
zjJL+)+SGHxKQ?@m{b5keVs|F!fwSJ7Sy>O7{lc>zRIA1JMfa{dRp<AVPoDGYB8ial
zO><P8nv6Cr@^WQkpFC$9PtA+1UfZp%Z`#kjSf(g@%HnssKW0e9ntoAupqa3ElU;OF
z&<D3Zi{&jACg(rpo`0^DaxgJz)rOnN9nQ<^zf~Pu+b@}arMRV_WARc$?MVlw2)J)u
zZeDUzWE)fC^^-Ff@A=3(cc$`^>rNI+t#y6pryif;DmwLNncLw;PmZOr!n3bhN8Eno
zJ)4!kBmYj&Bk35siv3=$4N4I!Pdv)oA$8Pw?;&e3rsv$w28q*-R%B^BTxw}@@Xya@
zOEhNXd#lxl7X|fZS@qrv3%WJ&A=`R>4xWOoRYJM<i*4l;ZKupH<6wzXpTGX()U?@3
zhh8bn$|!u!&(g=pXZ>saY3t{&K0RdZODlN#YMz$+)tIbx0u$C}cBy`>FkZgvm6`es
z?YafheTfU)LedosBKUOsFHO)rUDi-y^Tq4?{*UXTW~s9Z)!PU+J*<zazFjeWY3=X%
zVf_CuYyF<@%E-|Fe`Uv_C$E1oyxpi;_$FO^Zb7jB^oe#JhZ{2_Th8q>&z+a?SY_R`
zhrg}Y&Jyxi*3ETz*9WatU5|rWE5Gs`(%yOGh;}`1m*0BRjF-kMjI(%GJ^zytJMVri
z55LR9b<!7Srn3BaqiOZ0p7$!3j_};N5}qHQD_&=a@&4Fuvh-56CJ*bM)$Vf`ew1IW
zT4Q)&^R>lw93e0LCg!+YufLG}$a9%o7uTjQVe6~*zK9Jex$`(>ZXMtDqHk*(L(ZDj
zoPNCVTK0-Jo1)n_-17Xnf7UatMN%>`_hjz>4>dV(C+~o;_|i)(xBuIUSoFQD|L-*^
zSCY%b@%*&KY(MsIdd;v~{N?82*st6Etxwe0Bwdi3F|Xl3!q5FN+MyxtMQ<6I9!Y2~
zy~bj&>6drqPn%o8VLErO%g%EX+slwRW6|P=uP)qWXA(K>61jG%gM;9Iy<C^oZVIQC
z6&>VaIk0&4E+$5|IXO)XD_DKY1srY`3oz85n{Cd)u({scSo_<&iYA6Cg`fL)jsJdM
z;&gLKp;aQoFM--7h91QkGZ*j4&bfPGS^efhi>lLG*q{FOUX)?U7#^@Oc$&e=#qGx4
z=cnesJl(x1QsZ3CTMniNCU0%R_thC4dd8Rc;Y>o1XOOS{gU&0t&yVrlFw9$Sb~~Vk
z`Bd7o2ld}u^Y<{OJ(LS>JfHD@)BQO$dnC?>rLNDIbah(8{p3fo^L1Wc|HC$SQPZWL
z(>IHh<*5J7T5$Wv$@uT9A`b~PKHm3w;|z1Bm^pJA7leN~@9y3!_V`c%OKpmchcnwX
z7XSa78cmm+EZ?pzyrA!*tVN@zt4vT}x$a?c!RFqNovZe3sMk>rcvH*v(9gs!F=JQq
zi(?H_`#*9<nC4U$CGeX0o;R_3$hh!)dCk4-UVnz^=gLagO@8lZVHX(M{$=W_S#Kh=
z!r6ngHr-tLbjs7{ws*fQ7Oy)R9&h~mz?CA^340md{|kz#x+!q+-hH+0GTY{@-uY!$
zV&|mCn~G0A)Q;8Lo3-d({l@I=agU2$ZLPKb8pOXmPRsi1(KhAl=k^J`Xe(^V@ZasN
ze{xfzNU2mdclW8-*mLEnEBf~2Cw>3=wMg~s*3$OnA%0h{9#}A+F?471X5|m6YCJ#v
zUmKsY{55;d>uE}bHdA-4&e>+teRQt7^tw$UwoxAG-Y3}pO}N4IN4BJ*wV0v)kU{6A
z(9P;P|1^Z>@3n8sotYt|JEJtSNBpw;zqYO2a?IjUlUqbTXl~xq9W!6bIqAsPL-}5B
z8be|O-We3%FZ}Os_5Q@Rg^cMP%Qqcm?$OEBy6e5?arCO@=>Bz;?KXE<_xAf=Xm?#+
zyUx~m$5r0)o%OLtkDlC{J)P~>In6ug=GCv$RQxB!If3b-=GL_h+g<PK%qrq>H2nVH
zOWX0DgOw+jePR6NQ5yA1cuM5785Q4{0ycJRP><X3{Gn1!M|z>0``zGDrn81uwyiuQ
z|FK0>KyUHBg~b{oH;;WYQ2IFIT$*Oc*A1}-cE2t2^5c1z>F;*^-zNHHwfv$#mR&**
z70(um)Tg)dFLhF%IazN(uJWmMi@e_2F1!`@+(LHx*R8WvXV^s8>nr^9`)te`SiW;s
z#^$4KZ%>PzYQ53HTf|vA;r2WIm>aX0y-V3%oZs{{KC&+`KJ`PAw%P8NV&68ixlCKr
z@UZLLDb*)!tJ61?y_%{L7<TQ*CO(0@lb(0tS=#PQn}29h{k~iQ??90&r?PL<i27gr
zyGYn$x9_2ksxr4?yYFvRWjJqMt00(uXU3f7%`V6JbHX3q+VpYhfyMLZ6tK;4O}qbT
zW?SF74h#2*GH)Ij#Qk6P_UpOz{0mlHn&EtNm&iH&O}`sXP8Pe(o*RGN)b*Oj<bSh&
zTR*#}bl-<vXnhv@tklm>>bHBY%KMq_y@&bgq4m4NxLaR-{2j_u;&i3t;&qvIIZ3gh
zwIw2Q`&ydCwsxnx%}x1MyYln&vY(qIr%bm}znG&|bMDgBB!O5b!5w#dI1a`X2J{G4
zY+GAwx@t}2mXyigzGpnEySde7_unVG-fvTwsr!`syEuR1W2<RvCT-S`XfBvrU#xs;
z)1p@|TGr|v;^VpKsGeb(abpTEue|8yWr;<do`?UKy0QIbnKJF4Pyu(NknGx{$F%Y<
zte&(aOIUdB`wj1I8Bghu{=8#yRq};PbEjFUo(xw#98}p7oN#j5X1k>udJeymZE?PI
ze2G)llsm=}E+-S`dPwh-TJ<Qx^`H0RX4UHY62CANjx70b^AdNRXAagMTTBy=sU(!J
zUpSJgytC);&3Thr#pPTJMAXZcoKEFAU8eT(kD_0IM>glt^vmpb`4|fR9+-Ng>f4VK
z`fmbt7OT7$p4B7#ub|4b=j6v**HUgCGF|eGYj=gM7*Fwogk@28^D@^sTq{{yx=H?U
z`!>Ov$olIt5!24fFiwzAsrT$iZ2Dbd+Bt2ugjPv_wcp&^KVJMi%lbYf&u*GA%R{zz
zdHW7BZg{)k>f6QIi(Yjd)e9C|7(V&h;_q4wiC$NK-H6tEua;9SbmQ)<n|g1r6=hgA
z2x+P@I7Khh-lI6twLefyb#g~e@t#cnlO^Z2W*aZ&n*PMKo@4sy5~0P1`h4{?XTB6o
z*;!_K(y`U!@s{Xo)AFKH+Edp`ZP4uvIaHRQuPI{iXj5%?uDZ>+y<rMGQ8OFr&1Bv*
z{gf0tq*u6;|D3;?<IYoZk(Kpp-mY$2TP;_pBoY6&`={HF3Vn@dQ<YxT=d!Ly+~M`w
zPTJ1SnCV|Ai!j6VL-oPGbj~P>GT(7ijEFTpvz<RTdSZ3x+Hce3_6yqDas=@2(QBR_
zVHoPF)7h09{Mh{J@xpayj`6*kn!DAq{I}h_`|E}Nggg#-x8FW?mwlmXO6hHRk39iv
z)r~Aam72WQ*!ia<{+~h3>cdTW`cpLz-`&P@U%z12<X4inHc9j++vjlCA8gT|zVyO2
zZHH;60v?7GIQ}zGw&&O<msZc_Y;!nlz1pgI2Tzt<zVGCd`+!mX?6SaZF7->hnSL6q
zFBCgyKCS-$Kc>nXDW^`b&RMcqL`Ns)sixxX7lOY|HQ%fC_-r$mzt*}rvfeqqPnhBE
zi@Rm(|7YHKEzlVH!T!uW=KcF5e5cgce7Ti8NB8q?5e31ejSKrOzuter*L?2UXPz(X
zo;pq1SKJ_@KH1}vy=eW~=Nr_f$#1IQs{hMV*JQ}zVl}DL|L4lsu#~A!+CueCS4}hu
z+pBZhHDJ@1xqF-*gaq#KtI<7fY!ZFlceC=mkAmHw9D-{%tv5aRfAUj5j_d_rIZdxM
ze6CmFzqZO=#IvY$^70Ea)<u<gPF%q#cG1{o@vVf;7^lq(R*CHn-d?jq-0q0M!)XiS
z>NxfP|DColuFS?LbW#w*w)XR%JhD`hQ+=muwfnuxVAFM)zbyJ2SLqsEf!6ICSRdu*
zZ3~!l;FHgM^~w9yqXni+xw$C&|KmS57I|dGg}WTeseg7$%|`y#gg;N+W<TucQZD{F
z>usl5*TR~VErIF#*kf*G^Uqh!{AS(U&SUG<BG;uauu96wFCf2o^5nFNWf?Q7ConB{
z_p4qYV?OurnozS@5BUG+_~wTtdK%B+xq892V%lLr9`&1|re|H2^6WO7qt%%+XUoHR
zlDRjSXKbI(bv?BH<n~hGcUg0oKF^bG%6tDNYOl@mmbXhcxZb^9E4}rs?8Gbbp3a3=
zyt!X>y|ps4TK-ui=V9^Ti~gB%oNJ@s$1R9fdYQKAa{lH+53J|(1ZZz=-KZpP*Rvti
zC2HFYU8eZkEw1h?Ww#8~Vpe}$^Fi)U<NQ@V({67O{OV9Hd&BGXhh_D>FJ5=)6_+lT
z@tv&}YQd~hWPjQ1l8y2P0k862g+0!Hb_upz(qdOwIX_c5vhDFKC4c)KK~IsbKg&A0
zFPz^L7~?Liy3TFW^W55ZA*L_;vVN?)Inimi{M99^=B2Ov{Xx7=eAUUjyR(EugG)_1
zmtRRK5zAQh@}Iye&&i>_F3WZM>OEfdFY#aT`^CMjliz6eEk9**wxBABQ?7V%nDvuA
zLHlRVT*Y_Y=GR@_&41_qy6d20H7!*x)!T7uh<8}vxg!~OzNA;=H@Lh@zjI<=M&}0c
zH!IBB6<!EPGHiPMW7m<=f<qmP9#`C5^<>4o5RQ8nUq1`+nU{C7@#;dE4M|)gS4%yN
z>WyTRd;UbucwDicV=|M~(MBE}=Br<I`8hA>xrEJJ&YXV4CikGEm%@GV=)je^@_X9M
z+E(fA;{RS*_VxF@No*I(Qde&ODO9$;dHGQnPOXK`x);`Kd|K{Tu`5@j?ZpQPuOn-g
z3CJDg)Vld#{frM)H-1+yXzY9R@bAXknTAsTGCVKUzpdCh|Big3s#upg=lQ~aQ`QwR
zvg;-6-oGgA7X!z7qY_RQ(>`mKiJwb4F6bT)xqf!;@7e0p)}LkzeH4;<QPV@_-o#eJ
zTLulDWtN=FzBrZa$jsmCf3wu+iqL_QBJ){+r{)@NvD2}?&2%ckS#|yoe-*CekWKqF
z6W4mF>j*xW+FO5=ul|A6mRrvj=APPhy7V4ntlxpQTLtr5j;p9mzo5gxHlsr#tj<%s
ztt-=W(xQ(!&#(B@EDD`Db-76Su9S*Ptq%pY`~xearuZz@nRTyizOs7vA^nU$*P|!j
z%2^aDw>4<*oy4=H*E{8QpDR7xC#5?%cmA_wR|Qr(e)C;9+xKZ>Q+@Ews~d$bOxL(&
zJ?(f{oZRW;-KUdFdFS5<)b9_yF1zyv>-rsYoeMd;8Lm$YEy+u?WP0&qU;1y^NEwT1
zs_dJVPw<`Q<dXLO!c7g<7b!J740EDBve>?Ay{2~M2Gc$_-h9nPx>{KWiYK`JJ~f+b
z-zSSpwT<jY&a4!fJaOjxAH_bN^`3qFeao7n-#u^@tl|jQ<!}o*)thi}vW@?%QyXOk
z?=T-coV~9iwf&vsw8slSPc2un_DPoC<b2KKqG;+mlND<{S6^{!U+~*oM7NGf-nFB-
zeoy~o4!O+sP?4O=pKE08o(cYuaC@&mnMp;sVZX*=;p822jubw;Ww?ShrfyT?sx|eM
z4M#ux`FCI7(^~%xmclnqP0hCBt!JF{MZYxqt%R|()Q4MdOW6IsJTp%W_!zH$Q7?Pn
z%TsLE<sP4?Kg`t1K9_rmxR>fxztHZq&P}P&vVwDyXD@AABfU<w=<3#a#+mb1@HN&S
z=WGc7cri9Z^jV_k;;@I;wVGe8tI8Dh+{UA9P`~5Xhu)~CbGCSy_U#N_o0PHd#i|;~
zrN@gSOjHG!L)VFajo9`k+U``Mtk<n=D`r-xDQCX-{!tsPc>ibVFU2$SALx6|*ehcH
z@bKr?Cxg|_d^G>}_5Oc#nT$Hy?&t3>dlpts^{M~Yb%1l;Ib$LH6rE$d&-oMnG5tx%
z^qqP-dey@1^|Leg=&n!MS=C@)X0%)J^DO3$<M+J_f&{)~sO4Y39JNl&eg54=mgXia
zIeb6N^AY)@vZS!+#Jw}TzCTwSm~*|cb72LSi-)Qs)4s*$-&u3r;gFxa^K5tf^q<~}
zEn*oGMb9fH|KI<Aw_UYT)cn=ID|fAy{`Kt7HsxCfo4stRBM;81e|~b?+UrT$v+}b0
z*5towy2y2Jd;G5a%N3_xE;dJhIwj5M>$mp&I-f~LWNbS*C;#+Pnw|cT|4&J(=%3rF
zZVPTryCkS<Rp#TpwbnCuui-LRW!ZmwmIikV>+ZW5yoBSQFXtiNKH0bF({{73Fj#SH
z<Mgs{)nF0bx!Qj#ue^IM_&BRR@Cny$t=cW_+d|&o`gdj58;xb@9WhrMj-Oc|H1~kO
z_X!Ej|L1n;++1p;>*B82A?~O4gneHi*TcX2|Fhkf71))j{_n4A$+hkmvKL-jKVGQ!
zU;j%5t0PlYhfVdqhh1iQ(rzDin-=Fhv=4Rtb%NLZQJcQ9xlEsQ+g84};(IP1Jy2iJ
zeE7+pzJt3jMm$sfn|1ZFUvH}C$EOZ&tscL6acoom&#siK7Qulox38^Td+62TjW)B`
z7jLwYI=#ME+Hc0TeQPINo6)}cyGX}6n>m5HGC{f~Lj7J@j0>Nwxjf;Uqqn+`iS=^6
zD7DZ|t7Ch#*4^CtEHt#yDk^C=-xV2!(_WulZ`M!Dy83)l&aFP>wf&)T%%;f;o~>w{
zCDm@V?uGOz&jVitZiX(>tz1>|O0;>=E2~wn4x1m#y)kc6a*s~$#HH8s&-YARD*U|7
zto_%OoeyV!Hh%woewt?1r1xx$*CXW@SK80|?r!wHa!qQ|jWELn(YJnWXezz5NY+Bl
zZr8(>wW*pHbnEY~4|nQb_HgIAj$?^y_uMu3mU{QYnP!8;PLD%3pNA^_nefd2W6NxV
zeM_eQ|JipeBllN5cWV61CvCI*V;&ZtuUhfw&6&McCnG;+{WjZpag)g`m6SD^e=`<K
zm)7$oNT2kRSN^$X=98A(Gll!V9*9=n`#st{zQBIp1I<gXUf+8wyS={b<`&Dmx9^h5
zYPXlQcdWR`CzHi4m+(1!o9xZcN23JSX=Hb;y0)uuYxRBJ<6D0HSSTAT_Rh}EwU}XA
z<(wrm{nwcaSV*L|e($<))pn)TRh~5_&&w6r*qh6k<$lC&{`b?Ub8nw-Y<yVruZ6RA
zrDV=p=+O77$7AC6j|Wn$*kAQ2)_b-sT($LY^wCc$HxII}Hxs{^@aL86p&zeB-*$d1
zm%3uR|As?!#)961{VTT3<c-(XWB<p$--wO*Si%9uX2#?-6*t&EdY=0=TkUR0GVAvr
z`JY}pyt4E-l3=^nFyq5-E~z!0-&PbaF22-e=3Ss$de7!w&1%Iy;q4kbdKXO=*<{mL
zKf~}9ldabAVBX`a-v4>MFKLQjm%Xcf_IbgkN#12U`khjg<L|Fh*yplky}H@K``=IV
z*xa=I8pfs+(Q{ieOmfm4GqpL73y+K3p1b<w#>*GysaX4ly;GW=_N2}5;hcBey`0ZK
z9E`GE`TBO7M31^sUh}F+iZ(MczF9dg%doxMQk-1he{EArL;GLh{5pNjw#Ak~XQQt9
ziu>5F*frPIa<YZH`k|TC+ooNNU%qOx#J6)hf=gdnhc2FIxsCG?>yI018OyJH^eXba
zE-e2&wrq)Myyup~vdyx|8<Kg-LTvWQvRd=RPv;F>lRr6P*{W=|k}by)j!V3;I8>5-
zuurP}-??u_y!Gy8CV{UzJl@$^DR(C9Y166P_ixV8%fDS`tYOa)H{_aNu#fq%vF+oP
z>=V=qjaaXzT~z+jwQ`}zNA-rX4;uff(+{~m@OU5b?{4ei`76HMyw9a@KS6C*TH)WH
zYc*;V>KA9uJ38g1)uYG3Dk`ggI#1zzx!9mH&G=#Qv~^;q&5knq);}_;s=qmXk|mQv
zi-lkj!<4648$CAL%;K6{@=pM?6uDDC|6p%>{31`iCo4ak6jy@`eEchN4HWFXq80u<
zMSaE6u-;2W{FgrIcnEO2&D4}o%wzVSC{roY?Jm;2eSXaMy9_&jq)zE}&<j&^6Yp8I
z$YR|RK8yZrk)HM82K8Et6Ko2r4bN<{IQ5Ie^_bwxxzF{x51c)6;#~frlwEJxTaTLW
z7k{`%<PgujTTV8wcPY*^%KBZd78$(wjNPO&Veu=^EED=7+1(?*@yLvp(;WYpR9H4V
zyrd%Z&VH}{lV8#m>Mtjq`OU_f&75bkP_J*#qS$p$gM{W?zM7+DGv&yC(fW3u>9XG9
zwdd_7`J{8c+dT2)b8ln!8@jjj`7_pc%#V+rFvm0h0N31h{eT?bOHT~{?K#zO^OwwX
z|Mda)O=@~SyJ{U~tc-ZdG52B4OfLV2;Te%4dTd`Tr+t*kXgBD*bhYT4lFnsOgA3^w
zi#UCH1ZF86_;sL8b-MGS7wfWEo=MKFmo|9k+IK2npR2i2;O2@ye&IjwPCv4+pgMi`
z`wg!STl3yuCi&y7+#j~Oo%?tH=R1DA`-a~v22L*}yA$=^@(1TT+=#JjaX7W(n{=0)
z?c6n|)NYioF6^IBw$t9LNwarudPAZ9%MZo>umAsH-}`d&uXS_wORaR>=swRndfx53
z3zgmLAMQN%#eJT1<-04DdF5xN`wtcOzTBTSuet5niE`!pva{dst6bZ|9vwL^(V|IE
zZjM{G#>_)IBE0^D34L~z<cgTH!APXV{Q>LY1NU>p&bHWaPkw3>r+2VATvT9hYEE(H
zyv&yyZmrY~I$?Kwf{=*L)TACqN1KD6Q@4eSi=>D}^wv8bOkTItJ>bK?ck`r#WRe^A
zXGiDuc=r@`PJHh6KLf&g?>GJZu^mSwre{>?Y&~#rr_<U^d$xaRvhz2I4P337q4}We
zsmJTMIUEXaoK;pVFM0PspmT0Z?6>CD*!j11&J^8qvb5#-hDxU{2S(w|TN#dsiUxl@
z-~21y_hOuB{cVn;!V40dKepsgzi%j<m(5fDYZs@-lj}?7da+t1>N&_9>y8Z3b#LEV
zyObrXBm9SyL<-}Sus>TX_B0jz5DJX2>`|J?aLJ^ure~AJN$vSBq#dPvUtBg^$X_w*
z_{lXt?azI(51IBis$hGed_~JGLtVx;o7Ye0ykDeVAnX`@vb%m+;kM(;4q7X(JGN$L
z&Fj0IZHM*we%y=ZXb^Td_OsyE*{cH0uO0RYFK6HVxA4j0MK_eI&b2JNHRZNO9bZny
z784sw?@CtZ-r2tOZs|rIjkBNe9>~9K(YU#4{`8BD*7fXpC7f>4A2{e8%vLbnbDi0|
zN^7=Fxag;9@B3#BY`n{J>URWRnCcQ`o3rMz$D5$lpA9=j{6E-qW*4k_@ldCG_A$ms
z%lTfVJ$wE#xM*XpNrgV!v0L4Fbr1Ily-9y}c}}93WAv$4kx6^=g&us8lYh-t|L*FZ
zoBfHKXPW(;u>E3)^`ez-=USY<wBA$DpO>idefBS}i<>wY=l%aR<Ky~VCL?vNdhYL&
zvst&g?~a?2dUNtQm8VbGR{joO!Z4ff-_FO+|CK&me)Qg+Rck)$?iHN9p81Jq>G_E3
zCFiGI2rPA$-WcfjR_<<U(vCO#j7^g3YJ42ed06|G=e9=&xW}KaGZNdrsP}i-!~CeO
zWh<X6{&_hi=-cz%Jd=&%o@FuiZo1y*BgvjrAH~1LBJp*~;xqBqi#BW&dc<?x|2^yT
z^V`>V+_VbHyFD%Jd+hD-9rFIuw@vnp*RcA&bka&wv6pj2qsk+el;4{2%k9OID~8W1
z*UWg{d4BHnMaOI#C)yvgZ9F%%ubSz@|NBShF`Q;%n^C7x_by=97OpA#j@?bmvz{o|
zoX&bO%}<c8enP&0K)==I9VgH5XMd2bE_>tqch@~T^Q=4Vv$uXXeG?}8e$K*Kx5`r1
zPu%NQa!6VKZ(e2Q<J3d({xcWzJqYOYDedwJEeq{8-)`nETzv6%^3TNRA2Ci}UTe)c
zc-Mw=p1!Jif1Ldjr?NKz@01(g%qcq^7vFSghC|t{AClka%=E6m7JK)?`6pX`M=7^&
z*SO6e*Z7U;N8|dZ{Kh{@-gdCYPn~<9EwW~A`^H&atMiH&PP^-byn9d_{!w*H+beO?
zlOI#hms##x@hU6f#1{Qon-*qnWpNJu&LpeYqjE#{-RJ3%Ar4<+MGhq_^+|A_SLtzC
z|FgmGqWSHSd~&@PV;&sfRH&cvUjN@W!?_-lmUQn*dT=%1zQOAJ6U&xHZtC23w1T-T
zTxY-HtYvC5g(_Pggzle|`6w-RS;^z2=TGYR?YlBhYu}5-HJ+Q6or>`)=U5u}E`oK<
ztOya_`+}m)%U4|r7Y)1@qm_7Zv&&(phKS_Au+oc1CN1-re8hgz7c=k6$Auo|PpFSb
zzUy$g=F&I4{Xc(*a5BVgZfiNbefry=1vVl5DbB|>Gt8ZSY4-o~_qMC;>{ZE(V-4HC
z%j;Co9s$M^+b;2(pH!t%+BYNgZ*opf&&if4Z<IP^CPij_<^9G|-FSEM8{cm#(t-@0
z3`dk1m=tDk&VIQ_h=q~=`+50?4<9b<<PUi$ULTyigWqiP*FSz=CUI{$KB3*N#^#nz
z%)PVWMmP5fHePsq`1{$KV~Y>pJY2GsOYBp{Gu{5hVUHHPD>dNW@piX<)!LesoD1Zy
zzF5__?ZNYn7yLNcF1Pz#fADFYV%!?3Qi<B{vE1Qz>?<;xXGyBF9bsHE#drJV-J8@c
z?6%&y>%CK>xc-Yx<?f}yIj=vy4G(y;Bf9;&|K<e(d%IGm{gv5q=Bb!M)l02AUJFhJ
z)U5g&@Xh(M`?Q^lzZtjRUHs?i+waE@X*_;WyHwwzFv0WJr}}?y6YaF^PT6@M`f4@7
zJ~nsi&Pz-01nJM3_hZpsU(4Qw8(o9aGXDjiN)yg1@d-aZP30>KcfEt~yNc~MTQ>g;
zG0B}+?04}%YI^ik)`C@vfpr&i=k;rEvSig?ZQy+MNzwLgpF<D*mietwms`B&=D!l-
z2^)=<{<^wqVW*GlmCYH~AGKEt8h*RKr=<H+ty6NGU;Liy%igU0pSb5*p1PD--LZy$
z={FDSFdN<0s*c!FsUKmh+ic2S|J$vaV}GXiLd}@CgH!tTgdIx+nN#lU-E@Qd>gz2Z
z-u?>vyYu?Q73Tbxk2-Dee6i3Y@#vdNN1jGkAAUY>bAj&01N%jO%qUJdqqP3i|NqBI
z=ajo2&P<Z(^U&#8z&$^=@MZqIcV-98!ftI*>OL3pS^x8npBrQf<Q}HJ-@9z>gnC`4
z`j<heN;k|FM<&*qbv)j+tlPKZ!rLtk_n!FmmR$cV8(Ut#VxE+vQ_kc!`l`#^Ugpp8
z>*eBb;R&AJVVQk)XS9km$A;9-hq8fl_t|Fsl8^NYm^*LU{e9i(duFPfJ{afo?DWaY
zs}y8Et#8+U9(-xWo3^;N6|Db%rsvIQE-k1x{wq~~<8{I>caz2E*qch$rAqX_oTSVZ
z7QWwi(>Gt)cS4sxtlc*IR<?agP4unQw=Od2<v-m2oZ5Z#{!i<De?K*6N*LL0d}{yi
zY5VW`V`t-j@V<W<GXFl);$+#r_ZxPe<hmyPGyKO6>4$C68(kIZHvYV>Z|^bXn9k(g
zj_D7To-CeamwUKABI}^|#P1njZtpJG%{$$<e(${c+KfN1es*zdRFvO3U!VW2Sn-j`
z@AqQzhd%V*z0C0C$@zU}uFITWIAP(BV5RE+)BjxJ6!$+J`snwS*3cj4p8c~|OA`IJ
z?{~KUVL8A3>z;pqG(Y~QdVQOiBD<jwuX(8?quh*VG1oSp(M*+3n)R{%-rX~)$<rIX
z_SEeTmzBK#<Br-ZNoLXP;Hu4=`DY}D%vG({uFp5@l2yKRY%bgS`>dWig3im%vSjYI
zPP<rNUVrqKy5y4767N4c`wTa$N+&hN&J<p4bXHHbPjk}>&C5QDZ9AhkUEc0}Ij6G2
zU(w5Fb7<yle$MwbpKJC!i%Xf8Tz@LMZmz);$=@^6?UvejwP-wCqvyhYQ%^04{jbK6
zX)P6-y<IFWvi#AXEkFB2;0^gyOC#s&jK>zsL{E8qyukE++v`U*CEHo<tbMxn_UrTW
z^X{87O!yM=ih=2hi%9p4ES3OohE3lhcN#0dD4Q?b;~S)GA!+(Xu)t<^)w}Ku?y?H?
z$2q5;&Y5JVs`gRv?(XA3YXv5^KUtSv?Ie`jGULPnYl}DUGL(<Fw5g^&O^V!`D#F;L
zu(Dg}`JsR*?+R8fd-nB^`wQXwY|BgqO#gq7*>kLSuI8N}k(jj=6MxLG>RENgrKc%a
z<j!()Uy*fjGuSpHT03YaTA%U_%scyKZJbTZhIjR+mG(Z@+b;6?i2ISPdkwN(HBYEY
zyfmu6?={6KHFf5aEcLTBH?MUCDKzivO^v-PvD74BzuLKVag(}}l73G*y7*#_wwGY`
zf%SQJa;ENF^e2l|JLc92mv5iV=h>dh|MYLw^gTju0)f}1z2aU*o;dfn;HuBttY-y0
z4-8%6jyzdct6Bea{ki9I4<<cY;+ylJq){q@N8+R4-KmRO1)f>ow8*VDoM~eg`zj#*
zh!68QHAW_%j1#_6p%r@aKLzJqzVBgke7aDD-b+Q<f0AqV>m15A_g{T;9=C$fx`@>I
z91=U@9-7bJbpOu($C;b|>QqQw^;Ov~GUH5k#NYm#?~X+(^m%r!tLHL{?EZOu>IwIE
zc{^t3HL)(Yi&vig#<EKzdz0a=6MxiL3c{FL&si>L+_+BsanSAUyeemZbG%C~=r_n?
zJZAJPzv)TU#eGk2Y|C8fv)o;!-y^T>Nu8Sglo@hGk#nv}EwpjkS*u|lTe@QUUbP#F
zZmu=<k1CE7Cd-)5(Ni~5;4Q9a`LH$2s^g*LlrM5qr&tQgUyRjM)L8#q;LXk(No&_W
zFPW@<U|*&&b4kGNDVygN?+I3${e9P#uQJPh;)M_A{@kkia;;O$mDx$c-sT!#+JZjV
z=_y~G6OnYAZJ%T4qK|=fhi7>($_LB8UX*`cW7js9>xYGeHVPb@!n>5GBF5>}nd9|c
z{V&#u=)PL0=z3@GFVXJa=l4DzvpH{ab(?bkQaelT>!}-bXEM+IaZ4^N-lfT<eQ|wO
zVAkHYwq?&9Pk(c>zv>Y(f9+!fgT;3@DrfwE%KFye(AJ>c4~5z5q;=#Ee=YeNTdZaI
z;MbKmM|8!D9e(flne|<Gq49YIv%+Iir^PN_4X&RnuDGb_W~HENRc+>d&Cb`tfz92F
z%65x(f66J2TD7xR)8=qVOo-sKIlJZNJ9P^*Sl_t)#i;4aJL8o;$L%(+{#M!}md-HI
z@XO3ZM*j_4asyp6yOyk&;awWGOX8Ex9A^=2x%urAAI>@7@Z2+POV9jEqO;yDdpNnc
zce|Czca7fN^^?CT{!coYK4n3wT}_Ev__?c*5*LGog<mKz-oLn>r)uBA*9SiQVSBw|
z`rfx!m9Knpafua8d)?>H!@X@P_j`>qYG+z@@rN9lAzA6v^D=*l#A9#Oq&+)2b}d=5
zEI{I`Qr457XAGwb&Cun3tknEm?bC|u{k+UQwf6VaO;0UWk?3u%Pn4VO^1oxzl(<K%
zcQ^Y^)+jw&{V?d2rb$DC;qB|8+vddeI0V%Q$qP0IEo}5{%$L6vBHqr%^ZE=+gr-S?
zSek;J;vxkTgADPBjAjZ8dQvV~UvcJ<cpwn>oX1A-=oj<AssoC3HByI)J9)e&ZEZcY
zVy8*&i!*6`UMUwg85|3JQ9sL(*}&kMZoFz??aoC-yl-~RU1DB$_Uz{^{}=7@oSSjS
z>d?A-o>i}}ueK=5Vz@BttWAjz&kGy#hBAf8^B$?@1g03R64KvW(0GFN=9iOVnvEU(
zFRtyqca7`JicBfHkk#=a9B01dStU3)+00)Na$x_L13e}e?#{aHDRa3nN9TrJ{j0oE
zo@*|0U$YoAi(($Fo^miI@5@)2MFMO!Eh~eXI||(<Kjy5yVZbMuWVSW*(){EYp^P;V
z3Xha`ZB2}E+Q`Y{oOwF8X}9fqRmIy0c7>s<+~cbf7MTa!o8^1gw>?s1o}!nGv0Lft
z%{he)lP@0-T{g*3Tya02%h!qCtCssY$km%_n9hq9+r~We%T&vS63Lg(E!m_{WT0{`
zF)~xcLu8@EQ6@q8i)(q6Ma~;-e6Arnb-upSK`R5>>2nn?mlX<nc}`mkif56#A9&tK
zXw8i|e#P_HjJntQLa#JA7BvL56dn~x=V{^+Hs9;waAe7ew1A~C1w9SFIvQ5+N`83C
z*Ha&;a8+<Zq>jKx!Mn4%8YCEbQhLw+OVVPx#JOofP~S0uT((D5@r4hzen0<X{z~EP
zovNpEy8Q0g)kIc2+O}uugCn^+4u>{1)Gan#qI`LApO;CPUaPXh^8K9*vawzayOt=X
z`t^GpjQqEELDZ_{KOTR+tmGkgVBY;2+wRASCA~f$>L0IR@rsU|WTkr~>TH6r^qCZw
z1)nPua$TkdUAYs^#czFL`&7RK`G;R^9$XDBwca-2_=2xBbDngpY4nj5D9X93q;>O+
zVB~q`$Ra)#Z>yD^{V(qQ)i@#{Bd+W$wf4w?PnQlH_-w0E#jNeqbnb@qEdvh5n~PNp
zjZ2@MV&>v@pT|&N`AeGf(~lsgCpSGm-@J5SJO9B+#c7uDt=anf%2LX`uCa4j#&3@4
zn;~}DahuI))*GpTd#;K;ua22|Oz(>1xvM{#uB$G+kbdJ-$3dCv=T@!e%k`dqTqb^*
zg;deM<gWYfNv0ab5o@pIpKLEqJDdJz+Z*dyw(Ax|$4qB_kv~&bYF_=?*KTP>tUA|=
z6O|f&ItARkzRb6+(xSu2$6cUsdfz4Qm--Da0~cILT%Ph~X5ZVJYfTehs|7gm9c9q)
z6+EVvbNu19>kir3@3>1F^Jj!LUwgwDxo1vUw(pa`;?K)1p52j-7BCkM^0zeg%2`pp
z;Gm=uXQP1l?N1w?JlC0Ew@s3v{>%dBJ#4wxPcJR3VY%Dtom}9X<-P5CC%9BTa%`Wz
zO01~7<C|F?_t$RZH??b$`?zA_+?7w9%#A;(lmy>*xzMXO)#3XkC&m50A4_wv^gY%1
zd2d>);W4M5rxt$pI1ylF{yp{d;x7zLrWdc=J(1A$@7CfBE!WS{_BX{J7~kHqq27z<
z%OO^!!*hS#{pias>7lbIoK>zT;O^zseFrScA27<iQjQAEnPFUHE-<xq<`-`*L*@qE
z?FH9nUW<t_nRMv<cAxien$K<dbL(C8CujLrc`Xf9J3O2e{+^w|?{#`{7*ER)K~A%v
zmDAU&_HR61cc`b|_S@~Mk2jL-A06g?vGILU{q^sm-|p=#u)lbjr9p~C+5hE_b=`s<
zUH*@hlLDgJL$_adxO_>FP4wp6(|6)7pOtCcwXJvKTBenS)x8T|?F_5^b$5P@!-Cw`
z4w-gV*M#i(L%td}DlL02bUfsxrYY0wN`~zEX)atbZztHy)K~Xktz32V{%6tlqwmgj
z37*J4`07>tcF$iQ|5xoOvP(T4Y&ct5=ymm~_g7Z(%hun~`Ce4AYkj21rj^OTjYkC|
zA5UAhsdM?KCV_9Zv;W7tmfJmLRa{eE6veG@qD*&b+}fSW`YW&0a&kNEH#6`xJzv($
z<Jb{8%jTNMg%6hsc1|f**g7w&tDDVo*QJj<;`%#}OzwBSYf}IHv++0K|9g3qE_YU&
zZoHD?%DTb*>*BrHi=E~!2&kR%JLwVA;lsSs<n(;<G@o_$cVt&5%2{8Zq`_kJ%#lUw
zPj&C*6zfi>lEW8kS6sPbZ~N)grYRL0O*L=7`t{Dc+ezx@rl>%UvY4c#=%sRs;<I$y
zJCj8g=~y-yeA{qoW0B;^`mNeaE^SbVx1HrG^xppNHxbVl+uj;<g>W6$kJGd|P%rPu
zY}$|>?7Z~s)6{N_qc$G}3v=_&WU=y|IesC1?bS^}D#pf>R?Ri}oEoj4v4Mw+*IRXh
zy=_KC-OBF$W&1a`ov*9R$X%`SCOb>sol9krQ(}))h-!Ggz0r~r7h|XP%GDnYY-TU@
zUuMrHY8rF@o<+F%e&0pwR5bn6j?6I3U3LE5!pYnGVptY<m7bF*nWGt;xYH$lnOETH
z&-_-G+!m<3a5y6D>&zxsl5l$ajFllX=1czU?$$r$lYM43U)=4#agsiaUu(is*Zb)G
z=)Ss}wcXOqG<Bb_^6ibO%a1)fW&L7B{j%+G*$jy-rAv%9zMC>N`G{cZq6z6q(Wxsx
zZt@N4s(hsUC~985XTL#tZc9|(>#gAn8NYA+o$DskG21Pr>Z@t4Pcg?@>tKmXwJ&W7
zBv^KZ3+neLp7?M`z^hi;;>pc6#mVPw=d$|qbWCuUW>K8JoJn@U-u@FUmTfm4F5a~1
z|DSs0$rc?R{RZYn8*LmG+}io}dAE<nTLw#?qkT)=<=oXI%T!Wk|5KavHgU#z6{DZq
z)%upjh4UZ29Q$U4PQY`g%@IGPIA2_PlN|5KdH>jzz{q1ck%_zd+0@L#&#7p=$$65Q
z^_DM*XUC18NQswI`WN4yF3H(nxK5IHNnzqL^`kZ)>jevQ=gKTRx#F>AY5Dh!xydU+
z)~=ME@N%0>$$d#q=HF@?rhMyp(tAhcMiF28HWlr{;O*Ww)!Hig_Rn8-=bNE%_S}|*
zt0v0_N!{$dS04YN(dTlZ?4n8SZ<r-*pQ$9BvYrqX&nmy|ZH3qKIp6emvMuzCzB9k(
zg=W>mGuy9-iO%0!|4hbKr@p$zUp7oOXQo)#NyFG}2|;Is<pr;QyR=I5#EWT%H>8FZ
zyuH)N!)(PgO=m^rw??1S{$f$z7q;(PGUYRmwr`*NW{uWMm+ytOSI!6v?TGi=vEWrv
zP>=5kp9z|Ni?#$EH(zD@bwzy2TXhTDwcB3^7#zB??Ofk23B_geoqyEV%#9KYxg~AQ
zruqBAIX1qO4U49I{&Q2TCFjY?5aE8uQ-u>I+06;{{rT#shSkw4o985|*{(Vp^4e_u
z^5CQm*@~+|8fN4R<TS)=G=BLerhkXo&jr3sPyf79dys!Or}=8w>@#vrzn8aW{QpvI
z&nKrm-|>N#>Fi3*fOjWrg=beZ1k_LUWlg>nQaaK5%lAv&uIhG+qQ9$fHt8_Tc+(at
zYNH@6ec_{)*p73uHUXEv+T^TybKp#m&CW;Dem$FcjB~zJM&;R~g0FVy8HGM-nU-oO
z7WX77mZOxVG3aN|^iLV{)@kVN+3`|S*<m890^1{T<z=Veddo}u3tp&=f2ndh#y&i#
zzQj;RaOn(nJEe6?Z7O}@4{`{{BwyLSAhLAD!<&~GPMFmk4zYZxZ~Uo6;_`i)VE=#H
z)#fdpr(QlWBWIzJZ&sw(V)gTr*&oWC*kU_1Pjv37#oYdRUqkFVYucrq78wOsJe=|F
z@MLF>j)$jnt4)^%K0V2<wg}SMoT8r2aJ#5}<;GK!g%2m~Q{S7lRQ<ca$B&UVmMput
zeO4IppON3`=re6~-OJlcbA%6v6usYSJU!Cq$=vpZGtBcIslQ5nCEF7d?Ya8bgFl^?
zkJnkuuiJA=_?M-~)YW@lS7<2~xYf+{(0aIe_oUgMv{|0_pWAEMnDTMKJSP6G#(axU
ze?A6k3csoUU$QCa+r{|FQ4Qzk&5piu$YLq$!>_W-1!fB=70i)&_%G=azhPI7M*qY$
zJ;59f7p(msbMnXvYk$37!t+8db+XYLj}vCKyQkzP&DfapTDD7hy}9)Z!|FxXbAk&G
zoOu@CSK;?6O4oVfX?4*c{(gPC32Ov@`tRnJK6l+U;JDx4>Uw77;LRLYF7a;s{H&<K
z!f9RbiLaAGFZrHcC=om1?`z>Q-$nONGbOG~_<8KLW{jHEbE!i2kiKIEPI30!-DXz~
z>h;`h<Wus#KC$wJZkgkP)3b$2+FZX@t#FI5InJT->4{+1(Felv7uWs{>@w8){`_+8
zo8aj68W#fIH62zlQ3<rJPdyRJdwGI(b>`I{-@d)Bs<W@0v#zZFvbSkl;^`;Lb~)Z(
z+)`wcRj9&y#IWYmgCeac>+B=b>Nj;CiRQ}N{eR1^Pd?la|GCxv{pO=T@7~e+@}n1T
zv+FtM?&8~0YQX)hp6&as>7_fa<jqOgc;V31N59Worad{q{c~Mnr9-ORjJ(`>%U4%!
zmTKMQ-@HFi{PmOy-O|SAPj=rnGhM|p^Wq0yZo@mC@e{vm=L+jF=bGJRk$!9LKRbWP
zzCzbO@0b6arTxh4%Cv`B`*^S4`lfinjQ?TPYTf3uX}_jsAE^0l5PwoX=K0=lt@lfm
zpId7EI;AoHroob+gv{S3N{hd44V-+-@9WljW=&7sBTEBxU-#WmuS@nUJ<!6XesH3>
z;Qt^;xvdLMu2^@ecGa`CE|UONnaAoj<`$C@U6d`YE^vO^5j&?f`<=M_;j4e$B3x&t
z?%}%rz0qK@`F!2yu~OBNuTD+4Vt4ZAJDc|x4zDS4I976a?W#$gQ|2<=7qj;7|Duqd
z99Z*t=a>2g>idpMcYEcU8aJ%Hd@p42Ye8|X_HCAL0{#@dF7%#zuVCwM|J_ekSTJ<R
zI`C>S%<_qlWVpSrG9q8^c*U>TiV1tPEiNltl(*cqo%7>=cS6z0NZaQdUxqyneD?f2
zU;E80(KFlbJjmO2IX7Tg>Wvh;?cwQ8GtNnD;n~?U(?Q~bZ~gJDf0*aEwYYL5?^V{x
z-O$w_a?`K-#f4Xg62vPqx0h`GZWZxixm@CY^H(7v!LF_wUf7-ea^{)Ai{ynnlfo=k
zc?7xk6iYtbU23s+`lf#~V#ED2yymNI_urz;AgP$}BOv_d{P;O%@`X>hcU!6JIhC?z
z7BfuPC(E;>IBxfDk)4Y5Z=9K3JJ<TD#I0coOezq~3$8r)J3;Io`z@mw_FLZ)cU?LZ
zdO-O<Xa678$*kF1&&KTNda1=R->myRujSoqFQzD&ACoz6SdjM5+t&Gtki6JR){|`f
zpZ?@ze>!tC<*XJX&yrA^eUU+hZ1Gc<*(_VnyvE)8-~#<p`$OT^AA8@LcDBA~S4775
z%Wp$&uUo$QY|*}(XAWQ03z)Tj+2*%@&O6xpEn4?FX?E(huF|=`rup(YSi42o<Xm3d
z@pRsYX>G4QFw9g~ChTxXj$x%x*q3#jcMZ+zzv(f#m$C|k^L<r(^f<3@jo1k`tN->>
zpT_kkH1JKIbb9Z~{3UEV)DqIBXjre)zg6$J^+0LFvUEuW<Co`EPa4g9o+@j@a${4`
za?6*u7pJa!5wvmHzSTijc6jK<?|Q{rRG90>c`JW;dH1E%CqLeHPCCW2Cga=jqbf1~
zdy1l$h3OfE_dkAnQDTx3t6p&=%bd4crR$<U-uxwhi~r}A8osoHkHdbn-hSV|v~tar
z<j?b$Wz;86@^@`vYChsW_vvK`z4tpdaVeQwM(qms{wLV|GVf!#?y8AKiE+Im`yc*&
zWw(pfY2%-1#qGM&L(D#{|4`W8wc|se@WuQ3_9t?+J~pK8DihOl(mV3<a#{M?CjH;1
z*6rHAmwWr|f}3BizfK8^yeoe`JjUVR?#p%msyEMfZ!<gjuYUK@c7s33-%S{vW-iL#
zf9Cq^x6cDk1pL`KL+$^hc&mqxGIx5_%-^+v>;BElQ%|(H)&E5qT3)oH&)&KF)ED2j
zMGu$h2`hQO+hkt6;CbvamG5ae&tlppJdIsi5&yV+fl2J?87reZx-H68xlfiD9%NgP
z=el^wifQ$RhJjkEQeAc}iI}i$xAOiO7aKy%c7|xCiP%>kJf%6KQeyuf?Xul3*Qjl4
z@J%gsIJ;^Zk5yk%!<<`8_m5uC(=A=e!*r|rTJovry4akArl~RM`)Ahiz9}{Mt9&SV
z-{aXfwsvjV6T8cV^mOzmt1pdMBsbyYr$<^}JI-F5FW<aXqrR(4uc&C-<*#K|WKz#2
zs%_tLGnQrXvpGhxkGVGRgxyYh;VrKC{_f?F(}j%w=DrP6f6dwUaz~25f>e{u0#OYC
z-9hoGNoG#of>E2^t!wmkHA%ORZuHd^)A%6!$~j*oqnl~!*U7#rtD5)hjIr$ulK<V7
z*rRo8k;`nZEuJ@m)uQX$R{AlB%v}3i_jXvU_C}xE+g3?A9e;a@!*uVbveb@5)9X<a
z_MG&pf4`#Gl~4Bm)|{D9Z{wbHF8cg<N^Y5V#*g$3|2A>#ziYELt@_!CCx^FhS4x`O
z;c)FM*Z!YV{yzIB?_blT6d|j+eNuea3XA;zD-$}l&Zv*=e3A2g_U@Nz5e4;DpC8^#
zX-;=Jd~x2nyq?hYEL^kM>czOZ{(rsq_kpPNPA1(mu?godJDsuN@!6fUd|8q4ym?zE
zRJ0{(&20)gH$hp~=YCv$^0yVO%QF`GEy#X;-D{2`_rf$?v0M3Z_p}_n_XT`!e0O2h
zHnCHeedcOqCGO2!)>oRMG~?u>q*ELp*VKRU`uoyP|KTTnm%E>S=Wkuma`<EA+LfVo
zX_qg>X@>HZxLx}iziGX^*6KIL##y^|)eEoFZ;f7>eRP?t&NQo?B`miZOZjAEK7Zom
zy?$%|e~T<v=_tcJJ120@;@DQasZJ-DH!1mk{`%R0?4RnF{><FB@1_3H-UYdaO(AA0
z`wymc*B?w-9w7YVX06tP6}?BS3>c5HXuN$cSNKJzFhqTRe24NQPnpn?rE-NO6D_2?
z_FUMIzMv?(^o@m;V_fvjn7|`aug|>e<J-LRM)L$C>*7Aq(hcdiA6)2jyD)G5S>2q?
zk@mB1Z#yS@#Vez&p!QNyr@fuMZc~bC&}~lTwU?KzF}+mp^VaOUjVJ%Z^ByWCTYW>k
z)KxxToTu{5w@c1vQ{7&T1M6*<6`y(%J@@VYr^__LKl99$&@i+5-Nb(RgT<+-UUkAb
zHkM0!Ij8vBYf2<sQEKZB3Hx$d@=}wgl<{5J>c8L4bbURgYgxQkV3S8bU!!HA-P;)q
zVN-APPpCcjIdjeGH}z#2#|6`@lFw!Qiu+U(mGk%HqZJYc9SNsbuX-%SSp2g=lY?{W
zYMBl1I-U1S<@+ZUb5ve<YnRQAkcOWTAI)PN>ZLzviFaxLvx_|U_V@khRliQpofB!7
zB(}=pUDbln`V${=UY)$Au=w!K3}2a!OJ4&X9M4JHx$jNvWv|Bhdf&^#>$PL{GTASh
zeWbzYhj+kA;U@vyr&aHZ)!x~`b}9Cja=6alz)!h4AFOAFCO6(U*8LzXGIdLq_P6?@
z?eCt-ZjSZXzGqp*hGa)i1%c;*Jp5AE{T|*7ect5A7?ryA(a~37`<-gt+E%pP;=lM<
zHttIF)O6=}ycaUK=9b@_c!|I1y?A{6v0GL-R!Jq90R{?g@#(Lc=Fe68rP2G-Jc~Og
zSHvz@e5crj-!GX>{5Ji*DA_pw?sBHSD9xH}E0+J>=5VHC@$}BcWnA-T=0#e4aFgk6
zJue>p!$<Rz#=2Jx%i@0ci2mNPb<+aZ+#Km?LH`$Q{IFBx_{|eVIyL8BC`xh9->p}}
zcBlTe<*vy$j|C)jiEzyio&GCIv!<=@?W}ckGv9aX^hfQQX?Hze=b04u{PMJ?b6U?!
zylPu{dE<vlof^Bf4fhs5%6P%G+xAh0PR+fohn)i2=Uw8oRZf_f{IK}l&x*>c@d8q2
z4-6A6gRXGT58d6dJvh&E;yVX(V{6tE#W~zdx(}U_tS@5uJm>SxV~3vK<C^cy=V^U-
zi%7S#h4JP~hZ(bfZoYKPYL<Df^P@d!uQW@4na9W#PD!8eYwk9|6~X<fvlqOuGptQp
zetwo-dPi5Enw!1fZN1uCX(HJniYd2eZreG}^Uo@ilN>h-Z-q4~pI)>t`s$~YBNr`l
zZ!@xeuH5%@wF$3$J>%z1-&ZRgHsM`zWomW7WwRdH+t;Ul``G_Gw4M7!zVo{4M=#ce
zKF^<TS9xyjbFt%-makHBxU=I{-=((D*1yGU@0K{)9bLAZ+jzpVNma{y|CX?Co@JHJ
zc;L;JbB4?}Z0fvT+!4Gc*m?2andgQ}Q@5thc{kPPt8Ss^r&xJmna-W=^=)xR?y?6W
ze~CwToApgGGJm*z9%t0<Z9jkNlsn$++u&1EQno!*i(#%Jv-M$x63+TvDQjYOFa7<o
zjiodEwD&=lT_V%|J()HsEBL`}=9|+#uj;uV#kZsSbm%*4wj-BR|D8O#e%cfD=`mcz
z^&)mTEB}8H51&5!>x=%SzaO<N$zIx0&v(yvMZo_*H}4d7EE8h??!Eh0-`?<9;fCI;
zHkNNnj;b~MopCa6i`%QOt1dfivbFlVkn`s*lNQ+>_X<@0o%;X1@BjPtOw;OHAK5=m
zj681qDC)@nXvw9!7-!zwEGHWD%lzMd{-&eyEZz2E7iAe$ChlDAy7OAb=OdZ(4qj1Y
zyOxkqpS!U3%E>9Q^;*di%s0vo6tw*I+O4U)J+ta#iQcJ#+gG)3J(%MgeyZ6vEugC{
zQ7t%SU!eA$jK%$og*NRoH&y6vyx~|o`{lI4OG)i76*Eqs68Ly_NB)lEk$ngJ|NJ|*
z;qRkq1?M)Onh@8!SEsM(;m6H8SZ}g^_`F8=lHc`&s524uecigYfmV9kN;cjwN((%n
z>v3EsD|<^{<dxtTGMiT<i-j6?cX68R%IS*H?~ZbNdBA95RN}=0sVbGf)!*EF;(Po|
z3iFA*fp?-f{NDF;u6X|P%DJ4UOXsCZgxLR>sdatx?inY1julV-G277iChI(wfJfq`
z-!6&IV34Sq5bE|jvZ?-W_^rTMi{hDXdD)fd?3^DHqP6Ep*&O5Gs(Za+JB{>wo2Qxn
zUgNv0tmsDI{IKqX`Mk>&z7|!)ZaA%^x8m<cU77N;U;ZS@gx-uR3tHRr<DtrxRYCR1
zcDZMke-(Z8_EY9c!#xu}A6~pt#-QQ&roR~rQ#&^Ee-qxQ7QFMz+s7g6-0ROPxXA9W
zXE+r*@2YIuzA4KJZWb8XJ+(`-5Zzs`y}mN_%Ew>r3!VOM5cz&Uc<Dq%@o+2qT`U$Y
zuhvb@{LAAz`-X#Ai`?GBr@!#c6^dBF^=ZOnHNGX;Me8qS`wF$6<~cp#Rb*0?_xHYQ
z2A69u&dzI^zp&h+)z@(EmMQxKCBIF0QJ?xTeOGpB+dIYF=DwGvr(dj{sPMKsj^h-E
zTus}{tzJzz<(fr$%Q__XByuPUIldIB_xSy+e1>N1dWR{yG8EbsTv?p=DBHYEy|?A&
zb{}=VFEaxR-3|m;?^o`2n7pdw)S1+dm&xa?TOK}rSjwbO_~X09&ANCN>!5sH%T%j2
zf4TY?k%I~@bEUJ6P1*iuS99Ke-Jq#`@%DQsf6rTA|9Q(#x9HYITYuHtpS^x??>648
zNL`c0!zsUYujJk7{Oou8jH~%KXt!j&Yw_9_WriOsV$L7_T4VkHddQcI>-%h3R-bWR
z_J{xCk=lRi^}_X^FE6f`zaZ)7j(UzGJ}&>hJeTpcnfJwzK_~3uidvV{+>k|^7hmGP
zeydklM{K>S2^XVPW&PKmm*4)5$;i6@>^A#Z&pRKYs-tw9&brNBZ}Q(b<e)_35z`-k
z7HJ;SV75DDJMm*{GLLwY&z?KwhCVXk6CS0yZ_qwam9D>co=bX6fzA0tPw&>KzP;yG
zFLmPV<-7JxH_f*7Y<~IpOWwW*X{I+a;&-IovG*1K<GxySfBL`YHK*5p$k`QlzIm(J
z-(RJ&HB<YmWN$|%PCu-F>w4*l+_!IgmE*QQ39J4*Yxe60S`nu{o9ytrb3T3I&1q_X
ze`oEQbAC(b|9R0LuGXA>`|H)}u2V0reCf<w&BFchNPTQ*>Z2Vi8c&HV+GG9EEwh`;
zyDHBke|^3@Q*ZIBPUr4smsI<(lZUyO|39wgE;{l5ef>XqUJr(Ki|UJl_tpwXEnKAY
z=-iS+HU5uHjprGCnR;=>0_GVz`#3vKMY5#EoW0m+axwi<qw<{BORvpwWxe!rdzrM$
zmSy>!?9UG?nmrDxul`mx>9{l>zefs7ks5b_?N0^g`P<^-PfmC=$NQr8)oVx9<WCtq
zGs=lt`Bpp2fA6cdh;*ZhXWV^RlM^iW@8bL+J9}*e>nzA9U&Dv@&m|or8A9^TYyS-{
zuI}6LR`5~ED*sQ(vqH{K+OqNV#^;vOoO(4<6GECZav!d_>?vCBdV<9<piObZ<r|Ax
zRNopeviwjw*Xi(5Ez4<l6sNKn9p76Ucj?dcxwgkY)h&%o`X`Zb`=?RGT3_Ln(;m$4
zm$??WO0_Rb<v{H7y)zp1mPcC(icVkTk!~FAKfRp$U-$0!w|vdd`A^-SdGz#g)2}}F
zH_I%&lend;qEX;oMN#$@{$KU;Be~pv^d|i&j^A+U(A7t~%fvfM&QAY%U47y{9_8&y
zpJx}Xsx}t<AJVAXY<294(9Dfb+T5gtmniGA-acd3mh#My^@U$TG^cM`BHOOb>z-}+
zq{mrllCVgk!=uBSIkwj5U3Ttl*_tms?G7_FT?G#-dT0gz-4&f7ax8uqTb)pLz4RNW
zHwiU14`fbp6hC@==WqY**N2ogvAx{)$6Uwz^}b2ZtLNHzlnQS7U=(0=_H3%##gduE
z`Ln(^Pka07@#7|GsdtYowmxOvs<HL)V!hi-7d9}@x3|9AzyHDAk6Jzj_Bzx4Jj<<W
zIC8%0*QL`FZZB1}Jo7z%i|Vzyix08|x369Fr~bhIYg4ZE8*b8>JuA#A;jyAb(-|-E
zW!7vZ{OZ3|nYVp0uVu~Q-6=YqciK7LSHDk*En{R+cyaa8L9Yv;?y0FUE28UeO!4x2
za@T6kz3D<%nEKaER?L{fd)iZd$#bVlRq6Jog$1g4adw~1UoG2z?C;YP|F@i6`&HHR
zeTd`A83xa0*01x`nRvu|nU~IOQyt~KpN$SrZsGK<SN5IJFSsdKJ!yWzp)GT4eO;|*
z2$|{W?wvLHlGS@1k@cU#CW>k7(9=_W*UWhRsj9t<qRIY<{l4=rT)oj1bI`H;wAzbI
zLkDIB>-u^R=_Sj=tQF>e)7ZK_^YPi%-ATVB%72P%d6u)r{@1yBuBnsNrkBQFyUDXz
z)^U!(*_`%<yl1oCUMM@%vRTik?%I(SpZ`w{WFK!fTzXQXq~ysfwG%EzmECqPgASJ|
zNjip{>0ehN{C<Dsw>~re?4uV#KAcQiJ74=py=DFHFDHxhY{PHoEWiC=tJ(c7PoGU+
zZ0aH&ZNJm|>z(_*x9%H7M6T8^I$bI|`+u|jkF;ew_BiMC%xpBi*%Z@ojI&|Ot8_iJ
zfZ6|+tra<xvTu&fS}uj4Yv&E8=Ugq9W&XEg#-4+lPZhDN2Sm(O_Bi_?L~!O^frTxn
z%-g@$?716n6Hv56<D+$D{_%QFfz|5*ow{f4dz<Re!;!kp@$zxUr%e{ymSrid%dfAB
zZo605`kiAh^Osi-bMy7P7pSe1`Igkxzx${mL#MYyQ^=Zv@_g5R#t3&W>zHTRLR&Xx
z-?9pn@K`o&@16vki~5S;nfF+OH_pz#UZlG3xgO8|2*2y~KHDyDz0q#<e2awAw5r{+
z^md;9yZ&}T8&9zL#75=m6Ml2g{NneANurIfo>$_(iksfwpBcFy7ic{Cl~SjFYs#OA
zJF~yL`EPUU)bzPH=?B}pW94Tm*BBp9+!wSr-T7Sll{?o<`HxrErN*8NJJV7$v!GTa
zOWfY8Fi5g^@1LDh0;Mm^II^YcoqXU<l{pnht_GT|>Pbmdu+Ukj^?ioSy$cy`xy&cz
z1t(X?=cLuY{yEpfo&8qwj+<wWelZDUJC%F)@!{{h<{8|%-wd7xZ#n9gvfg!HmFlLu
z(i3-|d$O~9>e8PYdQT7EEzi7Me=YB6NaO~WkX3otAO5|pGvS)$t)r@3UzYt2&#=D(
z9#A`X{J{P_fB#7oq}ZKF=f9J8`Z#yCsjZavmV@QH9_8Q7sXuhHM(gsAM}e1~9WPg1
z>3FLzyNUn!ok!Qr9QiC~%NHf?eDLgZ@rGUeJGSqH4XkWq68szZE-ifV>7T;-c6ZqO
z|CAT~zOA*l^PPRq^;>UU4j(%&ap8YYo5Y%V(+<w!&tGiYa(~NG38Up^Gd`IgnKb2a
z+_~+KOfwswS*~7F@5^Y%a%D>F+v=^)w!imMUXYb7*lFf(wsCd0bN$b438i0et4Mc*
zNKBjZHajh1m3txUeX$vP=jwYVf4cJBb;h~+$NW1BI-4c`34VSH8ZFCul;HAwMcOQj
z57KW6&YC7G|L2k~ce+t<LBFN$^G=ig_9ymAzn1Ku*V;Xy-^oO>yZ(9J!RfW>oyY@a
z!QH#;=dEukt%j|?JOEyQS<$nxTB>2*pPSQv&bqxmz+{(jy;t_O|K4>kTe_D&)%sZb
z>X24Vdhurw;f-AN_y2BPp2$Bf{`%+YkNy{LH`_aR-+C>%dz!||(-IRc47_69M4rXD
zWr)}7sNS3Wu4U5h*+yZLx5%2<o~Vg5k4<7<?5EM`Hqmu%z>9gJyQR5)@o>3iNXiP?
zcjupx$j+H{B<-fFKWnn=^QW3e0=j$-HmSQs8gx!otH}?zed4<6AIH=F)7aiw+&E!a
zIqA2m(qx5DG09CeFK+6%-PYJNOKJB_);d3tOQsX1*Kd(K^{HuHJFil7f=zYh)1Z5e
z0&8n~;^g*~xE{P4b?;RAhToEhI17a0+$H^PEZp|=-JG*4{+UjE+~R$)W@UuWBjsJ&
z7v8>Jdv}4t<we_Xz1{xs?8K)#IbMsNy<T;=tvK<<1}Tkv+xxzn&1a_UulUB~72H?*
z<kvD4wYu1^PO}7#)t^05cF)vs)z%u5O&#YZyjfs-%GFbEGJiNPYw+pYf^!{i7tY_G
zaO<mM-_xM)E2eA}mv51DQd7J4qFL&v!tP>u7Y<|F2m3h~mulv}YgV)Uq0o^dEd4Lw
zS?BbJ%~D*;rtvSjvQg&p*&H|d+JK$>Nqhp?Pek7e?*8c5^J3S@Hw(|yPt-rlzVvnd
z?hlRvk<+@Lo;sNETBLTu<!1gv^Byt>np&Sqjj-eRF)h>i+5(qXs=pO7q-qp494an<
z+}w~|Id5hE=LKnYC*Iy+XXL)kHtF}-_!(=Jtr)^WPHW0^-*<01+raVb#n0R8mD7Hk
zIpl?`2tU2Crs}eaZ0z~fFT@%}9v9Z9DZlt!^MEmN-ReDD2@+CjD>P=kJ+}JA1yfrN
zV@9bbuijYs<hW|c_&Zs=%wJOQ$oqNF%xUv=t{i)Pr%~bj7Yi{XGv3)3KI>GhSgn5~
zET=76G@xzyrtcrmKKFe7ZSGIDX1TNHb_g1*m}*h5CE~o2_D$Z1)x3FjBJIg5W@ZPT
zpHR;gm%5_(W^~R9_B8HKJX?R9xcfip#kTjsqRUeDgzsFO6zJ~nS*ey%qOs^q*1T}0
z8=F2wuaTQ97_TMfochw$`?lEp_@`eA&Hub8Iy$dA%k7Hx-oA2m6?48P5tD3x6>}|9
zWKeh?@^WKRm2sF)c!-a&*O!bPFQqa9?YGps^h#yaA9rdwDtWh->DY5dhSKczQ`WmD
z*z!(#dwO$u`IG8!OEcZYVy*W?gP+{)TUzuZ^5==rGrtySnSjT87gpNHus%~U_kJp2
ztGQZ;@px$7D(#AfR|gEFmv41vERXS+nQmRt(o-Yc6&<}%w(XJUrms2dOuLpF?=9qd
z>XPwSkfpzV{txe^501#oKU<xk{9ovnx*2O-we{akyhlzeef*caE%{{6Y5j=j%^R;9
zcL{&6wbo@i*Q-{tRcF#W!NWH9R9;4{TOVz7D1%G-Hsd@S?Xs4$k$r~#=M_VAlA@AQ
zFHXr7?AqIP`P;c%LBmbDY^%BKKe@&AJ^FE}<56GWoh3{g=hc5ayvtMKY}hWv>#8S8
zE_Yrox+{@rEOVnI_m4!1oN0>tE=Fnl2Q$=+8yBRT)QNXIQd^q;wMV}HSx0KX#3KK7
zQa`Pn1?0jU+#Y{t>WP0B<-6|VXTi^%k=4%b`;YZkZDUrhnp8J8p{HT?y9ITR9)0O}
zq!(}}<>a1uAC(W33YHqz8%x|ZU|@cd@Y1b?=jOSk=T~K2J;!*x+&1DW=Z8CGd_qYw
z@~c)V-!<2r@$Qf1q8)Z!`ll9HiZExTWH<11`!iiwxc(bQtD^LlFNM5XY`&G(R?U`6
zJNE8X0`IJo4CcGFx+Na3F>QSG^!A6xCBGH^1+t~5@7nWN|H_fQSGO#Em*-ReD5&9g
z{T@O2NBci{eVPB__StW{xAn;V(x?}&Uu@HHZROwn|FvD;&9*K6T_(+1%QyRt$BO4Z
z3sMV|C*6I*EOWV4KxWyWt85upd{eJ{d7QZGOgEFT@tj;|dlP<Tt3$7Bw4`j$q%OKq
zceOTcO4PKx$g*V-JPh2wsur2mnAGbZ_*1_!oKy1u{Jq(cWw$*x?LSnIzbjd1=e^ch
z9NJ8+?z`1<+MPFr%gPF0**oFL$GuN@CxvI7J#{<se3IGS(?2Ku6s|wHfJ=e1p+cmK
z;YYwtS>0}hMyIDI*PIECoXNyFzf@w{Lt~!VM){$7PiHUxSyX>G?8Ib+2^^uPr3Lj?
zu-sD&tluN}>8xR@?>7N%iS!j5Kh-z8S<J2N&N`4VFD{Oww7sH0naf*3r^l)HAh*Ux
z{$<IV8clc1>V2zyb;Ayp#75uuGTF^t{ASzN`tH!n*^tYdvFt!%gwECZ)8E^hB;469
zTsKGFG+Cx}P3iMlYi4=P@lCy_wt3IuM{hJE_eZal_nuqNowncR<zw+(w=cibal7}-
zLH_8CJu)B9c5(dZ&f^J`iz!rCC4RWBC2Pme6XLUP-hT2~>+b?avsLX|UPXM_61V1B
z_uEJIlYhtBxBve7>gCGSA?s!aos++3ub*BYmus){E=P1T`>ki?<*T<|n;x>E{%myQ
zg0Rmo3i%!%UU|y9VrO`Lz4dOha6aGK#jzV!nK#)qR36bW`o3$=L@D+foBWjTw_d3(
z|6uXTzg6z{-sFDe$bCOHujyM{SUN9<BYEx4yx6z4{$<(EfA)sED{lx)Pv0(ff32MS
z{OR4h*SrilS$<--?OC(W$DP-&zWILt-^VXDJb%Cc@8&z|S#{gva?K)_A6i}?!j`|!
zzi@Y){-n~Y$}7!p3jDPAa>&a}veBqdz-~>`;Tdm|&pnyuQ?~cVR@wP1b?f_f@9wi=
zf4=Ez{;Zh$A^mo0ucKmT)JEmlXNZ?itxMeMx3=<O(ktoL#i!2uSNi`xn*6J})XFzI
z^75K%59)82*Jds_(88(`y2eY7Q^bKqr#{bD;YHAcjhnc{c5PlV>qGwjPb=)Y{wN*|
z`*ror`>kRNW|s3j<L}=7x^<uL#8R7iZ{jDH&J0_ftQx&wz5VTQ`GCuRH&oc)w_Cnz
zoqBxy{Us;%%wPY$x#YpiJvQn5_DAjRo>={U?ybju2cE8cptvM3EOt${mesvo*R^j+
zY|9CJys3V}&RP4JpRuq6?{$8zbmi&2ZIL^;E9`^K^=mhC*H8BId>Yu66nm@Y)xGtc
zhRzFbX3x97ujG7C*wj@%H80QISaqrD<zxH2e<^Q8@Af^-opEf>*SGCEUj@uNpIKG<
zb<e)w+?w#Gc_Cjdek-ZxEvS^dwrJ0(rh}`^8%z22Tu$6v&`|$SfYm9C|CfP)`+@^m
z{i1vl$3C+w^6d9s{%q;Bqj?{-kIS6D_kQVgsd=*{+CIv7ul}wZx%VI2dP&aZpKJa7
z`MY=Dsw!K|^C)y%+kN%#T~RV+lJ>PmMRi3dR5KeKzs4uD35Ct&44*F}H1mpl=~1nP
zN9?{Nw60itH~Mw>8YAxw^%s)Y+Zaw_Y`E$c6CiX@Inb^*GgxkO+B9!fv3kF?C&OL_
zh8o>`%GW6QW$ya@`R<CI$7ZtJe2~lSIW6yEfQQiQjJCZ;^iQ=vjml5kYOKQOlYg<S
z`FBe3_P8aCD>uFOUumX4rBM5=)0%r{=auD7e80(V+w3o|Kfl#$Uhq%%`pMY(O+Ri-
zjksWYC@AC4hstdyt}XK2H#z9{hc@fIf9FVDj$br=^BR?vlNkKoZ;xI7ZvOkA^4y)L
zEFQ+b*s}igPpjJL>b0ucr+&UX{OFx?R_oWjQ<u(=_q}*M>cr>NUngp<AL*1`<!$_&
zFWT;uW}Ud=rLCCLubXN!r$%>(KMOvaHt*hqdd>v<7k0r*?&WW;K4?-jC-+<E&r3H?
z&u#d<D0#zo4zZUz{=cfVJuPg=@H^U!{p6>ghLOwP@1I#4Q9H|am8ECA`po72Cj_?~
z_Hw;e%Jx-5SMkdqg)ZwZk8kEFX{=>eXJq}osJ~n%uRr?6w4=E$MekD`5AXh0n0oY9
z<Jz6O)+M~C|IyL>uSt8|>#o=D4g^KVh$l8}zOGx#V>>B-!c@CsF;8Y^Kh58qW35}A
z&>ys?G*-L7!{(v;)ZSyiTz)!f+@7Lw{zP8NZhaG3#*?qCB;=O2NiYkAaIFdUl5kp7
z!<5(6rc%XtWA@{)U5ddk?A{zZP;O^d!1_A!gUl)6umW{9xq9!_`9^OxM};+6O<sTT
zI#cA@CC4T#Rh(G&LNnv*7eASl;FEv06t8MN6T5A~^lkHweL7^cRo}$SvuJNs<<-6#
z7R6t`e=I-A&}-eaJ^Sy~x7u?HmM_+?ws;wOUBd4`x02eHgAyhdJ8gaCw<J`|ZFf$6
zQ{5vW>Zx62;cTJTVHDq0e}IW?am3W@>}#pP1-tJX@z%IG>A%xBQY6i!)qiK-;|*J~
zcy8ufhOrg}@T{rMubC*@{+8u(e&XXx-N_5IC*9kYz!ux=@q}4$rmVYs74IRJQs>V;
zSvfmQS1dbn^hLJ7LOWsiv(|IhW=A)_nYH?RbhAK{(vwd?PKRDcG{0fFoL|qosKwaf
z$y@C+8?pmB7P`b9uQ(pr%+S@+->mO6m2KbVORBS#ylr24@weL~$U1J?Tzb@0V@3k6
zlG2nz$=8<k9TUh7y1dr$@H)o>k7pd)rP~|b+;d%lM{wB*^Q=FXKctSZd}fX<J+apD
z?{$UQR?BO*pRs6i6c4QK&K6k6@MHIz`nB`#9TC>iv6-Us#p!jc>NEz?Rr+xkQ=Tj9
z+nmct%v{0zp*R2X@@@B{m=8;EEScPQCn~sl-_pFbjY<b>(%eJ~+T*e<c$^LuhVE{X
zP+WNXtI59eE8UCN<wkCNdGtblWTMdGj!#+DJ{=pMY?@;e(!KVM1;5sUQ**S|hswT}
zxcjQ!IsQjSmzR?0=LPad8V|;pME_+xa4Wa%^}-A>@g>u$N}PLV@JBlMn7L;y-+nV7
zeunM}<14TFcPZ<22#fvRm9zHhD^(8Vjhh~+tq%Vem>4wc{=c)2KA!!&>ETuBjnnu3
znmAFgfpw+W&3!5?H@@>vt!+OMAph^G{njV3uS0I7RL-woGx_)YXYYk~`aRqK>i7K$
zv6|zaa<6rM{L<flbLla2+fdud@_k8HRqE5^r}J-{Q?b3x=kz1XX_wxYSv@Iw6WjOq
zf#tUthgo~`EVJhwKBQZfa`w8}66c-+9_FR5dvx1$^w;0d4GZM{y1D9;gs_0Xm5XNQ
zADs0STiSgp^Q^nwhsX7^)_(f8A|}9QS9tR0Eo-7!)&yqhBpgwBVz?pg%$x@IbM9+%
zeH9n5D_PYZ{}i{)d8h7?&E*;XEz@4_GWrrBGH0#QgE%oxk7Ld}S}e`Yy`nj6rq(N+
zgf)M*?$)``U!%G7*HPcc&reyuh*EX^W~6e%z3{-QxeqxyQg%JCbT+cHt50*j`ehC4
z1iiUe?`=?iy|*Os@xLmwvgbxZ296E+B?`ZL7OHl!9&l%U$@6B5@yW+CQdY|xj8mIw
z$kD2=t!CLP#NgFf^VdlJe36B<&6gc(-xj>x7d=<DP3c%jM_=yh;@gRBdpfEekMbES
z2CrpdXI)eIdEe}A=hsg{CBs+iwAR%(7OOZr_H7DvdezM!cWZOZ0)@GSxu)xSqkFup
zE{jclml^Z2;+W`?g|a+5FGfx=o+K#OWcOvY)yWzBPS<X=EL|}9rhRxyjQs!7uZL@X
zJN`ZX?Cj^;cjw<+y8Erw{krS}*#`ABvlkUD%L{DCeVLpc{O4>U!+jC=sVn|l-(DdX
ztXlumk=Ns8m)vplpA*b(ewkaEvcby8c16vvPZw?9?b~;=Zk}zLf8{0fQ|ssC|M>si
z{_4l)FV;j{nr3u+@lICWO#a8#+$Jg(A2N;_JhV8n^6~RIt*=YPPMvg7^Im7TzQy&4
zq9*(8GMV&~PaRooxvVE`TR;EX)=iGRot3Sszc;P@$zPxQtY%B)>%`q_{Vcz3x+`;%
zp*!nnxkkl=fBxzYH<fJCv-DUVw!i%U$;`~{jvRx}DUQi~yx+KQ91C^l)lx7wf2=-F
zZ}&ah=3ALF@>ICCeu_VK-)n*3Gl}rmih?4^{v6wP-<CGpD`-@_n=Ae1-KC023*~+N
zYD}M8Ud{8;-2Zb?z4R8hwMTon8g_46*8Rye_yQC6tqGA)c@}A_rJNhh^VbDW`&N-!
zA}!tHYd@`M`Nn&}(rd1*T>W=T)V=%=?kh>>`R{V7SWj3qt4e>`uC?Ev{e0WKC(*0d
zFYIGyQkOED?<D^0nXl5;EjhOJ9rrCom4JrQ$|nuQ=hvt2_hX4q6R@tYy3_luqSyTE
zlXZP1Ki%1E%lOaMz6<N`Kkt1t-gjYM|Nhw1b8n{ot9W5%^84@h`YGpq{{Fjde|m1!
zrezshb5F!cEtt+=|7=OrrzuABn;ql=G8^wnEqlM{!tFzM`5NPD!oHZ<l<X=0#=n-!
zTAFcsVWh;8K05)`hJFK=?svO?M%14w_9&2&o9g9ql_%g_@X`ADz31*3+oV1<3aab0
z)pwfOu%hhcs<&$-7~lP7|F)v}+4+4j0soe*NZ+`}-#)lqb^qT}r~M8sfB(hqZ`QZ(
zQNoGBEN7P`{K;Pa=F|Oz{1YDGt8Z#oExvmyE}y$EC!$(8WNpRYx5q;jAGBxJ?(IKW
zRsY&v=id9~GgnMj<+c0Ywp)K+YC`P(lKTJa|L4{m)~=Bi-v6j7e*MAvmv5D~NSg|8
zZ**3WTzTzkY8}tS3#>=^H{YA=u3IfRUEFv}$~oD*>+8>)teg8b>t)@vhJ!C!*D|&3
z(DwSV>(2MBihu4o|LLD?zf$1W+3)&S<_4`<-L`AhrFs`drF&oN*KYZ7cYE_y?p=DH
z*X3>SdcXgg)?~h~g<JLRY1T%~UADrXuV(G9#p}!j)Al`hu)=J{N4roKi3ROG2g+s%
zCkgyEn(VXnfIO##liKZ6!6yOXB^FgmUln`5Rj18dJim5lVDf(NXI?9;`7HmRyM6W5
z_2WN(T{pk|Gs^I6cYTS@JiGfF=h!b(dw;(^?ye(;WP-)^SzGsCsPt%$SMU#Lw7Dz4
z>7n?qtz7#V;-r!*4E}reh27h>MI`RW0olJ39{sPJyCV4S?3<UXz7`hNXJqYpfA5dM
z|2b>4zu)UMx|A4T$G`NW__`;(w%P06R3%-LD0snYGS_`_^0x!bB658<>$iBzv2ZMn
ze>dq*-~0T@d)PiseB#<I^EIGPWxjao%D=63my=IEx4-H6@9D(u5TTO4o90BVGCT4^
zP^i!>@OF6r*6;P92mH+)wAM%NI3%_Hjh4j~h88aU4#`v7Cq^*}Emw1qW0;h$INe@n
z!Ll0H#cjeIixr}iqn3Sn9+EF7HhXP--ly96InTS>UH%^5eY9v{?VKyWw2ii{DDQ51
zwNz}cuZkQqN9vYy->%00yuNw<>!XWy_bPk*zgA`}knHvO)I8DcTj~xu%6L2|pSt4Q
zJKxU!tmoU-tgbZ)ioU|vd*!HxaqiCvU5Q)&)qFRozo)Wu+Z09zr!`&iuV%0GzRCCC
z@}Ir+|Efe!EPcMh(&AFhq+sXL{Q|cVZ7%)c`ry6)gqNLcX5GKZlQdKxx}@C{*t`1d
z&(o=&zK2~AeG;rY<)Ybg--8F_#l5$FHMqa*-I@7Te@;&7_~hzsx~gXWnWCR%v*(`j
zirEpg@k+b)BBhgurtWnwSR-%4YoYSSYS+T$BCInWI6pmB|0Y#iUTn&;8dq&z&PI15
zu3dIdeym%ex0`YM)TxK1t{hK&`fpKuoa&~8y)`#Q=UukX)jfDErefXi?YXVbch%nw
zs#y2?_}bRzKK1tkD%QPz&h^{9_4%{;;pboF8d>g*{$soIzPFZ#oLt?E?2FGY|MkpW
zf3>63T>b0y&Ftdu!q!&QPnUmV8M5bc{oU|A)-iWPcifKGuPw+rYR&&-R?M%El&|kr
z&3|X};nMOcnc_=6X*!m+evz{K8gTcq#SPtzUf1sh4XpM#u}{L7f^|L}D*e5Aol)sD
z+q@|U=dBBwaCBztIp(`2kM8WN^m=@?#%A}A?V5?pt_6h%rxk79y;u9d<~7UePq(Tq
zxFXygf2$_oo7&~JtA`)$Yxo`Ue{$mUhYQ!Re{FAY)NzWkWr@k2bX)!R36<&l*ZCGn
zt*DaNJ|pG-?;O7<<7f9??H94I-1}hl*R@l>o^+4edGTHP+B(PEU9)s$s_Ke!<}d`G
zQ*e&Eaq;taiDh3cLUz^i9A>JT_ikI{v($_AS?@MZPF~4tyK?i}-!d~r*3J{KO@6TC
zryCdd!<y)I&CyP-n-{J)SuGpp5D~zcG`G%nL&G5n&mH;h2J0HCPCjZ^xw*`)sn^0t
zEYqd%gUP&<h0mgoU7sG~G<}7>efw%_Zbr@U<QT?@SLPjE$@%XRZ`6ldPBy1@ONZM?
zc)LspyjXwh^3i71-pptQ%e~t#>20}w@_6?Ysjn^^whJuQYKN_=`(>QJlIe~PmncWC
zHn+sDz#}s`T`MoMy*vB)_J5Ce+wPZ4%gGa&#^vvRO<It-?V<Sl*#9$Xv`#KxZe+G$
zjr`?|T~)UqO%c1J{&=EZ*y%KtmG+z7Pn{R6ZKdY8!*~X}alMp~NJ;CfX4ZD~U~ZRX
z3w0K%<>ddCkvVAo{PypI+wWXlq2>D3Q*r6>1i=@@?5vzD3iXUpT*{_htBieYB17&Z
z#~f+&+}=KK(IJr^N%hB0^vtl|y&~?n{kCU3heD<l?z^7(y~Tcu#Fl;EKeg~K+m&>8
zvSDkjnnK^&z5CL9o?ecxKXS$8RlMB8mj)HQ@hjE{Gn}4!?~Kb0N#6N>eC!%swdbww
zWEpZVw^lGyGMRnjT*r$qovkWrGyIG<-YtCd`VFt|=|i!1OOo}!*c|dHda~@m$Ct}a
z6eL+6K6$y(C$l4YZfFsw-MyD$PPgN;tL4@`f4^<>+4U!#H|27(O1WN>44GH|ZfdX=
zr$N@eSZzZ&BZsW$TX9Rn`Qn!SH{<sDJF&Jnc8=tVNl#Y4EGjjW7HD*F_#G84xcQ>S
zgNfp^rev--<9N|tJ9g))4{PQq<}BO%xIKMFqGW|t|E}dBde^7gIfncz|IcTsvAw`l
z<m!t+{w^E-0NIcCdRClL-t%(>d$LmU`T7+i9`S;=B#zzgtYUbm<Ttm=Q|+3o?UX7e
z)&!p5l6kE63X+cZS2dhp^8Tk<q$zW?=X1V-y2@Si^FKF#eSh7v&f)Rhs;?H+XFlbb
zZGNs^+7ep0VzYew`9JTYycU^q75=g4I2IAg{g25uR`k=9mv1NeT#>(Ne*LIt$Ci(I
zMuM*OPdl6+9Y3&w^QLQy&4K9>43lL)o{|x12-sH?s@R{taNgr~8<C4195*I%W;dKs
z4t#ps|GMfb&ax#g+`N}~S}rncbqKIA8a-TH#q{tRlU7e<cVxcELzdefAsY+~7U@@Y
zT{!jT-UTxu@tz|ak6ubO+^DyJsqr}DnrZj#rTg+V7Stc=P~d-Ix#j!8HE(uuuV9;;
z=wQ+G#KiuW*3S&lfUoO+_i}brZq@LKb#PUB$mnK$=G;@35S`xtDi_k5cQpjeY<_gC
z>SY)A3bukn8e$jBy*_47IpurI@8|1zA6mS7=V~y?xZZIR_MUih?TU0UwTU6ZI!jFJ
zW*o8A|K^(FRJmn+z3gF+*JX!P-V}Fr7+A4Xe>$?ufYV#SBV|jy6>DF<=50N;o(BQ8
zX-Vf_IUQ1oy%jjst$zNQkeiC3GG7aqcG<0B@N)Nf8D--=?fdGlZpVYfrqwm=UEZ;D
z<D}lD`@gA5>}e5uSypEi-Fk1!*@Z&FLe2*wejkbRThZeBYVzrIRr2+v#Q_XXKTn0W
zda0_^XzhM_#@XV_SL+D7!>?k_{IHG=6mgL={pdC;wQh1_#J^qYHV=cC4VhXyCYVI-
z$oDzFc}90qit>z)$$E!c&u<7kdhVNRhD7GwfOYf5T{XV>Y|HhZA0b|x_-L_4pRP=2
zU_p(y;dG8vhMldg7W)k?&X%Ot8(yy!E3ws^mf14(j{Y|7tK!Q1hs><rMXAe+pIkmO
z@BbG5Lpo>v{%92MYT-{#b85PO!)guF`hTAFspgxrw7XtwIL9QdNlV>wV08%dXK(Gr
ztDl+#@dkL<GFli%uGVhI==X@=dAwum%hRbY3lAUd<XfcdIAwyyr`C&{@8SgZcGX9z
z)ISJ2E|k4Es{O*o9*wRU`b$oF{)l#d;h)*j-*C0@%rB1njU_4*lGs{A_Iax1?S5^3
z=2Ly#uEPs+RHiKCNM!0gS2?e5)9WX7GT)9xwY7G%q^SA)77=@Q#$`tHO%H>|E?)24
z(>Z3IKg+eMZ~pIpA*&WY&-(Udg3j~YTtlhE`snjNEU)ynM4gi4uU&lRjkI#rg4mhM
z*Svo9Zt=D~lMjebe6v7Z=5OW9Z8i<BK1}4?9>VDI@qyRNQ>GW^U(iZDba8&L(-hW)
z#}~ZQGShHbzgXn6JWEA(bD(>JUuU3`0(ZoUzU~{zBHVthn~pp`^1>=^rSpuJAvyd0
zo(jKx_h#*v`ZI6xmPlE=t5Tn36J$CgdY`|&_p{k+_eY-D|5qz=-;d4jUdA!)@tYyU
zar2FrrB!LriDdmJjR(|Tvb<gQ#Nb`a;^6$t8@Plz1+;B?`;FX}TfGsL^_Nrb+4ZSP
zZF51N-`>C^u1^XRHU_=zbpGJ&xyeMAAxV<ka#3m2@qk}VN{j1lE=w<-B69JT{dp4y
zk>l;ln9jOrvl#Mqy!g;6bFpe>^=8FyOj0vsf>tmrS|IvVJKSZ<v?-5*pFW9hNep;p
zdgD8%{sP~(_U@1URxxxi_ANB&d{U)(VC#h)k5@HJWwPk2@M3+SIDvt~l%<;W#KfX5
zra8S2bu%u%J8P8wT;kvJ^!kN@EJxHn-`RNK@;WU$CmvR{uVuIXzDcanJ??a0<rKp~
z6_G<vY7W2DDAw6>IPPNZhs!$xB}4`0A39_y_RB5NO_QrZF#Gyx>%A3XRo}iJZ@;_5
ze(hx6P5)-pY+b(0TkgT~qASzRH~(E6pBl<M>G;n_iDmn5+_B9!>t>8zZj~x|x?bL4
z-;}1AZ+PxW%w&(>mU-lY&9Tg6*~%4vgI76sNrYDRY<*o^=xvZ>_<TvTR)NodhhqyX
z16>*)GzMPo?#kh=irZHddh?>$SuvTldEI7bYF;WXW<Is|LYXel&c_>;yGV1VoaqRZ
zy?3j0^_$n?iYFNRmdMRCWzlb3yszb9#?yv+{aG?IeT072#Gk4VE$=&7psGLDs9*GG
zfPRndVvS{Yi)D{JExhwHcUiOIyeRdQz309-&3t@L_8Zp?Wr5=P5~4?D96DtA?vI=J
zr4!<gEk0%o*_UzZsWw;dp0##zIseJK-!vv#pTDz>^PS9UTg@K9Wim?Vm#jP8yD|2I
z>ZTPF6qW0{COl%>{US!_*T(Y+$6qG6KkxH<9whVI;d=f{rV7_>tPRuGXEZqT*y`v@
zKIl6)S1z~Xh|8~eO{W%QoVMSUu;t~Zy<u|`WM^k{nrYR}(lb)aE`Ph}>X~PJ4Gzbj
ziIqHFo}FC0`taT<6BXNJPu6^NWbM~@;xg;VvA6%e81&6ho>`y&xkR<s_DrwWq<MWt
z?nc&cM9nWL_b1-}u72>I$iiicr({mrXr2jp66m+vV~N#_xv$R5pPkaTsqfnN&HXR8
ztlD&XSKNuqm#e1!OZfI}^Xh{&x)RG2K3STcD%fr4#S(cnL5zuO<(yf1ZIZ_ulJsN_
z*Q80c8TN0OFw5YK=uI((xcYsQau*i#*;HI)DChb7B*Xiw1ox>cclQ}=v23fXy6hIw
z#kIic^t^=ENi*0T9^8~nGyY=vuJG20IZF(LHhMn&QRbg5al-1|N3+R1wT%nJR~_%`
zH#%->^+r@zhDYh7^W?izHFf4Zmdl^A-BBjtO|;IE=kdn&b44<K%u4fl$8GYs{>}IM
z|Ej-PRaP**%=tX!&e?g>U(PxI&Th%MB%|HlvKPO;ay*~9|H6^NiIbD0g5P;d@X6Uc
zm*1N3>&?w|SGszRNN_EZIHqSb$)HKO(>fx~-K|e{qwusYoo8z299_4uwoG52Vwdsl
zBCE0t<BPovHx6-F8(k91l@QddNqVGwN8(n!iDc&Gji)9~+?LSX^mCtyR$RdD9Mj1Y
zO*wch6W$tb_b$)%EN_)_IFOnZx5v4MiOc5AOqCg1b~hb<ab>6S{t(%mb2?T`IZ+or
z30xMKsd9fV->uIR*DrV6S@Y<?!yinAXC~?&QrPfep1a|`=VdFCrH@zzRYaZOI?sG0
zykFMj!S4Es>kIZJ6@1vWF`{7EqE?4%LE7)0`QLtC%GGf5o3qjH?LUR8HvjzF&A)p0
z&6>*f+PZ%n7sS5a<D|;Qom-Ui<;XXA!(&ccpZTrv5kKJO%xz}HA+c;#Nu;=RufZq7
zPU}^3EX}H#UM=%jCbxz0koG2Bg*`Q{$~OOBGzrezc=603j(TgeTVlBv_&wivPTuk1
z*o9>`GS0W<73XZxy<xFjCqXLA_TaUnZu|`n$KAwBoHt$+D4Lzj)z>@kisG_N2BDTq
zE8YgK*0$*Ji)E0`WsW-j$d@bX0JA>t!Jc17=JMRiPF;L=-KnFGB^2d41y&r{=54UA
zueWT%+a{gv6;lovdzMMpTRf0HVtRvD?ehFX&bo@yd%{dBH+~mlG|~Qi%+mKy*+$I`
z$8Kyq7S2&6#3`^L;O`UeWJ`suS~q9^_SxMZ#idnn?$3qy9&LqdKE7}b(d-jArpB@U
z#I_y!LAxeh4YWFOX5%rwT<c4YZ9)w<G-kipcHvpX<C6!Er80QFnK^-V>*=t1Z@$ll
zOuJ7q9hcg^`B=v^?T52gwkP*4FYORqGU5H4o&T(^PI60M94vl7>b6W%QO@lp3w&*3
z4onoxuCdD49V<Bf@SB31)?_)hL@rD2_mv_t^KUPmX4X-<=Hr%S=033nJU&L|Y<xf5
z5_-A+b*QGaC-kPjev)`y_}%pD_v_^&R@65xo~VA6X<FMN4n@%u^@&^BReR^`zwvHy
zn$*my9!B|+Giw%4bXWhjNaB;g>>W2mZ;A<kN?emop`}dj%F&B@ou#}OeEB}#mAGZn
zn=|>r<BXK(4T&wSA6A^)`1e-o>X~lqoE7~4zYV-|>f#C3yBAD~Jk9N7V}6`^6L+X#
zXZq&)9E)SsQTOL^sx@CwY0CW`)gJhtJ7Lf7%!vE!0^CyzY@RI9YJ5E5@{QnPyGAby
zySoo7mYnrDWUJG=>G!e&8y{bmP1&vBy^*6<@ruA3<#;)xQ>7L+nCB{gc=^-9uOWwd
z{=A9Lk1UOeU%bD!LTBZM0=>50HRlgao&F&5!t}(L*G2WKPGvT4RrtE|ywlzr*VZj$
z<xrS)i`BEo<L{qky1!m~{N?+b8<*pxDA=ZUJ52PZnE32VqHDWYXT}<>d_U{0Z<pDW
zU1mzf>iZ40WCq>m%IUkZokhA^StI%Wp)I8oe=#dunD~qJZom?|MJLqGUNd}p`I6>|
zX2CQuwmsc`8!xToPN+Y%k~Kq?-9j^?v4=Un<r??F0}D!HwjJIm+Gojbl;Xx25&Ymr
zs7N6w&GHC1$H{zcf8)A?ZTCT+L#fjr)Xd77GI!F$_$B+Qwm*KfzA&)!(G0J}Y~|)|
zCmnuGiU_!~Frje8rJ(HT*OYEPy&Q2Q_U~4sF6;f7-BA`@yPb8V1pVq2((JzZ_FV~0
ziFT;?+2vN@GNa>(s<I5DM$FI2nHP`Q@Yo8>tgta}GxIIHz2uu^*X>(dTs0aD%oaY*
zn!YxztD1qus%!TGY5TnmOiJ%g&5vP>*4TTaSuNf1R_wvVl2=R`dD85M0-Gl{O;hS{
zJfgFwFVNw&MqYcewu~@?vz%o8)p@g+j`uw(oA9<p=dq2@rIhfr2MzfwUAqrBM>?5B
zti68nZrR2ttyR^VD}3%3?DGjbziP$F*iTE=ueo>j(EId*?LUiV+E`5b@u@b+LrI`4
z^>tt4M{b4Am8_oNWGptx;*d=8x|WNcLgHzsCmt!wy4X>`W7}1mxrsMuVlHF-!>8JP
zDrFhj4${37<-N|U#OyZQVmYazXal4CWrh;d_qlP~W_dnh6nNqBs4354(vPeryUxld
z-jll&->!RLU{%VfvXM7KcBh$U#>OKpOlgxi3|TW|J;9lh?NLBb-b35|ret>^>l0-{
z84Q*4uCQ9F@%?a9=;Y?#+LH63s(xG4IkD5b+n@WMWB)I+r{eG9^Yt_LdjEZM{A$t0
z*tL^?-~6{Yeqa4ZwI6=d&+7fv`Jbir;--J%ggWQb{{O#jUH@;_oV}cr+S$2f3uQhp
z(P~n9S1lA(p?*E@fUael$YOzP&8us3UH9vM+aR!{`Sux$XMdLIW<9ic#<w-Y#UjNh
zt)6MN4ih9-Pg0aIl&hNFQ*hRBca-$xMZFag-VDBFCV4!$vy!*nx=<jqS@i}(Pvry2
z&<ks0x6QJA#whT@;#pIkMxUf%c=qE@vPl|B4>&U>MO=Bu#Rn>~46DT!l-{|sWWiFC
zZwVr?Tp9Y4lecBQdstwjEB%n&heczLK)sB0-%ZDD!kZctS!SL*xU`XZpP<DvCjL~;
zn?1>TstWnnKR<m|xbf!1<NIsucNbfp2xrO{`yJP_u+?>6*<<dP-f6)rl+u_qvM(Iv
zSANvY;F1+(_46UvlRqAUJ?Ur5*J-nOvP$YrvCVJ1!)5H>T)5U^cF0wrQti)&-ui`S
zKOVl^tlOvgeQDJD>ZY}u1CGArDEU+1Xl_&|aeK<vy@_={{xdQ#{Qv)EZfx;wvt14U
zxzwwR&$Y(Z)R&meWppg?PrJ14xZn3Tcl(mGH~Pt*RMtFOR{vK~!%{6UFhziK*_-Qg
zGjoC#DLkrd6z6yRP&T*r=g03S@4gOSee=)b`nz93zdnCuzdpY6*Yw{fzkd0u{iJ?>
z+#TI%>*Iax){C#N`|)zUea+9p+PV7E%r8HEwS0e={@%Sms}>x+`)1k4X`B9il(swb
z?p>2}jLzTP?>Fnu_*-Eg>38qWp1Pw;RCZ2#6}q?7_V4#w`BQJp*DrXQRaP7+$|Jh;
z!il=OB^$p;yVmz{x$Mi+SthSG|NXW4x}UB0pL9%UtqPr<6g^}0SNZbG?fIY2?fKfe
z_TI*uyMLWOET>~;9rl0Y>TlcRx2^y8@U@}&|KHz~Lzcf#Z<(;p|61DTXB%!`4&R^j
z`~Tb5>;G@Jx-$Khj?laPFMjp^Ud5IF&-P|b$m&bix_|Gu_5ai3yI-8^RV?G{YW7F0
z7oVSWfA5^yJ+g{hB8-9-K0p6cV!_@!bJqOdU3a6zcJBJ+r`6ql@>s}OAHS3EPG!0B
z!y9&n8!LG>c3s)Zc0w(y$}+r1>igRheM+Vk8fVn%wmtp9mvaBi_0vbsK7ai+$MC$E
z{HCwdkKW9(-FsfI_wRg;IjWy#N?e#z-;t)-p%&(zvVZx#Ye|KD=K{?a{+oZ+d;c+?
z$KAF$T)P&gT~AZJ-uLKl`nTA>|Bl^!#lhXWVB&pgvAHeAjgs3IEtf01n%n&0+dN^#
zISOme8+2y0n6lk^l;Iqf9Qu@9zki$e=hN|1IvRZq8+12iD$kknQ-FKJ=G#W0-&0=a
zJmjcn(K~&#=56K9xV?3$(>9;Hn^yieRQ_JetNHAEel@OsRc!P>B6qf5SoY>wzH$3!
zojd<;+V{tQQjV@ne*B0hTWLiP2fyB{SuGztO!l1Sbq@JiEGHU#*G)-l*}ORR$F`P5
z>RT3EywvwdCa^`*Rw%=fo%wj640qq0b5F}u{yvdu=BsB*sh%CDEu+45OOMmH*=~(N
z3+HsJZMuK_*||j~e7{XI=dD?nzPWwR&R@T;oe7Wsf3NP|_KmVtk0Y--*Tnc;yF5Fm
zsMa=i?WU7|)T-@w{(HO4O7Uj9mgoPcMoMxA)YW;}gElQ%Qo8=<p+D|R{HJ+!_)glm
zT0VQ0>KYxROW}Fz{p;_^mH0<A_`l=q&O4;4Y$@eZ5wM;2(CzB1Lpu*mdY-H%Q&UuR
zOG0?sr%mdw<}2qum^}G#8%w!r;lD4n$~m=mKN(n>{PT_+yQaRn;{(I|IVOs3A=;-4
z1f_%TdMRryb9*qkkY&q*hm)MTTLVQlD;BE-IdI!aTJ$}dAsQ6me*9AXk0yV%l<J9F
zuW07lMg^V-I{jwZu|*SjH{U+ob)W6n+o1f@Wq*V!{yBgB!~f}g#{q_ob1YOBay|b!
zp;FQ@W=_>=S<PvJZe3;f^e+pY771Kdt+y-h$fs`$st++8De{jf5z?D^W!;j08MQ6!
zrMDIK>}5Y<bm5w#bW*D(PnmwdnLz#NlZlUyS)Thb;aqk4_N%(n_#G5Fp1S4a94$H{
z#Go*3vf7ajK}8AEjfPj>nk`DoVo9=8Yh+NIel2f`>r8LsBY8_2#Euvh$Q(6HpC#OQ
z{E%Tss>3$^$dxjhBIlAPz7;#&4DyRh(oV%SebG$k>oun7=lzd3yH36Q=IxX6ms9JX
zJ=;F{=hrWnuhj3Gf8o#Tr?<D8DU~?bs`tutO`UedwI|MhZ)|_a`It$GuOeUVxssaO
zQ+@R_*nd}w{UhEe&Ji~{-yWmR|7x3KMK%A5^aX31%@SEQZL_P@?=cHFe=_m#x!{BC
zKB>2F=9aF0&Md(3YNw#I^w%><%i5c&{ASe)aBId+J;%G#x4K@dV#W-q#;Rq47n(MC
zzqU_FoOn%couLEwD{y>Gb;>@dd@RfTp~Mwqh1=~qjyn_Ab(|C5b@j7C%HcO=6RYw#
zRzIG0;IsUy=kxFX`SU&JmcTi|M5bAmxl=eEP7xFgEP9n&u_W!7#PdfRBX4XLF<ulC
zaP4w^YMu$iJ0`{c5hcs@CS9EKUAJ@Z>OAR|?cW>H*lt;6IPWyl+y0M3zsKytnUjga
zJC$7RGC~*3y&JYY&5ogQ!A$9m8?T-*%r(q-wuIAA>_*}|{>Z*GsBhBJ1TQqr^XAU=
z(-dxA_H=oJn2B+L%<+xKc5NuNvtaA1cQj_+wD*Kk5XZS%!9^3^UONMcuvxGOTQRe~
z^ybk=r%PtW`qr(tU^s2_OJSkcXRa^eFCHlAZq{2_{IBHgo4d1wd5W9_PFCk_xDoN}
zt>gC=)`(~GHoQ=o^JIyL<uYatNc{Nw?oeK{M0xuembnWxKC(}<oK$ai&QniQvMBnh
z@#3}H&b-QzYL>2e=5c39abLvC_CxnOqUWdEF*HAzDSff9&~25CV{^LP{~Z%U1Vd+Z
zsO|ADbvb&4!^XJ3LF`XVP)1~=h^3kHJnN}UQ6J76C=@yFF=bA8O>fKN<$kM9OC%jM
zUhaBL(6><X?w7LaXO1V@Zssc0pT83jc3ijd`F*`<r!#YBhTp7W+&t4?-=jh?=G-a8
z?a{Wz`b|eI`#zqL34XGi_syHTv+6EO9NqfXa&z?gJ(e%#ZD2TmM_6ob3wM%t^hJx7
z4`p}FIP>3HYMz(b?50uiEaKLVHJehh*<!C&#BI4cFS0q^uUoosij~mC9bc=Sy-%+f
zuqjry+LL~a>%FIi)nR6v;!9H^CN*(8m*n`o4U{!BG-W!kqr8CCetKBhg09FozM1Fe
zz<d`W)ZATj;_g*V3%2t-C%!43-l59DyK8xY`lf1|&jO%CIW@g51nj&06|-;NK3QY8
z_w<`bmw(@^2|AxMZ%tVD>znl{wrjPDx9vO2_4}CqjX#;2&Yjrr@-5SF{#x<VTvIwU
znE2<KbgT#t6A+xZe@>Nf|4~cJW$o+YR6XBoH9T0@zwgk>kXs5O3$qS5Gw0s-k>1X?
z`48*1<qsClXWG6<hs!!APp#s`^6wWk++H@e`^^%XEaVyX=|$nw&z~=7F-3hS&#X`F
zm|!`%;=&T^4;K!6%vir%z;UK?^RuIk?e~0jDqbu(5j3$l+tXyzgN4<H7g*k}G(7*~
z{-*i~^=YAr%gQ(Z%)i<GGUacjWozd7#r>k1trKM2&wQ*{xa+3GA-0!kN-}AD{IkC8
zu?0obU2rrNAI%6AZ~P{%s^{HwyM?{99#kkkFDWr@bDeYflAFd<w$}dq<Xai<Ty#U{
zYv;z9XS40H)Xn3v^f{!OuBqNM<Cj#C|M9D9XY(IWc=61wBxa7F&z6SF^N;22b_rd4
z<j)@$?6Nz?*6zp>dr2D~pI!&^<1riAe5Z@t(wZ9TB>8{ZzQP!;3`cY3<ApN8=X9Rc
z&%0q*z_Uwo!dqvt!lkQw4%X)|N5-uR68<<Nx?Jh@+S%-i(rMc}ONyN%Ww*1;t1mYH
zZSkEuGv(VQ|D1F8VpBur3X3S4A23ko<=0uwHgnPf*L0oZm$=@0&RZ*Z@X;JMiDz%c
zt8Z*cXH0M95jzr|w>~j%j;_I(!o-W77O7LF?y8qCdal3XHrvCuwh1#DxAEO_(d1av
zyzYjr@vQ>}_KbNq9v%{V<3HVy^^wy0_G7=T*X13qVPHFGV7@3hdq(oJwx&eCSwh}N
zw`%1Ym`{0rE@-9vQ`dT<iWxJist)@k>2){E^tSFRURQ89EW(RXS0e3!MOX47u9FdR
zDkfJgqC`{cl><GG8ro<GOnB=YR&cT?_*dGZnffAKR+09SYuRt+9k$`PX|UTjWrj4X
zWge%^|A);?_xD8Z`Tg+v>FxQs*|N`b%+A;C?BS4{aL!Y?P4bka7V}G=mTQ*lAHTTi
zC8V=>o|Hv|Rq%3-WwquO%_VlfwxlWkZO(67v!45s<dV<jm+IX%X!9EW%TSoh{+R8O
zrI2|`W@MqLN#2?!*Q{?R!qShkUb1v7eD&gnN4li3v&a%q_MGJukkk~-y1{Vif_1g4
z4{%RBXTcOTq11Dst5bsanrvt8OFLVQJf44)-V^jJ;zGwoDUb9C=lq*>oR(fwn~{EG
z#^P(Y7%eS4p5HUzSX%#i`74&Q>sI$z&pwmC>HF!UPnP}5HjVvfx4Z4B$uW5o8SRG!
zwi_xF=c!z>Ha__2*TfR*>fOsOF0U)T@9blKud@1rZ6f2no$5MPoasAtxU>H@Y_#lg
zIU^RfofB5%TC6^n?AJT9pozoM*<_I*+j|*r=|yW!IlsDmeF<;<bXD!{bl>Q=b1f^L
zFWE6MNbH1U{T8*nSyq~v)r*dv>sx)>TE4;Y!@1rUJ1mrs==(|v$~~BSxlg%RIrn7~
zC$mSepJio;)Lg@>YwbYIgL{_GE}8z=wdBI1ONs}tl<{mTmzmBK_2C?-G`{4yy5fh3
zV_%@7*I$lYt-eG{UWfW^{E;ahrhAI_&CGcA!0SoSrd=Oz^yh3ZmutzHaaZlrIo{OL
zJ3<xzX0JA1FMd;h`ps9L|Nc$d$;SCtN5w?k<@rP92@@AwF?hKCxlMFt(^Jd72^Yd1
zE-KV7-rn*oOgyYToAdp&3%8r!adJ<%V4HYx?llpe@K65UChxq?H}A?7xL1FVSx745
zb&mhm$FsYo^?S@FT<~v@o;qjBmN?6Gjk7lh8D`(+{BW*sb+nUwgX4q?y)SNfWaud!
zVpb`+<*54DLfmG;g*icsRI~ikZq4VBQ(NYyurN7@V?w#-!qisXmX#m>9NY6)<3dvl
zTi@fFmbDssua|y4e!$Qp1Jp{IXnm?aMDEFhB`L=qMg9Mxa9ejFB)hAWK(hP)2ZBxe
zW778TdXw}qb=R47YWu@gQr2z!;O26sq1117Qg))x!=~Td@lC3oTh6(9O0Lb$36H<N
zW$vmg9tHm|#l6;986q)Hr9#Bed5XqLC%N-!%NMNK7O^dzoekU=T&}jKJ}`apwsVH_
znAa5R<X(NzZ#nt5vdQAO=|SQ*Y}0RLy_R8>ws`JwV}ro7Qw4!8=ZgDsHYmM4m$&-1
zrF?@!!MWZWAI?d3uXx&jz;gLjxf2tdrax3N^9fdNN-uur&&n?xtXyzThbgL{)N|oc
z7v-Ld6SD7WE?|AmBjaA^Qygx(SF3&o^Y;{)=3+KbH^4VA?$V-Oor_zpq;7Dy?w7KA
zF0^e`d~e#@AF3bz96s_qUDolE;5kk1X=@yJ8N|%k+xu_j_V*F<CUda*wrIRkuX^+5
z*Q~oM56!TYTyg#`*Db+Fqp<n90T;jpUeUFQPOavbxgM_IoOew03aHyuKkw}?qxjc3
zXSk%oYW=*V#Vqw@c=cWyhDj}Y)#@@|H+O4On88O&&AdAa&MtFbnH4NPBbD5=w6&}(
z=C05$2A)1gr|tXSee>>J*i(9<!q|68$Hbb1WtUG09?jHqHMo?rQ}NELy^${SWjeXG
z2+qiqW7L({d!%`{Os0q(OZM`M-t~;SGcrMiqFzy>h`f=%9_#UhWsEoNQaFX&B7|a2
zip`2{?q>Yyw0$NdLl|AI{`6XT#!GK|&Z=7LvvudzZ+iNrsw#dx*IU8!>%`~ZJO6C*
z`g_0bhPqv~bTl+$nNo7=f~HRD?2FmPSGQ?>H=g^w_1sGHHxctLbExV{PC8NF?^E`z
z(pOYVXVPBp-25q9Hpxt%ZTV!ulE;S@O)hY2J+yA#g)1sC(WzRWqM==g<!XBZ(;tUk
z`oeS6<`>tL5}n)=Tv;nVx@Rm?>X2OO`B(D%RI~L<47)1RL?vEE++qp;xAMa}U+JsU
zrZa7S|KrBZ{8H(c%nBTCRSgsKTI;>78Rl=vnPRX{-SSmWs=-sw%jM6!(hGM;s6O2^
zW!t=)wkCzPKFkJ8!E(j%d=kf&TnM{qk)~4Pa@z0k45lc9rw0mUj`uuOHFp+E^gn1g
zqsU>KdHewj_jP_hrB6&+q_Jib^Mto%Pn#$38sBzVRujGL@$>_q|68=*mi~P8jQ{O<
z^=FoMpI_Z(7GHB>U03;6mv={hmF+sSE?@og&o6mX&puml-d|%L%QDsLjqfJ52%fok
z@6Ycm(bYQVBU|juJ65-`zrOao<@TiR4NGsWJs;Uw!s#6;?`oL1hWqHr2`?EO@0xMs
zzrU$7`*TG4g7b~r*cW<E6w{h>@vl$@bM3r&W|=W7>jP{TZ*i1f)D^ntO6Gxob*rQ2
z%QHA$c&@(ge5i{~vz`B}eZ@Wzjg>bTB4T>mJgf5#{5$g1fcF5$mz_NgVq21fGA2%4
zI_(<2#C}6}E3PdI&Nn{R=rb%=4R_tJOnUJ(uXDGvcP)MJV<kI7CC|nM=PxT|YZz7j
zIKq6;r*lW<`Du6SJ9D<%$+zUpxT$vO`Rz@z`&s7wKm9UxYP7%cnoZ^5(!BF_&c3<p
z)Xlf4-xmIQ{@mRDzpYZ(fu7~QYembtRwk?!&f`89qs&s)a%!ddlZd&CIgIB;KY8+4
zwK>oDoUXR%slS<w#{XKrt+3i%_xI+5b&Ka#Li^VTdYE%#w}{qn(fH)JUEJRI)`6bo
z%XzaW?T%ro<#F8PRd@5T>zi|Rt8e$nGdLPNUY>i*)#p58noZ`fa}l09To=xHI78s%
z-)jpyUnfd62+sI8<FsXh$Ydw)3922JJ9_?TTxgPH?|Up2AO2Up<bmJ3o@9f^4LSLj
zXHTE>(rC_()JVl*8-|;C#`Sg_Hw`xXrX1UzYF)=^^Z#$!o%P~&yJy}z`)q&e<@@60
z^Z&+t?s&i$I&a&~@K)V9JsI!(_onuTJoWOjxBizHyj^Shl~DV`^L$ovIa#hbSR^vn
zWx|H2l@4)Tb8gSsVfGN31n#JPj7U!od%8iVBIuTY+LkFd7GGMxt9^vGUi0<&igjCZ
zLn=jb7o1Yge7w^%HL|6Av-0~>Qq3FA)vb=6D$n2;@O1f_Q!7<Wl%_dORrkGbxV`Db
zuD>}Z3|oFVuBqU8bB=An?Q1Tk4F2m+WuDm@70~0kP_$KeSIR%#vpshNPW5D;Pw`=V
zYY}Q$U{hD_v@~#&+P0fNl5{0c#TM1~&97oeNa|n>ohKGKJzVCg<GuqsCiYEWJW;m#
zljDiDnYl`#`*d%A^nUO&fBnHvr&G(eUVS9^kkQB{w<IG(>*!%N*~FzAZb{27zgL!A
z+Sjjnc%rIP*4521k|#bSb1P>}&?(&U>cF|vwhD(G+v?ZP)5)1<{k;F%p$ulpdGpK?
zTkEGR`r@7uS0%_-JbC+dXJ$!0qd(H8+*U48IP5-uU-40&b%)ZQge7Y{;<{X?bdY=E
zLqk~sk0V=>{qklwo6In9$UV1Hn7L}LxZ>f&dp}-Sip`LUZRu|oTza(YmDN(;Q-)`x
z*;9T`P(5Mf?yBJ2qOF;JP`JmTt-cx*)t`_36{>%_`L)*nZ>#&`<8IaO{dx38dH%nL
zo32)En&;Lc=<#_;po?xg(`QBVeT^pH<IhYzp?Kf_-J5^2-U?4+-JE)UbC$y_`Pf<B
z>jZS0uB53$W1BIPZ%gTgXZib7iUO*CoylO%1m*RqEB<j8CCt0bG}|rw?#=vC@6XH%
z9GB|<Y*}jAb~NnRdwFm7UXy>iy=)C3x(r&c*E$PdKWxIFJmHyPc$I^hT5H&owx?R(
zd5@Yu^KfqzTVh!tldfQ@E&2Ph(LB|w=epcC&*)07=u%g>efx`&|1_>otbHO$r?%za
zyV;qu-2{@uCq0Aa@Vd;puixH0`lJ<Iudn|3>hhaePt|AteqI@8XY<5ZX@bLS^<HHk
z*|U<Lrq7I<@vQU8Gu4MYw)1A#a<7*N4T#bIb7l3}ohf_n=qzHnWuvEK#ksskeAAPI
zXT+2Qj#d_!yTV4$q)W}b4!u`f{q<={+}yb@&mMj9*d*RheShwXk4|&vE}S8##!>$@
zf={r_a$Ehb=e<FiD>L-ff3LBq5qK52r7@sC_iM(wHO6&wKe6TBDA4Pwnz?nJ+}h|B
z@(c$*a(;a~UCLSVb;nK%WuA{KSf@M2sQWIk)a8oZmGIQ*l%=R^E%);aS3VjvnlTlg
zRa(IMaJop;${C@lJGT5i7Z}f#;TQuNiWIKzo|>9=V$()V3${XDP#>Br%HT*-;)3tj
zqF(Mw_&xEh*mcN&$fGL%y1ycis!TtB-mCuk<)hDBI^T>rv|>V*6#t*LELzbgB=^s~
zx(8Qg*Tg)OP28w@=#t&N=u5d9S&Jkzg(_O^ibWT#`Mk$2`uwS+$5zf2Nei0=)hxNE
z*ROMIF?`tdO=`E}Q!|15_bUa}ML)Aj8$Xhc)K!XlczMdz#Rqz|oTnZ=$gy!1SM%qy
zk+th~<mQz29I^0?w&l8`l72){GNw(9b(Mt4+iqp+-RtsVYZ=&%FJHYqK8q)Hjq|bB
zSv@Xik~S`1xjLBb?DWH0!Z{{aT@9}H7tfORiLI|+QmQ?9g@<(G`e%Yy<}G_}9h;?k
zc-n~qq2qoRr#D3jTw^FwIU!`N#~xeh$!H{zov>|sWTW?SZ)uI}My}8ud-CkQ8Sj37
zaGKG5jfJ*n+9dNfu}5D66{R%{CSkAR^f#^DyzXB_?Y{l-I@fPLaxUBZ<4c{{{G6~a
z|0cxlJ-cRQy>02&+kcZxKU-_1C^5Y<N&U+Co$=6(L#<l<M(r~e*3Ilqe;u~LYPsOH
zTVhjIC~k|2nk5{&{`1!(+o~Amo{xG_uxR3huV<6b{%ntW(e>~6z6AHTW)k`DS4!%0
zZqEwNUVQDU;r2Q8{p;#p?okYtXbvox+r@FV)%2~|*O?!;eY|Q|FZ8&&=BxMfUd~3t
z)hDyQnJm+o$g!;U-J89;TE8+Xuq4{LL~K`DCfn>EW#P1Fon39eN4?_3KTi+v?-g<3
zeAc}8Cd-PTi%h%!sV#_p>SD^MZ1B?X`iuSU4ouq&7oWexwKp)Mmd%KD+k#7tPdp}R
zKXS=DymE`b+o$g0mkhZ*KHKZV3We=@y1d{0z2C5AVe~ToCNrM`;ger-Z}!DA$umyB
zyxVxqw%~oM7pL@>)rR{2uPfgg`bJRjv+X7sCbeg5r<;tYv>OX<TqyLpDF48ms^-3{
zmfbUN?-jhwKYLfM<<%WtEKBX%^{-F+zW?_Rqs{_1gK3REN-oK5Qk<vyPUXf4)Mvf5
zl#M(WKSf8LId77LpO1!KL}bLluZt(wwC#wUx&ML1Ex+KY{f%bFEQM#~{hqgevwKXU
z$ix1*rRnFlJ{S7&UD<*y*skr-kA{vyId!eM!pDwIKdg2*bd|zWXQl6@uR4_aGH%xS
z+;5!yZuzy-`X^4z)tS3O!0Cz70^#RQ!lC;0hLt8(2bGvN7oM5aB(%<@Af54)=CWII
zTU~2krC4;G6aIQsy?eRNQURSBizP=`xLuldF5126d7i3S$4vu$+gW;c|Jp0B)$iRu
zbItO)uF|(1rQfPUQ|p|**8k0UTkn44aObTDmwDUwh!icZo&Aa_FnHRIy=$JnJNU$K
z_J{BHg&FFtPydW%|89_6+sCGn^8ftr-S+eI)^wOHTDNeY_L?g}vG@1i{UfzzLCm|F
zOKHEKeto&jQh(>2y45eg{(AJ<@}>2@`r~g)f*8v!{~E;lJ-@!MZ=2)JpJ&2vemWi*
z`t9j#=}muo_`8o*J+Qv=*!tPk*=FhG|9)P6TC#PX&i-R3ONwgit3#*OM03BLe!qUt
zr>WtlQWCGvS{ZNOcS__o+eP26H>*B7ewVMG_ai@ktL#;8iFMBF_&opTMJ-pI{iE*B
z?d_ud^|^k>Ux#;HcfVf~UO(CK@AKp5U;jG4)^)|Ai^f}eGd;J>w%3_9UvFCX`uO$t
z{v39GdS(B<_&a|B{}*lE@$b({)>Av{f7MTJyem3=rLzFv=H#;l%&ciLPTt!m-?*S(
zFQZnbw(_!H-t2FTB8oi1B@&(zzu#&zGD*eoPx$3@Li^k0^XtkU{(I<B^t_|>jKtkt
z|9YQW%}^3c|FWU9{{OFiH+J-eOgLtcE|P!Oc-p(e?_N*;ZR8~pJVjGS_Ur;9Ez4WG
zRtDKHIV=(p*2wp0W0|ZIXx<@sV3LZn#1x++3Lolg*3T+DC@?K3{|E0S<tcg1&dUq8
zL>vgRmVO|#K!mH+U-s~Pp2$Y)4U$bwk_vxY8tq+LCkJLHgz!3?j#OBi<q&mNIksZQ
zSDPRMmAmJZTb(bZ$z>MIIq{*vxNl{g&kTitm-`<vaEQETjn6P>*Q!4K$!D(V<>+Xh
zGcSJgxZYY_uR6hCp3e@}mZKK@$=-?4+XYOr{)n&LKT+=Lzq3of9{9N<_iI@>!~77v
zsb58tlfrkum(=1|wyH}?z~XvgO2{(F{N2nmy)SpKY?WEh`oN>8I{xI}iA%q&zqW1t
z<qM3*PvjT5e~S89y6AoHiDiyk<dRkX>|Gw}^o8@C**wu%aRq<tC!Rgd)Wn$3`{Bg&
z*3=v+Hiy_!X+N_~Yjl>M>w1_HoA7s{?T5uP<P`MO6q_!*G7x`!FZpD|(U&H3&I%gr
z3pglpYFkEDlF=r?h)0{;*YHQ23%GrrN7$eH)+*L{p8D;7KJ0Y*s~B<6I_jI{E${1_
zygz?6X}l^@v!C<i`OZpT#q0HR=PY>f)L~{r{?4!^r>|cWeGwvQu-NXzvmoD(D>u#l
zeA#Ex9iK;s40SKw{ko57u2E3>x=x{DjRi@EuBrw5bMN0ZG5dUK%C)#HAItVveG5O|
zpO<>|eMD&F?D>UP=P&tqe(meZhxh7v=0{zYQ{QkRH2>&@R>@my55AKNF|+-YwZ%wT
z-^;CMRgc)!=x1LS*Sv4~_26UXTXz51|5yE6)Rpaj>9=w@IPJ2stgxK2b@JZJn)M~`
zJDPoE)XJ9i-bPQhg?R<W4zIWs*Wd8B-+im~cf0Y~xxy2*9_O&{kruaTNt8=}vx9km
zeQkVh<6@H}!&8>Z)&+`OHgD|fcj+r1N;+m5*ym^<I+f|Y!>X+v3LzG%A}gh5C7u-Z
z|HAuB<de!tgGSZq^@d+!&#z5$D)O0P^xIac%!$jyqGq03x2jg}X4Z9zeiN@exnsQN
z<^<k7&9@Ct#4{NAzo<?QF_0E@TAjrbrOzZ8s(dL;^5v<+t*1Ty%_!O3Bhuq}J9Wv$
z!bOjn&#@+^89tn(l>6_L@)D1IVo9k7m|nh7J#AcT=wqH^sFA#$TcloceV;>fh1o-~
zAi2pl9noC(8=W?-_!wXMFRUu$xB2vQ%}*|GGrfJ+vFO^)OKaE95!!w0?}0)`&dZ@!
z7O3p(R8H4A>vCQ8MbNBBpXd`W_cF}MxOXp{D?e%F-_=`XUzvP4p!sR`or`<I{@GnD
zR!_OCWYw=J^zYv0P^Yt;@66*wX4N|u?Cm_eob`vYL&wP@(_K&BFyra?{c78soTSLD
zbF8$dP5#>WRqfk>?MZA6J7%<gXbx-N8}Xgd`>2)Q^M|)&9A9^vcujoOwM<P-SlZD$
zhQH%0<7vxxzh<URs&u?va)4X*Nq~%f*a~|UeHZ`Jc5c<LqRTSv=cR3!)g*7G_|?#`
z{=BBpvdO_Rwz87bU#OaQ`TVq6#xz%pxy*$9X51=G%kZ-5x>=r+ZhOX9Y*THj{V(`p
z%EWyf%poocohH){%*e4^yXTc%qRjP)vDHS$&OVF&u!FNUAo8qsSoG`M&zsjhYc~0}
zlX;!A%pU!N*UyzGaB3CVMr<s}Y|qQyvA=!l*&ek;Qqohevm~D@Io9<`^z)yvJ=Lze
z%Q<TG|NQz9zcoJe<GJE5#asVR-zdo{%Ida>Gn{Mkylb-c(fPN{M2_EQe`C#2^-3<O
zYm&@ur?-bbt8#B(bbG@aabKcqTZ`8shfCY*e#yQ{TN=k5$#-jshn&i!M?XKlnDBgk
z@{_;q&N7nLYAGisFYL6Qmbsy&$lk}y`Sbg$f0pG-1x-AnF6e*bf{jsToy6Z?r+@Di
z|8a=nwM*6SDST%gR$i%h_PEq=)TQjv1TMXwQKcIFg7NE}PhFVDEq_yN+omuri`1Y)
zLBf;XbeM3<*qSW<+4SaM1aqtlvyzDQ9El%E9^$+K|7B9PGaQt@^+V=qOwTUYTdR#H
z+~#75n>%sWTaW*ri?+Y@_-48Cdf$^dlYhPSNI$M=XaB$^mD47sSWuWXwtl0dq2T;&
zFL@h+7acloz;#t{#<ko|#l%l5Tip{v1(S~NpK-z2@O$KwAF8eVXV-e{IrMMYrt9mf
zch(lJ)`_^16B&`$aKq6xXzI2_o^OgiYc9V%LshF=N=T>4;*$$oV$@3KJJZ<sn}s(0
z4r3@Q3^{ju`HsX@wZ6BO{WP1?;Cn~E{+nK3(fR$pN5Y!#Ug<cssFU0Ox{ht<Lzeks
zd7`tT3JRS+YBGM4d@yZOn|$}Ai~15B^EO2nA8N}CE&3FDan1*kh&uty`%9Lweb_Fp
zn6vzg^ZhNAy{ffO*{44|W$Scnnud@-q^hr9jJZhDq!-sZelm7jto`C+d^6SYb=ZOA
znA`P-ONwI`-Q(Qn>%ZlL)vdpto4)VL+tm=NQWM+qr1YE)^Q5Lp$FJ<ZuM#uiIEPKs
z7gGi1-W>KXqI{e7rTUhgIQI3YrqpW>^FM1Bs$P_F{5$<n!|SCk6RJ<lU1vPEbnU+b
zXWSlMUioE(?fk^@EN8);-xnPFBHy+w^=wZ0yFYQfzBlV<<tT?%Pr4}XbpKk@ksQSY
zJ(-l<@3U4vNL(G?Aj=WDYri;4$kYS>RQJ7me~gpYqNa3<_)JGj2Kg7gVjK3BOSnH|
z(~1+!Ysxt%xGG-wrK;<`O*1~bcvd*6@IEx@arIc#r1rX2>$G3mzZtPl)8BG5#vJ@x
z6H_-YBPy3iG<)Hz`rOZ_zsJchlKCmDQPHz%tJN}rhwjT3ho1Q}@n36aND}MrmBxoA
zYc6~ct2HAf>G+w68)N?de*PxW%-pVzH#h9;|BvhHRkuufc_i?SXUS_%qZtoo=BBRv
zpcAKFqi7Moy@J`a=$#mM@>Y{Mb?hv6f3Wjgg&jA#=eWE&nYZrp<oLY$6_P8QL{sNx
zSG^1Vc<9@axBbV@1%G@cTCX3rN8`#iACXsAtAp2vuZrvbc=wIz>1&&JR`j|h{{Fi9
z!hu(%vC*%6eOo^)X+Qp0udYz*%Q8!Q*P|ud|GZwCR#j(Pl_046;V>6}++Mw%%*T(O
z`@$W&YnPSR^j8uaFWt45__%P|$=H*eX;02Acq7ld_&=y6)_mbs`-NMK?H6t_wO_c!
z+<xH}OZ$aetnC+Wv9({g#om757DxMqTb%6|ZgI6=xW(Om;TBK(g<HJaFWllQ|I9gW
zo}H4dAOnNg^ni2xD)m~n>lf`ScNVQ)wZ1d`&5=|~>+=Gfks6a(xLI<wqkrH3&Y-fk
z+y2b4nO>8+PW-#TzWn@O6RAHR^W(3r-t6zYKYRDt?&5#U*7n(E-FY|j?Dkmve_Ok=
z;CcG)h4*w`hi=b|n{(fM_vih4|D6B-<KyS6>m=EB<&?XrnC)Ku_0iFvck2yu*6Zn)
zckfP%eP8!t+u61Qx7WpAXRdu~AG7ZDzsGge8(&<zU6ql4E7bZ<JlC-^KPUQpe^hfn
ze_fx7-nF-j-k05dYv8x@RbKvW*Wy3Lvo8Lh_*-t-lDv)a=Cc!bJx;y9rlUP?^X}~J
zyM?!YoxjxW!QIXi%j>sx_`kmTa@YAee;ezc*1h}n^txny%ih96+un<v&2#69TJOiR
zZFal#>U@scz2Esu-ai&n4>Fm%`Q_#0b6>v;3fvbk=zezk`^5rxsp%%44d<5qtSvbc
zZuPb~Vdu}{Gd4;a?$nEOD*D&-)E|4<pYe9u;ZN^=pO{-xVtH<D`>%EPYPX(XK6?A=
z_rQ>T4Y_)roU6KacFS(o%)js@yy^REcG-9PQn@3(GFIKa_ocMHL(+Y6oc_AF+8ZUc
zwVRd7PHZ`U!s@)*yJP<+PdaXY`S`QbCU&R3{dibrTBmSv|B?ycs;{p-Zx^3CyUwxB
z_v3_(pHB-$l<hBZ&MLnhTKj+7zY8sS{)!v(XWwPX|6OK0WnX<<&%-5gM|W5Kx%1_J
z&h*1~OI$ba%ba%o`|gF2)ANjFmR&s@aCy<R_p?m}<z{XEmu2$w-_=R-dAp_Gg`d4$
zb$N;>-_lsYz@!hszCx2*KNkNAS|4`(ajEsA@<01-yl=~^l}dVi!K235Msk+N`?%aX
z=}?Q_JEmVHzRi!>dQ<rR-_P|I6Q_KcJTdjQdH2HuPiKhK-Q9CL-;bx3d!OphnisF%
z-u<Jxf6hCb>Z89WcCYV_GcEn!Am`qmr(IvSYfa{#;N98t^%u&e?pAd4UA+IwYv!VW
z`r_#K;mmhG&hnTiqA;f_Mb`d){`*;lb!)b>f7Rpv_Nn;#_WhI3y$*f#`uveo{<ZbH
z&rB(P@89BAZW;6czOVh3?*IQ^eO$I=uX*YJg?Ha%yzN)}b8EZFkAk~jWB1PK(Fly0
z(Pn=4_TMzaH_Lf`@JCsmozrmiWX1aJUz^yJH{L5>efh3>(ZoyT{x_c<IC*!I`mft+
zvP-s9|4c65D$U-PbbD^?uB?3%tW8gsEhtL5t#`fNw6*Twmo3x0T^@bDntPphc3#H4
zZ^HY3f8E`tHL=d3@Y00$F_V2|W?01fzIfp*?#^FjR<rAV@U63HKhI82FPYaj?}W|F
ziWQk^+f`5YUi$a;tI79eaaU)%7W*EGF@0FK;80?H@sBS83Tiv|<-PLvT_G-CuHALF
zPG_fD!TzfI*WcD>T{)4Mu*<6cW>8{T?e%+n?`Jih>S+F_X10$1=*<_0Qcmihn=cZR
zzbn;v)}MX(|0{mX`;#`s`a1i>2hyb}t5O=yKAAgNyZ7U}U4Qp}oc;8^rpZ>9LZe;v
z&CT^k8Gb6<{TY5r{{8#yzL{B*ctqFMpFOK)J0)~??d5N^*Z1Gj(L7iGFV#FdKY#YT
zybD27DxLE#eY^YI^IUSzVyS=nyWZDb|1(){*%#fDJmT`s<^fHXM|ao9PoCai+io@S
z>63qt+smIUi$5$E99i&^e@@>EW~cuV*6l4NJu}V=F5KG8k?!Z#6ET_BrKL>fzI&1U
z2N})ORKuhYRmDXTtNZwV$Oy(H)~}n&Q^*#mC)zV($p#nG@=r5<%v0_<pW&#j&*H+f
zo%8eojkg=_^jxcB+jxhyBSO=1UUKfG8QD&6E*j5M>|4@s{c%W;q*7z%Y=c8H4{j4~
zow3t@OIYG-UmcB_`Ag;=zvOw$t7g87(F%o0m7W!T6Qi6Z3yKyn=}s)pYG0r#w8*A@
zi@>?c14WOWGPsu-D{)5Yi%-*jQ5>_W>FF`mpG<C3xJ38$8-=zNJw7#!CH1Unw}o86
zS>tb(>sZf6Shf8-yva-_O~-t(*D)ChX{Br1mewy+Ez*x^wd8oRJc8|@O6PT9O_g&B
zvqB@T1#-8(pFE?Z!f#sDv8v5nOy0{23=ca#tY<r}6*uMJ#xC)-X(drUeG%7=ZsPgs
z%sJP4Mee~)^Ao16%I(QYmeU>?cO+U|=F*gDE>l>tNv>ES^q{1(N3Pb2;Ds|f1MHnN
z(-zL>O1-i2V0B+a*G++o$D~%UcDbKQ=Ic?ZKhojJ<b8Z*lA)=;E9<n<gt}vVO;2_>
z2_CGFIa=?cvb(9!-DRc4vj%~f{hb~cS*xaX%&P8Dk<;2UPpA9HYstSN(H?@v7hQu?
zj?S9glOyt~eZq+ay<W?j4{>Z)IykjYP%ykKh%0`<nx!U}oL5ik66MxD5tx5t)e<4C
z$Ck5xa(!MlRq%rNiF=VP|CD<LcwV_nz1CVIuYRMaPD`(_{>{{>XBdlPC9Mxfr<!gv
zcj`K7^>xZigZz78Qp%3vy(?Yw(kwcTFkdU%a?UA!LZ@V7>8u^cLwkLiR&n)sWNy?~
z$^3LiCe>t%K-bB2O)?s*VvMHrEjrThblC|H;h2?5-TVREU0EADo(HHLFSWV3drE-)
zsR`xIyTjaAS)aOjY^=Xfb@j&Sgsf{`bGA=?uIBV#@ralXv%l|yGcs!nkGd|c5;9O)
zso;H3o7e5^35AG<2Kj+fcO+M?(75W@q9Yex)?y(uHFT2kTo%QJ2kuzx_g`paX%N#t
zPoz;Ww{XdYD>mXMwk|6ZKW=_DlCQuv%0p{|qEcvT0?)-svjQ{%ZJ!C%uV7u9Vm8Cv
zx7B1z_S%!;3tBUjBM%rE%(Cq1aci~Q(02aC@mURvyUXrmNv!7R<hy_TB$vC&+n&=G
z%1T_Ha$k#4RetARx>Re={>8S}E7ZNK4>DiNT#zMu%-o6TU1sToLO1S-S$Z$*t%58L
z6nQH>j+!~^SJUd~xjwDsoII(*Z|YS=qKxjP8qe9quMv3or1XMkD`u{?n8K*~AVKlE
zh+uJJ<S`9jU2_)`f4<MlmLB+cYk~mR7t7?xgLm5%m!}8ZJ~rvP<6jAHzji_OltV?u
zn;ujhw^;h2Zt6;pyo#kE`#F0mjw)|)dA#-D*A*YPDC8&xM!d0C3o5s058*ql5Osb*
z{dM6<%-8&rEYB%idtlWf&~Ci;q>=ZYOKX(xEi`SNE~zWB_T#<DsdwBm6sKh+#q1WC
zRJd;YBzJz*tfd_mU4fCN!KG22qWo?vLQ*gIuYDx*eM{KM0LPt?uAy&aCKbx&uk1K>
zb-~h)4Lh%yJ<LhtHQgPtXvu;e)AY#J#WQ5hZ8y~OXrw>7b;(N1ByQzn*7stO>Rjh~
zry1F<(_L$9{FlE{rnRWYqx*CJSG9+G5}3DsEA}`fyT3Q3b4GIpSKnf@!lxU>gq~LE
z%uO-5_jgg>wTHhNoa|?Q<9TGW<k^a-V>>T$8dYZO^^53z9jCr^j-$!!m&@iPCOxy8
zGSBo1ljsJi_w~oEyZ+vgUU{=jaQ&62zVBIgUho9mK3CY8UF4+mZnH*@ag3|=tHKX)
zy{jdTSN}+9D|EXpB3QVkWa9b2?x@NaU-gPtZgAQBRrJ{c3E2hgmy2xboFaOwZwFj!
ziFmU-`C)H)q1LM%XL*uJ8ZWK1_|EFpJ|X6Tm$p#f?3u}?@9uHdcUGvc7CM#Ky}m_n
z$Le)UrtBzFJo0eeu1O)s9)H!CnEot7Px8xrukHwgB$pW%w;2}C+B}DAg`~>kefu`4
zJ&L_>s<&Ul!0Du1u(I&Mn>$x&<R6~3Xw?^=r?HjCHck(TtK73Xy7I7Xk78)T#e{0^
zphtURUFQ|n#=1UaxhA7%aVV@l@{G;o6`MjNPwUQ}seUO`km>q{mWn$sLtUd}rgEOF
zaKGH|DI4_J$7oHimf4&J?XPo0*W{i*`Fz1M8+mgPMJ0{Gq;;GJ{ny$&^@&lhN}SU?
zYw@RzMKfkC`O~~o(@WhzsdA~zx{l*#J!|-mYHNR(T(iFN$i(x(N88Jl_k@KWlBo~S
zZvMVwMeY>ivIJL?8+;m*&%BsYRNL_H@yhaJk&}beQgwR*PWNwh2(`Hv{e9jc>1PL|
zPj6haSb%B033rioPrz-q;tl7od`ON=pZalgZ2XMi2=3NP%5nY^rQWj&JWAKuC!x3C
z4%1ejA0ZJ5RuA&D3XX1+Wm#u=uUpOK;bor2`bUg)QH~uKEBYn~UiWbP)Z_n+<tW!V
zRY%Fz`ICeuac#M`V9lY6_8F2J^jQvbott&ApzZghgdR)T^xC!vv2>I4sRyp+E;y9s
z#x`|L!F~a)#T&l1bukGo6N@}1Id_85;<qcf?>*<9xQaoi*+W)qn&YBrV#O1Dj#&39
zdbe3mo3OS1z(y?*!NOVxk3-fk76urzZk6Rcvb(a-hLd@x+L6a|*PK{#Sp6@DCf_Hq
zojN=CI-642mkPH|+mYWBWVC0tx!jfm=~9a?F@-Y*p6=LdIBR24dxxl)`)Lci<VG=}
zwOcu#wZ5o06lol`=z*O@*ro@ktgEy_64$A#JkHaqI@MrcT<_Y|Vf?!D)`RD>3}?-(
z?V9oAOkRc0<Ac#XGa9m|&*(LXxShK0xI$?4n${y47o8+k(qz|F3yVre6nh4Q^QO*a
zSt@^~F@<Y>g6}N#nr#{{C(5{qB+oEcmZ__orJ?b7?j)0U{YiW}TmQ_MB`P;x+&j!F
zCR!-yVX*GWtlr=ApXAgZ`)Cy^qOqam*jiSd?5x$g`}|X7`q*2}%t`RQ7sSmn@j&Xa
zD@^6axA+!4P=BMjqG7Kct8&^3*R047UW>hch$MMAJ&bLd(sO)yT<|lw@YkKYVw<13
zZ8cgK5M;(yzHVhlL;0qlg-04JHQXM2UfdEAv8dAKIa}$hnTbCtJ<~$$>K)GtKKjVX
zp|vpm+qKU>1f+u>U-3=6x8v!K%hFRD=JWJ&)QQbH$@3v{r4J)#_^Os0EV>iDRO}CL
z4Uv04C(W}}zsLHDk>Q=`Pd@TkJ)0F(5_2<ES-0+*+tC+2?CIIO?+$Qu?x;TRXZ=Ub
zVO_|D3(}j@nW_uEJ)Y^`bHiZsC5Nl^R@eNWNPa(frHO5hVNr;jefp_D?K{lRGSd9A
zQ@ytS=!x_*dZ05|^WB5bPj*hY?U>bC*qMF$P|GpHjRvV7-x*F-<a%Ev;^?b7#c$y}
zhW8i56ji^qDrww2f5y<}!Bj8B51krZ(*)l$h^eNPNpY91lROsa^oD2A%p(&QWn>rJ
zkErJ{Jv1Xbnd#ngjk0$yBR1MR`o6hK_2n4_qlbIHK3d{evGiQZobBF6)2vPfaprDl
z4SDkA2A_?wvC)S!54F}-=zY-KE;Qr7^Yz{vMAtp`ej+p9R_{vF*)t2*Rk>XY`SHlM
z-D$7!x)0tD+2(ucrQE34zkR~T2cNB5ejK*D|CI68-aqw)wWj}`o&53X^TW$O7QMgu
z%Ru?s>)HF(p1;?#zpKbHuJCd_&)a7$_f59P-nVgE_uX9m!=pzGZ?1jKGX1ya5a*03
zd*{VX@jv_aOJ!-r|GzgyZDnH3w%7mn%Kp>ZzU=?X_up<a|Ju2Jd%Jjr|MY^=4BhMD
zZ__I7uJeDq`|sJWnb(%pzk4t5`1|aq>;8XLma~7X*(ddG*4HmeUtU*LZ7kch?Pkm$
zkz?VBv#ZR1#76J8_<tkz&*Qs}|GGo#ZTGp~-m`N1?ZdBkn9Kj@{v7)IVV=dGU5vHS
z*Zv-VdHO7S;gKUPoy=ynE5#oty?ph!z1^@-s=8#ZP4CHe!&0eW{^VaK%jA#MpUFJa
zlU)A&_Oke`v-|bW?XQzLp1=F*s^52X>gx{dm$c$YtlIzUYTCwI`qD-dQn%mOdn<<j
z>qmBzW$}_fHrrbl?5PU9U;g{y$*sy}aeHdL694#$&a%_|e)7pl&3PY<{$(6lq{_T+
zU;KTWKQ)K=+e_Epx?6BJ=k-3hUH@ACUT-P%aa|p^N9q1z|H~Gv4?_I>9e2#DTbMZG
z-VwFU71!^yRbAKVzmd7h&gbv^HJdtJmvgVO{M(u&y8lDtZSI$wI<6dz%o6?g{l&HA
z@+DP2>p_Dv+b`bd*JYai+gpHdy7+T`f$1`j_;sg;Jm>#0J>FY@bvplJevRpsFZkbY
zw|K&Tk!kwBhx~HW)t~e0PWODoe{j0b3x3`0=8yRq8MkM=<iBRg5_iJn!*(ZSfftPR
z;Sv&xcR#bqoEDPe7puLptte-sk753Xm+ubfipwqU*sm|{E#+_XrP*+;$eqV6H_J8G
zl<oR<F;?H2i^pW%vVfeW3?EaL+8gRvw|<!&vO>}KvS}~7G;7qoh&;Z&T(igtmPt_@
z_YU!?AI!REwcw*X`(9SLZF`;@b=drRE7)?~tiI@z`8!MHi~jEv&giSV=c$Iz<H=-x
z@%EZUexzn?_Z{VTXESeInZ`Zs8)rt0+P4MQ8@?^wwXbRS>W$|A`8U1{$yuK`!Ro}!
z`kU_;Z|Zn9J91h|e#t!Tf1%0Ctq))AtkvpXzvFzkh}ehIcV`FPZ`oZr_tnzkBEJ4h
z@?3{qcPuEKW_@mXyJ6==ro90VmmRn5<*VwBy(y+$b42WLirz;K*`U>ItBjA{cGz*G
z{o%^ZC8pjVUi-g)bNII8eT8bq$&P15wwJ33urg2oZz!NVeZHoE?sk78fg_C5ZR`c)
zra#mY(A(Z&FOa}Ey+%_&dAgF3fbR4&ngZWh9t7Pzw_V&sK$Ee4wt-07d-b>W5iz-!
z4oFPelG@@GebqD6DwQ{M`Q4r0GCy#cZJHbK?{AOri}qEqQSXc8`pz5v{23g)*fv9X
zN(`Gy7c<KwO}!^&l6R&}f0)_OvvktsD9;Es+Zhk`{P>}mo_lOc+tx7M4DAC7>66rw
z-x{r0lWW|=yL!PwwW5}ihxN^uc(=Y=v3<wVHxu(1)eH9=;Eg!7a%qJ?+uIbLKCg3X
zsk<+46)$L8*v7WRbcPFy{bsR}<a@dwPI7M&+?M|^wuDFh*U1jfH@oy+1@-Ux$N8^z
z@2wxJ+no3|CkcpHrq2}J)~57#W8C7R=1Z3~Iw#BcA8eR$_iuc{wjY(BMROhA*RvKk
zmk8hB=99Vm{X|&9tySht7p1OlVehrOyJL1RbH^F`>bn|}Ek4X!y~M*$IYck)z0{Pa
z$#rqP?An=GQt^gChsER<zMt~!*w+uNYyXIQ^u@C=hXyCFzq`3Y;K2T`M_F6_Jhw1?
zcrm*~<H82<!;7`j)a7QLTb#SLb<2Tn6*YB#MMcHPDNl4}Pt@uA?{i7;RDZQ?$vd~l
z-zLg#Ggm8r`Mvv-QNcbHzQ>yl6jf)YzkKwdI{3d%PG9c-`xgA3LX&o=Y}WCOlXiD}
zIJ@qzzX*6Jb-SXufHcQ+H$MTs>0b5%cH8;21wOECf8ij|#KgsLG1%hCK?a80rPE8C
z1^k&7ot%EuSwK$)JpIkYz;KG2fkB7?87Q2c{P2Rv^rhYc0y5w!M<xb_BisxOg2;*(
z*e^`?cM(vQ0Z;F+Ffgo?U|<kHQStiH^gb5>8z$xJ)6anDmp7+#xe9nOX(>++a}{u8
z`fD?NrmKJ>^9yFC=^NYyjHdIu3Ft9NicPn76L4fwbD7=*qIXW3e$-9Cky-x|8$^}0
zyMR7Z#<uA>?gEZX${N$xy9*>SADzNEy>SMo*mNxq0WIcDT3isd3-q{Hr}u$`CQI;4
z_m|-jn|=|*6FSQSQDERHpwIOA(DX!40Y@gUlhc=Y3OF-6o#LPV?<~L6bQUiG0hxKQ
zp=K}iAeeP(x`CH~Efd?F>G@s)eoQQPr|<U?&;u2lQcTA8rgM4=s7|->5s+dEzCZn<
zw}2?9@YH9T^kDh~ZvlIzCl5h_TGQWp2}m&|Kb|h`BcM86&s#u>$>#}3Ms#|vkAObY
z+^5s$`3TrCiT<2^&qu(H+3hd?ba`I^E2biu>CwIdYBHe0m6?fwL5c-E28!jU&-E3s
zVOpa!{kE@w>hyL$0V$@9s?%ls1hl5x1PDkm9aW#6<R_py%}+pzxlmJJI-ik%<n+CM
z0s`P7WBNru0Wqc>TGL<p3D`0n)17|6UqEcSoxgwy(;0mT$J<hXYr2QOfWY)6{sK}=
umkmL}UQD-)LA2O(r2qjbP{NdAdT2U5B0#{NN!Au3vpzsTicQ%blw<&^UAi&=

diff --git a/38202.tex b/38202.tex
index 7cd7785..afc2de0 100644
--- a/38202.tex
+++ b/38202.tex
@@ -340,45 +340,45 @@ D & SRS & N/A & \\ \hline
 \centertablecaption{6.1-2}{Uplink ``Transmission Type'' combinations}
 { |p{10cm}|p{5cm}| }{
 \textbf{Supported Combinations} & \textbf{Comment} \\ \hline
-\emph{j} x A & Note 1 \\ \hline
-\emph{k} x B & Note 2 \\ \hline
-(\(\sum_{i = 1}^{p}{m_{i})} \times\)C & Note 3, Note 4, Note 10 \\ \hline
-\emph{p\textquotesingle{}} x D & Note 3, Note 5 \\ \hline
-\(\widetilde{j} \times\) A + \(\widetilde{k} \times\)B & Note 6 \\ \hline
-\(\widetilde{j} \times\) A + \(\widetilde{(\sum_{i = 1}^{\widetilde{p}}{{\widetilde{m}}_{i})}} \times\)C
+\(j \times \text{A}\) & Note 1 \\ \hline
+\(k \times \text{B}\) & Note 2 \\ \hline
+(\(\sum_{i = 1}^{p}{m_{i})} \times \text{C}\) & Note 3, Note 4, Note 10 \\ \hline
+\(p' \times \text{D}\) & Note 3, Note 5 \\ \hline
+\(\widetilde{j} \times \text{A} + \widetilde{k} \times \text{B}\) & Note 6 \\ \hline
+\(\widetilde{j} \times \text{A} + \widetilde{(\sum_{i = 1}^{\widetilde{p}}{{\widetilde{m}}_{i})}} \times \text{C}\)
 & Note 6, Note 10 \\ \hline
-\(\widetilde{j} \times\) A + \(\widetilde{p}' \times\)D & Note 6 \\ \hline
-\(\widehat{k} \times\) B + \(\widehat{(\sum_{i = 1}^{\widehat{p}}{{\widehat{m}}_{i})}} \times\)C &
+\(\widetilde{j} \times \text{A} + \widetilde{p}' \times \text{D}\) & Note 6 \\ \hline
+\(\widehat{k} \times \text{B} + \widehat{(\sum_{i = 1}^{\widehat{p}}{{\widehat{m}}_{i})}} \times \text{C}\) &
 Note 8, Note 10 \\ \hline
-B + \(\ (\sum_{i = 1}^{\widehat{p}'}{\widehat{m}}_{i}')\  \times\)C &
-Note 9, Note 10 \\
-\(\widetilde{k} \times\) B + \(\widetilde{p}' \times\)D & Note 7 \\ \hline
-\(\ (\sum_{i = 1}^{\widetilde{p}}{\widetilde{m}}_{i})\  \times\)C +
-\(\widetilde{p}' \times\)D & Note 7, Note 10 \\ \hline
+\(\text{B} + (\sum_{i = 1}^{\widehat{p}'}{\widehat{m}}_{i}') \times \text{C}\) &
+Note 9, Note 10 \\ \hline
+\(\widetilde{k} \times \text{B} + \widetilde{p}' \times \text{D}\) & Note 7 \\ \hline
+\(\ (\sum_{i = 1}^{\widetilde{p}}{\widetilde{m}}_{i}) \times \text{C} +
+\widetilde{p}' \times \text{D}\) & Note 7, Note 10 \\ \hline
 \multicolumn{2}{|p{16cm}|}{
-	Note 1: The number of cell groups \emph{j} in the supported combination is subject to UE capability.
+	Note 1: The number of cell groups \(j\) in the supported combination is subject to UE capability.
 	
-	Note 2: The number of PUCCH groups \emph{k} in the supported combination is subject to UE capability.
+	Note 2: The number of PUCCH groups \(k\) in the supported combination is subject to UE capability.
 	
-	Note 3: The number of carriers \emph{p,} and \emph{p\textquotesingle{}} in the supported combinations are subject to UE capability.
+	Note 3: The number of carriers \(p\), and \(p'\) in the supported combinations are subject to UE capability.
 	
-	Note 4: In the case there is one SUL carrier, then \emph{p}-1 would be supported.
+	Note 4: In the case there is one SUL carrier, then \(p-1\) would be supported.
 	
-	Note 5: UE may be configured with \emph{p\textquotesingle{}} but may also have capability to simultaneously sound less than this number.
+	Note 5: UE may be configured with \(p'\) but may also have capability to simultaneously sound less than this number.
 	
-	Note 6: Simultaneous PRACH with PUCCH (or PUSCH or SRS) is supported only in the case of inter-band CA, with \(\widetilde{j} \leq j\), \(\widetilde{k} \leq k\), \(\widetilde{p} \leq p\), and \(\widetilde{p}' \leq p\)\textquotesingle{} depending on the configuration, and subject to UE capability for parallel transmission.
+	Note 6: Simultaneous PRACH with PUCCH (or PUSCH or SRS) is supported only in the case of inter-band CA, with \(\widetilde{j} \leq j\), \(\widetilde{k} \leq k\), \(\widetilde{p} \leq p\), and \(\widetilde{p}' \leq p'\) depending on the configuration, and subject to UE capability for parallel transmission.
 	
-	Note 7: Simultaneous SRS with PUCCH (or PUSCH) is supported only in the case of inter-band CA, with \(\widetilde{k} \leq k\), \(\widetilde{p} \leq p\), and \(\widetilde{p}' \leq p\)\textquotesingle{} depending on the configuration, and subject to UE capability for parallel transmission.
+	Note 7: Simultaneous SRS with PUCCH (or PUSCH) is supported only in the case of inter-band CA, with \(\widetilde{k} \leq k\), \(\widetilde{p} \leq p\), and \(\widetilde{p}' \leq p'\) depending on the configuration, and subject to UE capability for parallel transmission.
 	
 	Note 8: Simultaneous PUCCH and PUSCH(s) for the case that multiple PUCCH groups are configured and the respective PUCCH and PUSCH(s) are transmitted in the different PUCCH groups, with \(\widehat{k} < k\) and
 	\(\widehat{p} \leq p\). \(k\) and \(p\) are subject to UE capability for supported number of PUCCH groups and UL carriers, respectively. \(\widehat{k}\) and \(\widehat{p}\) depend on configuration.
 	
-	Note 9: Simultaneous PUCCH and PUSCH(s) within the same PUCCH group in the case of inter-band CA, \({\widehat{p}}'\ \)depending on the configuration, and subject to UE capability for parallel transmission of
+	Note 9: Simultaneous PUCCH and PUSCH(s) within the same PUCCH group in the case of inter-band CA, \({\widehat{p}}'\)depending on the configuration, and subject to UE capability for parallel transmission of
 	PUCCH and PUSCH within the same PUCCH group.
 	
 	Note 10: Simultaneous transmission of up to 2 PUSCHs within the same active UL BWP, with
-	\(p,\ \widetilde{p},\ \ \widehat{p},\ \ {\widehat{p}}'\ \)depending on the configuration,
-	\(m_{i},\ {\widetilde{m}}_{i},\ {\widehat{m}}_{i},\ {\widehat{m}}_{i}'\) equal to 1 or 2, subject to UE capability for parallel transmission of PUSCH and PUSCH within the same active UL BWP.
+	\(p, \widetilde{p}, \widehat{p}, {\widehat{p}}'\)depending on the configuration,
+	\(m_{i}, {\widetilde{m}}_{i}, {\widehat{m}}_{i}, {\widehat{m}}_{i}'\) equal to 1 or 2, subject to UE capability for parallel transmission of PUSCH and PUSCH within the same active UL BWP.
 } \\
 }
 
diff --git a/3gpp.sty b/3gpp.sty
index 5b47c54..c5167e7 100644
--- a/3gpp.sty
+++ b/3gpp.sty
@@ -17,7 +17,7 @@
 \ifPDFTeX
 \usepackage{tgheros}
 \else
-\usepackage[paper=portrait,pagesize]{typearea}
+%\usepackage[paper=portrait,pagesize]{typearea}
 \usepackage{fontspec}
 \setmainfont{Times New Roman}
 \setsansfont{Arial}
@@ -28,8 +28,6 @@
 \titleformat*{\subsection}{\Large\bfseries\sffamily}
 \titleformat*{\subsubsection}{\large\bfseries\sffamily}
 
-
-
 \setcounter{secnumdepth}{6}
 \makeatletter
 \renewcommand\paragraph{\@startsection{paragraph}{4}{\z@}%
@@ -211,7 +209,7 @@
 	\fancyhead{}\fancyfoot{}
 	
 	\fancyhead[L]{\textsf{\textbf{#1}}}
-	\fancyhead[C]{\textsf{\thepage}}
+	\fancyhead[C]{\textsf{\textbf{\thepage}}}
 	\fancyhead[R]{\textsf{\textbf{3GPP #2 #3 (#4)}}}
 	\fancyfoot[C]{\textsf{\textbf{\textit{3GPP}}}}
 }
@@ -349,9 +347,9 @@
 	\end{ednotes}
 }
 
-
 % Color palette
-\definecolor{asnKeyword}{HTML}{8B0000} % maroon
+\definecolor{asnKeyword1}{HTML}{8B0000} % maroon
+\definecolor{asnKeyword2}{HTML}{808000} % maroon
 \definecolor{asnComment}{HTML}{6A737D} % gray
 \definecolor{asnString}{HTML}{032F62}  % dark blue
 \definecolor{asnBg}{HTML}{F6F8FA} % light gray background
@@ -360,6 +358,8 @@
 \lstdefinelanguage{ASN1}{
 	morekeywords={SEQUENCE,CHOICE,INTEGER,SIZE,OF},
 	sensitive=true,
+	morekeywords=[2]{OPTIONAL},             % separate group for OPTIONAL
+	morekeywords=[3]{NULL},             % separate group for OPTIONAL
 	morecomment=[l]{--}, % ASN.1 line comments
 	morestring=[b]"
 }
@@ -367,16 +367,18 @@
 \lstdefinestyle{asnStyle}{
 	language=ASN1,
 	basicstyle=\ttfamily\footnotesize,
-	keywordstyle=\color{asnKeyword}\bfseries,
+	keywordstyle=\color{asnKeyword1}\bfseries,
+	keywordstyle=[2]\color{asnKeyword2}\bfseries,
+	keywordstyle=[3]\color{asnString},
 	commentstyle=\color{asnComment}\itshape,
 	stringstyle=\color{asnString},
 	backgroundcolor=\color{asnBg},
-	numbers=left,
+	numbers=none,
 	numberstyle=\scriptsize\color{asnComment},
 	stepnumber=1,
 	numbersep=10pt,
 	showstringspaces=false,
-	tabsize=2,
+	tabsize=4,
 	frame=single,
 	rulecolor=\color{asnComment},
 	breaklines=true,
diff --git a/scripts/latexdiff-fast.pl b/scripts/latexdiff-fast.pl
new file mode 100644
index 0000000..44d6005
--- /dev/null
+++ b/scripts/latexdiff-fast.pl
@@ -0,0 +1,6031 @@
+#!/usr/bin/env perl
+##!/usr/bin/perl -w
+# latexdiff - differences two latex files on the word level
+#             and produces a latex file with the differences marked up.
+#
+#   Copyright (C) 2004-22  F J Tilmann (tilmann@gfz-potsdam.de)
+#
+# Repository/issue tracker:   https://github.com/ftilmann/latexdiff
+# CTAN page:          http://www.ctan.org/pkg/latexdiff
+#
+#    This program is free software: you can redistribute it and/or modify
+#    it under the terms of the GNU General Public License as published by
+#    the Free Software Foundation, either version 3 of the License, or
+#    (at your option) any later version.
+#
+#    This program is distributed in the hope that it will be useful,
+#    but WITHOUT ANY WARRANTY; without even the implied warranty of
+#    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+#    GNU General Public License for more details.
+#    You should have received a copy of the GNU General Public License
+#    along with this program.  If not, see <http://www.gnu.org/licenses/>.
+#
+# Detailed usage information at the end of the file
+#
+#
+# Note references to issue numbers are for the github repository of latexdiff: https://github.com/ftilmann/latexdiff
+#
+# Version 1.3.4:
+# New features:
+#  - Option to use lua-ul instead of ulem (for use with LuaLaTeX) (fix #17, #60, #188, #255, #270)
+# Enhancement
+#  - If amsmath is detected as one of the included (based on idea in PR #263 contributed by github user xlucn
+# Bug fix
+#  - sometimes the introduction of aux commands to show deleted lists or descriptions leaves in place empty auxiliary
+#    list environments, which would cause error messages (though skipping messages would result in a correct output pdf).
+#    These are now removed.
+#  - add more mboxsafecmd and safecmd commands for SIunitx to stay compatible with newer versionsof SIunitx (PR #283, fixing issue #282, contributed by github user joe6302413)
+#  - File added via --preamble option is no longer assumed to be ASCII, but read either with encoding as defined by --encoding option or using the encoding of the LOCALE (fixes issue #285 )
+#  - multicolumn argument is now treated as text
+#  - when tikz-dependency package is used, \& is no longer a safe command as it has special meaning inside dependency environment.  The fix is a little of a hack as really it should only be considered unsafe within dependency environment (fixes (mostly) issue #303 )
+#  - listings package had trouble with non-ASCII chars. The encoding is now set (thanks to github user anka-213 for finding this). Fixes #304
+#
+# Version 1.3.3:
+# New features:
+#  - Option --no-del to remove all deleted text (merge contributed by github user tdegeus PR #252, fixing issue #66
+#
+# Bug fixes:
+#  - Abbreviations involving punctuations within them. They need special treatment because otherwise in some
+#    circumstances the gnoring of white space differences in conjunction with merging according to MINWORDSBLOCK rule
+#    could turn 'i.e.' into 'i.\PAR e.' (see https://github.com/ftilmann/latexdiff/issues/269). A few abbreviations
+#    are now hard-coded and treated as atomic:
+#    English: i.e., e.g.   Deutsch: z.B.
+#    (fixes issue #269)
+#  - In WHOLE and COARSE math modes, now properly treat math environments with arguments such as \alignat. Fixes #251
+#  - For FINE math mode, multiple improvments to the processing work flow yield more robust outcomes. In particular, changes
+#    to the equation type, e.g. \begin{displaymath} -> \begin{equation}  without modifications  now usually no longer result
+#    in errors. (Partially) fixes issues #235  and #244.
+#  - When encountering deleted math array environments such as align or eqnarray, rather than replacing them with  a
+#    fixed replacement environment (e.g. align* or eqnarray*), an asterisk is now added to the original command, which
+#    in amsmath (and with eqnarray) will result in the same environment but without line numbers. Config variable MATHARRREPL
+#    is therefore (nearly) redundant, and a depracation warning is given when it is set. Reference to MATHARRREPL are have
+#    been removed from the manual (there is one exception, when it's still being used: sometimes latexdiff can figure out
+#    that there is a deleted array environment, but does not know which one. In this case, MATHARRREPL is still being used
+#    to encapsulate these parts of the source, and therefore it is still set internally.  But this is a quite rare situation).
+#    Fixes issue #216
+#  - Unlike 'array' environment, 'split' (amsmath) does not work in argument of \DIFadd or \DIFdl in UNDERLINE modes; therefore remove it from ARRENV configuration variable.
+#     Exclude \begin and \end in math environments in COARSE and WHOLE modes. Fixes #258. Fixes #109
+#  - --flatten now works for empty files. Fixes issue #242
+#  - improved processing of Chinese and Japanese texts in that splitting is done based on characters. Thanks to LuXu (Oliver Lew) in git for working this out. Fixes #229, fixes #145
+
+
+
+# Version 1.3.2
+# API adaptions:
+#  - latexdiff now completes with exit code 0 after --help or --version command (see issue #248)
+# New features / feature extensions
+#  - extend CUSTOMDIFCMD related postprocessing to deal properly with multiline commands, or a sequence of several commands in the same line (see github issue #204)
+#  - Support for additional macros from import package (\import, \inputfrom, \includefrom, \subimport,\subinputfrom, \subincludefrom). Provided by janniklasrose in PR #243 (fixes #239)
+#  - replace default driver dvips->pdftex
+# Bug fixes:
+#  - fix issue #206 affecting proper markup of text commands which are not also safe cmd's and have multiple arguments
+#  - fix issue #210 by adding \eqref (amsmath package) to the list of safe commands
+#  - fix bug reported in issue #168 mangled verbatim line environment
+#  - fix bug reported in issue #218 by replacing \hspace{0pt} after \mbox{..} auxiliary commands with \hskip0pt.
+#  - more ways to process \frac correctly with atomic arguments (committed by julianuu PR #246
+#  - fix a bug in biblatex mode, which prevented proper processing of modified \textcite (see: https://tex.stackexchange.com/questions/555157/latexdiff-and-biblatex-citation-commands)
+#  - -h string fix: add -driver option
+#
+# Version 1.3.1.1
+#  - remove spurious \n to fix error: Unknown regexp modifier "/n" at .../latexdiff line 1974, near "=~ " (see github issue #201)
+#
+# Version 1.3.1
+#    Bug fixes:
+#      - remove some uninitialised variable $2 warnings in string substitution in flatten function in case included file is not found
+#      - add minimal postprocessing to diff processing of preamble commands (replace \RIGHTBRACE by \} )
+#      - pre-processing: replace (contributed) routine take_comments_and_enter_from_frac() with take_comments_and_newline_from_frac(), which does the same thing
+#        (remove whitespace characters and comments between the argument of \frac commands) in an easier and more robust way. In addition, it
+#        will replace commands like \frac12 with \frac{1}{2} as pre-processing step.   Fixes issue #184
+#      - add "intertext" to list of unsafe math commands @UNSAFEMATHCMD . Fixes issue #179
+#      - provide citation command patterns for biblatex and protect them with mbox'es. Fixes issue #199
+#      - hardcode number of parameters for \href and \url commands to allow spaces between commands and arguments even if --allow-spaces option is not used (this
+#        is needed because some bibliography styles add such in-command-sequence spaces)  Fixes issues: #178 #198
+#      - bibitem is now kept even in deleted blocks such that deleted references show up properly (this implies that the actual numbers in numerical referencing schemes will change)
+#        (this is implemented by introducing a new class of commands KEEPCMD , which are kept as is in deleted environments (no effect in added environments). Currently
+#        \bibitem   is hardwired to be the only member of this class  (fixes issue #194, #174)
+#    Features:
+#      - add some special processing for revtex bibliography commands, so that the spaces between bibliography commands \bibfield and \bibinfo and their arguments are ignored.
+#         (fixes issue #194, should fix #174)
+#
+# Version 1.3.0 (7 October 2018):
+#    - treat options to \documentclass as potential package names (some packages allow implicit loading of or imply selected packages
+#    - improved pattern matching: now allows nested angular brackets, and is no longer confused by escaped curly braces
+#    - improved pattern matching in COARSE mode: occasionally, the closing bracket or some other elements would be matched in an 'unnatural' way due to another sequence being more minimal in the computational sense, sometimes even causing errors due to tokens moving in or out of the scope of math environments. This is now discouraged by adding internal \DIFANCHOR commands (which are removed again in post-processing) (fixes issues reported via email by li_ruomeng .
+#    - verbatim and lstlisting environments are marked-up with line-by-line in a similar style to non-verbatim text (requires the listing package to be installed)
+#       (see new configuration variable VERBATIMLINEENV) (several issues and pull requests by jprotze)
+#    - --flatten: now supports \verbatiminput and \lstlistinput
+#     - --flatten: if file is not found, do not fail, simply warn and leave command unexpanded (inspired by issue #112).  Don't warn if file name contains #[0-9] as it is then most likely an argument within a command definition rather than an actual file (applies to \input, \subfile, \include commands)
+#     - added to textcmds: \intertext
+#    - new config variable CUSTOMDIFCMD to allow defining special versions of  commands  in added or deleted blocks (Pull request by github user jprotze)
+#    - added option -no-links (mostly for use by latexdiff-vc in only-changes modes) (Pull request by github user jprotze)
+#    - new option --filter-script to run both input through a pre-processing script (PR jasonmccsmith  #167)
+#      new option --no-filter-stderr to hide stderr output from filter-script (potentially dangerous, as this might hide malfunctioning of filter scripts)
+#    - --flatten now can deal with imports made using the import package {PR jasonmccsmith #173)
+#   Bug fixes:
+#    - pattern matching of \verb and \lstinline commands had an error which meant they would trigger on commands beginning with \verb.
+#    - In description environments, mark up item descriptions by effectively reating the insides of item commannds as text commands (fixes #161)
+#
+#
+# Version 1.2.1 (22 June 2017)
+#    - add "DeclareOldFontCommand" to styles using \bf or \sf old style font commands (fixies issue #92 )
+#    - improved markup: process lstinline commands in listings package correctly
+#      for styles using colour, \verb and \lstinline arguments are marked up with colour (blue for added, red for deleted)
+#    - bug fix: protecting inline math expressions for mbox did not work as intended (see stack exchange question: http://tex.stackexchange.com/questions/359412/compiling-the-latexdiff-when-adding-a-subscript-before-a-pmatrix-environment-cau)
+#    - bug fix: when deleted \item commands are followed immediately by unsafe commands, they were not restored properly
+#      (thanks to J. Protze for pull request) (pull request #89)
+#    - treat lstlisting and comment as equivalent to verbatim environment
+#      make environments that are treated like verbatim environments configurable (config variable VERBATIMENV)
+#      treat lstinlne as equivalent to verb command
+#      partially addresses issue #38
+#    - refactoring: set default configuration variables in a hash, and those that correspond to lists
+#    - feature: option --add-to-config used to amend configuration variables, which are regex pattern lists
+#    - bug fix: deleted figures when endfloat package is activated
+#    - bug fix: alignat environment now always processed correctly (fix issues #65)
+#    - bug fix: avoid processing of commands as potential files in routine init_regex_arr (fix issue #70 )
+#    - minimal feature enhancement: treat '@' as allowed character in commands (strictly speaking requires prior \makeatletter statement, but always assuming it to be
+#       @       a letter if it is part of a command name will usually lead to the correct behaviour (see http://tex.stackexchange.com/questions/346651/latexdiff-and-let)
+#    - new feature/bug fix: --flatten option \endinput in included files now respected but only if \endinput stands right at the beginning of the line (issue #77)
+#    - bug fix: flatten would incorrectly attempt to process commented out \include commands (from discussion in issue #77 )
+#    - introduce an invisible space (\hspace{0pt} after \mbox{..} auxiliary commands (not in math mode), to allow line breaks between added and deleted citations (change should not cause adverse behaviour otherwise)
+#
+# Version 1.2.0:
+#    - highlight new and deleted figures
+#    - bug fix in title mark-up. Previously deleted commands in title (such as \title, \author or \date) were marked up erroneously
+#    - (minor) bug fixes in new 1.1.1 features: disabled label was commented out twice, additional spaces were introduced before list environment begin and end commands
+#    - depracation fix: left brace in RegEx now needs to be escaped
+#    - add type PDFCOMMENT based on issue #49 submitted by github user peci1 (Martin Pecka)
+#    - make utf8 the default encoding
+#
+# Version 1.1.1
+#    - patch mhchem: allow ce in equations
+#    - flatten now also expands \input etc. in the preamble (but not \usepackage!)
+#    - Better support for Japanese ( contributed by github user kshramt )
+#    - prevent duplicated verbatim hashes (patch contributed by github user therussianjig, issue #36)
+#    - disable deleted label commands (fixes issue #31)
+#    - introduce post-processing to reinstate most deleted environments and all needed item commands (fixes issue #1)
+#
+# Version 1.1.0
+#    - treat diacritics (\",\', etc) as safe commands
+#    - treat \_ and \& correctly as safe commands, even if used without spacing to the next word
+#    - Add a BOLD markup type that sets added text in bold face (Contribution by Victor Zabalza via pull request )
+#    - add append-mboxsafecmd list option to be able to specify special safe commands which need to be surrounded by mbox to avoid breaking (mostly this is needed with ulem package)
+#    - support for siunitx and cleveref packages: protect \SI command in siunitx package and \cref,\Cref{range}{*} in cleveref packages (thanks to Stefan Pinnow for testing)
+#    - experimental support for chemformula, mhchem packages: define \ch and \ce in packages as safe (but not \ch,\cee in equation array environments) - these unfortunately will not be marked up (thanks to Stefan Pinnow for testing)
+#    - bug fix: packages identified correctly even if \usepackage command options extend over several lines (previously \usepackage command needed to be fully contained in one line)
+#    - new subtype ONLYCHANGEDPAGE outputs only changed pages (might not work well for floating material)
+#    - new subtype ZLABEL operates similarly to LABEL but uses absolute page numbers (needs zref package)
+#    - undocumented option --debug/--nodebug to override default setting for debug mode (Default: 0 for release version, 1: for development version
+#
+# Version 1.0.4
+#    - introduce list UNSAFEMATHCMD, which holds list of commands which cannot be marked up with \DIFadd or \DIFdel commands  (only relevant for WHOLE and COARSE math markup modes)
+#    - new subtype LABEL which gives each change a label. This can later be used to only display pages where changes
+#      have been made (instructions for that are put as comments into the diff'ed file) inspired by answer on http://tex.stackexchange.com/questions/166049/invisible-markers-in-pdfs-using-pdflatex
+#    - Configuration variables take into accout some commands from additional packages:
+#      tikzpicture environment now treated as PICTUREENV, and \smallmatrix in ARRENV (amsmath)
+#    - --flatten: support for \subfile command (subfiles package)  (in response to http://tex.stackexchange.com/questions/167620/latexdiff-with-subfiles )
+#    - --flatten: \bibliography commands expand if corresponding bbl file present
+#    - angled bracket optional commands now parsed correctly (patch #3570) submitted by Dave Kleinschmidt (thanks)
+#    - \RequirePackage now treated as synonym of \usepackage with respect to setting packages
+#    - special rules for apacite package (redefine citation commands)
+#    - recognise /dev/null as 'file-like' arguments for --preamble and --config options
+#    - fix units package incompatibility with ulem for text maths statements $ ..$ (thanks to Stuart Prescott for reporting this)
+#    - amsmath environment cases treated correctly (Bug fix #19029) (thanks to Jalar)
+#    - {,} in comments no longer confuse latexdiff (Bug fix #19146)
+#    - \% in one-letter sub/Superscripts was not converted correctly
+#
+# Version 1.0.3
+#    - fix bug in add_safe_commands that made latexdiff hang on DeclareMathOperator
+#      command in preamble
+#    - \(..\) inline math expressions were not parsed correctly, if they contained a linebreak
+#    - applied patch contributed by tomflannaghan via Berlios: [ Patch #3431 ] Adds correct handling of \left< and \right>
+#    - \$ is treated correctly as a literal dollar sign (thanks to Reed Cartwright and Joshua Miller for reporting this bug
+#      and sketching out the solution)
+#    - \^ and \_ are correctly interpreted as accent and underlined space, respectively, not as superscript of subscript
+#      (thanks to Wail Yahyaoui for pointing out this bug)
+#
+# Version 1.0.1 - treat \big,\bigg etc. equivalently to \left and
+#              \right - include starred version in MATHENV - apply
+#            - flatten recursively and --flatten expansion is now
+#              aware of comments (thanks to Tim Connors for patch)
+#            - Change to post-processing for more reliability for
+#              deleted math environments
+#            - On linux systems, recognise  and remove DOS style newlines
+#            - Provide markup for some special preamble commands (\title,
+#              \author,\date,
+#            - configurable by setting context2cmd
+#            - for styles using ulem package, remove \emph and \text.. from list of
+#              safe commands in order to allow linebreaks within the
+#              highlighted sections.
+#            - for ulem style, now show citations by enclosing them in \mbox commands.
+#              This unfortunately implies linebreaks within citations no longer function,
+#              so this functionality can be turned off (Option --disable-citation-markup).
+#              With --enable-citation-markup, the mbox markup is forced for other styles)
+#            - new substyle COLOR.  This is particularly useful for marking up citations
+#              and some special post-processing is implemented to retain cite
+#              commands in deleted blocks.
+#            - four different levels of math-markup
+#            - Option --driver for choosing driver for modes employing changebar package
+#            - accept \\* as valid command (and other commands of form \.*). Also accept
+#              \<nl> (backslashed newline)
+#            - some typo fixes, include commands defined in preamble as safe commands
+#              (Sebastian Gouezel)
+#            - include compared filenames as comments as line 2 and 3 of
+#              the preamble (can be modified with option --label, and suppressed with
+#              --no-label), option --visible-label to show files in generated pdf or dvi
+#              at the beginning of main document
+#
+# Version 0.5  A number of minor improvements based on feedback
+#              Deleted blocks are now shown before added blocks
+#              Package specific processing
+#
+# Version 0.43 unreleased typo in list of styles at the end
+#              Add protect to all \cbstart, \cbend commands
+#              More robust substitution of deleted math commands
+#
+# Version 0.42 November 06  Bug fixes only
+#
+# Version 0.4   March 06 option for fast differencing using UNIX diff command, several minor bug fixes (\par bug, improved highlighting of textcmds)
+#
+# Version 0.3   August 05 improved parsing of displayed math, --allow-spaces
+#               option, several minor bug fixes
+#
+# Version 0.25  October 04 Fix bug with deleted equations, add math mode commands to safecmd, add | to allowed interpunctuation signs
+# Version 0.2   September 04 extension to utf-8 and variable encodings
+# Version 0.1   August 04    First public release
+
+# Inserted block for differenceing 
+# use Algorithm::Diff qw(traverse_sequences);
+# in standard version
+# The following BEGIN block contains a verbatim copy of
+# Ned Konz' Algorithm::Diff package version 1.15 except
+# that subroutine _longestCommonSubsequence has been replace by 
+# a routine which internally uses the UNIX diff command for
+# the differencing rather than the Perl routines if the 
+# length of the sequences exceeds some threshold.
+# Also, all POD documentation has been stripped out.
+#
+# (the distribution on which this modification is based is available
+#  from http://search.cpan.org/~nedkonz/Algorithm-Diff-1.15
+#  the most recent version can be found via  http://search.cpan.org/search?module=Algorithm::Diff )
+# Please note the LICENCE for Algorithm::Diff :
+#   "Copyright (c) 2000-2002 Ned Konz.  All rights reserved.
+#    This program is free software;
+#    you can redistribute it and/or modify it under the same terms
+#    as Perl itself."
+# The fast-differencing version of latexdiff is provided as a convenience
+# for latex users under Unix-like systems which have a 'diff' command.
+# If you believe
+# the inlining of Algorithm::Diff violates its license please contact
+# me and I will modify the latexdiff distribution accordingly.
+# Frederik Tilmann (tilmann@esc.cam.ac.uk)
+# Jonathan Paisley is acknowledged for the idea of using the system diff
+# command to achieve shorter running times
+BEGIN { 
+package Algorithm::Diff;
+use strict;
+use vars qw($VERSION @EXPORT_OK @ISA @EXPORT);
+use integer;    # see below in _replaceNextLargerWith() for mod to make
+                # if you don't use this
+require Exporter;
+@ISA       = qw(Exporter);
+@EXPORT    = qw();
+@EXPORT_OK = qw(LCS diff traverse_sequences traverse_balanced sdiff);
+$VERSION = sprintf('%d.%02d fast', (q$Revision: 1.15 $ =~ /\d+/g));
+
+# Global parameters
+
+use File::Temp qw/tempfile/;
+# if larger number of elements in longestCommonSubsequence smaller than
+# this number, then use internal algorithm, otherwise use UNIX diff
+use constant THRESHOLD => 100 ; 
+# Detect whether diff --minimal option is available
+# if yes we use it
+use constant MINIMAL => ( system('diff','--minimal','/dev/null','/dev/null') >> 8 ==0 ? "--minimal" : "" ) ;
+
+
+
+# McIlroy-Hunt diff algorithm
+# Adapted from the Smalltalk code of Mario I. Wolczko, <mario@wolczko.com>
+# by Ned Konz, perl@bike-nomad.com
+
+
+# Create a hash that maps each element of $aCollection to the set of positions
+# it occupies in $aCollection, restricted to the elements within the range of
+# indexes specified by $start and $end.
+# The fourth parameter is a subroutine reference that will be called to
+# generate a string to use as a key.
+# Additional parameters, if any, will be passed to this subroutine.
+#
+# my $hashRef = _withPositionsOfInInterval( \@array, $start, $end, $keyGen );
+
+sub _withPositionsOfInInterval
+{
+	my $aCollection = shift;    # array ref
+	my $start       = shift;
+	my $end         = shift;
+	my $keyGen      = shift;
+	my %d;
+	my $index;
+	for ( $index = $start ; $index <= $end ; $index++ )
+	{
+		my $element = $aCollection->[$index];
+		my $key = &$keyGen( $element, @_ );
+		if ( exists( $d{$key} ) )
+		{
+			unshift ( @{ $d{$key} }, $index );
+		}
+		else
+		{
+			$d{$key} = [$index];
+		}
+	}
+	return wantarray ? %d : \%d;
+}
+
+# Find the place at which aValue would normally be inserted into the array. If
+# that place is already occupied by aValue, do nothing, and return undef. If
+# the place does not exist (i.e., it is off the end of the array), add it to
+# the end, otherwise replace the element at that point with aValue.
+# It is assumed that the array's values are numeric.
+# This is where the bulk (75%) of the time is spent in this module, so try to
+# make it fast!
+
+sub _replaceNextLargerWith
+{
+	my ( $array, $aValue, $high ) = @_;
+	$high ||= $#$array;
+
+	# off the end?
+	if ( $high == -1 || $aValue > $array->[-1] )
+	{
+		push ( @$array, $aValue );
+		return $high + 1;
+	}
+
+	# binary search for insertion point...
+	my $low = 0;
+	my $index;
+	my $found;
+	while ( $low <= $high )
+	{
+		$index = ( $high + $low ) / 2;
+
+		#		$index = int(( $high + $low ) / 2);		# without 'use integer'
+		$found = $array->[$index];
+
+		if ( $aValue == $found )
+		{
+			return undef;
+		}
+		elsif ( $aValue > $found )
+		{
+			$low = $index + 1;
+		}
+		else
+		{
+			$high = $index - 1;
+		}
+	}
+
+	# now insertion point is in $low.
+	$array->[$low] = $aValue;    # overwrite next larger
+	return $low;
+}
+
+# This method computes the longest common subsequence in $a and $b.
+
+# Result is array or ref, whose contents is such that
+# 	$a->[ $i ] == $b->[ $result[ $i ] ]
+# foreach $i in ( 0 .. $#result ) if $result[ $i ] is defined.
+
+# An additional argument may be passed; this is a hash or key generating
+# function that should return a string that uniquely identifies the given
+# element.  It should be the case that if the key is the same, the elements
+# will compare the same. If this parameter is undef or missing, the key
+# will be the element as a string.
+
+# By default, comparisons will use "eq" and elements will be turned into keys
+# using the default stringizing operator '""'.
+
+# Additional parameters, if any, will be passed to the key generation routine.
+
+sub _longestCommonSubsequence
+{
+	my $a      = shift;    # array ref
+	my $b      = shift;    # array ref
+	my $keyGen = shift;    # code ref
+	my $compare;           # code ref
+
+	# set up code refs
+	# Note that these are optimized.
+	if ( !defined($keyGen) )    # optimize for strings
+	{
+		$keyGen = sub { $_[0] };
+		$compare = sub { my ( $a, $b ) = @_; $a eq $b };
+	}
+	else
+	{
+		$compare = sub {
+			my $a = shift;
+			my $b = shift;
+			&$keyGen( $a, @_ ) eq &$keyGen( $b, @_ );
+		};
+	}
+
+	my ( $aStart, $aFinish, $bStart, $bFinish, $matchVector ) =
+	  ( 0, $#$a, 0, $#$b, [] );
+
+	# Check whether to use internal routine (small number of elements)
+	# or use it as a wrapper for UNIX diff
+	if ( ( $#$a > $#$b ?  $#$a : $#$b) < THRESHOLD ) {
+	  ###	  print STDERR "DEBUG: regular longestCommonSubsequence\n";
+	  # First we prune off any common elements at the beginning
+	  while ( $aStart <= $aFinish
+		  and $bStart <= $bFinish
+		  and &$compare( $a->[$aStart], $b->[$bStart], @_ ) )
+	    {
+	      $matchVector->[ $aStart++ ] = $bStart++;
+	    }
+
+	  # now the end
+	  while ( $aStart <= $aFinish
+		and $bStart <= $bFinish
+		and &$compare( $a->[$aFinish], $b->[$bFinish], @_ ) )
+	    {
+	      $matchVector->[ $aFinish-- ] = $bFinish--;
+	    }
+
+	  # Now compute the equivalence classes of positions of elements
+	  my $bMatches =
+	    _withPositionsOfInInterval( $b, $bStart, $bFinish, $keyGen, @_ );
+	  my $thresh = [];
+	  my $links  = [];
+	  
+	  my ( $i, $ai, $j, $k );
+	  for ( $i = $aStart ; $i <= $aFinish ; $i++ )
+	    {
+	      $ai = &$keyGen( $a->[$i], @_ );
+	      if ( exists( $bMatches->{$ai} ) )
+		{
+		  $k = 0;
+		  for $j ( @{ $bMatches->{$ai} } )
+		    {
+		      
+		      # optimization: most of the time this will be true
+		      if ( $k and $thresh->[$k] > $j and $thresh->[ $k - 1 ] < $j )
+			{
+			  $thresh->[$k] = $j;
+			}
+		      else
+			{
+			  $k = _replaceNextLargerWith( $thresh, $j, $k );
+			}
+		      
+		      # oddly, it's faster to always test this (CPU cache?).
+		      if ( defined($k) )
+			{
+			  $links->[$k] =
+			    [ ( $k ? $links->[ $k - 1 ] : undef ), $i, $j ];
+			}
+		    }
+		}
+	    }
+
+	  if (@$thresh)
+	    {
+	      for ( my $link = $links->[$#$thresh] ; $link ; $link = $link->[0] )
+		{
+		  $matchVector->[ $link->[1] ] = $link->[2];
+		}
+	    }
+	}
+	else {
+	  my ($fha,$fhb,$fna,$fnb,$ele,$key);
+	  my ($alines,$blines,$alb,$alf,$blb,$blf);
+	  my ($minimal)=MINIMAL;
+	  # large number of elements, use system diff
+	  ###	  print STDERR "DEBUG: fast (diff) longestCommonSubsequence\n";
+
+	  ($fha,$fna)=tempfile("DiffA-XXXX") or die "_longestCommonSubsequence: Cannot open tempfile for sequence A";
+	  ($fhb,$fnb)=tempfile("DiffB-XXXX") or die "_longestCommonSubsequence: Cannot open tempfile for sequence B";
+	  # prepare sequence A
+	  foreach $ele ( @$a ) {
+	    $key=&$keyGen( $ele, @_ );
+	    $key =~ s/\\/\\\\/g ;
+	    $key =~ s/\n/\\n/sg ;
+	    print $fha "$key\n" ;
+	  }
+	  close($fha);
+	  # prepare sequence B
+	  foreach $ele ( @$b ) {
+	    $key=&$keyGen( $ele, @_ );
+	    $key =~ s/\\/\\\\/g ;
+	    $key =~ s/\n/\\n/sg ;
+	    print $fhb "$key\n" ;
+	  }
+	  close($fhb);
+	  
+	  open(DIFFPIPE, "diff $minimal $fna $fnb |") or die "_longestCommonSubsequence: Cannot launch diff process. $!" ;
+	  # The diff line numbering begins with 1, but Perl subscripts start with 0
+	  # We follow the diff numbering but substract 1 when assigning to matchVector
+	  $aStart++; $bStart++ ; $aFinish++ ; $bFinish++ ;
+	  while( <DIFFPIPE> ) {
+	    if ( ($alines,$blines) = ( m/^(\d*(?:,\d*)?)?c(\d*(?:,\d*)?)?$/ ) ) {
+	      ($alb,$alf)=split(/,/,$alines);
+	      ($blb,$blf)=split(/,/,$blines);
+	      $alf=$alb unless defined($alf);
+	      $blf=$blb unless defined($blf);
+	      while($aStart < $alb ) {
+		$matchVector->[ -1 + $aStart++ ] = -1 + $bStart++ ;
+	      }
+	      # check for consistency
+	      $bStart==$blb or die "_longestCommonSubsequence: Fatal error in interpreting diff output: Inconsistency in changed sequence";
+	      $aStart=$alf+1;
+	      $bStart=$blf+1;
+	    }
+	    elsif ( ($alb,$blines) = ( m/^(\d*)a(\d*(?:,\d*)?)$/ ) ) {
+	      ($blb,$blf)=split(/,/,$blines);
+	      $blf=$blb unless defined($blf);
+	      while ( $bStart < $blb ) {
+		$matchVector->[ -1 + $aStart++ ] = -1 + $bStart++ ;
+	      }
+	      $aStart==$alb+1 or die "_longestCommonSubsequence: Fatal error in interpreting diff output: Inconsistency in appended sequence near elements $aStart and $bStart";
+	      $bStart=$blf+1;
+	    }
+	    elsif ( ($alines,$blb) = ( m/^(\d*(?:,\d*)?)d(\d*)$/ ) ) {
+	      ($alb,$alf)=split(/,/,$alines);
+	      $alf=$alb unless defined($alf);
+	      while ( $aStart < $alb ) {
+		$matchVector->[ -1 + $aStart++ ] = -1 + $bStart++ ;
+	      }
+	      $bStart==$blb+1 or die "_longestCommonSubsequence: Fatal error in interpreting diff output: Inconsistency in deleted sequence near elements $aStart and $bStart";
+	      $aStart=$alf+1;
+	    }
+	    elsif ( m/^Binary files/ ) {
+	      # if diff reports it is a binary file force --text mode. I do not like
+	      # to always use this option because it is probably only available in GNU diff
+	      open(DIFFPIPE, "diff --text $fna $fnb |") or die "Cannot launch diff process. $!" ;
+	    }
+	    # Default: just skip line
+	  }
+	  while ($aStart <= $aFinish ) {
+	    $matchVector->[ -1 + $aStart++ ] = -1 + $bStart++ ;
+	  }
+	  $bStart==$bFinish+1  or die "_longestCommonSubsequence: Fatal error in interpreting diff output: Inconsistency at end";
+	  close DIFFPIPE;
+	  # check whether a system error has occurred or return status is greater than or equal to 5
+	  if ( $! || ($? >> 8) > 5) {
+	    print STDERR "diff process failed with exit code ", ($? >> 8), " $!\n";
+	    die;
+	  }
+	  unlink $fna,$fnb ;
+	}
+	return wantarray ? @$matchVector : $matchVector;
+}
+
+sub traverse_sequences
+{
+	my $a                 = shift;                                  # array ref
+	my $b                 = shift;                                  # array ref
+	my $callbacks         = shift || {};
+	my $keyGen            = shift;
+	my $matchCallback     = $callbacks->{'MATCH'} || sub { };
+	my $discardACallback  = $callbacks->{'DISCARD_A'} || sub { };
+	my $finishedACallback = $callbacks->{'A_FINISHED'};
+	my $discardBCallback  = $callbacks->{'DISCARD_B'} || sub { };
+	my $finishedBCallback = $callbacks->{'B_FINISHED'};
+	my $matchVector = _longestCommonSubsequence( $a, $b, $keyGen, @_ );
+
+	# Process all the lines in @$matchVector
+	my $lastA = $#$a;
+	my $lastB = $#$b;
+	my $bi    = 0;
+	my $ai;
+
+	for ( $ai = 0 ; $ai <= $#$matchVector ; $ai++ )
+	{
+		my $bLine = $matchVector->[$ai];
+		if ( defined($bLine) )    # matched
+		{
+			&$discardBCallback( $ai, $bi++, @_ ) while $bi < $bLine;
+			&$matchCallback( $ai,    $bi++, @_ );
+		}
+		else
+		{
+			&$discardACallback( $ai, $bi, @_ );
+		}
+	}
+
+	# The last entry (if any) processed was a match.
+	# $ai and $bi point just past the last matching lines in their sequences.
+
+	while ( $ai <= $lastA or $bi <= $lastB )
+	{
+
+		# last A?
+		if ( $ai == $lastA + 1 and $bi <= $lastB )
+		{
+			if ( defined($finishedACallback) )
+			{
+				&$finishedACallback( $lastA, @_ );
+				$finishedACallback = undef;
+			}
+			else
+			{
+				&$discardBCallback( $ai, $bi++, @_ ) while $bi <= $lastB;
+			}
+		}
+
+		# last B?
+		if ( $bi == $lastB + 1 and $ai <= $lastA )
+		{
+			if ( defined($finishedBCallback) )
+			{
+				&$finishedBCallback( $lastB, @_ );
+				$finishedBCallback = undef;
+			}
+			else
+			{
+				&$discardACallback( $ai++, $bi, @_ ) while $ai <= $lastA;
+			}
+		}
+
+		&$discardACallback( $ai++, $bi, @_ ) if $ai <= $lastA;
+		&$discardBCallback( $ai, $bi++, @_ ) if $bi <= $lastB;
+	}
+
+	return 1;
+}
+
+sub traverse_balanced
+{
+	my $a                 = shift;                                  # array ref
+	my $b                 = shift;                                  # array ref
+	my $callbacks         = shift || {};
+	my $keyGen            = shift;
+	my $matchCallback     = $callbacks->{'MATCH'} || sub { };
+	my $discardACallback  = $callbacks->{'DISCARD_A'} || sub { };
+	my $discardBCallback  = $callbacks->{'DISCARD_B'} || sub { };
+	my $changeCallback    = $callbacks->{'CHANGE'};
+	my $matchVector = _longestCommonSubsequence( $a, $b, $keyGen, @_ );
+
+	# Process all the lines in match vector
+	my $lastA = $#$a;
+	my $lastB = $#$b;
+	my $bi    = 0;
+	my $ai    = 0;
+	my $ma    = -1;
+	my $mb;
+
+	while (1)
+	{
+
+		# Find next match indices $ma and $mb
+		do { $ma++ } while ( $ma <= $#$matchVector && !defined $matchVector->[$ma] );
+
+		last if $ma > $#$matchVector;    # end of matchVector?
+		$mb = $matchVector->[$ma];
+
+		# Proceed with discard a/b or change events until
+		# next match
+		while ( $ai < $ma || $bi < $mb )
+		{
+
+			if ( $ai < $ma && $bi < $mb )
+			{
+
+				# Change
+				if ( defined $changeCallback )
+				{
+					&$changeCallback( $ai++, $bi++, @_ );
+				}
+				else
+				{
+					&$discardACallback( $ai++, $bi, @_ );
+					&$discardBCallback( $ai, $bi++, @_ );
+				}
+			}
+			elsif ( $ai < $ma )
+			{
+				&$discardACallback( $ai++, $bi, @_ );
+			}
+			else
+			{
+
+				# $bi < $mb
+				&$discardBCallback( $ai, $bi++, @_ );
+			}
+		}
+
+		# Match
+		&$matchCallback( $ai++, $bi++, @_ );
+	}
+
+	while ( $ai <= $lastA || $bi <= $lastB )
+	{
+		if ( $ai <= $lastA && $bi <= $lastB )
+		{
+
+			# Change
+			if ( defined $changeCallback )
+			{
+				&$changeCallback( $ai++, $bi++, @_ );
+			}
+			else
+			{
+				&$discardACallback( $ai++, $bi, @_ );
+				&$discardBCallback( $ai, $bi++, @_ );
+			}
+		}
+		elsif ( $ai <= $lastA )
+		{
+			&$discardACallback( $ai++, $bi, @_ );
+		}
+		else
+		{
+
+			# $bi <= $lastB
+			&$discardBCallback( $ai, $bi++, @_ );
+		}
+	}
+
+	return 1;
+}
+
+sub LCS
+{
+	my $a = shift;                                           # array ref
+	my $matchVector = _longestCommonSubsequence( $a, @_ );
+	my @retval;
+	my $i;
+	for ( $i = 0 ; $i <= $#$matchVector ; $i++ )
+	{
+		if ( defined( $matchVector->[$i] ) )
+		{
+			push ( @retval, $a->[$i] );
+		}
+	}
+	return wantarray ? @retval : \@retval;
+}
+
+sub diff
+{
+	my $a      = shift;    # array ref
+	my $b      = shift;    # array ref
+	my $retval = [];
+	my $hunk   = [];
+	my $discard = sub { push ( @$hunk, [ '-', $_[0], $a->[ $_[0] ] ] ) };
+	my $add = sub { push ( @$hunk, [ '+', $_[1], $b->[ $_[1] ] ] ) };
+	my $match = sub { push ( @$retval, $hunk ) if scalar(@$hunk); $hunk = [] };
+	traverse_sequences( $a, $b,
+		{ MATCH => $match, DISCARD_A => $discard, DISCARD_B => $add }, @_ );
+	&$match();
+	return wantarray ? @$retval : $retval;
+}
+
+sub sdiff
+{
+	my $a      = shift;    # array ref
+	my $b      = shift;    # array ref
+	my $retval = [];
+	my $discard = sub { push ( @$retval, [ '-', $a->[ $_[0] ], "" ] ) };
+	my $add = sub { push ( @$retval, [ '+', "", $b->[ $_[1] ] ] ) };
+	my $change = sub {
+		push ( @$retval, [ 'c', $a->[ $_[0] ], $b->[ $_[1] ] ] );
+	};
+	my $match = sub {
+		push ( @$retval, [ 'u', $a->[ $_[0] ], $b->[ $_[1] ] ] );
+	};
+	traverse_balanced(
+		$a,
+		$b,
+		{
+			MATCH     => $match,
+			DISCARD_A => $discard,
+			DISCARD_B => $add,
+			CHANGE    => $change,
+		},
+		@_
+	);
+	return wantarray ? @$retval : $retval;
+}
+
+1;
+}
+import Algorithm::Diff qw(traverse_sequences);
+# End of inserted block for stand-alone version
+
+
+use Getopt::Long ;
+use strict ;
+use warnings;
+use utf8 ;
+
+use File::Spec ;
+
+
+my ($algodiffversion)=split(/ /,$Algorithm::Diff::VERSION);
+
+
+my ($versionstring)=<<EOF ;
+This is LATEXDIFF 1.3.4 (Algorithm::Diff $Algorithm::Diff::VERSION, Perl $^V)
+  (c) 2004-2024 F J Tilmann
+EOF
+
+# Hash with defaults for configuration variables. These marked undef have default values constructed from list defined in the DATA block
+# (under tag CONFIG)
+my %CONFIG=(
+   MINWORDSBLOCK => 3, # minimum number of tokens to form an independent block
+                        # shorter identical blocks will be merged to the previous word
+   SCALEDELGRAPHICS => 0.5, # factor with which deleted figures will be scaled down (i.e. 0.5 implies they are shown at half linear size)
+                             # this is only used for --dgraphics-markup=BOTH option
+   FLOATENV => undef ,   # Environments in which FL variants of defined commands are used
+   PICTUREENV => undef ,   # Environments in which all change markup is removed
+   MATHENV => undef ,           # Environments turning on display math mode (code also knows about \[ and \])
+   MATHREPL => 'displaymath',  # Environment introducing deleted maths blocks
+   MATHARRENV => undef ,           # Environments turning on eqnarray math mode
+   MATHARRREPL => 'eqnarray*',  # Environment introducing deleted maths blocks (note that now the starred varieties are being used, so this is only used to replace MATHMODE environments (where original environment is unknown)
+   ARRENV => undef , # Environments making arrays in math mode.  The underlining style does not cope well with those - as a result in-text math environments are surrounded by \mbox{ } if any of these commands is used in an inline math block
+   COUNTERCMD => undef,
+                                        # COUNTERCMD textcmds which are associated with a counter
+                                        # If any of these commands occur in a deleted block
+                                        # they will be followed by an \addtocounter{...}{-1}
+                                        # for the associated counter such that the overall numbers
+                                        # should be the same as in the new file
+   LISTENV => undef ,  # list making environments - they will generally be kept
+   VERBATIMENV => undef,      # Environments whose content should be treated as verbatim text and not be touched
+   VERBATIMLINEENV => undef,  # Environments whose content should be treated as verbatim text and processed in line diff mode
+   CUSTOMDIFCMD => undef,# Custom dif command. Is defined in the document as a \DELcommand and \ADDcommand version to be replaced by the diff
+   ITEMCMD => 'item'                    # command marking item in a list environment
+);
+# Configuration variables: these have to be visible from the subroutines
+my ($ARRENV,
+    $COUNTERCMD,
+    $FLOATENV,
+    $ITEMCMD,
+    $LISTENV,
+    $MATHARRENV,
+    $MATHARRREPL,
+    $MATHENV,
+    $MATHREPL,
+    $MINWORDSBLOCK,
+    $PICTUREENV,
+    $SCALEDELGRAPHICS,
+    $VERBATIMENV,
+    $VERBATIMLINEENV,
+    $CUSTOMDIFCMD
+    );
+
+# my $MINWORDSBLOCK=3; # minimum number of tokens to form an independent block
+#                      # shorter identical blocks will be merged to the previous word
+# my $SCALEDELGRAPHICS=0.5; # factor with which deleted figures will be scaled down (i.e. 0.5 implies they are shown at half linear size)
+#                       # this is only used for --graphics-markup=BOTH option
+# my $FLOATENV='(?:figure|table|plate)[\w\d*@]*' ;   # Environments in which FL variants of defined commands are used
+# my $PICTUREENV='(?:picture|tikzpicture|DIFnomarkup)[\w\d*@]*' ;   # Environments in which all change markup is removed
+# my $MATHENV='(?:equation[*]?|displaymath|DOLLARDOLLAR)[*]?' ;           # Environments turning on display math mode (code also knows about \[ and \])
+# my $MATHREPL='displaymath';  # Environment introducing deleted maths blocks
+# my $MATHARRENV='(?:eqnarray|align|alignat|gather|multline|flalign)[*]?' ;           # Environments turning on eqnarray math mode
+# my $MATHARRREPL='eqnarray*';  # Environment introducing deleted maths blocks
+# my $ARRENV='(?:aligned|gathered|multlined|array|[pbvBV]?matrix|smallmatrix|cases|split)'; # Environments making arrays in math mode.  The underlining style does not cope well with those - as a result in-text math environments are surrounded by \mbox{ } if any of these commands is used in an inline math block
+# my $COUNTERCMD='(?:footnote|part|chapter|section|subsection|subsubsection|paragraph|subparagraph)';  # textcmds which are associated with a counter
+#                                         # If any of these commands occur in a deleted block
+#                                         # they will be succeeded by an \addtocounter{...}{-1}
+#                                         # for the associated counter such that the overall numbers
+#                                         # should be the same as in the new file
+# my $LISTENV='(?:itemize|description|enumerate)'; # list making environments - they will generally be kept
+# my $ITEMCMD='item';   # command marking item in a list environment
+
+
+
+my $LABELCMD='(?:label)';                # matching commands are disabled within deleted blocks - mostly useful for maths mode, as otherwise it would be fine to just not add those to SAFECMDLIST
+my @UNSAFEMATHCMD=('qedhere','intertext','begin','end');           # Commands which are definitely unsafe for marking up in math mode (amsmath qedhere only tested to not work with UNDERLINE markup) (only affects WHOLE and COARSE math markup modes). Note that unlike text mode (or FINE math mode0 deleted unsafe commands are not deleted but simply taken outside \DIFdel
+my $MBOXINLINEMATH=0; # if set to 1 then surround marked-up inline maths expression with \mbox ( to get around compatibility
+                      # problems between some maths packages and ulem package
+
+
+# Markup strings
+# If at all possible, do not change these as parts of the program
+# depend on the actual name (particularly post-processing)
+# At the very least adapt subroutine postprocess to new tokens.
+my $ADDMARKOPEN='\DIFaddbegin ';   # Token to mark begin of appended text
+my $ADDMARKCLOSE='\DIFaddend ';   # Token to mark end of appended text
+my $ADDOPEN='\DIFadd{';  # To mark begin of added text passage
+my $ADDCLOSE='}';        # To mark end of added text passage
+my $ADDCOMMENT='DIF > ';   # To mark added comment line
+my $DELMARKOPEN='\DIFdelbegin ';   # Token to mark begin of deleted text
+my $DELMARKCLOSE='\DIFdelend ';   # Token to mark end of deleted text
+my $DELOPEN='\DIFdel{';  # To mark begin of deleted text passage
+my $DELCLOSE='}';        # To mark end of deleted text passage
+my $DELCMDOPEN='%DIFDELCMD < ';  # To mark begin of deleted commands (must begin with %, i.e., be a comment
+my $DELCMDCLOSE="%%%\n";    # To mark end of deleted commands (must end with a new line)
+my $AUXCMD='%DIFAUXCMD' ; #  follows auxiliary commands put in by latexdiff to make difference file legal
+                          # auxiliary commands must be on a line of their own
+                          # Note that for verbatim environment openings the %DIFAUXCMD cannot be placed in
+                          # the same line as this would mean they are shown
+                          # so the special form "%DIFAUXCMD NEXT" is used to indicate that the next line
+                          # is an auxiliary command
+                          # Similarly "%DIFAUXCMD LAST" would indicate the auxiliary command is in previous line (not currently used)
+my $DELCOMMENT='DIF < ';   # To mark deleted comment line
+my $VERBCOMMENT='DIFVRB ';  # to mark lines which are within a verbatim environment
+
+# main local variables:
+my @TEXTCMDLIST=();  # array containing patterns of commands with text arguments
+my @TEXTCMDEXCL=();  # array containing patterns of commands without text arguments (if a pattern
+                     # matches both TEXTCMDLIST and TEXTCMDEXCL it is excluded)
+my @CONTEXT1CMDLIST=();  # array containing patterns of commands with text arguments (subset of text commands),
+                         # but which cause confusion if used out of context (e.g. \caption).
+                         # In deleted passages, the command will be disabled but its argument is marked up
+                         # Otherwise they behave exactly like TEXTCMD's
+my @CONTEXT1CMDEXCL=();  # exclude list for above, but always empty
+my @CONTEXT2CMDLIST=();  # array containing patterns of commands with text arguments, but which fail or cause confusion
+                         # if used out of context (e.g. \title). They and their arguments will be disabled in deleted
+                         # passages
+my @CONTEXT2CMDEXCL=();  # exclude list for above, but always empty
+my @MATHTEXTCMDLIST=();  # treat like textcmd.  If a textcmd is in deleted or added block, just wrap the
+                         # whole content with \DIFadd or \DIFdel irrespective of content.  This functionality
+                         # is useful for pseudo commands \MATHBLOCK.. into which math environments are being
+                         # transformed
+my @MATHTEXTCMDEXCL=();  #
+
+# Note I need to declare this with "our" instead of "my" because later in the code I have to "local"ise these
+our @SAFECMDLIST=();  # array containing patterns of safe commands (which do not break when in the argument of DIFadd or DIFDEL)
+our @SAFECMDEXCL=();
+my @MBOXCMDLIST=();   # patterns for commands which are in principle safe but which need to be surrounded by an \mbox
+my @MBOXCMDEXCL=();           # all the patterns in MBOXCMDLIST will be appended to SAFECMDLIST
+
+my @KEEPCMDLIST=( qr/^bibitem$/ );   # patterns for commands which should not be deleted in nominally delete text passages
+my @KEEPCMDEXCL=();
+
+my ($i,$j,$l);
+my ($old,$new);
+my ($line,$key);
+my (@dumlist);
+my ($newpreamble,$oldpreamble);
+my (@newpreamble,@oldpreamble,@diffpreamble,@diffbody);
+my ($latexdiffpreamble);
+my ($oldbody, $newbody, $diffbo);
+my ($oldpost, $newpost);
+my ($diffall);
+# Option names
+my ($type,$subtype,$floattype,$config,$preamblefile,$encoding,$nolabel,$visiblelabel,
+    $filterscript,$ignorefilterstderr,
+    $showpreamble,$showsafe,$showtext,$showconfig,$showall,
+    $replacesafe,$appendsafe,$excludesafe,
+    $replacetext,$appendtext,$excludetext,
+    $replacecontext1,$appendcontext1,
+    $replacecontext2,$appendcontext2,
+    $help,$verbose,$driver,$version,$ignorewarnings,
+    $onlyadditions,
+    $enablecitmark,$disablecitmark,$allowspaces,$flatten,$nolinks,$debug,$earlylatexdiffpreamble);  ###$disablemathmark,
+my ($mboxsafe);
+# MNEMNONICS for mathmarkup
+my $mathmarkup;
+use constant {
+  OFF => 0,
+  WHOLE => 1,
+  COARSE => 2,
+  FINE => 3
+};
+# MNEMNONICS for graphicsmarkup
+my $graphicsmarkup;
+use constant {
+  NONE => 0,
+  NEWONLY => 1,
+  BOTH => 2
+};
+
+my ($mboxcmd);
+
+my (@configlist,@addtoconfiglist,@labels,
+    @appendsafelist,@excludesafelist,
+    @appendmboxsafelist,@excludemboxsafelist,
+    @appendtextlist,@excludetextlist,
+    @appendcontext1list,@appendcontext2list,
+    @packagelist);
+my ($assign,@config);
+# Hash where keys corresponds to the names of  all included packages (including the documentclass as another package
+# the optional arguments to the package are the values of the hash elements
+my ($pkg,%packages);
+
+# Defaults
+$mathmarkup=COARSE;
+$verbose=0;
+$onlyadditions=0;
+# output debug and intermediate files, set to 0 in final distribution
+$debug=0;
+# insert preamble directly after documentclass - experimental feature, set to 0 in final distribution
+# Note that this failed with mini example (or other files, where packages used in latexdiff preamble
+# are called again with incompatible options in preamble of resulting file)
+$earlylatexdiffpreamble=0;
+
+
+# define character properties
+sub IsNonAsciiPunct { return <<'END'    # Unicode punctuation but excluding ASCII punctuation
++utf8::IsPunct
+-utf8::IsASCII
+END
+}
+sub IsNonAsciiS { return <<'END'       # Unicode symbol but excluding ASCII
++utf8::IsS
+-utf8::IsASCII
+END
+}
+
+
+my %verbhash;
+
+Getopt::Long::Configure('bundling');
+GetOptions('type|t=s' => \$type,
+	   'subtype|s=s' => \$subtype,
+	   'floattype|f=s' => \$floattype,
+	   'config|c=s' => \@configlist,
+	   'add-to-config=s' => \@addtoconfiglist,
+	   'preamble|p=s' => \$preamblefile,
+	   'encoding|e=s' => \$encoding,
+	   'label|L=s' => \@labels,
+	   'no-label' => \$nolabel,
+	   'visible-label' => \$visiblelabel,
+	   'exclude-safecmd|A=s' => \@excludesafelist,
+	   'replace-safecmd=s' => \$replacesafe,
+	   'append-safecmd|a=s' => \@appendsafelist,
+	   'exclude-textcmd|X=s' => \@excludetextlist,
+	   'replace-textcmd=s' => \$replacetext,
+	   'append-textcmd|x=s' => \@appendtextlist,
+	   'replace-context1cmd=s' => \$replacecontext1,
+	   'append-context1cmd=s' => \@appendcontext1list,
+	   'replace-context2cmd=s' => \$replacecontext2,
+	   'append-context2cmd=s' => \@appendcontext2list,
+	   'exclude-mboxsafecmd=s' => \@excludemboxsafelist,
+	   'append-mboxsafecmd=s' => \@appendmboxsafelist,
+	   'show-preamble' => \$showpreamble,
+	   'show-safecmd' => \$showsafe,
+	   'show-textcmd' => \$showtext,
+	   'show-config' => \$showconfig,
+	   'show-all' => \$showall,
+           'packages=s' => \@packagelist,
+	   'allow-spaces' => \$allowspaces,
+           'math-markup=s' => \$mathmarkup,
+           'graphics-markup=s' => \$graphicsmarkup,
+           'enable-citation-markup|enforce-auto-mbox' => \$enablecitmark,
+           'disable-citation-markup|disable-auto-mbox' => \$disablecitmark,
+	   'verbose|V' => \$verbose,
+	   'ignore-warnings' => \$ignorewarnings,
+	   'driver=s'=> \$driver,
+	   'flatten' => \$flatten,
+	   'filter-script=s' => \$filterscript,
+	   'ignore-filter-stderr' => \$ignorefilterstderr,
+	   'no-links' => \$nolinks,
+	   'no-del' => \$onlyadditions,
+	   'version' => \$version,
+	   'help|h' => \$help,
+	   'debug!' => \$debug ) or die "Use latexdiff -h to get help.\n" ;
+
+if ( $help ) {
+  usage() ;
+}
+
+
+if ( $version ) {
+  print STDERR $versionstring ;
+  exit 0;
+}
+
+print STDERR $versionstring if $verbose;
+
+if (defined($showall)){
+  $showpreamble=$showsafe=$showtext=$showconfig=1;
+}
+# Default types
+$type='UNDERLINE' unless defined($type);
+$subtype='SAFE' unless defined($subtype);
+# set floattype to IDENTICAL for LABEL and ONLYCHANGEDPAGE subtype, unless it has been set explicitly on the command line
+$floattype=($subtype eq 'LABEL' || $subtype eq 'ONLYCHANGEDPAGE') ? 'IDENTICAL' : 'FLOATSAFE' unless defined($floattype);
+if ( $subtype eq 'LABEL' ) {
+  print STDERR "Note that LABEL subtype is deprecated. If possible, use ZLABEL instead (requires zref package)";
+}
+
+if (defined($mathmarkup)) {
+  $mathmarkup=~tr/a-z/A-Z/;
+  if ( $mathmarkup eq 'OFF' ){
+    $mathmarkup=OFF;
+  } elsif ( $mathmarkup eq 'WHOLE' ){
+    $mathmarkup=WHOLE;
+  } elsif ( $mathmarkup eq 'COARSE' ){
+    $mathmarkup=COARSE;
+  } elsif ( $mathmarkup eq 'FINE' ){
+    $mathmarkup=FINE;
+  } elsif ( $mathmarkup !~ m/^[0123]$/ ) {
+    die "latexdiff Illegal value: ($mathmarkup)  for option--math-markup. Possible values: OFF,WHOLE,COARSE,FINE,0-3\n";
+  }
+  # else use numerical value
+}
+
+# Give filterscript a default empty string
+$filterscript="" unless defined($filterscript);
+
+# setting extra preamble commands
+if (defined($preamblefile)) {
+  $latexdiffpreamble=join "\n",(extrapream($preamblefile),"");
+} else {
+  $latexdiffpreamble=join "\n",(extrapream($type,$subtype,$floattype),"");
+}
+
+if ( defined($driver) ) {
+  # for changebar only
+  $latexdiffpreamble=~s/\[pdftex\]/[$driver]/sg;
+}
+# setting up @SAFECMDLIST and @SAFECMDEXCL
+if (defined($replacesafe)) {
+  init_regex_arr_ext(\@SAFECMDLIST,$replacesafe);
+} else {
+  init_regex_arr_data(\@SAFECMDLIST, "SAFE COMMANDS");
+}
+foreach $appendsafe ( @appendsafelist ) {
+  init_regex_arr_ext(\@SAFECMDLIST, $appendsafe);
+}
+foreach $excludesafe ( @excludesafelist ) {
+  init_regex_arr_ext(\@SAFECMDEXCL, $excludesafe);
+}
+# setting up @MBOXCMDLIST and @MBOXCMDEXCL
+foreach $mboxsafe ( @appendmboxsafelist ) {
+  init_regex_arr_ext(\@MBOXCMDLIST, $mboxsafe);
+}
+foreach $mboxsafe ( @excludemboxsafelist ) {
+  init_regex_arr_ext(\@MBOXCMDEXCL, $mboxsafe);
+}
+
+
+
+# setting up @TEXTCMDLIST and @TEXTCMDEXCL
+if (defined($replacetext)) {
+  init_regex_arr_ext(\@TEXTCMDLIST,$replacetext);
+} else {
+  init_regex_arr_data(\@TEXTCMDLIST, "TEXT COMMANDS");
+}
+foreach $appendtext ( @appendtextlist ) {
+  init_regex_arr_ext(\@TEXTCMDLIST, $appendtext);
+}
+foreach $excludetext ( @excludetextlist ) {
+  init_regex_arr_ext(\@TEXTCMDEXCL, $excludetext);
+}
+
+
+# setting up @CONTEXT1CMDLIST ( @CONTEXT1CMDEXCL exist but is always empty )
+if (defined($replacecontext1)) {
+  init_regex_arr_ext(\@CONTEXT1CMDLIST,$replacecontext1);
+} else {
+  init_regex_arr_data(\@CONTEXT1CMDLIST, "CONTEXT1 COMMANDS");
+}
+foreach $appendcontext1 ( @appendcontext1list ) {
+  init_regex_arr_ext(\@CONTEXT1CMDLIST, $appendcontext1);
+}
+
+
+# setting up @CONTEXT2CMDLIST ( @CONTEXT2CMDEXCL exist but is always empty )
+if (defined($replacecontext2)) {
+  init_regex_arr_ext(\@CONTEXT2CMDLIST,$replacecontext2);
+} else {
+  init_regex_arr_data(\@CONTEXT2CMDLIST, "CONTEXT2 COMMANDS");
+}
+foreach $appendcontext2 ( @appendcontext2list ) {
+  init_regex_arr_ext(\@CONTEXT2CMDLIST, $appendcontext2);
+}
+
+# setting configuration variables
+@config=();
+foreach $config ( @configlist ) {
+  if (-f $config || lc $config eq '/dev/null' ) {
+    open(FILE,$config) or die ("Couldn't open configuration file $config: $!");
+    while (<FILE>) {
+      chomp;
+      next if /^\s*#/ || /^\s*%/ || /^\s*$/ ;
+      push (@config,$_);
+    }
+    close(FILE);
+  }
+  else {
+#    foreach ( split(",",$config) ) {
+#      push @config,$_;
+#    }
+     push @config,split(",",$config)
+  }
+}
+foreach $assign ( @config ) {
+  $assign=~ m/\s*(\w*)\s*=\s*(\S*)\s*$/ or die "Illegal assignment $assign in configuration list (must be variable=value)";
+  exists $CONFIG{$1} or die "Unknown configuration variable $1.";
+  $CONFIG{$1}=$2;
+}
+
+my @addtoconfig=();
+foreach $config ( @addtoconfiglist ) {
+  if (-f $config || lc $config eq '/dev/null' ) {
+    open(FILE,$config) or die ("Couldn't open addd-to-config file $config: $!");
+    while (<FILE>) {
+      chomp;
+      next if /^\s*#/ || /^\s*%/ || /^\s*$/ ;
+      push (@addtoconfig,$_);
+    }
+    close(FILE);
+  }
+  else {
+#    foreach ( split(",",$config) ) {
+#      push @addtoconfig,$_;
+#    }
+     push @addtoconfig,split(",",$config)
+  }
+}
+
+# initialise default lists from DATA
+# for those configuration variables, which have not been set explicitly, initiate from list in document
+foreach $key ( keys(%CONFIG) ) {
+  if (!defined $CONFIG{$key}) {
+    @dumlist=();
+    init_regex_arr_data(\@dumlist,"$key CONFIG");
+    $CONFIG{$key}=join(";",@dumlist)
+  }
+}
+
+
+foreach $assign ( @addtoconfig ) {
+  ###print STDERR "assign:|$assign|\n";
+  $assign=~ m/\s*(\w*)\s*=\s*(\S*)\s*$/ or die "Illegal assignment $assign in configuration list (must be variable=value)";
+  exists $CONFIG{$1} or die "Unknown configuration variable $1.";
+  $CONFIG{$1}.=";$2";
+}
+
+# Map from hash to variables (we do this to have more concise code later, change from comma-separated list)
+foreach  ( keys(%CONFIG) ) {
+  if ( $_ eq "MINWORDSBLOCK" ) { $MINWORDSBLOCK = $CONFIG{$_}; }
+  elsif ( $_ eq "FLOATENV" ) { $FLOATENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "ITEMCMD" ) { $ITEMCMD = $CONFIG{$_} ; }
+  elsif ( $_ eq "LISTENV" ) { $LISTENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "PICTUREENV" ) { $PICTUREENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "MATHENV" ) { $MATHENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "MATHREPL" ) { $MATHREPL = $CONFIG{$_} ; }
+  elsif ( $_ eq "MATHARRENV" ) { $MATHARRENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "MATHARRREPL" ) { $MATHARRREPL = $CONFIG{$_} ;
+				  print STDERR "WARNING: Setting MATHARRREPL is depracated. Generally deleted math array environments will be set to their starred varieties and the setting of MATHARREPL is ignored.\n\n" unless $MATHARRREPL =~ /eqnarray\*/ ;
+				}
+  elsif ( $_ eq "ARRENV" ) { $ARRENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "VERBATIMENV" ) { $VERBATIMENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "VERBATIMLINEENV" ) { $VERBATIMLINEENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "CUSTOMDIFCMD" ) { $CUSTOMDIFCMD = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "COUNTERCMD" ) { $COUNTERCMD = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "SCALEDELGRAPHICS" ) { $SCALEDELGRAPHICS = $CONFIG{$_} ; }
+  else { die "Unknown configuration variable $_.";}
+}
+
+if ( $mathmarkup == COARSE || $mathmarkup == WHOLE ) {
+  push(@MATHTEXTCMDLIST,qr/^MATHBLOCK(?:$MATHENV|$MATHARRENV|SQUAREBRACKET)$/);
+}
+
+
+
+
+
+foreach $pkg ( @packagelist ) {
+  map { $packages{$_}="" } split(/,/,$pkg) ;
+}
+
+
+if ($showconfig || $showtext || $showsafe || $showpreamble) {
+
+  show_configuration();
+  exit 0;
+}
+
+
+if ( @ARGV != 2 ) {
+  print STDERR "2 and only 2 non-option arguments required.  Write latexdiff -h to get help\n";
+  exit(2);
+}
+
+# Are extra spaces between command arguments permissible?
+my $extraspace;
+if ($allowspaces) {
+  $extraspace='\s*';
+} else {
+  $extraspace='';
+}
+
+# append context lists to text lists (as text property is implied)
+push @TEXTCMDLIST, @CONTEXT1CMDLIST;
+push @TEXTCMDLIST, @CONTEXT2CMDLIST;
+
+push @TEXTCMDLIST, @MATHTEXTCMDLIST if $mathmarkup==COARSE;
+
+# internal additions to SAFECMDLIST
+push(@SAFECMDLIST, qr/^QLEFTBRACE$/, qr/^QRIGHTBRACE$/);
+
+
+# Patterns. These are used by some of the subroutines, too
+# I can only define them down here because value of extraspace depends on an option
+
+  my $pat0 = '(?:[^{}])*';
+  my $pat_n = $pat0;
+# if you get "undefined control sequence MATHBLOCKmath" error, increase the maximum value in this loop
+  for (my $i_pat = 0; $i_pat < 20; ++$i_pat){
+    $pat_n = '(?:[^{}]|\{'.$pat_n.'\}|\\\\\{|\\\\\})*';
+    # Actually within the text body, quoted braces are replaced in pre-processing. The only place where
+    # the last part of the pattern matters is when processing the arguments of context2cmds in the preamble
+    # and these contain a \{ or \} combination, probably rare.
+    # It should thus be fine to use the simpler version below.
+    ###  $pat_n = '(?:[^{}]|\{'.$pat_n.'\})*';
+  }
+
+  my $brat0 = '(?:[^\[\]]|\\\[|\\\])*';
+  my $brat_n = $brat0;
+  for (my $i_pat = 0; $i_pat < 4; ++$i_pat){
+    $brat_n = '(?:[^\[\]]|\['.$brat_n.'\]|\\\[|\\\])*';
+    ###  $brat_n = '(?:[^\[\]]|\['.$brat_n.'\])*';   # Version not taking into account escaped \[ and \]
+  }
+  my $abrat0 = '(?:[^<>])*';
+
+
+# variable definitions are in order that they are matched
+  my $and = '&';
+  my $quotemarks = '(?:\'\')|(?:\`\`)';
+  # some common abbreviations involving punctuations within them. They need special treatment because otherwise in some
+  # circumstances the gnoring of white space differences in conjunction with merging according to MINWORDSBLOCK rule
+  # could turn 'i.e.' into 'i.\PAR e.' (see https://github.com/ftilmann/latexdiff/issues/269)
+  # English: i.e., e.g.   Deutsch: z.B.
+  my $abbreviation='(?:i\. ?e\.|e\. ?g\.|z\. ?B\.)' ;
+  my $number='-?\d*\.\d*';
+
+  # word: sequence of letters or accents followed by letter
+  my $word_cj='\p{Han}|\p{InHiragana}|\p{InKatakana}';
+  my $word='(?:' . $word_cj . '|(?:(?:[-\w\d*]|\\\\[\"\'\`~^][A-Za-z\*])(?!(?:' . $word_cj . ')))+)';
+
+  # quoted underscore - this needs special treatment as perl treats _ as a letter (\w) but latex does not
+  # such that a\_b would otherwise be interpreted as a{\_}b by latex but a{\_b} by latexdiff
+  my $quotedunderscore='\\\\_';
+  # Handle tex \def macro: \def\MAKRONAME#1[#2]#3{DEFINITION}
+  my $defseq='\\\\def\\\\[\w\d@\*]+(?:#\d+|\[#\d+\])+(?:\{'. $pat_n . '\})?';
+  my $cmdleftright='\\\\(?:left|right|[Bb]igg?[lrm]?|middle)\s*(?:[<>()\[\]|\.]|\\\\(?:[|{}]|\w+))';
+
+  # for selected commands, the number of arguments is known, and we can therefore allow spaces between command and its argument
+  # Note that it is still expected that the arguments are blocks marked by parentheses rather than single characters, and that intervening comments will inhibit the association
+  my $predefinedcmdoptseq12='\\\\(?:href|bibfield|bibinfo)\s*(?:\['.$brat_n.'\])?\s*(?:\{'. $pat_n . '\}\s*){2}';  # Commands with one optional and two non-optional arguments
+  my $predefinedcmdoptseq01='\\\\(?:url|BibitemShut)\s*\s*(?:\{'. $pat_n . '\}\s*){1}';  # Commands with one non-optional argument
+  # \bibitem in revtex styles appears to be always followed by \BibItemOpen. We bind \BibItemOpen to the bibitem (if present) in order to prevent the comparison algorithm to interpret the \BibItemOpen as an identical part of the sequence; this interpretation can lead to added and removed entries to the reference list to become mixed.
+  my $predefinedbibitem='\\\\(?:bibitem)\s*(?:\['.$brat_n.'\])?\s*(?:\{'. $pat_n . '\})(?:%?\s*\\\\BibitemOpen)?';  # Commands with one optional and one non-optional arguments
+  my $predefinedcmdoptseq='(?:'.$predefinedcmdoptseq12.'|'.$predefinedcmdoptseq01.'|'.$predefinedbibitem.')';
+
+  # standard $cmdoptseq (default: no intrevening spaces, controlled by extraspcae) - a final open parentheses is merged to the commend if it exists to deal properly with multi-argument text command
+  my $coords= '[\-.,\s\d]*';
+  my $cmdoptseq='\\\\[\w\d@\*]+'.$extraspace.'(?:(?:<'.$abrat0.'>|\['.$brat_n.'\]|\{'. $pat_n . '\}|\(' . $coords .'\))'.$extraspace.')*\{?';
+
+  # inline math $....$ or \(..\)
+  ### the commented out version is simpler but for some reason cannot cope with newline (in spite of s option) - need to include \newline explicitly
+  ###  my $math='\$(?:[^$]|\\\$)*?\$|\\\\[(].*?\\\\[)]';
+  my $math='\$(?:[^$]|\\\$)*?\$|\\\\[(](?:.|\n)*?\\\\[)]';
+  ### test version (this seems to give the same results as version above)
+  ## the current maths command cannot cope with newline within the math expression
+  ### my $math='\$(?:[^$]|\\\$)*?\$|\\[(].*?\\[)]';
+  ###  my $math='\$(?:[^$]|\\\$)*\$';
+
+  my $backslashnl='\\\\\n';
+  my $oneletcmd='\\\\.\*?(?:\['.$brat_n.'\]|\{'. $pat_n . '\})*';
+  my $comment='%[^\n]*\n';
+  my $punct='[0.,\/\'\`:;\"\?\(\)\[\]!~\p{IsNonAsciiPunct}\p{IsNonAsciiS}]';
+  my $mathpunct='[+=<>\-\|]';
+
+  # Assembled pattern
+  my $pat=qr/(?:\A\s*)?(?:${abbreviation}|${and}|${quotemarks}|${number}|${word}|$quotedunderscore|${defseq}|$cmdleftright|${predefinedcmdoptseq}|${cmdoptseq}|${math}|${backslashnl}|${oneletcmd}|${comment}|${punct}|${mathpunct}|\{|\})\s*/ ;
+
+
+
+
+# now we are done setting up and can start working
+my ($oldfile, $newfile) = @ARGV;
+# check for existence of input files
+if ( ! -e $oldfile ) {
+  die "Input file $oldfile does not exist";
+}
+if ( ! -e $newfile ) {
+  die "Input file $newfile does not exist";
+}
+
+
+# set the labels to be included into the file
+# first find out which file name is longer for correct alignment
+my ($diff,$oldlabel_n_spaces,$newlabel_n_spaces);
+$oldlabel_n_spaces = 0;
+$newlabel_n_spaces = 0;
+$diff = length($newfile) - length($oldfile);
+if ($diff > 0) {
+  $oldlabel_n_spaces = $diff;
+}
+if ($diff < 0) {
+  $newlabel_n_spaces = abs($diff);
+}
+
+my ($oldtime,$newtime,$oldlabel,$newlabel);
+if (defined($labels[0])) {
+  $oldlabel=$labels[0] ;
+} else {
+  $oldtime=localtime((stat($oldfile))[9]);
+  $oldlabel="$oldfile   " . " "x($oldlabel_n_spaces) . $oldtime;
+}
+if (defined($labels[1])) {
+  $newlabel=$labels[1] ;
+} else {
+  $newtime=localtime((stat($newfile))[9]);
+  $newlabel="$newfile   " . " "x($newlabel_n_spaces) . $newtime;
+}
+
+$encoding=guess_encoding($newfile) unless defined($encoding);
+
+$encoding = "utf8" if $encoding =~ m/^utf8/i ;
+print STDERR "Encoding $encoding\n" if $verbose;
+if (lc($encoding) eq "utf8" ) {
+  binmode(STDOUT, ":utf8");
+  binmode(STDERR, ":utf8");
+}
+
+# filter($text)
+# Runs $text through the script provided in $filterscript argument, if set
+# If not set, just returns $text unchanged.
+# If flatten was set, defer filtering to flatten.  flatten will run the filter
+# on all incoming text prior to its own processing.
+# If flatten was not set, filter each of old and new once (see just below this def)
+sub filter {
+  my ($text)=@_;
+  my ($textout,$pid);
+  if ($filterscript ne "") {
+    print STDERR "Passing " . length($text) . " chars to filter script " . $filterscript . "\n" if $verbose;
+
+    if ($ignorefilterstderr) {
+        # If we need to capture and bury STDERR, use the Open3 version, and close CHLD_ERR below.
+        use IPC::Open3;
+        # We consume STDERR from the process, and hide it
+        $pid = open3(\*CHLD_IN, \*CHLD_OUT, \*CHLD_ERR, $filterscript) or die "open3() failed $!";
+    }
+    else {
+        # Capture STDOUT and use as our new $text.  Allow STDERR to go to console.
+        use IPC::Open2;
+        $pid = open2(\*CHLD_OUT, \*CHLD_IN, $filterscript) or die "open2() failed $!";
+    }
+    # Send in $text
+    print CHLD_IN $text."\n";  # Adding a newline just to make sure there is one.
+    close CHLD_IN;
+    # Wait for output and gather it up
+    while (<CHLD_OUT>) {
+      $textout = $textout.$_;
+    }
+    if ($ignorefilterstderr) {
+        close CHLD_ERR;       # Enable only if Open3 used above
+    }
+    # On the off chance a very long running and/or frequently called script is used.
+    waitpid( $pid, 0 );
+    $text = $textout;
+    print STDERR "Received " . length($text) . " chars after filtering\n" if $verbose;
+    print STDERR $text if $verbose;
+  }
+  return $text;
+}
+
+$old=read_file_with_encoding($oldfile,$encoding);
+$new=read_file_with_encoding($newfile,$encoding);
+
+if (not defined($flatten)) {
+    $old=filter($old);
+    $new=filter($new);
+}
+
+
+
+
+# reset time
+exetime(1);
+($oldpreamble,$oldbody,$oldpost)=splitdoc($old,'\\\\begin\{document\}','\\\\end\{document\}');
+
+($newpreamble,$newbody,$newpost)=splitdoc($new,'\\\\begin\{document\}','\\\\end\{document\}');
+
+
+if ($flatten) {
+  $oldbody=flatten($oldbody,$oldpreamble,File::Spec->rel2abs($oldfile),$encoding);
+  $newbody=flatten($newbody,$newpreamble,File::Spec->rel2abs($newfile),$encoding);
+  # flatten preamble
+  $oldpreamble=flatten($oldpreamble,$oldpreamble,File::Spec->rel2abs($oldfile),$encoding);
+  $newpreamble=flatten($newpreamble,$newpreamble,File::Spec->rel2abs($newfile),$encoding);
+
+}
+
+
+
+
+my @auxlines;
+
+# boolean variab
+my ($ulem)=0;
+
+if ( length $oldpreamble && length $newpreamble ) {
+  # pre-process preamble by looking for commands used in \maketitle (title, author, date etc commands)
+  # and marking up content with latexdiff markup
+  @auxlines=preprocess_preamble($oldpreamble,$newpreamble);
+
+  @oldpreamble = split /\n/, $oldpreamble;
+  @newpreamble = split /\n/, $newpreamble;
+
+  # If a command is defined in the preamble of the new file, and only uses safe commands, then it can be considered to be safe) (contribution S. Gouezel)
+  # Base this assessment on the new preamble
+  add_safe_commands($newpreamble);
+
+  # get a list of packages from preamble if not predefined
+  %packages=list_packages($newpreamble) unless %packages;
+  if ( %packages && $debug ) { my $key ; foreach $key (keys %packages) { print STDERR "DEBUG \\usepackage[",$packages{$key},"]{",$key,"}\n" ;} }
+}
+
+# have to return to all processing to properly add preamble additions based on packages found
+if (defined($graphicsmarkup)) {
+  $graphicsmarkup=~tr/a-z/A-Z/;
+  if ( $graphicsmarkup eq 'OFF' or $graphicsmarkup eq 'NONE' ) {
+    $graphicsmarkup=NONE;
+  } elsif ( $graphicsmarkup eq 'NEWONLY' or $graphicsmarkup eq 'NEW-ONLY' ) {
+    $graphicsmarkup=NEWONLY;
+  } elsif ( $graphicsmarkup eq 'BOTH' ) {
+    $graphicsmarkup=BOTH;
+  } elsif ( $graphicsmarkup !~ m/^[012]$/ ) {
+    die "latexdiff Illegal value: ($graphicsmarkup)  for option --highlight-graphics. Possible values: OFF,WHOLE,COARSE,FINE,0-2\n";
+  }
+  # else use numerical value
+} else {
+  # Default: no explicit setting in menu
+  if ( defined $packages{"graphicx"} or defined $packages{"graphics"} ) {
+    $graphicsmarkup=NEWONLY;
+  } else {
+    $graphicsmarkup=NONE;
+  }
+}
+
+if (defined $packages{"hyperref"} ) {
+  # deleted lines should not generate or appear in link names:
+  print STDERR "hyperref package detected.\n" if $verbose ;
+  $latexdiffpreamble =~ s/\{\\DIFadd\}/{\\DIFaddtex}/g;
+  $latexdiffpreamble =~ s/\{\\DIFdel\}/{\\DIFdeltex}/g;
+  $latexdiffpreamble .= join "\n",(extrapream("HYPERREF"),"");
+  if($nolinks){
+    $latexdiffpreamble .= "\n\\hypersetup{bookmarks=false}";
+  }
+  ###    $latexdiffpreamble .= '%DIF PREAMBLE EXTENSION ADDED BY LATEXDIFF FOR HYPERREF PACKAGE' . "\n";
+  ###    $latexdiffpreamble .= '\providecommand{\DIFadd}[1]{\texorpdfstring{\DIFaddtex{#1}}{#1}}' . "\n";
+  ###    $latexdiffpreamble .= '\providecommand{\DIFdel}[1]{\texorpdfstring{\DIFdeltex{#1}}{}}' . "\n";
+  ###    $latexdiffpreamble .= '%DIF END PREAMBLE EXTENSION ADDED BY LATEXDIFF FOR HYPERREF PACKAGE' . "\n";
+}
+
+# add commands for figure highlighting to preamble
+if ($graphicsmarkup != NONE ) {
+  my @matches;
+  # Check if \DIFaddbeginFL definition calls \DIFaddbegin - if so we will issue an error message that graphics highlighting is
+  # is not compatible with this.
+  # (A more elegant solution would be to suppress the redefinitions of the \DIFaddbeginFL etc commands, but for this narrow use case
+  #  I currently don't see this as an efficient use of time)
+  ### The foreach loop does not make sense here. I don't know why I put this in -  (F Tilmann)
+  ###foreach my $cmd ( "DIFaddbegin","DIFaddend","DIFdelbegin","DIFdelend" ) {
+  @matches=( $latexdiffpreamble =~ m/command\{\\DIFaddbeginFL}\{($pat_n)}/sg ) ;
+  # we look at the last one of the list to take into account possible redefinition but almost always matches should have exactly one element
+  if ( $matches[$#matches] =~ m/\\DIFaddbegin/ ) {
+    die "Cannot combine graphics markup with float styles defining \\DIFaddbeginFL in terms of \\DIFaddbegin. Use --graphics-markup=none option or choose a different float style.";
+    exit 10;
+  }
+  ###}
+  $latexdiffpreamble .= join "\n",("\\newcommand{\\DIFscaledelfig}{$SCALEDELGRAPHICS}",extrapream("HIGHLIGHTGRAPHICS"),"");
+
+  # only change required for highlighting both is to declare \includegraphics safe, as preamble already contains commands for deleted environment
+  if ( $graphicsmarkup == BOTH ) {
+    init_regex_arr_list(\@SAFECMDLIST,'includegraphics');
+  }
+}
+
+$ulem = ($latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{ulem\}/ || defined $packages{"ulem"});
+
+# If amsmath is defined and $ulem is used for markup, redefine the \sout command to also work (mostly) in math mode
+# See stack exchange https://tex.stackexchange.com/questions/20609/strikeout-in-math-mode/308647#308647 based on comment by Taylor Raine
+if ( defined($packages{'amsmath'}) and $ulem ) {
+    $latexdiffpreamble .= join "\n",(extrapream('AMSMATHULEM'),"");
+}
+
+# If listings is being used or can be found in the latexdiff search path, add to the preamble auxiliary code to enable line-by-line markup
+if ( defined($packages{"listings"}) or `kpsewhich listings.sty` ne "" ) {
+  my @listingpreamble=extrapream("LISTINGS");
+  if ($latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/ ) {
+    @listingpreamble=extrapream("COLORLISTINGS");
+  }
+  my @listingDIFcode=();
+  my $replaced;
+  # note that in case user supplies preamblefile the type might not reflect well the actual markup style
+  @listingDIFcode=extrapream("-nofail","DIFCODE_" . $type) unless defined($preamblefile);
+  if (!(@listingDIFcode)) {
+    # if listingDIFcode is empty try to guess a suitable one from the preamble
+    if ($latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/ and $ulem ) {
+      @listingDIFcode=extrapream("DIFCODE_UNDERLINE");
+    } elsif ( $latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/ ) {
+      # only colour used
+      @listingDIFcode=extrapream("DIFCODE_CFONT");
+    } else {
+      # fall-back solution
+      @listingDIFcode=extrapream("DIFCODE_BOLD");
+    }
+  }
+  # add configuration so that listings work with utf-8
+  push @listingpreamble, '\lstset{extendedchars=\true,inputencoding='.$encoding."}\n";
+
+
+  # now splice it in
+  $replaced=0;
+  ###print STDERR "DEBUG: listingDIFcode: ",join("\n",@listingDIFcode),"|||\n" if $debug;
+
+  @listingpreamble=grep {
+    # only replace if this has not been done already (use short-circuit property of and)
+    if (!$replaced and $_ =~ s/^.*%DIFCODE TEMPLATE.*$/join("\n",@listingDIFcode)/e ) {
+      ###print STDERR "DEBUG: Replaced text $_\n" if $debug;
+      $replaced=1;
+      1;
+    } else {
+      # return false for those lines matching %DIFCODE TEMPLATE (so that they are not included in output)
+      not m/%DIFCODE TEMPLATE/;
+    }
+  } @listingpreamble;
+  ###  print STDERR "DEBUG: listingpreamble @listingpreamble\n";
+  $latexdiffpreamble .= join "\n",(@listingpreamble,"");
+} else {
+  print STDERR "WARNING: listings package not detected. Disabling mark-up in verbatim environments \n" ;
+  # if listings does not exist disable line-by-line markup and treat all verbatim environments as opaque
+  $VERBATIMENV = liststringtoregex($CONFIG{VERBATIMENV}.";".$CONFIG{VERBATIMLINEENV});
+  $VERBATIMLINEENV = "";
+}
+
+# adding begin and end marker lines to preamble
+$latexdiffpreamble = "%DIF PREAMBLE EXTENSION ADDED BY LATEXDIFF\n" . $ latexdiffpreamble . "%DIF END PREAMBLE EXTENSION ADDED BY LATEXDIFF\n";
+
+# and return to preamble specific processing
+if ( length $oldpreamble && length $newpreamble ) {
+  print STDERR "Differencing preamble.\n" if $verbose;
+
+  # insert dummy first line such that line count begins with line 1 (rather than perl's line 0) - just so that line numbers inserted by linediff are correct
+  unshift @newpreamble,'';
+  unshift @oldpreamble,'';
+  @diffpreamble = linediff(\@oldpreamble, \@newpreamble);
+  # remove dummy line again
+  shift @diffpreamble;
+  # add filenames, modification time and latexdiff mark
+  defined($nolabel) or splice @diffpreamble,1,0,
+      "%DIF LATEXDIFF DIFFERENCE FILE",
+      ,"%DIF DEL $oldlabel",
+      "%DIF ADD $newlabel";
+  if ( @auxlines ) {
+    push @diffpreamble,"%DIF DELETED TITLE COMMANDS FOR MARKUP";
+    push @diffpreamble,join("\n",@auxlines);
+  }
+  if ( $earlylatexdiffpreamble) {
+    # insert latexdiff command directly after documentclass at beginning of preamble
+    # note that grep is only run for its side effect
+    ( grep { s/^([^%]*\\documentclass.*)$/$1$latexdiffpreamble/ } @diffpreamble )==1 or die "Could not find documentclass statement in preamble";
+  } else {
+    # insert latexdiff commands at the end of preamble (default behaviour)
+    push @diffpreamble,$latexdiffpreamble;
+  }
+  push @diffpreamble,'\begin{document}';
+  if (defined $packages{"hyperref"} && $nolinks) {
+    push @diffpreamble, '\begin{NoHyper}';
+  }
+}
+elsif ( !length $oldpreamble && !length $newpreamble ) {
+  @diffpreamble=();
+} else {
+  print STDERR "Either both texts must have preamble or neither text must have the preamble.\n";
+  exit(2);
+}
+
+# Special: treat all cite commands as safe except in UNDERLINE and FONTSTRIKE mode
+# (there is a conflict between citation and ulem package, see
+# package documentation)
+# Use post-processing
+# and $packages{"apacite"}!~/natbibpapa/
+
+
+if (defined $packages{"units"}  && $ulem ) {
+  # protect inlined maths environments by surrounding with an \mbox
+  # this is done to get around an incompatibility between the ulem and units package
+  # where spaces in the argument to underlined or crossed-out \unit commands cause an error message
+  print STDERR "units package detected at the same time as style using ulem.\n" if $verbose ;
+  $MBOXINLINEMATH=1;
+}
+
+if (defined $packages{"siunitx"} ) {
+  # protect SI command by surrounding them with an \mbox
+  # this is done to get around an incompatibility between the ulem and siunitx package
+  print STDERR "siunitx package detected.\n" if $verbose ;
+  my $mboxcmds='SI,ang,numlist,numrange,SIlist,SIrange,qty,qtylist,qtyproduct,qtyrange,complexqty';
+  init_regex_arr_list(\@SAFECMDLIST,'num,si,numproduct,unit,complexnum');
+  if ( $enablecitmark || ( $ulem  && ! $disablecitmark )) {
+    init_regex_arr_list(\@MBOXCMDLIST,$mboxcmds);
+  } else {
+    init_regex_arr_list(\@SAFECMDLIST,$mboxcmds);
+  }
+}
+
+if (defined $packages{"cleveref"} ) {
+  # protect selected command by surrounding them with an \mbox
+  # this is done to get around an incompatibility between ulem and cleveref package
+  print STDERR "cleveref package detected.\n" if $verbose ;
+  my $mboxcmds='[Cc]ref(?:range)?\*?,labelcref,(?:lc)?name[cC]refs?' ;
+  if ( $enablecitmark || ( $ulem  && ! $disablecitmark )) {
+    init_regex_arr_list(\@MBOXCMDLIST,$mboxcmds);
+  } else {
+    init_regex_arr_list(\@SAFECMDLIST,$mboxcmds);
+  }
+}
+
+if (defined $packages{"glossaries"} ) {
+  # protect selected command by surrounding them with an \mbox
+  # this is done to get around an incompatibility between ulem and glossaries package
+  print STDERR "glossaries package detected.\n" if $verbose ;
+  my $mboxcmds='[gG][lL][sS](?:|pl|disp|link|first|firstplural|desc|user[iv][iv]?[iv]?),[aA][cC][rR](?:long|longpl|full|fullpl),[aA][cC][lfp]?[lfp]?';
+  init_regex_arr_list(\@SAFECMDLIST,'[gG][lL][sS](?:(?:entry)?(?:text|plural|name|symbol)|displaynumberlist|entryfirst|entryfirstplural|entrydesc|entrydescplural|entrysymbolplural|entryuser[iv][iv]?[iv]?|entrynumberlist|entrydisplaynumberlist|entrylong|entrylongpl|entryshort|entryshortpl|entryfull|entryfullpl),[gG]lossentry(?:name|desc|symbol),[aA][cC][rR](?:short|shortpl),[aA]csp?');
+  if ( $enablecitmark || ( $ulem  && ! $disablecitmark )) {
+    init_regex_arr_list(\@MBOXCMDLIST,$mboxcmds);
+  } else {
+    init_regex_arr_list(\@SAFECMDLIST,$mboxcmds);
+  }
+}
+
+if (defined $packages{"chemformula"} or defined $packages{"chemmacros"} ) {
+  print STDERR "chemformula package detected.\n" if $verbose ;
+  init_regex_arr_list(\@SAFECMDLIST,'ch');
+  push(@UNSAFEMATHCMD,'ch');
+  # The next command would be needed to allow highlighting the interior of \ch commands in math environments
+  # but the redefinitions in chemformula are too deep to make this viable
+  # push(@MATHTEXTCMDLIST,'ch');
+}
+
+if (defined $packages{"mhchem"} ) {
+  print STDERR "mhchem package detected.\n" if $verbose ;
+  init_regex_arr_list(\@SAFECMDLIST,'ce');
+  push(@UNSAFEMATHCMD,'ce','cee');
+  # The next command would be needed to allow highlighting the interior of \cee commands in math environments
+  # but the redefinitions in chemformula are too deep to make this viable
+  # push(@MATHTEXTCMDLIST,'cee');
+}
+
+if ( defined $packages{"tikz-dependency"} ) {
+  init_regex_arr_ext(\@SAFECMDEXCL, 'AMPERSAND');
+}
+
+
+my ( $citpat);
+
+if ( defined $packages{"apacite"}  ) {
+  print STDERR "apacite package detected.\n" if $verbose ;
+  $citpat='(?:mask)?(?:full|short|no)?cite(?:A|author|year|meta)?(?:NP)?';
+} elsif (defined $packages{"biblatex"}) {
+  print STDERR "biblatex package detected.\n" if $verbose ;
+  $citpat='(?:[cC]ites?|(?:[pP]aren|foot|[Tt]ext|[sS]mart|super)cites?\*?|footnotecitetex)';
+  push(@TEXTCMDEXCL, qr/^textcite$/);
+} else {
+  # citation command pattern for all other citation schemes
+  $citpat='(?:cite\w*|nocite)';
+};
+
+if ( ! $ulem ) {
+  # modes not using ulem: citation is safe
+  push (@SAFECMDLIST, $citpat);
+} else {
+  ### Experimental: disable text and emph commands
+  push(@SAFECMDEXCL, qr/^emph$/, qr/^text..$/);
+  # replace \cite{..} by \mbox{\cite{..}} in added or deleted blocks in post-processing
+  push(@MBOXCMDLIST,$citpat) unless $disablecitmark;
+  if ( uc($subtype) eq "COLOR" or uc($subtype) eq "DVIPSCOL" ) {
+    # remove \cite command again from list of safe commands
+    pop @MBOXCMDLIST;
+    # deleted cite commands
+  }
+}
+push(@MBOXCMDLIST,$citpat) if $enablecitmark ;
+
+
+if (defined $packages{"amsmath"}  or defined $packages{"amsart"} or defined $packages{"amsbook"} ) {
+  print STDERR "amsmath package detected.\n" if $verbose ;
+  $MATHARRREPL='align*';
+}
+
+# add commands in MBOXCMDLIST to SAFECMDLIST
+foreach $mboxcmd ( @MBOXCMDLIST ) {
+  init_regex_arr_list(\@SAFECMDLIST, $mboxcmd);
+}
+
+# check if \label is in SAFECMDLIST, and if yes replace "label" in $LABELCMD by something that never matches (we hope!)
+if ( iscmd("label",\@SAFECMDLIST,\@SAFECMDEXCL) ) {
+  $LABELCMD=~ s/label/NEVERMATCHLABEL/;
+}
+
+
+
+print STDERR "Preprocessing body.  " if $verbose;
+preprocess($oldbody,$newbody);
+writedebugfile($oldbody,'old-preprocess');
+writedebugfile($newbody,'new-preprocess');
+
+# run difference algorithm
+@diffbody=bodydiff($oldbody, $newbody);
+$diffbo=join("",@diffbody);
+writedebugfile($diffbo,"bodydiff");
+
+print STDERR "(",exetime()," s)\n","Postprocessing body. \n" if $verbose;
+postprocess($diffbo);
+$diffall =join("\n",@diffpreamble) ;
+# add visible labels
+if (defined($visiblelabel)) {
+  # Give information right after \begin{document} (or at the beginning of the text for files without preamble
+  ### if \date command is used, add information to \date argument, otherwise give right after \begin{document}
+  ###  $diffall=~s/(\\date$extraspace(?:\[$brat0\])?$extraspace)\{($pat_n)\}/$1\{$2 \\ LATEXDIFF comparison \\ Old: $oldlabel \\ New: $newlabel \}/  or
+  $diffbo = "\\begin{verbatim}LATEXDIFF comparison\nOld: $oldlabel\nNew: $newlabel\\end{verbatim}\n$diffbo"   ;
+}
+
+$diffall .= "$diffbo" ;
+if (defined $packages{"hyperref"} && $nolinks) {
+  $diffall .= "\\end{NoHyper}\n";
+}
+$diffall .= "\\end{document}$newpost" if length $newpreamble ;
+if ( lc($encoding) ne "utf8" && lc($encoding) ne "ascii" ) {
+  print STDERR "Encoding output file to $encoding\n" if $verbose;
+  $diffall=Encode::encode($encoding,$diffall);
+  binmode STDOUT;
+}
+print $diffall;
+
+
+print STDERR "(",exetime()," s)\n","Done.\n" if $verbose;
+
+
+# liststringtoregex(liststring)
+# expands string with semi-colon separated list into a regular expression corresponding
+# matching any of the elements
+sub liststringtoregex {
+  my ($liststring)=@_;
+  my @elements=grep /\S/,split(";",$liststring);
+  if ( @elements) {
+    return('(?:(?:' . join(')|(?:',@elements) .'))');
+  } else {
+    return "";
+  }
+}
+
+# show_configuration
+# note that this is not encapsulated but uses variables from the main program
+# It is provided for convenience because in the future it is planned to allow output
+# to be modified based on what packages are read etc - this works only if the input files are actually read
+# whether or not additional files are provided
+sub show_configuration {
+  if ($showpreamble) {
+    print "\nPreamble commands:\n";
+    print $latexdiffpreamble ;
+  }
+
+  if ($showsafe) {
+    print "\nsafecmd: Commands safe within scope of $ADDOPEN $ADDCLOSE and $DELOPEN $DELCLOSE (unless excluded):\n";
+    print_regex_arr(@SAFECMDLIST);
+    print "\nsafecmd-exlude: Commands not safe within scope of $ADDOPEN $ADDCLOSE and $DELOPEN $DELCLOSE :\n";
+    print_regex_arr(@SAFECMDEXCL);
+    print "\nmboxsafecmd:  Commands safe only if they are surrounded by \\mbox command:\n";
+    print_regex_arr(@MBOXCMDLIST);
+    print "\nnmboxsafecmd: Commands not safe:\n";
+    print_regex_arr(@MBOXCMDEXCL);
+  }
+
+  if ($showtext) {
+    print "\nCommands with last argument textual (unless excluded) and safe in every context:\n";
+    print_regex_arr(@TEXTCMDLIST);
+    print "\nContext1 commands (last argument textual, command will be disabled in deleted passages, last argument will be shown as plain text):\n";
+    print_regex_arr(@CONTEXT1CMDLIST);
+    print "\nContext2 commands (last argument textual, command and its argument will be disabled in deleted passages):\n";
+    print_regex_arr(@CONTEXT2CMDLIST);
+    print "\nExclude list of Commands with last argument not textual (overrides patterns above):\n";
+    print_regex_arr(@TEXTCMDEXCL);
+  }
+
+
+  if ($showconfig) {
+    print "Configuration variables:\n";
+    print "ARRENV=$ARRENV\n";
+    print "COUNTERCMD=$COUNTERCMD\n";
+    print "FLOATENV=$FLOATENV\n";
+    print "ITEMCMD=$ITEMCMD\n";
+    print "LISTENV=$LISTENV\n";
+    print "MATHARRENV=$MATHARRENV\n";
+#    print "MATHARRREPL=$MATHARRREPL\n";     # this is not deprecated and thus no longer shown
+    print "MATHENV=$MATHENV\n";
+    print "MATHREPL=$MATHREPL\n";
+    print "MINWORDSBLOCK=$MINWORDSBLOCK\n";
+    print "PICTUREENV=$PICTUREENV\n";
+    print "SCALEDELGRAPHICS=$SCALEDELGRAPHICS\n";
+    print "VERBATIMENV=$VERBATIMENV\n";
+    print "VERBATIMLINEENV=$VERBATIMLINEENV\n";
+    print "CUSTOMDIFCMD=$CUSTOMDIFCMD\n";
+  }
+}
+
+
+
+## guess_encoding(filename)
+## reads the first 20 lines of filename and looks for call of inputenc package
+## if found, return the option of this package (encoding), otherwise return utf8
+sub guess_encoding {
+  my ($filename)=@_;
+  my ($i,$enc);
+  open (FH, $filename) or die("Couldn't open $filename: $!");
+  $i=0;
+  while (<FH>) {
+    next if /^\s*%/;    # skip comment lines
+    if (m/\\usepackage\[(\w*?)\]\{inputenc\}/) {
+      close(FH);
+      return($1);
+    }
+    last if (++$i > 20 ); # scan at most 20 non-comment lines
+  }
+  close(FH);
+  ### return("ascii");
+  return("utf8");
+}
+
+
+sub read_file_with_encoding {
+  my ($output);
+  my ($filename, $encoding) = @_;
+
+  if (lc($encoding) eq "utf8" ) {
+    open (FILE, "<:utf8",$filename) or die("Couldn't open $filename: $!");
+    local $/ ; # locally set record operator to undefined, ie. enable whole-file mode
+    $output=<FILE>;
+  } elsif ( lc($encoding) eq "ascii") {
+    open (FILE, $filename) or die("Couldn't open $filename: $!");
+    local $/ ; # locally set record operator to undefined, ie. enable whole-file mode
+    $output=<FILE>;
+  } else {
+    require Encode;
+    open (FILE, "<",$filename) or die("Couldn't open $filename: $!");
+    local $/ ; # locally set record operator to undefined, ie. enable whole-file mode
+    $output=<FILE>;
+    print STDERR "Converting $filename from $encoding to utf8\n" if $verbose;
+    $output=Encode::decode($encoding,$output);
+  }
+  close FILE;
+  if ($^O eq "linux" ) {
+    $output =~ s/\r\n/\n/g ;
+  }
+  return $output;
+}
+
+## %packages=list_packages(@preamble)
+## scans the arguments for \documentclass,\RequirePackage and \usepackage statements and constructs a hash
+## whose keys are the included packages, and whose values are the associated optional arguments
+#sub list_packages {
+#  my (@preamble)=@_;
+#  my %packages=();
+#  foreach $line ( @preamble ) {
+#    # get rid of comments
+#    $line=~s/(?<!\\)%.*$// ;
+#    if ( $line =~ m/\\(?:documentclass|usepackage|RequirePackage)(?:\[(.+?)\])?\{(.*?)\}/ ) {
+##      print STDERR "Found something: |$line|\n" if $debug;
+#      if (defined($1)) {
+#	$packages{$2}=$1;
+#      } else {
+#	$packages{$2}="";
+#      }
+#    }
+#  }
+#  return (%packages);
+#}
+
+
+# %packages=list_packages($preamble)
+# scans the arguments for \documentclass,\RequirePackage and \usepackage statements and constructs a hash
+# whose keys are the included packages, and whose values are the associated optional arguments
+# if argument of \usepackage or \RequirePackage is comma separated list, treat as different packages
+sub list_packages {
+  my ($preamble)=@_;
+  my %packages=();
+  my $pkg;
+
+  # remove comments
+  $preamble=~s/(?<!\\)%.*$//mg ;
+
+  while ( $preamble =~  m/\\(?:documentclass|usepackage|RequirePackage)(?:\[($brat_n)\])?\{(.*?)\}/gs ) {
+    if (defined($1)) {
+      foreach $pkg ( split /,/,$2 ) {
+	$packages{$pkg}=$1;
+      }
+    } else {
+      foreach $pkg ( split /,/,$2 ) {
+	$packages{$pkg}="";
+      }
+    }
+  }
+
+  # sometimes, class options are defined in such a way that they imply the loading and/or presence of a package
+  # so we also treat all class options as 'packages.
+  if ( $preamble =~  m/\\documentclass\s*\[($brat_n)\]\s*\{.*?\}/s ) {
+     foreach $pkg ( split /,/,$1 ) {
+       $pkg =~ s/\s//g ;   # remove space and newline characters
+       $packages{$pkg}="" unless exists($packages{$pkg});
+     }
+  }
+  return (%packages);
+}
+
+# Subroutine add_safe_commands modified from version provided by S. Gouezel
+# add_safe_commands($preamble)
+# scans the argument for \newcommand and \DeclareMathOperator,
+# and adds the created commands which are clearly safe to @SAFECMDLIST
+sub add_safe_commands {
+  my ($preamble)=@_;
+
+  # get rid of comments
+  $preamble=~s/(?<!\\)%.*$//mg ;
+
+  my $to_test = "";
+  # test for \DeclareMathOperator{\foo}{myoperator}
+  while ( $preamble =~ m/\DeclareMathOperator\s*\*?\{\\(\w*?)\}/osg) {
+    $to_test=$1;
+    if ($to_test ne "" and not iscmd($to_test,\@SAFECMDLIST,\@SAFECMDEXCL) and not iscmd($to_test, \@SAFECMDEXCL, [])) {
+      # one should add $to_test to the list of safe commands.
+      init_regex_arr_list(\@SAFECMDLIST, $to_test);
+      print STDERR "Adding $to_test to the list of safe commands\n" if $verbose;
+    }
+  }
+
+  while ( $preamble =~ m/\\(?:new|renew|provide)command\s*{\\(\w*)\}(?:|\[\d*\])\s*\{(${pat_n})\}/osg ) {
+    my $maybe_to_test  = $1;
+    my $should_be_safe = $2;
+    print STDERR "DEBUG Checking new command: maybe_to_test, should_be_safe: $1 $2\n" if $debug;
+    my $success = 0;
+    # skip custom diff commands
+    next if ($maybe_to_test =~ m/^(?:ADD|DEL)?${CUSTOMDIFCMD}$/);
+    ###print STDERR "DEBUG: really test it. \n";
+    # test if all latex commands inside it are safe
+    $success = 1;
+    if ($should_be_safe =~ m/\\\\/) {
+      $success = 0;
+    } else {
+      while ($should_be_safe =~ m/\\(\w+)/g) {
+	###	  print STDERR "DEBUG: Testing command $1 " if $debug;
+	$success = 0 unless iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL); ### or $1 eq "";
+	###        print STDERR " success=$success\n" if $debug;
+      }
+    }
+    ###      }
+    if ($success) {
+      $to_test = $maybe_to_test;
+      if (  not iscmd($to_test,\@SAFECMDLIST,\@SAFECMDEXCL) and not iscmd($to_test, \@SAFECMDEXCL, [])) {
+	#        # one should add $to_test to the list of safe commands.
+	init_regex_arr_list(\@SAFECMDLIST, $to_test);
+	print STDERR "Adding $to_test to the list of safe commands\n" if $verbose;
+      }
+    }
+  }
+}
+
+
+# helper function for flatten
+# remove \endinput at beginning of line and everything
+# following it, # if \endinput is not at the beginning of
+# the line, nothing will be removed. It is assumed that
+# this case is most common when \endinput is part of a
+# conditional clause.  The file will only be processed
+# correctly if the conditional is always false,
+# i.e. \endinput # not actually reached
+sub remove_endinput {
+  # s/// operates on default input
+  $_[0] =~ s/^\\endinput.*\Z//ms ;
+  return($_[0]);
+}
+
+# flatten($text,$preamble,$filename,$encoding)
+# expands \input and \include commands within text
+# expands \bibliography command with corresponding bbl file if available
+# expands \subfile command (from subfiles package - not part of standard text distribution)
+# expands \import etc commands (from import package - not part of standard text distribution)
+# preamble is scanned for includeonly commands
+# encoding is the encoding
+sub flatten {
+  my ($text,$preamble,$filename,$encoding)=@_;
+  my ($includeonly,$dirname,$fname,$newpage,$fullfile,$filecontent,$replacement,$begline,$inputcmd,$bblfile,$subfile,$command,$verbenv,$verboptions,$ignore,$fileonly);
+  my ($subpreamble,$subbody,$subpost);
+  my ($subdir,$subdirfull,$importfilepath);
+  require File::Basename ;
+  ###  require File::Spec ;    # now this is needed even if flatten option not given
+  $filename = File::Spec->rel2abs( $filename ) ;
+  ($ignore, $dirname, $fileonly) = File::Spec->splitpath($filename) ;
+  $bblfile = $filename;
+  $bblfile=~s/\.tex$//;
+  $bblfile.=".bbl";
+
+  if ( ($includeonly) = ($preamble =~ m/\\includeonly\{(.*?)\}/ ) ) {
+    $includeonly =~ s/,/|/g;
+  } else {
+    $includeonly = '.*?';
+  }
+
+  print STDERR "DEBUG: includeonly $includeonly\n" if $debug;
+
+
+  # Run through filter, to let filterscript have a pass if it was set
+  $text = filter($text);
+
+  # Recursively replace \\import, \\subimport, and related import commands
+  $text =~ s/(^(?:[^%\n]|\\%)*)(\\(sub)?(?:import|inputfrom|includefrom))\{(.*?)\}(?:[\s]*)\{(.*?)\}/{
+          #  (--------1-------)(--(=3=)-------------2-------------------)  (-4-)             (-5-)
+          # $1 is begline
+          # $2 is the import macro name
+          # $3 is (optional) prefix "sub"
+          # $4 is directory
+          # $5 is filename
+          $begline = (defined($1)? $1 : "");
+          $subdir = $4;
+          $fname = $5;
+          $fname .= ".tex" unless $fname =~ m|\.\w{3,4}$|;
+          print STDERR "DEBUG begline:", $begline, "\n" if $debug;
+          print STDERR "DEBUG", (defined($3)? "subimport_file:" : "import_file:"), $subdir, "\n" if $debug;
+          print STDERR "DEBUG file:", $fname, "\n" if $debug;
+
+          # subimport appends $subdir to the current $dirname.  import replaces it with an absolute path.
+          $subdirfull = (defined($3) ? File::Spec->catdir($dirname,$subdir) : $subdir);
+
+          $importfilepath = File::Spec->catfile($subdirfull, $fname);
+
+          print STDERR "importing importfilepath:", $importfilepath,"\n" if $verbose;
+          if ( -f $importfilepath ) {
+              # If file exists, replace input or include command with expanded input
+              #TODO: need remove_endinput & newpage similar to other replacements inside flatten
+              $replacement=flatten(read_file_with_encoding($importfilepath, $encoding), $preamble,$importfilepath,$encoding);
+          } else {
+              # if file does not exist, do not expand include or input command (do not warn if fname contains #[0-9] as it is then likely part of a command definition
+              # and is not meant to be expanded directly
+              print STDERR "WARNING: Could not find included file ",$importfilepath,". I will continue but not expand |$2|\n";
+              $replacement = $2;
+              $replacement .= "{$subdir}{$fname} % Processed";
+          }
+          "$begline$replacement";
+  }/exgm;
+
+  # recursively replace \\input and \\include files
+  $text =~ s/(^(?:[^%\n]|\\%)*)(\\input\{(.*?)\}|\\include\{(${includeonly}(?:\.tex)?)\})/{
+	    $begline=(defined($1)? $1 : "") ;
+	    $inputcmd=$2;
+	    $fname = $3 if defined($3) ;
+	    $fname = $4 if defined($4) ;
+            $newpage=(defined($4)? " \\newpage " : "") ;
+            #      # add tex extension unless there is a three or four letter extension already
+            $fname .= ".tex" unless $fname =~ m|\.\w{3,4}$|;
+            $fullfile = File::Spec->catfile($dirname,$fname);
+            print STDERR "DEBUG Beg of line match |$1|\n" if defined($1) && $debug ;
+            print STDERR "Include file $fname\n" if $verbose;
+            print STDERR "DEBUG looking for file ",$fullfile, "\n" if $debug;
+            # content of file becomes replacement value (use recursion), add \newpage if the command was include
+            if ( -f $fullfile ) {
+	      # If file exists, replace input or include command with expanded input
+	      $replacement=flatten(read_file_with_encoding($fullfile, $encoding), $preamble,$filename,$encoding);
+	      $replacement = remove_endinput($replacement);
+	      # \include always starts a new page; use explicit \newpage command to simulate this
+	    } else {
+	      # if file does not exist, do not expand include or input command (do not warn if fname contains #[0-9] as it is then likely part of a command definition
+              # and is not meant to be expanded directly
+	      print STDERR "WARNING: Could not find included file ",$fullfile,". I will continue but not expand |$inputcmd|\n" unless $fname =~ m(#[0-9]) ;
+	      $replacement = $inputcmd ;   # i.e. just the original command again -> make no change file does not exist
+	      $newpage="";
+	    }
+	    "$begline$newpage$replacement$newpage";
+          }/exgm;
+
+  # replace bibliography with bbl file if it exists
+  $text=~s/(^(?:[^%\n]|\\%)*)\\bibliography\{(.*?)\}/{
+           if ( -f $bblfile ){
+	     $replacement=read_file_with_encoding(File::Spec->catfile($bblfile), $encoding);
+	   } else {
+	     warn "Bibliography file $bblfile cannot be found. No flattening of \\bibliography done. Run bibtex on old and new files first";
+	     $replacement="\\bibliography{$2}";
+	   }
+	   $begline=(defined($1)? $1 : "") ;
+	   "$begline$replacement";
+  }/exgm;
+
+  # replace subfile with contents (subfile package)
+  $text=~s/(^(?:[^%\n]|\\%)*)\\subfile\{(.*?)\}/{
+           $begline=(defined($1)? $1 : "") ;
+     	   $fname = $2;
+           #      # add tex extension unless there is a three or four letter extension already
+           $fname .= ".tex" unless $fname =~ m|\.\w{3,4}|;
+           print STDERR "Include file as subfile $fname\n" if $verbose;
+           # content of file becomes replacement value (use recursion)
+           # now strip away everything outside and including \begin{document} and \end{document} pair#
+	   #             # note: no checking for comments is made
+           $fullfile=File::Spec->catfile($dirname,$fname);
+           if ( -f $fullfile) {
+	     # if file exists, expand \subfile command by contents of file
+	     $subfile=read_file_with_encoding($fullfile,$encoding) or die "Could not open included subfile ",$fullfile,": $!";
+	     ($subpreamble,$subbody,$subpost)=splitdoc($subfile,'\\\\begin\{document\}','\\\\end\{document\}');
+	     ###           $subfile=~s|^.*\\begin{document}||s;
+	     ###           $subfile=~s|\\end{document}.*$||s;
+	     $replacement=flatten($subbody, $preamble,$fullfile,$encoding);
+	     ### $replacement = remove_endinput($replacement);
+	   } else {
+	      # if file does not exist, do not expand subfile
+	      print STDERR "WARNING: Could not find subfile ",$fullfile,". I will continue but not expand |$2|\n" unless $fname =~ m(#[0-9]) ;
+	      $replacement = "\\subfile\{$2\}" ;   # i.e. just the original command again -> make no change file does not exist
+	    }
+
+	   "$begline$replacement";
+  }/exgm;
+
+  # replace \verbatiminput and \lstlistinginput
+  $text=~s/(^(?:[^%\n]|\\%)*)\\(verbatiminput\*?|lstinputlisting)$extraspace(\[$brat_n\])?$extraspace\{(.*?)\}/{
+     $begline=(defined($1)? $1 : "") ;
+     $command = $2 ;
+     $fname = $4 ;
+     $verboptions = defined($3)? $3 : "" ;
+     if ($command eq 'verbatiminput' ) {
+       $verbenv = "verbatim" ;
+     } elsif ($command eq 'verbatiminput*' ) {
+       $verbenv = "verbatim*" ;
+     } elsif ($command eq 'lstinputlisting' ) {
+       $verbenv = "lstlisting" ;
+     } else {
+       die "Internal errorL Unexpected verbatim input type $command.\n";
+     }
+     print STDERR "DEBUG Beg of line match |$begline|\n" if $debug ;
+     print STDERR "Include file $fname  verbatim\n" if $verbose;
+     print STDERR "DEBUG looking for file ",File::Spec->catfile($dirname,$fname), "\n" if $debug;
+     # content of file becomes replacement value (do not use recursion), add \newpage if the command was include
+     ###$replacement=read_file_with_encoding(File::Spec->catfile($dirname,$fname), $encoding) or die "Couldn't find file ",File::Spec->catfile($dirname,$fname),": $!";
+     $replacement=read_file_with_encoding(File::Spec->catfile($dirname,$fname), $encoding); # (cannot on apparent failure as this triggers for empty fie. Original:  or die "Couldn't find file ",File::Spec->catfile($dirname,$fname),": $!";
+     # Add a new line if it not already there (note that the matching operator needs to use different delimiters, as we are still inside an outer scope that takes precedence
+     $replacement .= "\n" unless  $replacement =~ m(\n$)  ;
+     "$begline\\begin{$verbenv}$verboptions\n$replacement\\end{$verbenv}\n";
+    }/exgm;
+
+  return($text);
+}
+
+
+# print_regex_arr(@arr)
+# prints regex array without x-ism expansion put in by pearl to stdout
+sub print_regex_arr {
+  my $dumstring;
+  $dumstring = join(" ",@_);     # PERL generates string (?-xism:^ref$) for quoted refex ^ref$
+  $dumstring =~ s/\(\?-xism:\^(.*?)\$\)/$1/g;   # remove string and ^,$ marks before output
+  print $dumstring,"\n";
+}
+
+
+# @lines=extrapream($type,...)
+# reads line from appendix or external file
+# (end of file after __END__ token)
+# if $type is a filename, it will read the file instead of reading from the appendix
+# otherwise it will screen appendix for line "%DIF $TYPE" and copy everything up to line
+# '%DIF END $TYPE' (where $TYPE is upcased version of $type)
+# extrapream('-nofail',$type) will---instead of failing---simply return nothing if
+# it does not find the matching line in appendix (do not use -nofail option with multiple types!)
+sub extrapream {
+  my @types=@_;
+  my ($type,$arg);
+  my $nofail=0;
+  ###my @retval=("%DIF PREAMBLE EXTENSION ADDED BY LATEXDIFF") ;
+  my @retval=();
+  my ($copy);
+
+  foreach $arg ( @types ) {
+    if ( $arg eq '-nofail' ) {
+      $nofail=1;
+      next;
+    }
+    $type=$arg;
+    $copy=0;
+    if ( -f $type || lc $type eq '/dev/null' ) {
+      print STDERR "Reading preamble file $type\n" if $verbose ;
+      open (FILE,$type) or die "Cannot open preamble file $type: $!";
+      if (defined($encoding)) {
+	binmode(FILE,":encoding($encoding)");
+      } else {
+	require Encode::Locale;
+	binmode(FILE,":encoding(locale)");
+      }
+      while (<FILE>) {
+	chomp ;
+	if ( $_ =~ m/%DIF PREAMBLE/ ) {
+	  push (@retval,"$_");
+	} else {
+	  push (@retval,"$_ %DIF PREAMBLE");
+	}
+      }
+    } else {    # not (-f $type)
+       $type=uc($type);   # upcase argument
+       print STDERR "Preamble Internal Type $type\n" if $verbose;
+       # save filehandle position (before first read this points to line after __END__)
+       # but seek DATA,0,0 resets it to the beginning of the file
+       # see https://stackoverflow.com/questions/4459601/how-can-i-use-data-twice
+       my $data_start = tell DATA;
+       while (<DATA>) {
+	 if ( m/^%DIF $type/ ) {
+	   $copy=1;
+	 } elsif ( m/^%DIF END $type/ ) {
+	   last;
+	 }
+	 chomp;
+	 push (@retval,"$_ %DIF PREAMBLE") if $copy;
+       }
+       if ( $copy == 0 ) {
+	 unless ($nofail) {
+	   print STDERR "\nPreamble style $type not implemented.\n";
+	   print STDERR "Write latexdiff -h to get help with available styles\n";
+	   exit(2);
+	 }
+       }
+       seek DATA,$data_start,0;    # rewind DATA handle to beginning of data record
+     }
+  }
+  ###push (@retval,"%DIF END PREAMBLE EXTENSION ADDED BY LATEXDIFF")  ;
+  return @retval;
+}
+
+
+# ($part1,$part2,$part3)=splitdoc($text,$word1,$word2)
+# splits $text into 3 parts at $word1 and $word2.
+# if neither $word1 nor $word2 exist, $part1 and $part3 are empty, $part2 is $text
+# If only $word1 or $word2 exist but not the other, output an error message.
+
+# NB this version avoids $` and $' for performance reason although it only makes a tiny difference
+# (in one test gain a tenth of a second for a 30s run)
+sub splitdoc {
+  my ($text,$word1,$word2)=@_;
+  my ($part1,$part2,$part3)=("","","");
+  my ($rest,$pos);
+
+  if ( $text =~ m/(^[^%]*)($word1)/mg ) {
+    $pos=pos $text;
+    $part1=substr($text,0,$pos-length($2));
+    $rest=substr($text,$pos);
+    if ( $rest =~ m/(^[^%]*)($word2)/mg ) {
+      $pos=pos $rest;
+      $part2=substr($rest,0,$pos-length($2));
+      $part3=substr($rest,$pos);
+    }
+    else {
+      die "$word1 and $word2 not in the correct order or not present as a pair." ;
+    }
+  } else {
+    $part2=$text;
+    die "$word2 present but not $word1." if ( $text =~ m/(^[^%]*)$word2/ms );
+  }
+  return ($part1,$part2,$part3);
+}
+
+
+
+
+
+# bodydiff($old,$new)
+sub bodydiff {
+  my ($oldwords, $newwords) = @_;
+  my @retwords;
+
+  print STDERR "(",exetime()," s)\n","Splitting into latex tokens \n" if $verbose;
+  print STDERR "Parsing $oldfile \n" if $verbose;
+  my @oldwords = splitlatex($oldwords);
+  print STDERR "Parsing $newfile \n" if $verbose;
+  my @newwords = splitlatex($newwords);
+
+  if ( $debug ) {
+    open(TOKENOLD,">","latexdiff.debug.tokenold");
+    print TOKENOLD join("***\n",@oldwords);
+    close(TOKENOLD);
+    open(TOKENNEW,">","latexdiff.debug.tokennew");
+    print TOKENNEW join("***\n",@newwords);
+    close(TOKENNEW);
+  }
+
+  print STDERR "(",exetime()," s)\n","Pass 1: Expanding text commands and merging isolated identities with changed blocks  " if $verbose;
+  pass1(\@oldwords, \@newwords);
+
+
+  print STDERR "(",exetime()," s)\n","Pass 2: inserting DIF tokens and mark up.  " if $verbose;
+  if ( $debug ) {
+    open(TOKENOLD,">","latexdiff.debug.tokenold2");
+    print TOKENOLD join("***\n",@oldwords);
+    close(TOKENOLD);
+    open(TOKENNEW,">","latexdiff.debug.tokennew2");
+    print TOKENNEW join("***\n",@newwords);
+    close(TOKENNEW);
+  }
+
+  @retwords=pass2(\@oldwords, \@newwords);
+
+  return(@retwords);
+}
+
+
+
+
+# @words=splitlatex($string)
+# split string according to latex rules
+# Each element of words is either
+# a word (including trailing spaces and punctuation)
+# a latex command
+# if there is white space in the beginning return that as first token
+sub splitlatex {
+  my ($inputstring) = @_ ;
+  my $string=$inputstring ;
+  # if input is empty, return empty list
+  length($string)>0 or return ();
+  $string=~s/^(\s*)//s;
+  my $leadin=$1;
+  length($string)>0 or return ($leadin);
+
+  my @retval=($string =~ m/$pat/osg);
+
+  if (length($string) != length(join("",@retval))) {
+    print STDERR "\nWARNING: Inconsistency in length of input string and parsed string:\n     This often indicates faulty or non-standard latex code.\n     In many cases you can ignore this and the following warning messages.\n Note that character numbers in the following are counted beginning after \\begin{document} and are only approximate." unless $ignorewarnings;
+    print STDERR "DEBUG Original length ",length($string),"  Parsed length ",length(join("",@retval)),"\n" if $debug;
+    print STDERR "DEBUG Input string:  |$string|\n" if (length($string)<500) && $debug;
+    print STDERR "DEBUG Token parsing: |",join("+",@retval),"|\n" if (length($string)<500) && $debug ;
+    @retval=();
+    # slow way only do this if other m//sg method fails
+    my $last = 0;
+    while ( $string =~ m/$pat/osg ) {
+      my $match=$&;
+      if ($last + length $& != pos $string  ) {
+	my $pos=pos($string);
+	my $offset=30<$last ? 30 : $last;
+	my $dum=substr($string,$last-$offset,$pos-$last+2*$offset);
+	my $dum1=$dum;
+	my $cnt=$#retval;
+	my $i;
+	$dum1 =~ s/\n/ /g;
+	unless ($ignorewarnings) {
+	  print STDERR "\n$dum1\n";
+	  print STDERR " " x 30,"^" x ($pos-$last)," " x 30,"\n";
+	  print STDERR "Missing characters near word " . (scalar @retval) . " character index: " . $last . "-" .  pos($string) . " Length: " . length($match) . " Match: |$match| (expected match marked above).\n";
+	}
+	  # put in missing characters `by hand'
+	push (@retval, substr($dum,$offset,$pos-$last-length($match)));
+#       Note: there seems to be a bug in substr with utf8 that made the following line output substr which were too long,
+#             using dum instead appears to work
+#	push (@retval, substr($string,$last, pos($string)-$last-length($match)));
+      }
+      push (@retval, $match);
+      $last=pos $string;
+    }
+
+  }
+
+  unshift(@retval,$leadin) if (length($leadin)>0);
+  return @retval;
+}
+
+
+# pass1( \@seq1,\@seq2)
+# Look for differences between seq1 and seq2.
+# Where an common-subsequence block is flanked by deleted or appended blocks,
+# and is shorter than $MINWORDSBLOCK words it is appended
+# to the last deleted or appended word.  If the block contains tokens other than words
+# or punctuation it is not merged.
+# Deleted or appended block consisting of words and safe commands only are
+# also merged, to prevent break-up in pass2 (after previous isolated words have been removed)
+# If there are commands with textual arguments (e.g. \caption) both in corresponding
+# appended and deleted blocks split them such that the command and opening bracket
+# are one token, then the rest is split up following standard rules, and the closing
+# bracket is a separate token, ie. turn
+# "\caption{This is a textual argument}" into
+# ("\caption{","This ","is ","a ","textual ","argument","}")
+# No return value.  Destructively changes sequences
+sub pass1 {
+  my $seq1 = shift ;
+  my $seq2 = shift ;
+
+  my $len1 = scalar @$seq1;
+  my $len2 = scalar @$seq2;
+  my $wpat=qr/^(?:[a-zA-Z.,'`:;?()!]*)[\s~]*$/;   #'
+
+  my ($last1,$last2)=(-1,-1) ;
+  my $cnt=0;
+  my $block=[];
+  my $addblock=[];
+  my $delblock=[];
+  my $todo=[];
+  my $instruction=[];
+  my $i;
+  my (@delmid,@addmid,@dummy);
+
+  my ($addcmds,$delcmds,$matchindex);
+  my ($addtextblocks,$deltextblocks);
+  my ($addtokcnt,$deltokcnt,$mattokcnt)=(0,0,0);
+  my ($addblkcnt,$delblkcnt,$matblkcnt)=(0,0,0);
+
+  my $adddiscard = sub {
+                      if ($cnt > 0 ) {
+			$matblkcnt++;
+			# just after an unchanged block
+#			print STDERR "Unchanged block $cnt, $last1,$last2 \n";
+                        if ($cnt < $MINWORDSBLOCK
+			    && $cnt==scalar (
+				     grep { /^$wpat/ || ( /^\\((?:[`'^"~=.]|[\w\d@*]+))((?:\[$brat_n\]|\{$pat_n\})*)/o
+							   && iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL)
+							   && scalar(@dummy=split(" ",$2))<3 ) }
+					     @$block) )  {
+			  # merge identical blocks shorter than $MINWORDSBLOCK
+			  # and only containing ordinary words
+			  # with preceding different word
+			  # We cannot carry out this merging immediately as this
+			  # would change the index numbers of seq1 and seq2 and confuse
+			  # the algorithm, instead we store in @$todo where we have to merge
+                          push(@$todo, [ $last1,$last2,$cnt,@$block ]);
+			}
+			$block = [];
+			$cnt=0; $last1=-1; $last2=-1;
+		      }
+		    };
+  my $discard=sub { $deltokcnt++;
+                    &$adddiscard; #($_[0],$_[1]);
+		    push(@$delblock,[ $seq1->[$_[0]],$_[0] ]);
+		    $last1=$_[0] };
+
+  my $add =   sub { $addtokcnt++;
+                    &$adddiscard; #($_[0],$_[1]);
+		    push(@$addblock,[ $seq2->[$_[1]],$_[1] ]);
+		    $last2=$_[1] };
+
+  my $match = sub { $mattokcnt++;
+                    if ($cnt==0) {   # first word of matching sequence after changed sequence or at beginning of word sequence
+		      $deltextblocks = extracttextblocks($delblock);
+		      $delblkcnt++ if scalar @$delblock;
+		      $addtextblocks = extracttextblocks($addblock);
+		      $addblkcnt++ if scalar @$addblock;
+
+		      # make a list of all TEXTCMDLIST commands in deleted and added blocks
+		      $delcmds = extractcommands($delblock);
+      		      $addcmds = extractcommands($addblock);
+		      # now find those text commands, which are found in both deleted and added blocks, and expand them
+		      # keygen(third argument of _longestCommonSubsequence) implies to sort on command (0th elements of $addcmd elements)
+		      # the calling format for longestCommonSubsequence has changed between versions of
+		      # Algorithm::Diff so we need to check which one we are using
+		      if ( $algodiffversion  > 1.15 ) {
+			### Algorithm::Diff 1.19
+			$matchindex=Algorithm::Diff::_longestCommonSubsequence($delcmds,$addcmds, 0, sub { $_[0]->[0] } );
+		      } else {
+			### Algorithm::Diff 1.15
+			$matchindex=Algorithm::Diff::_longestCommonSubsequence($delcmds,$addcmds, sub { $_[0]->[0] } );
+		      }
+
+		      for ($i=0 ; $i<=$#$matchindex ; $i++) {
+			if (defined($matchindex->[$i])){
+			  $j=$matchindex->[$i];
+			  @delmid=splitlatex($delcmds->[$i][3]);
+			  @addmid=splitlatex($addcmds->[$j][3]);
+			  while (scalar(@$deltextblocks)  && $deltextblocks->[0][0]<$delcmds->[$i][1]) {
+			    my ($index,$block,$cnt)=@{ shift(@$deltextblocks) };
+			    push(@$todo, [$index,-1,$cnt,@$block]);
+			  }
+			  push(@$todo, [ $delcmds->[$i][1],-1,-1,$delcmds->[$i][2],@delmid,$delcmds->[$i][4]]);
+
+			  while (scalar(@$addtextblocks) && $addtextblocks->[0][0]<$addcmds->[$j][1]) {
+			    my ($index,$block,$cnt)=@{ shift(@$addtextblocks) };
+			    push(@$todo, [-1,$index,$cnt,@$block]);
+			  }
+			  push(@$todo, [ -1,$addcmds->[$j][1],-1,$addcmds->[$j][2],@addmid,$addcmds->[$j][4]]);
+			}
+		      }
+		      # mop up remaining textblocks
+		      while (scalar(@$deltextblocks)) {
+			my ($index,$block,$cnt)=@{ shift(@$deltextblocks) } ;
+			push(@$todo, [$index,-1,$cnt,@$block]);
+		      }
+		      while (scalar(@$addtextblocks)) {
+			my ($index,$block,$cnt)=@{ shift(@$addtextblocks) };
+			push(@$todo, [-1,$index,$cnt,@$block]);
+		      }
+
+		      $addblock=[];
+		      $delblock=[];
+		    }
+		    push(@$block,$seq2->[$_[1]]);
+		    $cnt++  };
+
+  my $keyfunc = sub { join("  ",split(" ",shift())) };
+
+  traverse_sequences($seq1,$seq2, { MATCH=>$match, DISCARD_A=>$discard, DISCARD_B=>$add }, $keyfunc );
+
+
+  # now carry out the merging/splitting.  Refer to elements relative from
+  # the end (with negative indices) as these offsets don't change before the instruction is executed
+  # cnt>0: merged small unchanged groups with previous changed blocks
+  # cnt==-1: split textual commands into components
+  foreach $instruction ( @$todo) {
+    ($last1,$last2,$cnt,@$block)=@$instruction ;
+    if ($cnt>=0) {
+      splice(@$seq1,$last1-$len1,1+$cnt,join("",$seq1->[$last1-$len1],@$block)) if $last1>=0;
+      splice(@$seq2,$last2-$len2,1+$cnt,join("",$seq2->[$last2-$len2],@$block)) if $last2>=0;
+    } else {
+      splice(@$seq1,$last1-$len1,1,@$block) if $last1>=0;
+      splice(@$seq2,$last2-$len2,1,@$block) if $last2>=0;
+    }
+  }
+
+  if ($verbose) {
+    print STDERR "\n";
+    print STDERR "  $mattokcnt matching  tokens in $matblkcnt blocks.\n";
+    print STDERR "  $deltokcnt discarded tokens in $delblkcnt blocks.\n";
+    print STDERR "  $addtokcnt appended  tokens in $addblkcnt blocks.\n";
+  }
+}
+
+
+# extracttextblocks(\@blockindex)
+# $blockindex has the following format
+# [ [ token1, index1 ], [token2, index2],.. ]
+# where index refers to the index in the original old or new word sequence
+# Returns: reference to an array of the form
+# [[ $index, $textblock, $cnt ], ..
+# where $index index of block to be merged
+#       $textblock contains all the words to be merged with the word at $index (but does not contain this word)
+#       $cnt   is length of block
+#
+# requires: iscmd
+#
+sub extracttextblocks {
+  my $block=shift;
+  my ($i,$token,$index);
+  my $textblock=[];
+  my $last=-1;
+  my $wpat=qr/^(?:[a-zA-Z.,'`:;?()!]*)[\s~]*$/;  #'
+  my $retval=[];
+
+  # we redefine locally $extraspace (shadowing the global definition) to capture command sequences with intervening spaces no matter what the global setting
+  # this is done so we can capture those commands with a predefined number of arguments without having to introduce them again explicitly here
+  my $extraspace='\s*';
+
+  for ($i=0;$i< scalar @$block;$i++) {
+    ($token,$index)=@{ $block->[$i] };
+    # store pure text blocks
+    if ($token =~ /$wpat/ ||  ( $token =~/^\\((?:[`'^"~=.]|[\w\d@\*]+))((?:${extraspace}\[$brat_n\]${extraspace}|${extraspace}\{$pat_n\})*)/
+				&& iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL)
+				&& !iscmd($1,\@TEXTCMDLIST,\@TEXTCMDEXCL))) {
+      # we have text or a command which can be treated as text
+      if ($last<0) {
+	# new pure-text block
+	$last=$index;
+      } else {
+	# add to pure-text block
+	push(@$textblock, $token);
+      }
+    } else {
+      # it is not text
+      if (scalar(@$textblock)) {
+	push(@$retval,[ $last, $textblock, scalar(@$textblock) ]);
+      }
+      $textblock=[];
+      $last=-1;
+    }
+  }
+  # finish processing a possibly unfinished block before returning
+  if (scalar(@$textblock)) {
+    push(@$retval,[ $last, $textblock, scalar(@$textblock) ]);
+  }
+  return($retval)
+}
+
+
+
+# extractcommands( \@blockindex )
+# $blockindex has the following format
+# [ [ token1, index1 ], [token2, index2],.. ]
+# where index refers to the index in the original old or new word sequence
+# Returns: reference to an array of the form
+# [ [ "\cmd1", index, "\cmd1[optarg]{arg1}{", "arg2" ,"} " ],..
+# where index is just taken from input array
+# command must have a textual argument as last argument
+#
+# requires: iscmd
+#
+sub extractcommands {
+  my $block=shift;
+  my ($i,$token,$index,$cmd,$open,$mid,$closing);
+  my $retval=[];
+
+  # we redefine locally $extraspace (shadowing the global definition) to capture command sequences with intervening spaces no matter what the global setting
+  # this is done so we can capture those commands with a predefined number of arguments without having to introduce them again explicitly here
+  my $extraspace='\s*';
+
+  for ($i=0;$i< scalar @$block;$i++) {
+    ($token,$index)=@{ $block->[$i] };
+    # check if token is an alphanumeric command sequence with at least one non-optional argument
+    # \cmd[...]{...}{last argument}
+    # Capturing in the following results in these associations
+    # $1: \cmd[...]{...}{
+    # $2: \cmd
+    # $3: last argument
+    # $4: }  + trailing spaces
+    if ( ( $token =~ m/^(\\([\w\d\*]+)(?:${extraspace}\[$brat_n\]|${extraspace}\{$pat_n\})*${extraspace}\{)($pat_n)(\}\s*)$/so )
+	 && iscmd($2,\@TEXTCMDLIST,\@TEXTCMDEXCL) ) {
+      print STDERR "DEBUG EXTRACTCOMMANDS Match |$1|$2|$3|$4|$index \n" if $debug;
+      #      push(@$retval,[ $2,$index,$1,$3,$4 ]);
+      ($cmd,$open,$mid,$closing) = ($2,$1,$3,$4) ;
+      $closing =~ s/\}/\\RIGHTBRACE/ ;
+      push(@$retval,[ $cmd,$index,$open,$mid,$closing ]);
+    }
+  }
+  return $retval;
+}
+
+# iscmd($cmd,\@regexarray,\@regexexcl) checks
+# return 1 if $cmd matches any of the patterns in the
+# array $@regexarray, and none of the patterns in \@regexexcl, otherwise return 0
+sub iscmd {
+  my ($cmd,$regexar,$regexexcl)=@_;
+  my ($ret)=0;
+  ### print STDERR "DEBUG: iscmd($cmd)=" if $debug;
+  foreach $pat ( @$regexar ) {
+    if ( $cmd =~ m/^${pat}$/ ) {
+      $ret=1 ;
+      last;
+    }
+  }
+  ### print STDERR "0\n" if ($debug && !$ret) ;
+  return 0 unless $ret;
+  foreach $pat ( @$regexexcl ) {
+    ### print STDERR "0\n" if ( $debug && $cmd =~ m/^${pat}$/) ;
+    return 0 if ( $cmd =~ m/^${pat}$/ );
+  }
+  ### print STDERR "1\n" if $debug;
+  return 1;
+}
+
+
+# pass2( \@seq1,\@seq2)
+# Look for differences between seq1 and seq2.
+# Mark begin and end of deleted and appended sequences with tags $DELOPEN and $DELCLOSE
+# and $ADDOPEN and $ADDCLOSE, respectively, however exclude { } & and all comands, unless
+# they match an element of the whitelist (SAFECMD)
+# For words in TEXTCMD but not in SAFECMD, enclose interior with $ADDOPEN and $ADDCLOSE brackets
+# Deleted comment lines are marked with %DIF <
+# Added comment lines are marked with %DIF >
+sub pass2 {
+  my $seq1 = shift ;
+  my $seq2 = shift ;
+
+  my ($addtokcnt,$deltokcnt,$mattokcnt)=(0,0,0);
+  my ($addblkcnt,$delblkcnt,$matblkcnt)=(0,0,0);
+
+  my $retval = [];
+  my $delhunk   = [];
+  my $addhunk   = [];
+
+  my $discard = sub { $deltokcnt++;
+                      push ( @$delhunk, $seq1->[$_[0]]) };
+
+  my $add = sub { $addtokcnt++;
+                  push ( @$addhunk, $seq2->[$_[1]]) };
+
+  my $match = sub { $mattokcnt++;
+		    if ( scalar @$delhunk ) {
+                      $delblkcnt++;
+		      # mark up changes, but comment out commands
+                      push @$retval,marktags($DELMARKOPEN,$DELMARKCLOSE,$DELOPEN,$DELCLOSE,$DELCMDOPEN,$DELCMDCLOSE,$DELCOMMENT,$delhunk);
+		      $delhunk = [];
+		    }
+                    if ( scalar @$addhunk ) {
+                      $addblkcnt++;
+                      # we mark up changes, but simply quote commands
+                      push @$retval,marktags($ADDMARKOPEN,$ADDMARKCLOSE,$ADDOPEN,$ADDCLOSE,"","",$ADDCOMMENT,$addhunk);
+		      $addhunk = [];
+		    }
+		    push(@$retval,$seq2->[$_[1]]) };
+
+  my $keyfunc = sub { join("  ",split(" ",shift())) };
+
+  traverse_sequences($seq1,$seq2, { MATCH=>$match, DISCARD_A=>$discard, DISCARD_B=>$add }, $keyfunc );
+  # clear up unprocessed hunks
+  push @$retval,marktags($DELMARKOPEN,$DELMARKCLOSE,$DELOPEN,$DELCLOSE,$DELCMDOPEN,$DELCMDCLOSE,$DELCOMMENT,$delhunk) if scalar @$delhunk;
+  push @$retval,marktags($ADDMARKOPEN,$ADDMARKCLOSE,$ADDOPEN,$ADDCLOSE,"","",$ADDCOMMENT,$addhunk) if scalar @$addhunk;
+
+
+  if ($verbose) {
+    print STDERR "\n";
+    print STDERR "  $mattokcnt matching  tokens. \n";
+    print STDERR "  $deltokcnt discarded tokens in $delblkcnt blocks.\n";
+    print STDERR "  $addtokcnt appended  tokens in $addblkcnt blocks.\n";
+  }
+  return(@$retval);
+}
+
+# marktags($openmark,$closemark,$open,$close,$opencmd,$closecmd,$comment,\@block)
+# returns ($openmark,$open,$block,$close,$closemark) if @block contains no commands (except white-listed ones),
+# braces, ampersands, or comments
+# mark comments with $comment
+# exclude all other exceptions from scope of open, close like this
+# ($openmark, $open,...,$close, $opencmd,command, command,$closecmd, $open, ..., $close, $closemark)
+# If $opencmd begins with "%" marktags assumes it is operating on a deleted block, otherwise on an added block
+sub marktags {
+  my ($openmark,$closemark,$open,$close,$opencmd,$closecmd,$comment,$block)=@_;
+  my $word;
+  my (@argtext);
+  my $retval=[];
+  my $noncomment=0;
+  my $cmd=-1;    # -1 at beginning 0: last token written is a ordinary word
+                 # 1: last token written is a command
+                # for keeping track whether we are just in a command sequence or in a word sequence
+  my $cmdcomment= ($opencmd =~ m/^%/);  # Flag to indicate whether opencmd is a comment (i.e. if we intend to simply comment out changed commands)
+  my ($command,$commandword,$closingbracket) ; # temporary variables needed below to remember sub-pattern matches
+
+
+
+# split this block to split sequences joined in pass1
+  ### print STDERR "DEBUG: marktags before splitlatex blocksplit ",join("|",@$block),"\n" if $debug;
+  @$block=splitlatex(join "",@$block);
+  ### print STDERR "DEBUG: marktags $openmark,$closemark,$open,$close,$opencmd,$closecmd,$comment\n" if $debug;
+  print STDERR "DEBUG: after splitlatex ",join("|",@$block),"\n" if $debug;
+
+  # we redefine locally $extraspace (shadowing the global definition) to capture command sequences with intervening spaces no matter what the global setting
+  # this is done so we can capture those commands with a predefined number of arguments without having to introduce them again explicitly here
+  my $extraspace_mt='\s*';
+
+
+  foreach (@$block) {
+    $word=$_;
+    if ( $word =~ s/^%/%$comment/ ) {
+      # a comment
+      if ($cmd==1) {
+	push (@$retval,$closecmd) ;
+	$cmd=-1;
+      }
+      push (@$retval,$word);
+      next;
+    }
+    if ( $word =~ m/^\s*$/ ) {
+      ### print STDERR "DEBUG MARKTAGS: whitespace detected |$word| cmdcom |$cmdcomment| |$opencmd|\n" if $debug;
+      # a sequence of white-space characters - this should only ever happen for the first element of block.
+      # in deleted block, omit, otherwise just copy it in
+      if ( ! $cmdcomment) {   # ignore in deleted blocks
+	push(@$retval,$word);
+      }
+      next;
+    }
+    if (! $noncomment) {
+      push (@$retval,$openmark);
+      $noncomment=1;
+    }
+    # negative lookahead pattern (?!) in second clause is put in to avoid matching \( .. \) patterns
+    # also note that second pattern will match \\
+    if (  $word =~ /^[&{}\[\]]/ || ( $word =~ /^\\(?!\()(\\|[`'^"~=.]|[\w*@]+)/ &&  !iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL)) ) {
+      ###    if ( $word =~ /^[&{}\[\]]/ || ( $word =~ /^\\([\w*@\\% ]+)/ && !iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL)) ) {
+      # word is a command or other significant token (not in SAFECMDLIST)
+	## same conditions as in subroutine extractcommand:
+	# check if token is an alphanumeric command sequence with at least one non-optional argument
+	# \cmd[...]{...}{last argument}
+	# Capturing in the following results in these associations
+	# $1: \cmd[...]{...}{
+	# $2: cmd
+	# $3: last argument
+	# $4: }  + trailing spaces
+	### pre-0.3    if ( ( $token =~ m/^(\\([\w\d\*]+)(?:\[$brat0\]|\{$pat_n\})*\{)($pat_n)(\}\s*)$/so )
+      if ( ( $word =~ m/^(\\([\w\d\*]+)(?:${extraspace_mt}\[$brat_n\]|${extraspace_mt}\{$pat_n\})*${extraspace_mt}\{)($pat_n)(\}\s*)$/so )
+	   && (iscmd($2,\@TEXTCMDLIST,\@TEXTCMDEXCL)|| iscmd($2,\@MATHTEXTCMDLIST,\@MATHTEXTCMDEXCL))
+           && ( !$cmdcomment || !iscmd($2,\@CONTEXT2CMDLIST, \@CONTEXT2CMDEXCL) )  ) {
+	# Condition 1: word is a command? - if yes, $1,$2,.. will be set as above
+        # Condition 2: word is a text command - we mark up the interior of the word. There is a separate check for MATHTEXTCMDLIST
+        #              because for $mathmarkup=WHOLE, the commands should not be split in pass1 (ie. math mode commands are not in
+        #              TEXTCMDLIST, but the interior of MATHTEXT commnds should be highlighted in both deleted and added blocks
+        # Condition 3: But if we are in a deleted block ($cmdcomment=1) and
+        #            $2 (the command) is in context2, just treat it as an ordinary command (i.e. comment it open with $opencmd)
+        # Because we do not want to disable this command
+	# here we do not use $opencmd and $closecmd($opencmd is empty)
+	print STDERR "DEBUG: Detected text |$word| but not safe command \$2: $2 \$3: $3\n." if $debug;
+	if ($cmd==1) {
+	  push (@$retval,$closecmd) ;
+	} elsif ($cmd==0) {
+	  push (@$retval,$close) ;
+	}
+        $command=$1; $commandword=$2; $closingbracket=$4;
+	@argtext=splitlatex($3);   # split textual argument into tokens
+	# and mark it up (but we do not need openmark and closemark)
+        # insert command with initial arguments, marked-up final argument, and closing bracket
+	if ( $cmdcomment && iscmd($commandword,\@CONTEXT1CMDLIST, \@CONTEXT1CMDEXCL) ) {
+	  # context1cmd in a deleted environment; delete command itself but keep last argument, marked up
+	  push (@$retval,$opencmd);
+	  $command =~ s/\n/\n${opencmd}/sg ; # repeat opencmd at the beginning of each line
+	  # argument, note that the additional comment character is included
+          # to suppress linebreak after opening parentheses, which is important
+          # for latexrevise
+          push (@$retval,$command,"%\n{$AUXCMD\n",marktags("","",$open,$close,$opencmd,$closecmd,$comment,\@argtext),$closingbracket);
+        } elsif ( iscmd($commandword,,\@MATHTEXTCMDLIST, \@MATHTEXTCMDEXCL) ) {
+	  # MATHBLOCK pseudo command: consider all commands safe, except & and \\, \begin and \end and a few package sprcific one (look at UNSAFEMATHCMD definition)
+	  # Keep these commands even in deleted blocks, hence set $opencmd and $closecmd (5th and 6th argument of marktags) to
+	  # ""
+	  local @SAFECMDLIST=(".*");
+	  local @SAFECMDEXCL=('\\','\\\\',@UNSAFEMATHCMD);
+	  push(@$retval,$command,marktags("","",$open,$close,"","",$comment,\@argtext)#@argtext
+                       ,$closingbracket);
+        } else {
+	  # normal textcmd or context1cmd in an added block
+	  push (@$retval,$command,marktags("","",$open,$close,$opencmd,$closecmd,$comment,\@argtext),$closingbracket);
+	}
+	push (@$retval,$AUXCMD,"\n") if $cmdcomment ;
+	$cmd=-1 ;
+      } elsif ( $cmdcomment &&
+	       ( $word =~ m/^(\\([\w\d\*]+)(?:${extraspace_mt}\[$brat_n\]|${extraspace_mt}\{$pat_n\})*${extraspace_mt}\{)($pat_n)(\}\s*)/so )
+	       && iscmd($2,\@KEEPCMDLIST, \@KEEPCMDEXCL) ) {
+	# 'keepcmd' in a deleted environment: keep  the command as is
+	push (@$retval,$close) if $cmd==0 ;
+	push (@$retval,$word);
+	$cmd=-1;  # pretend we are at the beginning of a sequence because we do not want to add an additional $closecmd or $close before the next token, no matter what it is
+      } else {
+	# ordinary command
+	push (@$retval,$opencmd) if $cmd==-1 ;
+	push (@$retval,$close,$opencmd) if $cmd==0 ;
+	$word =~ s/\n/\n${opencmd}/sg if $cmdcomment ;   # if opencmd is a comment, repeat this at the beginning of every line
+        ### print STDERR "MARKTAGS: Add command |$word|\n";
+	push (@$retval,$word);
+	$cmd=1;
+      }
+    } else {
+      ###print STDERR "DEBUG MARKTAGS is an ordinary word or SAFECMD command \n" if $debug;
+      # just an ordinary word or command in SAFECMD
+      push (@$retval,$open) if $cmd==-1 ;
+      push (@$retval,$closecmd,$open) if $cmd==1 ;
+      ###TODO:  check here if it is a command in MBOXCMD list, and surround it with \mbox{...}
+      ### $word =~ /^\\(?!\()(\\|[`'^"~=.]|[\w*@]+)/ &&  iscmd($1,\@MBOXCMDLIST,\@MBOXCMDEXCL))
+      ### but actually this check has been carried out already so can simply check if word begins with backslash
+      if ( $word =~ /^\\(?!\()(\\|[`'^"~=.]|[\w*@]+)(.*?)(\s*)$/s &&  iscmd($1,\@MBOXCMDLIST,\@MBOXCMDEXCL)) {
+	# $word is a safe command in MBOXCMDLIST
+	###print STDERR "DEBUG Mboxsafecmd detected:$word:\n" if $debug ;
+	push(@$retval,"\\mbox{$AUXCMD\n\\" . $1 . $2 . $3 ."}\\hskip0pt$AUXCMD\n" );
+      } else {
+	# $word is a normal word or a safe command (not in MBOXCMDLIST)
+	push (@$retval,$word);
+      }
+      $cmd=0;
+    }
+  }
+  push (@$retval,$close) if $cmd==0;
+  push (@$retval,$closecmd) if $cmd==1;
+
+  push (@$retval,$closemark) if ($noncomment);
+  return @$retval;
+}
+
+#used in preprocess
+sub take_comments_and_newline_from_frac() {
+  # some special magic for common usage of frac, which does not conform to the latexdiff requirements but can be made to fit
+  # note that this is a rare exception to the general rule that the new tex can be reconstructed from the diff file
+
+  # regex that matches space and comment characters
+  my $space = qr/\s|%[^\n]*?/;
+  # \frac {abc} -> \frac{abc}
+  # \frac1 -> \frac{1}
+  # \frac a -> \frac{a}
+  # \frac \lambda -> \frac{\lambda}
+  s/\\frac(?|${space}+\{($pat_n)\}|${space}*(\d)|${space}+(\w)|${space}*(\\[a-zA-Z]+))/\\frac\{$1\}/g;
+  # same as above for the second argument of frac
+  s/\\frac(\{$pat_n\})(?|${space}*\{($pat_n)\}|${space}*(\d)|${space}+(\w)|${space}*(\\[a-zA-Z]+))/\\frac$1\{$2\}/g;
+}
+
+# preprocess($string, ..)
+# carry out the following pre-processing steps for all arguments:
+# 1. Remove leading white-space
+#    Change \{ to \QLEFTBRACE and \} to \QRIGHTBRACE and \& to \AMPERSAND
+# #.   Change {,},\frac in comments to \CLEFTBRACE, \CRIGHTBRACE, \CFRAC
+# 2. mark all first empty line (in block of several) with \PAR tokens
+# 3. Convert all '\%' into '\PERCENTAGE ' and all '\$' into \DOLLAR to make parsing regular expressions easier
+# 4. Convert all \verb|some verbatim text| commands (where | can be an arbitrary character)
+#    into \verb{hash}  (also lstinline)
+# 5. Convert \begin{verbatim} some verbatim text \end{verbatim} into \verbatim{hash}  (not only verbatim, all patterns matching VERBATIMENV)
+# 6. Convert _n into \SUBSCRIPTNB{n} and _{nnn} into \SUBSCRIPT{nn}
+# 7. Convert ^n into \SUPERSCRIPTNB{n} and ^{nnn} into \SUPERSCRIPT{nn}
+# 8. a. Convert $$ $$ into \begin{DOLLARDOLLAR} \end{DOLLARDOLLAR}
+#    b. Convert \[ \] into \begin{SQUAREBRACKET} \end{SQUAREBRACKET}
+# 9. Convert all picture environmentent (\begin{PICTUREENV} .. \end{PICTUREENV} \PICTUREBLOCKenv
+#     For math-mode COARSE,WHOLE or NONE option -convert all \begin{MATH} .. \end{MATH}
+#    into \MATHBLOCKmath{...} commands, where MATH/math is any valid math environment
+
+# 10. Add final token STOP to the very end.  This is put in because the algorithm works better if the last token is identical.  This is removed again in postprocessing.
+#
+# NB: step 6 and 7 is likely to  convert some "_" inappropriately, e.g. in file
+#     names or labels but it does not matter because they are converted back in the postprocessing step
+# Returns: leading white space removed in step 1
+sub preprocess {
+  for (@_) {
+
+
+    # change in \verb and similar commands - note that I introduce an extra space here so that the
+    #       already hashed variants do not trigger again
+    # transform \lstinline{...}
+#    s/\\lstinline(\[$brat0\])?(\{(?:.*?)\})/"\\DIFlstinline". $1 ."{". tohash(\%verbhash,"$2") ."}"/esg;
+#    s/\\lstinline(\[$brat0\])?((\S).*?\2)/"\\DIFlstinline". $1 ."{". tohash(\%verbhash,"$2") ."}"/esg;
+    s/\\lstinline((?:\[$brat_n\])?)(\{(?:.*?)\})/"\\DIFlstinline". $1 ."{". tohash(\%verbhash,"$2") ."}"/esg;
+    s/\\lstinline((?:\[$brat_n\])?)(([^\s\w]).*?\3)/"\\DIFlstinline". $1 ."{". tohash(\%verbhash,"$2") ."}"/esg;
+    s/\\(verb\*?|lstinline)([^\s\w])(.*?)\2/"\\DIF${1}{". tohash(\%verbhash,"${2}${3}${2}") ."}"/esg;
+
+    #    Change \{ to \QLEFTBRACE, \} to \QRIGHTBRACE, and \& to \AMPERSAND
+    s/(?<!\\)\\\{/\\QLEFTBRACE /sg;
+    s/(?<!\\)\\\}/\\QRIGHTBRACE /sg;
+    s/(?<!\\)\\&/\\AMPERSAND /sg;
+# replace {,}, \frac in comments with \\CLEFTBRACE,\\CRIGHTBRACE, CFRAC to protect from special treatment
+    1 while s/((?<!\\)%.*)\{(.*)$/$1\\CLEFTBRACE $2/mg ;
+    1 while s/((?<!\\)%.*)\}(.*)$/$1\\CRIGHTBRACE $2/mg ;
+    1 while s/((?<!\\)%.*)\\frac(.*)$/$1\\CFRAC $2/mg ;
+
+    s/(?<!\\)\\%/\\PERCENTAGE /g ;  # (?<! is negative lookbehind assertion to prevent \\% from being converted
+    s/(?<!\\)\\\$/\\DOLLAR /g ;  # (?<! is negative lookbehind assertion to prevent \\$ from being converted
+    s/\\begin\{($VERBATIMENV)\}(.*?)\\end\{\1\}/"\\${1}{". tohash(\%verbhash,"${2}") . "}"/esg;
+    s/\\begin\{($VERBATIMLINEENV)\}(.*?)\\end\{\1\}/"\\begin{$1}". linecomment($2) . "\\end{$1}"/esg;
+
+    # mark all first empty line (in block of several) with \PAR tokens
+    s/\n(\s*?)\n((?:\s*\n)*)/\n$1\\PAR\n$2/g ;
+    # Convert _n or _\cmd into \SUBSCRIPTNB{n} or \SUBSCRIPTNB{\cmd} and _{nnn} into \SUBSCRIPT{nn}
+    1 while s/(?<!\\)_(\s*([^{\\\s]|\\\w+))/\\SUBSCRIPTNB{$1}/g ;
+    1 while s/(?<!\\)_(\s*{($pat_n)})/\\SUBSCRIPT$1/g ;
+    # Convert ^n into \SUPERSCRIPTNB{n} and ^{nnn} into \SUPERSCRIPT{nn}
+    1 while s/(?<!\\)\^(\s*([^{\\\s]|\\\w+))/\\SUPERSCRIPTNB{$1}/g ;
+    1 while s/(?<!\\)\^(\s*{($pat_n)})/\\SUPERSCRIPT$1/g ;
+    # Convert  \sqrt{n} into \SQRT{n}  and  \sqrt nn into SQRTNB{nn}
+    1 while s/(?<!\\)\\sqrt(\s*([^{\\\s]|\\\w+))/\\SQRTNB{$1}/g ;
+    1 while s/(?<!\\)\\sqrt(\s*{($pat_n)})/\\SQRT$1/g ;
+    # Convert $$ $$ into \begin{DOLLARDOLLAR} \end{DOLLARDOLLAR}
+    s/\$\$(.*?)\$\$/\\begin{DOLLARDOLLAR}$1\\end{DOLLARDOLLAR}/sg;
+    # Convert \[ \] into \begin{SQUAREBRACKET} \end{SQUAREBRACKET}
+    s/(?<!\\)\\\[/\\begin{SQUAREBRACKET}/sg;
+    s/\\\]/\\end{SQUAREBRACKET}/sg;
+    # Convert all picture environmentent (\begin{PICTUREENV} .. \end{PICTUREENV} \PICTUREBLOCKenv
+    s/\\begin\{($PICTUREENV)}(.*?)\\end\{\1}/\\PICTUREBLOCK$1\{$2\}/sg;
+    #    For math-mode COARSE,WHOLE or NONE option -convert all \begin{MATH} .. \end{MATH}
+    #    into \MATHBLOCKMATH{...} commands, where MATH is any valid math environment
+    #    Also convert all array environments into ARRAYBLOCK environments.
+    #    Where these environments have arguments they become optional arguments to the MATHBLOCK command enclosed in < > brackets
+    #    Example: \begin{alignat}{3} ... \end{alignat} will turn into \MATHBLOCKalignat[{3}]{ ... }
+
+    if ( $mathmarkup != FINE ) {
+      # DIFANCHORARRB and DIFANCHORARRE, DIFANCHORMATHB and DIFANCHORMATHE markers are inserted here to encourage the matching algorithm
+      # to always match up the closing brace. Otherwise sometimes one ends up with a situation where
+      # the closing brace is deleted and added at another point. The deleted closing brace is then
+      # prevented by a %DIFDELCMD, leading to material leaking in or out of the math environment.
+      # The anchors are removed in post-processing again. (note that they are simple text to cause least amount of complications
+      # Admittedly, this is something of a hack and will not always work. If it does not, then one needs to
+      # resort to WHOLE or FINE, or NONE math mode processing.
+      s/\\begin\{($ARRENV)}(.*?)\\end\{\1}/\\ARRAYBLOCK$1\{$2\\DIFANCHORARRB \}\\DIFANCHORARRE /sg;
+
+      take_comments_and_newline_from_frac();
+
+      # Convert Math environments with arguments
+      s/\\begin\{($MATHENV|$MATHARRENV|SQUAREBRACKET)\}((?:\[$brat_n\])|(?:\{$pat_n\}))+(.*?)\\end\{\1\}/\\MATHBLOCK$1\[$2\]\{$3\\DIFANCHORMATHB \}\\DIFANCHORMATHE /sg;
+      # Convert Math environments without arguments
+      s/\\begin\{($MATHENV|$MATHARRENV|SQUAREBRACKET)\}(.*?)\\end\{\1\}/\\MATHBLOCK$1\{$2\\DIFANCHORMATHB \}\\DIFANCHORMATHE /sg;
+    }
+
+    # add final token " STOP"
+    $_ .= " STOP"
+  }
+}
+
+
+# $expanded=linecomment($string)
+#preface all lines with verbatim marker (usually DIFVRB)
+sub linecomment {
+  my @verbatimlines=split("\n",$_[0]);
+  # the first line needs special treatment - we do want to retain optional arguments as is but wrap the remainder also with VERBCOMMENT
+  ### print STDERR "DEBUG: before verbatimlines[0] = ",$verbatimlines[0],"\n";
+  $verbatimlines[0]=~s/^((?:\s*\[$brat_n\])?\s*)([^\s\[].*)/ defined($2) ? ( "$1\%$VERBCOMMENT$2" ) : ( $1 )/e;
+  ### print STDERR "DEBUG: after  verbatimlines[0] = ",$verbatimlines[0],"\n";
+  return(join("\n%$VERBCOMMENT",@verbatimlines)."\n");
+}
+
+# $simple=reverselinecomment($env $string)
+# remove DIFVRB comments but leave changed lines marked
+sub reverselinecomment {
+  my ($environment, $verbatimtext)=@_;
+  ### print STDERR "DEBUG REVERSELINECOMMENT input: $environment,|$verbatimtext|\n" if $debug;
+  # remove markup added by latexdiff
+  # (this should occur only if the type of verbatim environment was changed)
+  # (note that this destroys some information in old file)
+  #  in theory I could save it by moving it out of the verbatim environment
+  #  but this requires more bookkeeping and is probably not necessary)
+  $verbatimtext =~ s/\\DIFaddbegin //g;
+  $verbatimtext =~ s/\\DIFaddend //g;
+  $verbatimtext =~ s/\\DIFdelbegin //g;
+  $verbatimtext =~ s/\\DIFdelend //g;
+  $verbatimtext =~ s/$DELCMDOPEN.*//g;
+
+  # remove DIFVRB mark
+  $verbatimtext=~ s/%$VERBCOMMENT//g;
+
+  # remove part of the markup in changed lines
+  # if any of these substitution was made, then there was at least
+  # one changed line, and we have to extend the style
+  if ( $verbatimtext=~ s/$VERBCOMMENT//g ) {
+    # in the next line we add ~alsolanguage~ modifier, but also deletes the rest of the line after the optional argument, as lstlisting commands gets sometimes
+    # very confused by what is there   (and othertimes seems to ignore this anyway)
+    unless ( $verbatimtext =~ s/^(\s*)\[($brat_n)\](.*)\n/$1\[$2,alsolanguage=DIFcode\]\n/ ) {
+      if ( $verbatimtext =~ m/^\s*\n/ ) {
+	$verbatimtext = "[alsolanguage=DIFcode]" . $verbatimtext;
+      } else {
+	$verbatimtext = "[alsolanguage=DIFcode]\n" . $verbatimtext;
+      }
+    }
+    # There is a bug in listings package (at least v1.5b) for empty comments where the actual comment command is not made invisible
+    # I therefore have to introduce an artificial '-' character at the end of empty added or deleted lines
+    $verbatimtext =~ s/($DELCOMMENT\s*)$/$1-/mg;
+    $verbatimtext = "\\DIFmodbegin\n\\begin{${environment}}${verbatimtext}\\end{${environment}}\n\\DIFmodend"
+  } else {
+    $verbatimtext = "\\begin{${environment}}${verbatimtext}\\end{${environment}}"
+  }
+  ### print STDERR "DEBUG REVERSELINECOMMENT output: |$verbatimtext|\n" if $debug;
+  return($verbatimtext);
+}
+
+
+#hashstring=tohash(\%hash,$string)
+# creates a hash value based on string and stores in %hash
+sub tohash {
+  my ($hash,$string)=@_;
+  my (@arr,$val);
+  my ($sum,$i)=(0,1);
+  my ($hstr);
+
+  @arr=unpack('c*',$string);
+
+  while (1) {
+    foreach $val (@arr) {
+      $sum += $i*$val;
+      $i++;
+    }
+    $hstr= "$sum";
+    last unless (defined($hash->{$hstr}) && $string ne $hash->{$hstr});
+    # else found a duplicate HASH need to repeat for a higher hash value
+  }
+  $hash->{$hstr}=$string;
+  ###  print STDERR "Hash:$hstr: Content:$string:\n";
+  return($hstr);
+}
+
+#string=fromhash(\%hash,$fromstring)
+# restores string value stored in hash
+#string=fromhash(\%hash,$fromstring,$prependstring)
+# additionally begins each line with prependstring
+sub fromhash {
+  my ($hash,$hstr)=($_[0],$_[1]);
+  my $retstr=$hash->{$hstr};
+  if ( $#_ >= 2) {
+    $retstr =~ s/^/$_[2]/mg;
+  }
+  return $retstr;
+}
+
+# stripdelcmpopen(string)
+# return string with $DELCMDOPEN removed
+sub stripdelcmdopen {
+  my ($str) = $_[0];
+  $str =~ s/${DELCMDOPEN}//mg;
+  return $str;
+}
+
+# writedebugfile(string, label)
+# if $debug set writes <string> to file latexdiff.debug.<label>
+# otherwise do nothing
+sub writedebugfile {
+  my ($string,$label)=@_;
+  if ( $debug ) {
+    open(RAWDIFF,">","latexdiff.debug." . $label);
+    print RAWDIFF $string;
+    close(RAWDIFF);
+  }
+}
+
+
+# postprocess($string, ..)
+# carry out the following post-processing steps for all arguments:
+# * Remove STOP token from the end
+# * Replace \RIGHTBRACE by }
+# * If option --no-del is set delete all deleted blocks
+# 1. Check all deleted blocks:
+#    a.where a deleted block contains a matching \begin and
+#      \end environment (these will be disabled by a %DIFDELCMD statements), for selected environments enable
+#      these commands again (such that for example displayed math in a deleted equation
+#      is properly within math mode.  For math mode environments replace numbered equation
+#      environments with their display only variety (so that equation numbers in new file and
+#      diff file are identical).  Where the correct type of math environment cannot be determined
+#      use a place holder MATHMODE
+#    b.where one of the commands matching $COUNTERCMD is used as a DIFAUXCMD, add a statement
+#      subtracting one from the respective counter to keep numbering consistent with new file
+#    Replace all MATHMODE environment commands by the correct environment to achieve matching
+#    pairs
+#    c. Convert MATHBLOCKmath commands to their uncounted numbers (e.g. convert equation -> displaymath
+#       (environments defined in $MATHENV will be replaced by $MATHREPL, and  environments in $MATHARRENV
+#       will be replaced by their equivalent with a * appended, e.g. align -> align*
+#    d. If in-line math mode contains array environment, enclose the whole environment in \mbox'es
+#    d. place \cite commands in mbox'es (for UNDERLINE style)
+#
+#   For added blocks:
+#    c. If in-line math mode contains array environment, enclose the whole environment in \mbox'es
+#    d. place \cite commands in mbox'es (for UNDERLINE style)
+#
+# 2.   If math-mode COARSE,WHOLE or NONE option set: Convert \MATHBLOCKmath{..} commands back to environments
+#
+#      Convert all PICTUREblock{..} commands back to the appropriate environments
+# 3. Convert DIFadd, DIFdel, DIFaddbegin , ... into FL varieties
+#    within floats (currently recognised float environments: plate,table,figure
+#    plus starred varieties).
+# 4. Remove empty %DIFDELCMD < lines
+# 4. Convert \begin{SQUAREBRACKET} \end{SQUAREBRACKET} into \[ \]
+#    Convert \begin{DOLLARDOLLAR} \end{DOLLARDOLLAR} into $$ $$
+# 5. Convert  \SUPERSCRIPTNB{n} into ^n  and  \SUPERSCRIPT{nn} into ^{nnn}
+# 6. Convert  \SUBSCRIPTNB{n} into _n  and  \SUBCRIPT{nn} into _{nnn}
+# 7. Expand hashes of verb and verbatim environments
+# 8. Convert '\PERCENTAGE ' back into '\%' and '\DOLLAR ' into '\$'
+# 9.. remove all \PAR tokens
+# 10.  package specific processing:  endfloat: make sure \begin{figure} and \end{figure} are always
+#      on a line by themselves, similarly for table environment
+#  4, undo renaming of the \begin, \end,{,}  in comments
+#    Change \QLEFTBRACE, \QRIGHTBRACE,\AMPERSAND to \{,\},\&
+#
+# Note have to manually synchronize substitution commands below and
+# DIF.. command names in the header
+sub postprocess {
+  my ($begin,$len,$float,$delblock,$addblock); ### $cnt
+  # second level blocks
+  my ($begin2,$len2,$eqarrayblock,$eqblock,$mathblock); ### $cnt2
+
+  my (@textparts,@newtextparts,@liststack,$listtype,$listlast);
+
+  my (@itemargs, $itemarg);
+
+
+  for (@_) {
+    # change $'s in comments to something harmless
+    1 while s/(%.*)\$/$1DOLLARDIF/mg ;
+
+    # Remove final STOP token
+    s/ STOP$//;
+    # Replace \RIGHTBRACE in comments by \MBLOCKRIGHTBRACE
+    # the only way to get these is as %DIFDELCMD < \RIGHTBRACE construction
+    # This essentially marks closing right braces of MATHBLOCK environments, which did not get matched
+    # up. This case should be rare, so I just leave this in the diff file output. Not really elegant
+    # but can still be dealt with later if it results in problems.
+    s/%DIFDELCMD < \\RIGHTBRACE/%DIFDELCMD < \\MBLOCKRIGHTBRACE/g ;
+    # Replace \RIGHTBRACE by }
+    s/\\RIGHTBRACE/}/g;
+
+    # Optional: Remove deleted block entirely
+    if ($onlyadditions) {
+        s/\\DIFdelbegin.*?\\DIFdelend//sg ;
+    }
+
+    # Check all deleted blocks: where a deleted block contains a matching \begin and
+    #    \end environment (these will be disabled by a %DIFDELCMD statements), enable
+    #    these commands again (such that for example displayed math in a deleted equation
+    #    is properly within math mode).  For math mode environments replace numbered equation
+    #    environments with their display only variety (so that equation numbers in new file and
+    #    diff file are identical)
+    writedebugfile($_,'postprocess1');
+
+
+    while ( m/\\DIFdelbegin.*?\\DIFdelend/sg ) {
+      # special processing within each deleted block
+      ###    while ( m/\\DIFdelbegin.*?\\DIFdelend/sg ) {
+      ###      print STDERR "DEBUG Match delblock \n||||$&||||\n at ",pos,"\n";
+      my $arrenv;    # dummy variable only needed within this block
+      my @symbols;   # another temporary variable
+      $len=length($&);
+      $begin=pos($_) - $len;
+      $delblock=$&;
+      ###   A much simpler method for math replacement might follow this strategy (can recycle part of the commands below for following
+      ###   this strategy:
+      ###   1. a Insert aux commands \begin{MATHMODE} or \end{MATHMODE} for all deleted commands opening or closing displayed math mode
+      ###      b Insert aux commands \begin{MATHARRMODE} or \end{MATHARRMODE} for all deleted commands opening or closing math array mode
+      ###   2  Replace MATHMODE and MATHARRMODE by correct pairing if appropriate partner  math command is found in text
+      ###   3  a Replace remaining \begin{MATHMODE}...\end{MATHMODE} pairs with \begin{$MATHREPL}..\end{$MATHREPL}
+      ###      b Replace remaining \begin{MATHARRMODE}...\end{MATHARRMODE} pairs with \begin{$MATHREPL}..\end{$MATHREPL}
+      ###   4  Delete all aux command math mode pairs which have simply comments or empty lines between them
+      ###   As written this won't actually work!
+
+
+      ###   Most general case: allow all included environments
+      ###      $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{(\w*\*?)\}\s*?\n)(.*?)(\%DIFDELCMD < \s*\\end\{\2\})/$1\\begin{$2}$AUXCMD\n$3\n\\end{$2}$AUXCMD\n$4/sg;
+      ### (.*?[^\n]?)\n? construct is necessary to avoid empty lines in math mode, which result in
+      ### an error
+      # displayed math environments
+      ###0.5:     $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{((?:$MATHENV)|SQUAREBRACKET)\}\s*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$MATHREPL}$AUXCMD\n$1$3\n\\end{$MATHREPL}$AUXCMD\n$4/sg;
+
+      if ($mathmarkup == FINE ) {
+        # if block contains both beginning and and markers for an explicit math environment, restore opening and closing commands
+	$delblock=~ s/(\%DIFDELCMD < \s*\\begin\{((?:$MATHENV)|SQUAREBRACKET)\}.*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$MATHREPL}$AUXCMD\n$1$3\n\\end{$MATHREPL}$AUXCMD\n$4/sg;
+	# also transform the opposite pair \end{displaymath} .. \begin{displaymath} but we have to be careful not to interfere with the results of the transformation in the line directly above
+	### pre-0.42 obsolete version which did not work on eqnarray test      $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\end\{($MATHENV)\}\s*?\n)(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\begin\{\2\})/$1\\end{$MATHREPL}$AUXCMD\n$3\n\\begin{$MATHREPL}$AUXCMD\n$4/sg;
+	###0.5:      $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\end\{((?:$MATHENV)|SQUAREBRACKET)\}\s*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\begin\{\2\})/\\end{MATHMODE}$AUXCMD\n$1$3\n\\begin{MATHMODE}$AUXCMD\n$4/sg;
+	$delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\end\{((?:$MATHENV)|SQUAREBRACKET)\}.*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\begin\{\2\})/\\end\{MATHMODE\}$AUXCMD\n$1$3\n\\begin\{MATHMODE\}$AUXCMD\n$4/sg;
+
+        # now look for unpaired %DIFDELCMD < \begin{MATHENV}; if found add \begin{$MATHREPL} and insert \end{$MATHREPL}
+        # just before end of block; again we use look-behind assertion to avoid matching constructions which have already been converted
+        if ($delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\begin\{((?:$MATHENV)|SQUAREBRACKET)\}\s*?(?:$DELCMDCLOSE|\n))/$1\\begin{$MATHREPL}$AUXCMD\n/sg ) {
+	  ### print STDERR "BINGO: begin block: \nBefore: |" . substr($_,$begin,$len) . "|\n" if $debug ;
+	  $delblock =~ s/(\\DIFdelend$)/\\end{$MATHREPL}$AUXCMD\n$1/s ;
+	  ### print STDERR "After: |" . $delblock . "|\n\n" if $debug ;
+        }
+        # now look for unpaired %DIFDELCMD < \end{MATHENV}; if found add \end{MATHMODE} and insert \begin{MATHMODE}
+        # just before end of block; again we use look-behind assertion to avoid matching constructions which have already been converted
+        if ($delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\end\{((?:$MATHENV)|SQUAREBRACKET)\}\s*?(?:$DELCMDCLOSE|\n))/$1\\end\{MATHMODE\}$AUXCMD\n/sg ) {
+	  ### print STDERR "BINGO: end block:\nBefore: |" . substr($_,$begin,$len) . "|\n" if $debug;
+	  $delblock =~ s/(\\DIFdelend$)/\\begin\{MATHMODE\}$AUXCMD\n$1/s ;
+	  ### print STDERR "After: |" . $delblock . "|\n\n" if $debug ;
+	}
+
+	### pre-0.42      # same as above for special case \[.\] (latex abbreviation for displaymath)
+        ### pre-0.42      $delblock=~ s/(\%DIFDELCMD < \s*\\\[\s*?\n())(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\\])/$1\\\[$AUXCMD\n$3\n\\\]$AUXCMD\n$4/sg;
+        ### pre-0.42      $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\\]\s*?\n())(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\\[)/$1\\\]$AUXCMD\n$3\n\\\[$AUXCMD\n$4/sg;
+        # equation array environment
+        ###pre-0.3      $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($MATHARRENV)\}\s*?\n)(.*?)(\%DIFDELCMD < \s*\\end\{\2\})/$1\\begin{$MATHARRREPL}$AUXCMD\n$3\n\\end{$MATHARRREPL}$AUXCMD\n$4/sg;
+        ###0.5      $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($MATHARRENV)\}\s*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$MATHARRREPL}$AUXCMD\n$1$3\n\\end{$MATHARRREPL}$AUXCMD\n$4/sg;
+	###1.3.2         $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($MATHARRENV)\}.*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$MATHARRREPL}$AUXCMD\n$1$3\n\\end{$MATHARRREPL}$AUXCMD\n$4/sg;
+	# Example Input: %DIFDELCMD < \begin{alignat}{2} \n\exp(ix)&=&\cos(x)+i\sin(x)\n%DIFDECMD < \end{alignat}
+        #                < $1                <$2   >><$3>  < $4                      ><$5                       >
+	# Example Output:\begin{alignat*}{2}%DIFAUXCMD\n\%DIFDELCMD\begin{align}\n\exp(ix)&=&\cos(x)+i\sin(x)\n\end{align*}%DIFAUXCMD\n%DIFDELCMD > \end{alignat*}
+        $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($MATHARRENV)\}(.*?)(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$2*}$3$AUXCMD\n$1$4\n\\end{$2*}$AUXCMD\n$5/sg;
+        ###  pre-0.42 obsolete version which did not work on eqnarray test     $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\end\{($MATHARRENV)\}\s*?\n)(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\begin\{\2\})/$1\\end{$MATHARRREPL}$AUXCMD\n$3\n\\begin{$MATHARRREPL}$AUXCMD\n$4/sg;
+        $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\end\{($MATHARRENV)\}\s*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\begin\{\2\})/\\end{MATHMODE}$AUXCMD\n$1$3\n\\begin{MATHMODE}$AUXCMD\n$4/sg;
+        ### print STDERR "STEP1: |" . $delblock . "|\n\n" if $debug ;
+
+        # now look for unpaired %DIFDELCMD < \begin{MATHARRENV}; if found add \begin{$MATHARRREPL} and insert \end{$MATHARRREPL}
+        # just before end of block; again we use look-behind assertion to avoid matching constructions which have already been converted
+        if ($delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\begin\{($MATHARRENV)\}(.*?)(?:$DELCMDCLOSE|\n))/$1\\begin{$2*}$3$AUXCMD\n/sg ) {
+	  $arrenv=$2;
+	  $delblock =~ s/(\\DIFdelend$)/\\end{$arrenv*}$AUXCMD\n$1/s ;
+        }
+        ### print STDERR "STEP2: |" . $delblock . "|\n\n" if $debug ;
+
+        # now look for unpaired %DIFDELCMD < \end{MATHENV}; if found add \end{MATHMODE} and insert \begin{MATHMODE}
+        # just before end of block; again we use look-behind assertion to avoid matching constructions which have already been converted
+        if ($delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\end\{($MATHARRENV)\}\s*?(?:$DELCMDCLOSE|\n))/$1\\end{MATHMODE}$AUXCMD\n/sg ) {
+	  $delblock =~ s/(\\DIFdelend$)/\\begin{MATHMODE}$AUXCMD\n$1/s ;
+        }
+
+	# parse $delblock for deleted and reinstated eqnarray* environments - within those reinstate \\ and & commands
+	###      while ( $delblock =~ m/\\begin{$MATHARRREPL}$AUXCMD\n.*?\n\\end{$MATHARRREPL}$AUXCMD\n/sg ) {
+        ### while ( $delblock =~ m/\\begin\Q{$MATHARRREPL}$AUXCMD\E\n.*?\n\\end\Q{$MATHARRREPL}$AUXCMD\E\n/sg ) {
+	while ( $delblock =~ m/\\begin\{$MATHARRENV\*}[^\n]*?$AUXCMD\n.*?\n\\end\{$MATHARRENV\*\}$AUXCMD\n/sg ) {
+	  ###	      print STDERR "DEBUG Match eqarrayblock $& at ",pos,"\n";
+	  $len2=length($&);
+	  $begin2=pos($delblock) - $len2;
+	  $eqarrayblock=$&;
+	  # reinstate deleted & and \\ commands
+	  ### $eqarrayblock=~ s/(\%DIFDELCMD < \s*(\&|\\\\)\s*?(?:$DELCMDCLOSE|\n))/$1$2$AUXCMD\n/sg ;
+	  $eqarrayblock    =~ s/(\%DIFDELCMD < (.*?(?:\&|\\\\).*)(?:$DELCMDCLOSE|\n))/
+        	  # The pattern captures comments with at least one of & or \\
+                  @symbols= split(m@((?:&|\\\\)\s*)@,$2);   #  extract & and \\ and other material from sequence
+                  @symbols= grep ( m@&|\\\\\s*@, @symbols); #  select & and \\ (and subsequent spaces)
+                  "$1@symbols$AUXCMD\n"
+            /eg ;
+	  ### print STDERR "Modified block:|$eqarrayblock|\n" if  $debug;
+	  # splice in modified block
+	  substr($delblock,$begin2,$len2)=$eqarrayblock;
+	  pos($delblock) = $begin2 + length($eqarrayblock);
+	}
+
+      } elsif ( $mathmarkup == COARSE || $mathmarkup == WHOLE ) {
+	#       Convert MATHBLOCKmath commands to their uncounted numbers (e.g. convert equation -> displaymath
+	#       (environments defined in $MATHENV will be replaced by $MATHREPL, and  environments in $MATHARRENV
+	#       will be replaced by their starred variety
+	$delblock=~ s/\\MATHBLOCK($MATHENV)((?:\[$brat_n\])?)\{($pat_n)\}/\\MATHBLOCK$MATHREPL$2\{$3\}/sg;
+	$delblock=~ s/\\MATHBLOCK($MATHARRENV)((?:\[$brat_n\])?)\{($pat_n)\}/\\MATHBLOCK$1\*$2\{$3\}/sg;
+      }
+
+      # Replacing math array environments with their starred varieties in deleted blocks as implemented now causes double asterisks
+      # in situation where there had been starred already, e.g. \begin{alignat*} is turned into \begin{alignat**}
+      # The following command seeks to undo this double-starring. It is a little bit of a hack because it relies on the fact that **
+      # double-starred math array environment does not occur naturally
+      $delblock =~ s/($MATHARRENV)(?<=\*)\*/$1/sg;
+
+      print STDERR "DELBLOCK after maths processing: |" . $delblock . "|\n\n" if $debug ;
+
+
+      # Reinstate completely deleted list environments. note that items within the
+      # environment will still be commented out.  They will be restored later
+      $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($LISTENV)\}\s*?(?:\n|$DELCMDCLOSE))(.*?)(\%DIFDELCMD < \s*\\end\{\2\})/{
+															###   # block within the search; replacement environment
+															###   "$1\\begin{$2}$AUXCMD\n". restore_item_commands($3). "\n\\end{$2}$AUXCMD\n$4";
+															"$1\\begin{$2}$AUXCMD\n$3\n\\end{$2}$AUXCMD\n$4";
+														       }/esg;
+      ###      $delblock=~ s/\\begin\{$MATHENV}$AUXCMD/\\begin{$MATHREPL}$AUXCMD/g;
+      ###      $delblock=~ s/\\end\{$MATHENV}$AUXCMD/\\end{$MATHREPL}$AUXCMD/g;
+      ###      $delblock=~ s/\\begin\{$MATHARRENV}$AUXCMD/\\begin{$MATHARRREPL}$AUXCMD/g;
+      ###      $delblock=~ s/\\end\{$MATHARRENV}$AUXCMD/\\end{$MATHARRREPL}$AUXCMD/g;
+
+      #    b.where one of the commands matching $COUNTERCMD is used as a DIFAUXCMD, add a statement
+      #      subtracting one from the respective counter to keep numbering consistent with new file
+      $delblock=~ s/\\($COUNTERCMD)((?:${extraspace}\[$brat_n\]${extraspace}|${extraspace}\{$pat_n\})*\s*${AUXCMD}\n)/\\$1$2\\addtocounter{$1}{-1}${AUXCMD}\n/sg ;
+
+      #    bb. disable active labels within deleted blocks (i.e. those not commented out) (as these are not safe commands, this should normally only
+      #        happen within deleted maths blocks
+      ###      $delblock=~ s/(?<!$DELCMDOPEN)(\\$LABELCMD(?:${extraspace})\{(?:[^{}])*\}[\t ]*)\n?/${DELCMDOPEN}$1${DELCMDCLOSE}/smg ;
+      ###      previous line caused trouble as by issue #90 I might need to modify this
+      $delblock=~ s/^([^%]*)(\\$LABELCMD(?:${extraspace})\{(?:[^{}])*\}[\t ]*)\n?/$1${DELCMDOPEN}$2${DELCMDCLOSE}/smg ;
+      ###      print STDERR "<<<$delblock>>>\n" if $debug;
+
+
+      #     c. If in-line math mode contains array environment, enclose the whole environment in \mbox'es
+      while ( $delblock =~ m/($math)(\s*)/sg ) {
+##		      print STDERR "DEBUG Delblock Match math $& at ",pos,"\n";
+	$len2=length($&);
+	$begin2=pos($delblock) - $len2;
+	$mathblock="%\n\\mbox{$AUXCMD\n$1\n}$AUXCMD\n";
+        next unless ( $mathblock =~ /ARRAYBLOCK/ or $mathblock =~ m/\{$ARRENV\}/ );
+	substr($delblock,$begin2,$len2)=$mathblock;
+	pos($delblock) = $begin2 + length($mathblock);
+      }
+      ###      if ($CITE2CMD) {
+      ######   ${extraspace}(?:\[$brat0\]${extraspace}){0,2}\{$pat_n\}))  .*?%%%\n
+      ###	$delblock=~s/($DELCMDOPEN\s*\\($CITE2CMD)(.*)$DELCMDCLOSE)/
+      ###	  # Replacement code
+      ###	  {my ($aux,$all);
+      ###	   $aux=$all=$1;
+      ###	   $aux=~s#\n?($DELCMDOPEN|$DELCMDCLOSE)##g;
+      ###	   $all."$aux$AUXCMD\n";}/sge;
+      ###      }
+      ###      # or protect \cite commands with \mbox
+      ###      if ($CITECMD) {
+      ######	$delblock=~s/(\\($CITECMD)${extraspace}(?:\[$brat0\]${extraspace}){0,2}\{$pat_n\})(\s*)/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/msg ;
+      ###	$delblock=~s/(\\($CITECMD)${extraspace}(?:<$abrat0>${extraspace})?(?:\[$brat0\]${extraspace}){0,2}\{$pat_n\})(\s*)/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/msg ;
+      ###      }
+      # if MBOXINLINEMATH is set, protect inlined math environments with an extra mbox
+
+
+      if ( $MBOXINLINEMATH ) {
+	# note additional \newline after command is omitted from output if right at the end of deleted block (otherwise a spurious empty line is generated)
+	$delblock=~s/($math)(?:[\s\n]*)?/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/sg;
+      }
+      ###if ( defined($packages{"listings"} and $latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/))   {
+      ###  #     change included verbatim environments
+      ###  $delblock =~ s/\\DIFverb\{/\\DIFDIFdelverb\{/g;
+      ###  $delblock =~ s/\\DIFlstinline/\\DIFDIFdellstinline/g;
+      ###}
+      # Mark deleted verbatim commands
+      $delblock =~ s/(${DELCMDOPEN}\\DIF((?:verb\*?|lstinline(?:\[$brat_n\])?)\{([-\d]*?)\}\s*).*)$/%\n\\DIFDIFdel$2${AUXCMD}\n$1/gm;
+      if ( $CUSTOMDIFCMD ) {
+        ###$delblock =~ s/(${DELCMDOPEN}.*)\\($CUSTOMDIFCMD)((?:\[${brat_n}\])*?(?:\s*\{${pat_n}\})*)/"$1${DELCMDCLOSE}\\DEL$2". stripdelcmdopen($3) ." ${DELCMDOPEN}"/egms;
+        ###$delblock =~ s/(${DELCMDOPEN}.*)\\($CUSTOMDIFCMD)/$1${DELCMDCLOSE}\\DEL$2/gm;
+        # ($1 ? "${DELCMDOPEN}$1${DELCMDCLOSE}":"") : only add the DELCMDOPEN / DELCMDCLOSE pair if there are actually any commands in between, otherwise this is redundant
+	$delblock    =~ s/${DELCMDOPEN}(.*?)\\($CUSTOMDIFCMD)((?:\[${brat_n}\])*?(?:\s*\{${pat_n}\})*)/($1 ? "${DELCMDOPEN}$1${DELCMDCLOSE}":"") ."\\DEL$2". stripdelcmdopen($3)/egs;
+        # if there is a sequence of several commands in the same row only the first will be converted due to the need to be connected to the DELCMDOPEN. To mop these up, just add the DEL to the front of any remaining cmd's in the deleted block
+        $delblock =~ s/\\($CUSTOMDIFCMD)/\\DEL$1/g;     # this will also convert comments but I guess it does not matter much
+      }
+
+      #     splice in modified delblock
+      substr($_,$begin,$len)=$delblock;
+      pos = $begin + length($delblock);
+    }
+
+    ### print STDERR "<<<$_>>>\n" if $debug;
+
+
+    # make the array modification in added blocks
+    while ( m/\\DIFaddbegin.*?\\DIFaddend/sg ) {
+      $len=length($&);
+      $begin=pos($_) - $len;
+      $addblock=$&;
+      while ( $addblock =~ m/($math)(\s*)/sg ) {
+	$len2=length($&);
+	$begin2=pos($addblock) - $len2;
+	$mathblock="%\n\\mbox{$AUXCMD\n$1\n}$AUXCMD\n";
+        next unless ( $mathblock =~ /ARRAYBLOCK/ or $mathblock =~ m/\{$ARRENV\}/) ;
+	substr($addblock,$begin2,$len2)=$mathblock;
+	pos($addblock) = $begin2 + length($mathblock);
+      }
+      # if MBOXINLINEMATH is set, protect inlined math environments with an extra mbox
+      if ( $MBOXINLINEMATH ) {
+	##$addblock=~s/($math)/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/sg;
+	$addblock=~s/($math)(?:[\s\n]*)?/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/sg;
+      }
+      ###if ( defined($packages{"listings"} and $latexdiffpreamble =~ /\\RequirePackage(?:\[$brat0\])?\{color\}/))   {
+	# mark added verbatim commands
+      $addblock =~ s/\\DIFverb/\\DIFDIFaddverb/g;
+      $addblock =~ s/\\DIFlstinline/\\DIFDIFaddlstinline/g;
+      if( $CUSTOMDIFCMD ) {
+        $addblock =~ s/\\($CUSTOMDIFCMD)/\\ADD$1/g;     # this will also convert comments but I guess it does not matter much
+      }
+      # markup the optional arguments of \item
+      $addblock =~ s/(\\$ITEMCMD$extraspace(?:<$abrat0>)?$extraspace)\[($brat_n)\]/
+	@itemargs=splitlatex(substr($2,0,length($2)));
+        $itemarg="[".join("",marktags("","",$ADDOPEN,$ADDCLOSE,"","",$ADDCOMMENT,\@itemargs))."]";
+      "$1$itemarg"/sge;   # old substitution: $1\[$ADDOPEN$2$ADDCLOSE\]
+      ###}
+#     splice in modified addblock
+      substr($_,$begin,$len)=$addblock;
+      pos = $begin + length($addblock);
+    }
+
+    # Go through whole text, and by counting list environment commands, find out when we are within a list environment.
+    # Within those restore deleted \item commands
+    @textparts=split /(?<!$DELCMDOPEN)(\\(?:begin|end)\{$LISTENV\})/ ;
+    @liststack=();
+    @newtextparts=map {
+      ### print STDERR ":::::::: $_\n";
+      if ( ($listtype) = m/^\\begin\{($LISTENV)\}$/ ) {
+	print STDERR "DEBUG: postprocess \\begin{$listtype}\n" if $debug;
+	push @liststack,$listtype;
+      } elsif ( ($listtype) = m/^\\end\{($LISTENV)\}$/ ) {
+ 	print STDERR "DEBUG: postprocess \\end{$listtype}\n" if $debug;
+	if (scalar  @liststack > 0) {
+ 	  $listlast=pop(@liststack);
+ 	  ($listtype eq $listlast) or warn "Invalid nesting of list environments: $listlast environment closed by \\end{$listtype}.";
+ 	} else {
+ 	  warn "WARNING: Invalid nesting of list environments: \\end{$listtype} encountered without matching \\begin{$listtype}.\n";
+ 	}
+      } else {
+	print STDERR "DEBUG: postprocess \@liststack=(",join(",",@liststack),")\n" if $debug;
+	if (scalar  @liststack > 0 ) {
+	  # we are within a list environment and should replace all item commands
+	  $_=restore_item_commands($_);
+	}
+	# else: we are outside a list environment and do not need to do anything
+      }
+      $_ } @textparts;     # end of map command
+    # replace the main text with the modified version
+    $_= join("",@newtextparts);
+
+
+
+
+    # Replace MATHMODE environments from step 1a above by the correct Math environment and remove unncessary pairings
+
+    if ( $mathmarkup == FINE ) {
+      # look for AUXCMD math-mode pairs which have only comments (or empty lines between them), and remove the added commands
+      # \begin{..} ... \end{..} pairs
+      s/\\begin\{((?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET|MATHMODE)\}$AUXCMD\n((?:\s*%.[^\n]*\n)*)\\end\{\1\}$AUXCMD\n/$2/sg;
+      # \end{..} .... \begin{..} pairs
+      s/\\end\{((?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET|MATHMODE)\}$AUXCMD\n((?:\s*%.[^\n]*\n)*)\\begin\{\1\}$AUXCMD\n/$2/sg;
+
+      writedebugfile($_,'postprocess15');
+      # The next line is complicated.  The negative look-ahead insertion makes sure that no \end{$MATHENV} (or other mathematical
+      # environments) are between the \begin{$MATHENV} and \end{MATHMODE} commands. This is necessary as the minimal matching
+      # is not globally minimal but only 'locally' (matching is beginning from the left side of the string)
+      # [NB: Do not be tempted to prettify the expression with /x modified. It seems this is applied after strings are expanded so will ignore spaces in strings
+      1 while s/(?<!$DELCMDOPEN)\\begin\{((?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET)}((?:${DELCMDOPEN}[^\n]*|.(?!(?:\\end\{(?:(?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET)}|\\begin\{MATHMODE})))*?)\\end\{MATHMODE}/\\begin{$1}$2\\end{$1}/s;
+      writedebugfile($_,'postprocess16');
+      1 while s/\\begin\{MATHMODE}((?:.(?!\\end\{MATHMODE}))*?)(?<!$DELCMDOPEN)\\end\{((?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET)}/\\begin{$2}$1\\end{$2}/s;
+      # convert remaining \begin{MATHMODE} \end{MATHMODE} (and not containing & or \\ )into MATHREPL environments
+      s/\\begin\{MATHMODE\}((?:(.(?!(?<!\\)\&|\\\\))*)?)\\end\{MATHMODE\}/\\begin{$MATHREPL}$1\\end{$MATHREPL}/sg;
+      # others into MATHARRREPL
+      s/\\begin\{MATHMODE\}(.*?)\\end\{MATHMODE\}/\\begin{$MATHARRREPL}$1\\end{$MATHARRREPL}/sg;
+      # Final cleanup of all equation environments: Sometimes I end up with %DIFDELCMD < \PAR lines, which I need to make sure do not get expanded into
+      # actual empty lines within equation environments (later part)
+      while (  m/\\begin\{((?:$MATHENV|$MATHARRENV|SQUAREBRACKET)\*?)}.*?(?<!\%DIFDELCMD < )\\end\{\1}/sg ) {
+	$len=length($&);
+	$begin=pos($_) - $len;
+	$eqblock=$&;
+	$eqblock =~ s/(%DIFDELCMD < )([^\n]*?)\\PAR\n/$1$2\n$1\n/sg;
+	#     splice in modified delblock
+	print STDERR "DEBUG EQBLOCK: |$eqblock|\n\n" if $debug;
+	substr($_,$begin,$len)=$eqblock;
+	pos = $begin + length($eqblock);
+      }
+    } else {
+      #   math modes OFF,WHOLE,COARSE: Convert \MATHBLOCKmath{..} commands back to environments
+      # with optionl argument to MATHBLOCK, e.g \MATHBLOCKalignat[{3}]{ ...}
+      s/\\MATHBLOCK($MATHENV|$MATHARRENV\*?|SQUAREBRACKET)\[($brat0)\]\{($pat_n)\}/\\begin{$1}$2$3\\end{$1}/sg;
+      # without optional argument e.g \MATHBLOCKalign{ ...}
+      s/\\MATHBLOCK($MATHENV|$MATHARRENV\*?|SQUAREBRACKET)\{($pat_n)\}/\\begin{$1}$2\\end{$1}/sg;
+      # convert ARRAYBLOCK.. commands back to environments
+      s/\\ARRAYBLOCK($ARRENV)\{($pat_n)\}/\\begin{$1}$2\\end{$1}/sg;
+      # get rid of the DIFANCHOR markers, first the delete comments, then everywhere
+      s/%DIFDELCMD < \\DIFANCHOR(?:MATH|ARR)[BE] (?:\n%DIFDELCMD < )?%%%\n//g ;
+      s/\\DIFANCHOR(?:MATH|ARR)[BE] //g;
+    }
+    writedebugfile($_,'postprocess2');
+
+    #  Convert all PICTUREblock{..} commands back to the appropriate environments
+    s/\\PICTUREBLOCK($PICTUREENV)\{($pat_n)\}/\\begin{$1}$2\\end{$1}/sg;
+#0.5:    # Remove all mark up within picture environments
+#     while ( m/\\begin\{($PICTUREENV)\}.*?\\end\{\1\}/sg ) {
+#       $cnt=0;
+#       $len=length($&);
+#       $begin=pos($_) - $len;
+#       $float=$&;
+#       $float =~ s/\\DIFaddbegin //g;
+#       $float =~ s/\\DIFaddend //g;
+#       $float =~ s/\\DIFadd\{($pat_n)\}/$1/g;
+#       $float =~ s/\\DIFdelbegin //g;
+#       $float =~ s/\\DIFdelend //g;
+#       $float =~ s/\\DIFdel\{($pat_n)\}//g;
+#       $float =~ s/$DELCMDOPEN.*//g;
+#       substr($_,$begin,$len)=$float;
+#       pos = $begin + length($float);
+#     }
+    # Convert DIFadd, DIFdel, DIFFaddbegin , ... into  varieties
+    #    within floats (currently recognised float environments: plate,table,figure
+    #    plus starred varieties).
+    while ( m/\\begin\{($FLOATENV)\}.*?\\end\{\1\}/sg ) {
+      $len=length($&);
+      $begin=pos($_) - $len;
+      $float=$&;
+      $float =~ s/\\DIFaddbegin /\\DIFaddbeginFL /g;
+      $float =~ s/\\DIFaddend /\\DIFaddendFL /g;
+      $float =~ s/\\DIFadd\{/\\DIFaddFL{/g;
+      $float =~ s/\\DIFdelbegin /\\DIFdelbeginFL /g;
+      $float =~ s/\\DIFdelend /\\DIFdelendFL /g;
+      $float =~ s/\\DIFdel\{/\\DIFdelFL{/g;
+      substr($_,$begin,$len)=$float;
+      pos = $begin + length($float);
+    }
+    ### former location of undo renaming of \begin and \end in comments
+
+    # remove empty DIFCMD < lines
+    s/^\Q${DELCMDOPEN}\E\n//msg;
+
+    # Expand hashes of verb and verbatim environments
+    s/${DELCMDOPEN}\\($VERBATIMENV)\{([-\d]*?)\}/"${DELCMDOPEN}\\begin{${1}}".fromhash(\%verbhash,$2,$DELCMDOPEN)."${DELCMDOPEN}\\end{${1}}"/esg;
+    # revert changes to verbatim environments for line diffs (and add code to mark up changes) (note negative look behind assertions to not leak out of DIFDELCMD comments)
+    # Example:
+    # < \begin{verbatim}
+    # < %DIF < DIFVRB old verbatim line
+    # < %DIF > DIFVRB new verbatim line
+    # < \end{verbatim}
+    # ---
+    # > \DIFmodbegin
+    # > \begin{verbatim}[alsolanguage=DIFcode]
+    # > %DIF < old verbatim line
+    # > %DIF > new verbatim line
+    # > \end{verbatim}
+    # > \DIFmodend
+    s/(?<!$DELCMDOPEN)\\begin\{($VERBATIMLINEENV)\}(.*?)(?<!$DELCMDOPEN)\\end\{\1\}/"". reverselinecomment($1, $2) .""/esg;
+#    # we do the same for deleted environments but additionally reinstate the framing commands
+#   s/$DELCMDOPEN\\begin\{($VERBATIMLINEENV)\}$extraspace(?:\[$brat0\])?$DELCMDCLOSE(.*?)$DELCMDOPEN\\end\{\1\}$DELCMDCLOSE/"\\begin{$1}". reverselinecomment($2) . "\\end{$1}"/esg;
+##    s/$DELCMDOPEN\\begin\{($VERBATIMLINEENV)\}($extraspace(?:\[$brat0\])?\s*)(?:\n|$DELCMDOPEN)*$DELCMDCLOSE((?:\%$DELCOMMENT$VERBCOMMENT.*?\n)*)($DELCMDOPEN\\end\{\1\}(?:\n|\s|$DELCMDOPEN)*$DELCMDCLOSE)/"SUBSTITUTION: \\begin{$1}$2 INTERIOR: |$3| END: |$4|"/esg;
+     s/ # Deleted \begin command of verbatim environment (Captures $1: whole deleted command, $2: environment, $3: optional arguments with white space
+          (\Q$DELCMDOPEN\E\\begin\{($VERBATIMLINEENV)\}(\Q$extraspace\E(?:\[$brat_n\])?\s*)(?:\n|\Q$DELCMDOPEN\E)*\Q$DELCMDCLOSE\E)
+        # Interior of deleted verbatim environment should consist entirely of delete DIFVRB comments, i.e. match only lines beginning with % DIF < DIFVRB
+        #   Captures: $4: all lines combined
+          ((?:\%\Q$DELCOMMENT$VERBCOMMENT\E[^\n]*?\n)*)
+        # Deleted \end command of verbatim environment. Note that the type is forced to match the opening. Captures: $5: Whole deleted environment  (previous way this line was written: (\Q$DELCMDOPEN\E\\end\{\2\}(?:\n|\s|\Q$DELCMDOPEN\E)*\Q$DELCMDCLOSE\E)
+          (\Q$DELCMDOPEN\E\\end\{\2\})
+      / # Substitution part
+            $1                   # Leave expression as is
+            . "$AUXCMD NEXT\n"   # Mark the following line as an auxiliary command
+            . ""    # reinstate the original environment without options
+            . reverselinecomment($2, "$3$4")   # modify the body to change the markup; reverselinecomment parses for options
+            . " $AUXCMD\n"  # close the auxiliary environment
+            . $5               # and again leave the original deleted closing environment as is
+      /esgx;  # Modifiers of substitution command
+    writedebugfile($_,'postprocess2');
+    # where changes have occurred in verbatim environment, change verbatim to DIFverbatim to allow mark-up
+    # (I use the presence of optional paramater to verbatim environment as the marker - normal verbatim
+    # environment does not take optional arguments)
+    s/(?<!$DELCMDOPEN)\\begin\{(verbatim[*]?)\}(\[$brat_n\].*?)\\end\{\1\}/\\begin{DIF$1}$2\\end{DIF$1}/sg;
+
+    s/\\($VERBATIMENV)\{([-\d]*?)\}/"\\begin{${1}}".fromhash(\%verbhash,$2)."\\end{${1}}"/esg;
+
+    # remove all \PAR tokens (taking care to properly keep commented out PAR's
+    # from introducing uncommented newlines - next line)
+    s/(%DIF < )([^\n]*?)\\PAR\n/$1$2\n$1\n/sg;
+    # convert PAR commands which are on a line by themselves
+    s/\n(\s*?)\\PAR\n/\n\n/sg;
+    # convert remaining PAR commands (which are preceded by non-white space characters, usually "}" ($ADDCLOSE)
+    s/\\PAR\n/\n\n/sg;
+
+    #  package specific processing:
+    if ( defined($packages{"endfloat"})) {
+      #endfloat: make sure \begin{figure} and \end{figure} are always
+      #      on a line by themselves, similarly for table environment
+      print STDERR "endfloat package detected.\n" if $verbose ;
+      # eliminate whitespace before and after
+      s/^(\s*)(\\(?:end|begin)\{(?:figure|table)\})(\s*?)$/$2/mg;
+      # split lines with remaining characters before float environment conmmand
+      s/^([^%]+)(\\(?:begin|end)\{(?:figure|table)\})/$1\n$2/mg;
+      # split lines with remaining characters after float environment conmmand
+      s/^((?:[^%]+)\\(?:begin|end)\{(?:figure|table)\}(?:\[[a-zA-Z]+\])?)(.+)((?:%.*)?)$/$1\n$2$3/mg;
+    }
+
+    # Remove empty auxiliary LISTENV  (sometmes these are generated if list environment occurs within the argument of a deleted comment)
+    #   (slightly hacky but I could not see an easy way to see if in argument of another command when this extra markup is added)
+    s/\\begin\{($LISTENV)\}$AUXCMD\n((?:\s*\%[^\n]*\n)*\n?)\\end\{\1\}$AUXCMD\n/$2\n/msg ;
+
+    # Convert '\PERCENTAGE ' back into '\%' (the final question mark catches a special situation where due to a latter pre-processing step the ' ' becomes separated
+    s/\\PERCENTAGE ?/\\%/g;
+    # Convert '\DOLLAR ' back into '\$'
+    s/\\DOLLAR /\\\$/g;
+
+    # undo renaming of the \begin and \end,{,}  and dollars in comments
+
+    # although we only renamed $ in comments to DOLLARDIFF, we might have lost the % in unchanged verbatim blocks, so rename all
+    s/DOLLARDIF/\$/g;
+#   Convert \begin{SQUAREBRACKET} \end{SQUAREBRACKET} into \[ \]
+    s/\\end\{SQUAREBRACKET\}/\\\]/sg;
+    s/\\begin\{SQUAREBRACKET\}/\\\[/sg;
+# 4. Convert \begin{DOLLARDOLLAR} \end{DOLLARDOLLAR} into $$ $$
+    s/\\begin\{DOLLARDOLLAR\}(.*?)\\end\{DOLLARDOLLAR\}/\$\$$1\$\$/sg;
+# 5. Convert  \SUPERSCRIPTNB{n} into ^n  and  \SUPERSCRIPT{nn} into ^{nnn}
+    1 while s/\\SUPERSCRIPT(\s*\{($pat_n)\})/^$1/g ;
+    1 while s/\\SUPERSCRIPTNB\{(\s*$pat0)\}/^$1/g ;
+    # Convert  \SUBSCRIPNB{n} into _n  and  \SUBCRIPT{nn} into _{nnn}
+    1 while s/\\SUBSCRIPT(\s*\{($pat_n)\})/_$1/g ;
+    1 while s/\\SUBSCRIPTNB\{(\s*$pat0)\}/_$1/g ;
+    # Convert  \SQRT{n} into \sqrt{n}  and  \SQRTNB{nn} into \sqrt nn
+    1 while s/\\SQRT(\s*\{($pat_n)\})/\\sqrt$1/g ;
+    1 while s/\\SQRTNB\{(\s*$pat0)\}/\\sqrt$1/g ;
+
+    1 while s/(%.*)\\CFRAC (.*)$/$1\\frac$2/mg ;
+    1 while s/(%.*)\\CRIGHTBRACE (.*)$/$1\}$2/mg ;
+    1 while s/(%.*)\\CLEFTBRACE (.*)$/$1\{$2/mg ;
+
+
+#    Change \QLEFTBRACE, \QRIGHTBRACE to \{,\}
+    s/\\QLEFTBRACE /\\\{/sg;
+    s/\\QRIGHTBRACE /\\\}/sg;
+    s/\\AMPERSAND /\\&/sg;
+    # Highlight added inline verbatim commands if possible
+    if ( $latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/ )   {
+      # wrap added verb commands with color commands
+      s/\\DIFDIFadd((?:verb\*?|lstinline(?:\[$brat_n\])?)\{[-\d]*?\}[\s\n]*)/\{\\color{blue}$AUXCMD\n\\DIF$1%\n\}$AUXCMD\n/sg;
+      s/\\DIFDIFdel((?:verb\*?|lstinline(?:\[$brat_n\])?)\{[-\d]*?\}[\s\n]*$AUXCMD)/\{\\color{red}${AUXCMD}\n\\DIF$1\n\}${AUXCMD}/sg;
+    } else {
+      # currently if colour markup is not used just remove the added mark
+      s/\\DIFDIFadd(verb\*?|lstinline)/\\DIF$1/sg;
+      s/\\DIFDIFdel((?:verb\*?|lstinline(?:\[$brat_n\])?)\{[-\d]*?\}[\s\n]*$AUXCMD\n)//sg;
+    }
+    # expand \verb and friends inline arguments
+    s/\\DIF((?:DIFadd|DIFdel)?(?:verb\*?|lstinline(?:\[$brat_n\])?))\{([-\d]*?)\}/"\\${1}". fromhash(\%verbhash,$2)/esg;
+    # add basicstyle color{blue} to added lstinline commands
+    # finally add the comment to the ones not having an optional argument before
+    ###s/\\DIFaddlstinline(?!\[)/\\lstinline\n[basicstyle=\\color{blue}]$AUXCMD\n/g;
+
+  }
+  return;
+}
+
+# $out = restore_item_commands($listenviron)
+# short helper function for post-process, which restores deleted \item commands in its argument (as DIFAUXCMDs)
+sub restore_item_commands {
+  my ($string)=@_ ;
+  my ($itemarg,@itemargs);
+  $string =~ s/(\%DIFDELCMD < \s*(\\$ITEMCMD$extraspace)((?:<$abrat0>)?$extraspace)((?:\[$brat_n\])?)\s*((?:${cmdoptseq}\s*?)*)(?:\n|$DELCMDCLOSE))/
+     # if \item has an []argument, then mark up the argument as deleted)
+     if (length($4)>0) {
+       # use substr to exclude square brackets at end points
+       @itemargs=splitlatex(substr($4,1,length($4)-2));
+       $itemarg="[".join("",marktags("","",$DELOPEN,$DELCLOSE,$DELCMDOPEN,$DELCMDCLOSE,$DELCOMMENT,\@itemargs))."]";
+     } else {
+       $itemarg="";
+     }
+     "$1$2$3$itemarg$AUXCMD\n";  ###.((length($5)>0) ? "%DIFDELCMD $5 $DELCMDCLOSE\n" : "")
+     /sge;
+  return($string);
+}
+
+
+# @auxlines=preprocess_preamble($oldpreamble,$newpreamble);
+  # pre-process preamble by looking for commands used in \maketitle (title, author, date etc commands)
+  # the list of commands is defined in CONTEXT2CMD
+  # if found then use a bodydiff to mark up content, and replace the corresponding commands
+  # in both preambles by marked up version to 'fool' the linediff (such that only body is marked up).
+  # A special case are e.g. author commands being added (or removed)
+  # 1. If commands are added, then the entire content is marked up as new, but also the lines are marked as new in the linediff
+  # 2. If commands are removed, then the linediff will mark the line as deleted.  The program returns
+  #    with $auxlines a text to be appended at the end of the preamble, which shows the respective fields as deleted
+sub preprocess_preamble {
+  my ($oldpreambleref,$newpreambleref)=(\$_[0],\$_[1]) ;
+  my @auxlines=();
+  # Remember to use $$oldpreambleref to refer to oldpreamble
+  my ($titlecmd,$titlecmdpat);
+  my (@oldtitlecommands,@newtitlecommands );
+  my  %oldhash  = ();
+  my  %newhash  = ();
+  my ($line,$cmd,$optarg,$arg,$optargnew,$optargold,$optargdiff,$argold,$argnew,$argdiff,$auxline);
+
+  my $warnmsgdetail = <<EOF ;
+     This should not occur for standard styles, but can occur for some specific styles, document classes,
+     e.g. journal house styles.
+     Workaround: Use --replace-context2cmd option to specifically set those commands, which are not repeated.
+EOF
+
+  # resuse context2cmdlist to define these commands to  look out for in preamble
+  $titlecmd = "(?:".join("|",@CONTEXT2CMDLIST).")";
+  # as context2cmdlist is stored as regex, e.g. ((?-xism:^title$), we need to remove ^- fo
+  # resue in a more complex regex
+  $titlecmd =~ s/[\$\^]//g;
+  # make sure to not match on comment lines:
+  $titlecmdpat=qr/^(?:[^%\n]|\\%)*(\\($titlecmd)$extraspace(?:\[($brat_n)\])?(?:\{($pat_n)\}))/ms;
+  ###print STDERR "DEBUG:",$titlecmdpat,"\n";
+  @oldtitlecommands= ( $$oldpreambleref =~ m/$titlecmdpat/g );
+  @newtitlecommands= ( $$newpreambleref =~ m/$titlecmdpat/g );
+
+
+  while ( @oldtitlecommands ) {
+    $line=shift @oldtitlecommands;
+    $cmd=shift @oldtitlecommands;
+    $optarg=shift @oldtitlecommands;
+    $arg=shift @oldtitlecommands;
+
+    if ( defined($oldhash{$cmd})) {
+      warn "WARNING: $cmd is used twice in preamble of old file. Reverting to pure line diff mode for preamble.\n";
+      print STDERR $warnmsgdetail;
+      return;
+    }
+    $oldhash{$cmd}=[ $line, $optarg, $arg ];
+  }
+  while ( @newtitlecommands ) {
+    $line=shift @newtitlecommands;
+    $cmd=shift @newtitlecommands;
+    $optarg=shift @newtitlecommands;
+    $arg=shift @newtitlecommands;
+
+    if ( defined($newhash{$cmd})) {
+      warn "$cmd is used twice in preamble of new file. Reverting to pure line diff mode for preamble.\n";
+      print STDERR $warnmsgdetail;
+      return;
+    }
+    $newhash{$cmd}=[ $line, $optarg, $arg ];
+  }
+  foreach $cmd ( keys %newhash ) {
+    if ( defined($newhash{$cmd}->[1])) {
+       $optargnew=$newhash{$cmd}->[1];
+    } else {
+      $optargnew="";
+    }
+    if ( defined($oldhash{$cmd}->[1])) {
+       $optargold=$oldhash{$cmd}->[1];
+    } else {
+      $optargold="";
+    }
+
+    if ( defined($oldhash{$cmd}->[2]) ) {
+      $argold=$oldhash{$cmd}->[2];
+    } else {
+      $argold="";
+    }
+    $argnew=$newhash{$cmd}->[2];
+    $argdiff="{" . join("",bodydiff($argold,$argnew)) ."}";
+    # Replace \RIGHTBRACE by }
+    $argdiff =~ s/\\RIGHTBRACE/}/g;
+
+    if ( length $optargnew ) {
+      $optargdiff="[".join("",bodydiff($optargold,$optargnew))."]" ;
+      $optargdiff =~ s/\\DIFaddbegin /\\DIFaddbeginFL /g;
+      $optargdiff =~ s/\\DIFaddend /\\DIFaddendFL /g;
+      $optargdiff =~ s/\\DIFadd\{/\\DIFaddFL{/g;
+      $optargdiff =~ s/\\DIFdelbegin /\\DIFdelbeginFL /g;
+      $optargdiff =~ s/\\DIFdelend /\\DIFdelendFL /g;
+      $optargdiff =~ s/\\DIFdel\{/\\DIFdelFL{/g;
+    } else {
+      $optargdiff="";
+    }
+    ### print STDERR "DEBUG s/\\Q$newhash{$cmd}->[0]\\E/\\$cmd$optargdiff$argdiff/s\n";
+    # Note: \Q and \E force literal interpretation of what it between them but allow
+    #      variable interpolation, such that e.g. \title matches just that and not TAB-itle
+    $$newpreambleref=~s/\Q$newhash{$cmd}->[0]\E/\\$cmd$optargdiff$argdiff/s;
+    # replace this in old preamble if necessary
+    if ( defined($oldhash{$cmd}->[0])) {
+      $$oldpreambleref=~s/\Q$oldhash{$cmd}->[0]\E/\\$cmd$optargdiff$argdiff/s ;
+    }
+    ### print STDERR "DEBUG NEW PRE ".$$newpreambleref."\n";
+  }
+
+  foreach $cmd ( keys %oldhash ) {
+    # if this has already been dealt with above can just skip
+    next if defined($newhash{$cmd}) ;
+    $argold=$oldhash{$cmd}->[2];
+    $argdiff="{" . join("",bodydiff($argold,"")) ."}";
+    if ( defined($oldhash{$cmd}->[1])) {
+      $optargold=$oldhash{$cmd}->[1];
+      $optargdiff="[".join("",bodydiff($optargold,""))."]" ;
+      $optargdiff =~ s/\\DIFdelbegin /\\DIFdelbeginFL /g;
+      $optargdiff =~ s/\\DIFdelend /\\DIFdelendFL /g;
+      $optargdiff =~ s/\\DIFdel\{/\\DIFdelFL{/g;
+    } else {
+      $optargdiff="";
+    }
+    $auxline = "\\$cmd$optargdiff$argdiff";
+    $auxline =~s/$/$AUXCMD/sg;
+    push @auxlines,$auxline;
+  }
+  # add auxcmd comment to highlight added lines
+  return(@auxlines);
+}
+
+
+
+# @diffs=linediff(\@seq1, \@seq2)
+# mark up lines like this
+#%DIF mm-mmdnn
+#%< old deleted line(s)
+#%DIF -------
+#%DIF mmann-nn
+#new appended line %<
+#%DIF -------
+# Future extension: mark change explicitly
+# Assumes: traverse_sequence traverses deletions before insertions in changed sequences
+#          all line numbers relative to line 0 (first line of real file)
+sub linediff {
+  my $seq1 = shift ;
+  my $seq2 = shift ;
+
+  my $block = [];
+  my $retseq = [];
+  my @begin=('','',''); # dummy initialisation
+  my $instring ;
+
+  my $discard = sub { @begin=('d',$_[0],$_[1]) unless scalar @$block ;
+                      push(@$block, "%DIF < " . $seq1->[$_[0]]) };
+  my $add = sub { if (! scalar  @$block) {
+		    @begin=('a',$_[0],$_[1]) ;}
+		  elsif ( $begin[0] eq 'd' ) {
+                    $begin[0]='c'; $begin[2]=$_[1];
+		    push(@$block, "%DIF -------") }
+                  push(@$block,  $seq2->[$_[1]] . " %DIF > " ) };
+  my $match = sub { if ( scalar @$block ) {
+                      if ( $begin[0] eq 'd' && $begin[1]!=$_[0]-1) {
+			$instring = sprintf "%%DIF %d-%dd%d",$begin[1],$_[0]-1,$begin[2]; }
+		      elsif ( $begin[0] eq 'a' && $begin[2]!=$_[1]-1) {
+			$instring = sprintf "%%DIF %da%d-%d",$begin[1],$begin[2],$_[1]-1; }
+		      elsif ( $begin[0] eq 'c' ) {
+			$instring = sprintf "%%DIF %sc%s",
+			                     ($begin[1]==$_[0]-1) ? "$begin[1]" : $begin[1]."-".($_[0]-1)  ,
+			                     ($begin[2]==$_[1]-1) ? "$begin[2]" : $begin[2]."-".($_[1]-1)  ; }
+		      else {
+			$instring = sprintf "%%DIF %d%s%d",$begin[1],$begin[0],$begin[2]; }
+		      push @$retseq, $instring,@$block, "%DIF -------" ;
+		      $block = [];
+		    }
+		    push @$retseq, $seq2->[$_[1]]
+		  };
+  # key function: remove multiple spaces (such that insertion or deletion of redundant white space is not reported)
+  my $keyfunc = sub { join("  ",split(" ",shift())) };
+
+  traverse_sequences($seq1,$seq2, { MATCH=>$match, DISCARD_A=>$discard, DISCARD_B=>$add }, $keyfunc );
+  push @$retseq, @$block if scalar @$block;
+
+  return wantarray ? @$retseq : $retseq ;
+}
+
+
+
+# init_regex_arr_data(\@array,"TOKEN INIT")
+# scans DATA file handel for line "%% TOKEN INIT" line
+# then appends each line not beginning with % into array (as a quoted regex)
+# This is used for command lists and configuration variables, but the processing is slightly
+# different:
+# For lists, the regular expression is extended to include beginning (^) and end ($) markers, to require full-string matching
+# For configuration variables (and all others), simply an unadorned list is copied
+sub init_regex_arr_data {
+  my ($arr,$token)=@_;
+  my $copy=0;
+  my ($mode);
+  if ($token =~ m/COMMANDS/ ) {
+    $mode=0;  # Reading command list
+  } else {
+    $mode=1;  # Reading configuration variables
+  }
+
+  while (<DATA>) {
+    if ( m/^%%BEGIN $token\s*$/ ) {
+      $copy=1;
+      next;
+    } elsif ( m/^%%END $token\s*$/ )  {
+      last; }
+    chomp;
+    if ( $mode==0 ) {
+#      print STDERR "DEBUG init_regex_arr_data regex >$_<\n" if ($debug && $copy);
+      push (@$arr,qr/^$_$/) if ( $copy && !/^%/ ) ;
+    } elsif ($mode==1) {
+      push (@$arr,"$_") if ( $copy && !/^%/ ) ;
+    }
+  }
+  seek DATA,0,0;    # rewind DATA handle to file begin
+}
+
+
+# init_regex_arr_ext(\@array,$arg)
+# appends array with regular expressions.
+# if arg is a file name, then read in list of regular expressions from that file
+# (one expression per line)
+# Otherwise treat arg as a comma separated list of regular expressions
+sub init_regex_arr_ext {
+  my ($arr,$arg)=@_;
+  if ( -f $arg ) {
+    init_regex_arr_file($arr,$arg);
+  } else {
+    init_regex_arr_list($arr,$arg);
+  }
+}
+
+# init_regex_arr_file(\@array,$fname)
+# appends array with regular expressions.
+# Read in list of regular expressions from $fname
+# (one expression per line)
+sub init_regex_arr_file {
+  my ($arr,$fname)=@_;
+  open(FILE,"$fname") or die ("Couldn't open $fname: $!");
+  while (<FILE>) {
+    chomp;
+    next if /^\s*#/ || /^\s*%/ || /^\s*$/ ;
+    push (@$arr,qr/^$_$/);
+  }
+  close(FILE);
+}
+
+# init_regex_arr_list(\@array,$arg)
+# appends array with regular expressions.
+# read from comma separated list of regular expressions ($arg)
+sub init_regex_arr_list {
+  my ($arr,$arg)=@_;
+  my $regex;
+  ###    print STDERR "DEBUG init_regex_arr_list arg >$arg<\n" if $debug;
+  foreach $regex (split(qr/(?<!\\),/,$arg)) {
+    $regex =~ s/\\,/,/g;
+    print STDERR "DEBUG init_regex_arr_list regex >$regex<\n" if $debug;
+    push (@$arr,qr/^$regex$/);
+  }
+}
+
+
+#exetime() returns time since last execution of this command
+#exetime(1) resets this time
+my $lasttime=-1;   # global variable for persistence
+sub exetime {
+  my $reset=0;
+  my $retval;
+  if ((scalar @_) >=1) {
+    $reset=shift;
+  }
+  if ($reset) {
+    $lasttime=times();
+  }
+  else {
+    $retval=times()-$lasttime;
+    $lasttime=$lasttime+$retval;
+    return($retval);
+  }
+}
+
+
+sub usage {
+  print STDERR <<"EOF";
+Usage: $0 [options] old.tex new.tex > diff.tex
+
+Compares two latex files and writes tex code to stdout, which has the same format as new.tex but
+has all changes relative to old.tex marked up or commented. Note that old.tex and new.tex need to
+be real files (not pipes or similar) as they are opened twice.
+
+--type=markupstyle
+-t markupstyle         Add code to preamble for selected markup style
+                       Available styles: UNDERLINE CTRADITIONAL TRADITIONAL CFONT FONTSTRIKE INVISIBLE
+                                         CHANGEBAR CCHANGEBAR CULINECHBAR CFONTCHBAR BOLD PDFCOMMENT
+                                         LUAUNDERLINE
+                       [ Default: UNDERLINE ]
+
+--subtype=markstyle
+-s markstyle           Add code to preamble for selected style for bracketing
+                       commands (e.g. to mark changes in  margin)
+                       Available styles: SAFE MARGIN DVIPSCOL COLOR ZLABEL ONLYCHANGEDPAGE (LABEL)*
+                       [ Default: SAFE ]
+                       * LABEL subtype is deprecated
+
+--floattype=markstyle
+-f markstyle           Add code to preamble for selected style which
+                       replace standard marking and markup commands within floats
+                       (e.g., marginal remarks cause an error within floats
+                       so marginal marking can be disabled thus)
+                       Available styles: FLOATSAFE IDENTICAL
+                       [ Default: FLOATSAFE ]
+
+--encoding=enc
+-e enc                 Specify encoding of old.tex and new.tex. Typical encodings are
+                       ascii, utf8, latin1, latin9.  A list of available encodings can be
+                       obtained by executing
+                       perl -MEncode -e 'print join ("\\n",Encode->encodings( ":all" )) ;'
+                       [Default encoding is utf8 unless the first few lines of the preamble contain
+                       an invocation "\\usepackage[..]{inputenc} in which case the
+                       encoding chosen by this command is asssumed. Note that ASCII (standard
+                       latex) is a subset of utf8]
+
+--preamble=file
+-p file                Insert file at end of preamble instead of auto-generating
+                       preamble.  The preamble must define the following commands
+                       \\DIFaddbegin,\\DIFaddend,\\DIFadd{..},
+                       \\DIFdelbegin,\\DIFdelend,\\DIFdel{..},
+                       and varieties for use within floats
+                       \\DIFaddbeginFL,\\DIFaddendFL,\\DIFaddFL{..},
+                       \\DIFdelbeginFL,\\DIFdelendFL,\\DIFdelFL{..}
+                       (If this option is set -t, -s, and -f options
+                       are ignored.)
+
+--exclude-safecmd=exclude-file
+--exclude-safecmd="cmd1,cmd2,..."
+-A exclude-file
+--replace-safecmd=replace-file
+--append-safecmd=append-file
+--append-safecmd="cmd1,cmd2,..."
+-a append-file         Exclude from, replace or append to the list of regex
+                       matching commands which are safe to use within the
+                       scope of a \\DIFadd or \\DIFdel command.  The file must contain
+                       one Perl-RegEx per line (Comment lines beginning with # or % are
+                       ignored). A literal comma within the comma-separated list must be
+                       escaped thus "\\,",   Note that the RegEx needs to match the whole of
+                       the token, i.e., /^regex\$/ is implied and that the initial
+                       "\\" of the command is not included. The --exclude-safecmd
+                       and --append-safecmd options can be combined with the --replace-safecmd
+                       option and can be used repeatedly to add cumulatively to the lists.
+
+--exclude-textcmd=exclude-file
+--exclude-textcmd="cmd1,cmd2,..."
+-X exclude-file
+--replace-textcmd=replace-file
+--append-textcmd=append-file
+--append-textcmd="cmd1,cmd2,..."
+-x append-file         Exclude from, replace or append to the list of regex
+                       matching commands whose last argument is text.  See
+                       entry for --exclude-safecmd directly above for further details.
+
+--replace-context1cmd=replace-file
+--append-context1cmd=append-file
+--append-context1cmd="cmd1,cmd2,..."
+                       Replace or append to the list of regex matching commands
+                       whose last argument is text but which require a particular
+                       context to work, e.g. \\caption will only work within a figure
+                       or table.  These commands behave like text commands, except when
+                       they occur in a deleted section, when they are disabled, but their
+                       argument is shown as deleted text.
+
+--replace-context2cmd=replace-file
+--append-context2cmd=append-file
+--append-context2cmd="cmd1,cmd2,..."
+                       As corresponding commands for context1.  The only difference is that
+                       context2 commands are completely disabled in deleted sections, including
+                       their arguments.
+                       context2 commands are also the only commands in the preamble, whose argument will
+                       be processed in word-by-word mode (which only works, if they occur no more than
+		       once in the preamble).
+
+--exclude-mboxsafecmd=exclude-file
+--exclude-mboxsafecmd="cmd1,cmd2,..."
+--append-mboxsafecmd=append-file
+--append-mboxsafecmd="cmd1,cmd2,..."
+                       Define safe commands, which additionally need to be protected by encapsulating
+                       in an \\mbox{..}. This is sometimes needed to get around incompatibilities
+                       between external packages and the ulem package, which is  used for highlighting
+                       in the default style UNDERLINE as well as CULINECHBAR CFONTSTRIKE
+
+
+
+--config var1=val1,var2=val2,...
+-c var1=val1,..        Set configuration variables.
+-c configfile           Available variables:
+                          ARRENV (RegEx)
+                          COUNTERCMD (RegEx)
+                          FLOATENV (RegEx)
+                          ITEMCMD (RegEx)
+                          LISTENV (RegEx)
+                          MATHARRENV (RegEx)
+                          MATHENV (RegEx)
+                          MATHREPL (String)
+                          MINWORDSBLOCK (Integer)
+                          PICTUREENV (RegEx)
+                          SCALEDELGRAPHICS (Float)
+                          VERBATIMENV (RegEx)
+                          VERBATIMLINEENV (RegEx)
+                          CUSTOMDIFCMD (RegEx)
+                       This option can be repeated.
+
+--add-to-config  varenv1=pattern1,varenv2=pattern2
+                       For configuration variables containing a regular expression (essentially those ending
+                       in ENV, and COUNTERCMD) this provides an alternative way to modify the configuration
+                       variables. Instead of setting the complete pattern, with this option it is possible to add an
+                       alternative pattern. varenv must be one of the variables listed above that take a regular
+                       expression as argument, and pattern is any regular expression (which might need to be
+                       protected from the shell by quotation). Several patterns can be added at once by using semi-colons
+                       to separate them, e.g. --add-to-config "LISTENV=myitemize;myenumerate,COUNTERCMD=endnote"
+
+--packages=pkg1,pkg2,..
+                       Tell latexdiff that .tex file is processed with the packages in list
+                       loaded.  This is normally not necessary if the .tex file includes the
+                       preamble, as the preamble is automatically scanned for \\usepackage commands.
+                       Use of the --packages option disables automatic scanning, so if for any
+                       reason package specific parsing needs to be switched off, use --packages=none.
+                       The following packages trigger special behaviour:
+                       endfloat hyperref amsmath apacite siunitx cleveref glossaries mhchem chemformula/chemmacros
+                       biblatex
+                       [ Default: scan the preamble for \\usepackage commands to determine
+                         loaded packages.]
+
+--show-preamble        Print generated or included preamble commands to stdout.
+
+--show-safecmd         Print list of regex matching and excluding safe commands.
+
+--show-textcmd         Print list of regex matching and excluding commands with text argument.
+
+--show-config          Show values of configuration variables
+
+--show-all             Show all of the above
+
+   NB For all --show commands, no old.tex or new.tex file needs to be given, and no
+      differencing takes place.
+
+Other configuration options:
+
+--allow-spaces         Allow spaces between bracketed or braced arguments to commands
+                       [Default requires arguments to directly follow each other without
+                                intervening spaces]
+
+--math-markup=level    Determine granularity of markup in displayed math environments:
+                      Possible values for level are (both numerical and text labels are acceptable):
+                      off or 0: suppress markup for math environments.  Deleted equations will not
+                               appear in diff file. This mode can be used if all the other modes
+                               cause invalid latex code.
+                      whole or 1: Differencing on the level of whole equations. Even trivial changes
+                               to equations cause the whole equation to be marked changed.  This
+                               mode can be used if processing in coarse or fine mode results in
+                               invalid latex code.
+                      coarse or 2: Detect changes within equations marked up with a coarse
+                               granularity; changes in equation type (e.g.displaymath to equation)
+                               appear as a change to the complete equation. This mode is recommended
+                               for situations where the content and order of some equations are still
+                               being changed. [Default]
+                      fine or 3: Detect small change in equations and mark up and fine granularity.
+                               This mode is most suitable, if only minor changes to equations are
+                               expected, e.g. correction of typos.
+
+--graphics-markup=level   Change highlight style for graphics embedded with \\includegraphics commands
+                      Possible values for level:
+                      none,off or 0: no highlighting for figures
+                      new-only or 1: surround newly added or changed figures with a blue frame [Default]
+                      both or 2:     highlight new figures with a blue frame and show deleted figures
+                                at reduced scale, and crossed out with a red diagonal cross. Use configuration
+                                variable SCALEDELGRAPHICS to set size of deleted figures.
+                      Note that changes to the optional parameters will make the figure appear as changed
+                      to latexdiff, and this figure will thus be highlighted.
+                      In some circumstances "Misplaced \\noalign" errors can occur if there are certain types
+                      of changes in tables. In this case please use option --graphics-markup=none as a
+                      work-around.
+
+--no-del               Suppress deleted text from the diff. It is similar in effect to the BOLD style,
+                       but the deleted text ist not just invisible in the output, it is also not included in the
+                       diff text file. This can be more robust than just making it invisible.
+
+--disable-citation-markup
+--disable-auto-mbox    Suppress citation markup and markup of other vulnerable commands in styles
+                       using ulem (UNDERLINE,FONTSTRIKE, CULINECHBAR)
+                       (the two options are identical and are simply aliases)
+
+--enable-citation-markup
+--enforce-auto-mbox    Protect citation commands and other vulnerable commands in changed sections
+                       with \\mbox command, i.e. use default behaviour for ulem package for other packages
+                       (the two options are identical and are simply aliases)
+
+
+
+Miscellaneous options
+
+--label=label
+-L label               Sets the labels used to describe the old and new files.  The first use
+                       of this option sets the label describing the old file and the second
+                       use of the option sets the label for the new file.
+                       [Default: use the filename and modification dates for the label]
+
+--no-label             Suppress inclusion of old and new file names as comment in output file
+
+--visible-label         Include old and new filenames (or labels set with --label option) as
+                       visible output
+
+--flatten              Replace \\input and \\include commands within body by the content
+                       of the files in their argument.  If \\includeonly is present in the
+                       preamble, only those files are expanded into the document. However,
+                       no recursion is done, i.e. \\input and \\include commands within
+                       included sections are not expanded.  The included files are assumed to
+                       be located in the same directories as the old and new master files,
+                       respectively, making it possible to organise files into old and new directories.
+                       --flatten is applied recursively, so inputted files can contain further
+                       \\input statements.  Also handles files included by the import package
+                       (\\import and \\subimport), and \\subfile command.
+
+--filter-script=filterscript    Run files through this filterscript (full path preferred) before processing.
+                       The filterscript must take STDIN input and output to STDOUT.
+                       When coupled with --flatten, each file will be run through the filter as it is brought in.
+
+--ignore-filter-stderr When running with --filter-script, STDERR from the script may cause readability issues.
+                       Turn this flag on to ignore STDERR from the filter script.
+
+--driver=type          Choose driver for changebar package (only relevant for styles using
+                       changebar: CCHANGEBAR CFONTCHBAR CULINECHBAR CHANGEBAR). Possible
+                       drivers are listed in changebar manual, e.g. pdftex,dvips,dvitops
+                       [Default: pdftex]
+
+--help
+-h                     Show this help text.
+
+--ignore-warnings      Suppress warnings about inconsistencies in length between input
+                       and parsed strings and missing characters.
+
+--verbose
+-V                     Output various status information to stderr during processing.
+                       Default is to work silently.
+
+--version              Show version number.
+
+Internal options:
+These options are mostly for automated use by latexdiff-vc. They can be used directly, but
+the API should be considered less stable than for the other options.
+
+--no-links             Suppress generation of hyperreferences, used for minimal diffs
+                       (option --only-changes of latexdiff-vc).
+EOF
+  exit 0;
+}
+
+=head1 NAME
+
+latexdiff - determine and markup differences between two latex files
+
+=head1 SYNOPSIS
+
+B<latexdiff> [ B<OPTIONS> ] F<old.tex> F<new.tex> > F<diff.tex>
+
+=head1 DESCRIPTION
+
+Briefly, I<latexdiff> is a utility program to aid in the management of
+revisions of latex documents. It compares two valid latex files, here
+called C<old.tex> and C<new.tex>, finds significant differences
+between them (i.e., ignoring the number of white spaces and position
+of line breaks), and adds special commands to highlight the
+differences.  Where visual highlighting is not possible, e.g. for changes
+in the formatting, the differences are
+nevertheless marked up in the source. Note that old.tex and new.tex need to
+be real files (not pipes or similar) as they are opened twice (unless C<--encoding> option is used)
+
+The program treats the preamble differently from the main document.
+Differences between the preambles are found using line-based
+differencing (similarly to the Unix diff command, but ignoring white
+spaces).  A comment, "S<C<%DIF E<gt>>>" is appended to each added line, i.e. a
+line present in C<new.tex> but not in C<old.tex>.  Discarded lines
+ are deactivated by prepending "S<C<%DIF E<lt>>>". Changed blocks are preceded  by
+comment lines giving information about line numbers in the original files.  Where there are insignificant
+differences, the resulting file C<diff.tex> will be similar to
+C<new.tex>.  At the end of the preamble, the definitions for I<latexdiff> markup commands are inserted.
+In differencing the main body of the text, I<latexdiff> attempts to
+satisfy the following guidelines (in order of priority):
+
+=over 3
+
+=item 1
+
+If both C<old.tex> and C<new.tex> are valid LaTeX, then the resulting
+C<diff.tex> should also be valid LateX. (NB If a few plain TeX commands
+are used within C<old.tex> or C<new.tex> then C<diff.tex> is not
+guaranteed to work but usually will).
+
+=item 2
+
+Significant differences are determined on the level of
+individual words. All significant differences, including differences
+between comments should be clearly marked in the resulting source code
+C<diff.tex>.
+
+=item 3
+
+If a changed passage contains text or text-producing commands, then
+running C<diff.tex> through LateX should produce output where added
+and discarded passages are highlighted.
+
+=item 4
+
+Where there are insignificant differences, e.g. in the positioning of
+line breaks, C<diff.tex> should follow the formatting of C<new.tex>
+
+=back
+
+For differencing the same algorithm as I<diff> is used but words
+instead of lines are compared.  An attempt is made to recognize
+blocks which are completely changed such that they can be marked up as a unit.
+Comments are differenced line by line
+but the number of spaces within comments is ignored. Commands including
+all their arguments are generally compared as one unit, i.e., no mark-up
+is inserted into the arguments of commands.  However, for a selected
+number of commands (for example, C<\caption> and all sectioning
+commands) the last argument is known to be text. This text is
+split into words and differenced just as ordinary text (use options to
+show and change the list of text commands, see below). As the
+algorithm has no detailed knowledge of LaTeX, it assumes all pairs of
+curly braces immediately following a command (i.e. a sequence of
+letters beginning with a backslash) are arguments for that command.
+As a restriction to condition 1 above it is thus necessary to surround
+all arguments with curly braces, and to not insert
+extraneous spaces.  For example, write
+
+  \section{\textem{This is an emphasized section title}}
+
+and not
+
+  \section {\textem{This is an emphasized section title}}
+
+or
+
+  \section\textem{This is an emphasized section title}
+
+even though all varieties are the same to LaTeX (but see
+B<--allow-spaces> option which allows the second variety).
+
+For environments whose content does not conform to standard LaTeX or
+where graphical markup does not make sense all markup commands can be
+removed by setting the PICTUREENV configuration variable, set by
+default to C<picture> and C<DIFnomarkup> environments; see B<--config>
+option).  The latter environment (C<DIFnomarkup>) can be used to
+protect parts of the latex file where the markup results in illegal
+markup. You have to surround the offending passage in both the old and
+new file by C<\begin{DIFnomarkup}> and C<\end{DIFnomarkup}>. You must
+define the environment in the preambles of both old and new
+documents. I prefer to define it as a null-environment,
+
+C<\newenvironment{DIFnomarkup}{}{}>
+
+but the choice is yours.  Any markup within the environment will be
+removed, and generally everything within the environment will just be
+taken from the new file.
+
+It is also possible to difference files which do not have a preamble.
+ In this case, the file is processed in the main document
+mode, but the definitions of the markup commands are not inserted.
+
+All markup commands inserted by I<latexdiff> begin with "C<\DIF>".  Added
+blocks containing words, commands or comments which are in C<new.tex>
+but not in C<old.tex> are marked by C<\DIFaddbegin> and C<\DIFaddend>.
+Discarded blocks are marked by C<\DIFdelbegin> and C<\DIFdelend>.
+Within added blocks all text is highlighted with C<\DIFadd> like this:
+C<\DIFadd{Added text block}>
+Selected `safe' commands can be contained in these text blocks as well
+(use options to show and change the list of safe commands, see below).
+All other commands as well as braces "{" and "}" are never put within
+the scope of C<\DIFadd>.  Added comments are marked by prepending
+"S<C<%DIF E<gt> >>".
+
+Within deleted blocks text is highlighted with C<\DIFdel>.  Deleted
+comments are marked by prepending "S<C<%DIF E<lt> >>".  Non-safe command
+and curly braces within deleted blocks are commented out with
+"S<C<%DIFDELCMD E<lt> >>".
+
+
+
+=head1 OPTIONS
+
+=head2 Preamble
+
+The following options determine the visual markup style by adding the appropriate
+command definitions to the preamble. See the end of this section for a description of
+available styles.
+
+=over 4
+
+=item B<--type=markupstyle> or
+B<-t markupstyle>
+
+Add code to preamble for selected markup style. This option defines
+C<\DIFadd> and C<\DIFdel> commands.
+Available styles:
+
+C<UNDERLINE CTRADITIONAL TRADITIONAL CFONT FONTSTRIKE INVISIBLE
+CHANGEBAR CCHANGEBAR CULINECHBAR CFONTCHBAR BOLD PDFCOMMENT
+LUAUNDERLINE>
+
+[ Default: C<UNDERLINE> ]
+
+=item B<--subtype=markstyle> or
+B<-s markstyle>
+
+Add code to preamble for selected style for bracketing
+commands (e.g. to mark changes in  margin). This option defines
+C<\DIFaddbegin>, C<\DIFaddend>, C<\DIFdelbegin> and C<\DIFdelend> commands.
+Available styles: C<SAFE MARGIN COLOR DVIPSCOL  ZLABEL ONLYCHANGEDPAGE (LABEL)*>
+
+[ Default: C<SAFE> ]
+* Subtype C<LABEL> is deprecated
+
+=item B<--floattype=markstyle> or
+B<-f markstyle>
+
+Add code to preamble for selected style which
+replace standard marking and markup commands within floats
+(e.g., marginal remarks cause an error within floats
+so marginal marking can be disabled thus). This option defines all
+C<\DIF...FL> commands.
+Available styles: C<FLOATSAFE TRADITIONALSAFE IDENTICAL>
+
+[ Default: C<FLOATSAFE> ]
+
+=item B<--encoding=enc> or
+B<-e enc>
+
+Specify encoding of old.tex and new.tex. Typical encodings are
+C<ascii>, C<utf8>, C<latin1>, C<latin9>.  A list of available encodings can be
+obtained by executing
+
+C<perl -MEncode -e 'print join ("\n",Encode->encodings( ":all" )) ;' >
+
+If this option is used, then old.tex, new.tex are only opened once.
+[Default encoding is utf8 unless the first few lines of the preamble contain
+an invocation C<\usepackage[..]{inputenc}> in which case the
+encoding chosen by this command is asssumed. Note that ASCII (standard
+latex) is a subset of utf8]
+
+=item B<--preamble=file> or
+B<-p file>
+
+Insert file at end of preamble instead of generating
+preamble.  The preamble must define the following commands
+C<\DIFaddbegin, \DIFaddend, \DIFadd{..},
+\DIFdelbegin,\DIFdelend,\DIFdel{..},>
+and varieties for use within floats
+C<\DIFaddbeginFL, \DIFaddendFL, \DIFaddFL{..},
+\DIFdelbeginFL, \DIFdelendFL, \DIFdelFL{..}>
+(If this option is set B<-t>, B<-s>, and B<-f> options
+are ignored.)
+
+=item B<--packages=pkg1,pkg2,..>
+
+Tell latexdiff that .tex file is processed with the packages in list
+loaded.  This is normally not necessary if the .tex file includes the
+preamble, as the preamble is automatically scanned for C<\usepackage> commands.
+Use of the B<--packages> option disables automatic scanning, so if for any
+reason package specific parsing needs to be switched off, use B<--packages=none>.
+The following packages trigger special behaviour:
+
+=over 8
+
+=item C<endfloat>
+
+Ensure that C<\begin{figure}> and C<\end{figure}> always appear by themselves on a line.
+
+=item C<hyperref>
+
+Change name of C<\DIFadd> and C<\DIFdel> commands to C<\DIFaddtex> and C<\DIFdeltex> and
+define new C<\DIFadd> and C<\DIFdel> commands, which provide a wrapper for these commands,
+using them for the text but not for the link defining command (where any markup would cause
+errors).
+
+=item C<apacite>, C<biblatex>
+
+Redefine the commands recognised as citation commands.
+
+=item C<siunitx>
+
+Treat C<\SI> as equivalent to citation commands (i.e. protect with C<\mbox> if markup style uses ulem package.
+
+=item C<cleveref>
+
+Treat C<\cref,\Cref>, etc as equivalent to citation commands (i.e. protect with C<\mbox> if markup style uses ulem package.
+
+=item C<glossaries>
+
+Define most of the glossaries commands as safe, protecting them with \mbox'es where needed
+
+=item C<mhchem>
+
+Treat C<\ce> as a safe command, i.e. it will be highlighted (note that C<\cee> will not be highlighted in equations as this leads to processing errors)
+
+=item C<chemformula> or C<chemmacros>
+
+Treat C<\ch> as a safe command outside equations, i.e. it will be highlighted (note that C<\ch> will not be highlighted in equations as this leads to processing errors)
+
+
+=back
+
+[ Default: scan the preamble for C<\usepackage> commands to determine
+  loaded packages. ]
+
+
+
+=item B<--show-preamble>
+
+Print generated or included preamble commands to stdout.
+
+=back
+
+=head2 Configuration
+
+=over 4
+
+=item B<--exclude-safecmd=exclude-file> or
+B<-A exclude-file> or  B<--exclude-safecmd="cmd1,cmd2,...">
+
+=item B<--replace-safecmd=replace-file>
+
+=item B<--append-safecmd=append-file> or
+B<-a append-file> or B<--append-safecmd="cmd1,cmd2,...">
+
+Exclude from, replace or append to the list of regular expressions (RegEx)
+matching commands which are safe to use within the
+scope of a C<\DIFadd> or C<\DIFdel> command.  The file must contain
+one Perl-RegEx per line (Comment lines beginning with # or % are
+ignored).  Note that the RegEx needs to match the whole of
+the token, i.e., /^regex$/ is implied and that the initial
+"\" of the command is not included.
+The B<--exclude-safecmd> and B<--append-safecmd> options can be combined with the -B<--replace-safecmd>
+option and can be used repeatedly to add cumulatively to the lists.
+ B<--exclude-safecmd>
+and B<--append-safecmd> can also take a comma separated list as input. If a
+comma for one of the regex is required, escape it thus "\,". In most cases it
+will be necessary to protect the comma-separated list from the shell by putting
+it in quotation marks.
+
+=item B<--exclude-textcmd=exclude-file> or
+B<-X exclude-file> or B<--exclude-textcmd="cmd1,cmd2,...">
+
+=item B<--replace-textcmd=replace-file>
+
+=item B<--append-textcmd=append-file> or
+B<-x append-file> or B<--append-textcmd="cmd1,cmd2,...">
+
+Exclude from, replace or append to the list of regular expressions
+matching commands whose last argument is text.  See
+entry for B<--exclude-safecmd> directly above for further details.
+
+
+=item B<--replace-context1cmd=replace-file>
+
+=item B<--append-context1cmd=append-file> or
+
+=item B<--append-context1cmd="cmd1,cmd2,...">
+
+Replace or append to the list of regex matching commands
+whose last argument is text but which require a particular
+context to work, e.g. C<\caption> will only work within a figure
+or table.  These commands behave like text commands, except when
+they occur in a deleted section, when they are disabled, but their
+argument is shown as deleted text.
+
+=item B<--replace-context2cmd=replace-file>
+
+=item B<--append-context2cmd=append-file> or
+
+=item B<--append-context2cmd="cmd1,cmd2,...">
+
+As corresponding commands for context1.  The only difference is that
+context2 commands are completely disabled in deleted sections, including
+their arguments.
+
+context2 commands are also the only commands in the preamble, whose argument will be processed in
+word-by-word mode (which only works, if they occur no more than once in the preamble). The algorithm currently cannot cope with repeated context2 commands in the preamble, as they occur e.g. for the C<\author> argument in some journal styles (not in the standard styles, though
+If such a repetition is detected, the whole preamble will be processed in line-by-line mode. In such a case, use C<--replace-context2cmd> option to just select the commands, which should be processed and are not used repeatedly in the preamble.
+
+
+
+=item B<--exclude-mboxsafecmd=exclude-file> or B<--exclude-mboxsafecmd="cmd1,cmd2,...">
+
+=item B<--append-mboxsafecmd=append-file> or B<--append-mboxsafecmd="cmd1,cmd2,...">
+
+Define safe commands, which additionally need to be protected by encapsulating
+in an C<\mbox{..}>. This is sometimes needed to get around incompatibilities
+between external packages and the ulem package, which is  used for highlighting
+in the default style UNDERLINE as well as CULINECHBAR CFONTSTRIKE
+
+
+
+
+
+=item B<--config var1=val1,var2=val2,...> or B<-c var1=val1,..>
+
+=item B<-c configfile>
+
+Set configuration variables.  The option can be repeated to set different
+variables (as an alternative to the comma-separated list).
+Available variables (see below for further explanations):
+
+C<ARRENV> (RegEx)
+
+C<COUNTERCMD> (RegEx)
+
+C<CUSTODIFCMD> (RegEx)
+
+C<FLOATENV> (RegEx)
+
+C<ITEMCMD> (RegEx)
+
+C<LISTENV>  (RegEx)
+
+C<MATHARRENV> (RegEx)
+
+C<MATHENV> (RegEx)
+
+C<MATHREPL> (String)
+
+C<MINWORDSBLOCK> (Integer)
+
+C<PICTUREENV> (RegEx)
+
+C<SCALEDELGRAPHICS> (Float)
+
+
+=item B<--add-to-config varenv1=pattern1,varenv2=pattern2,...>
+
+For configuration variables, which are a regular expression (essentially those ending
+in ENV, COUNTERCMD and CUSTOMDIFCMD, see list above) this option provides an alternative way to modify the configuration
+variables. Instead of setting the complete pattern, with this option it is possible to add an
+alternative pattern. C<varenv> must be one of the variables listed above that take a regular
+expression as argument, and pattern is any regular expression (which might need to be
+protected from the shell by quotation). Several patterns can be added at once by using semi-colons
+to separate them, e.g. C<--add-to-config "LISTENV=myitemize;myenumerate,COUNTERCMD=endnote">
+
+=item B<--show-safecmd>
+
+Print list of RegEx matching and excluding safe commands.
+
+=item B<--show-textcmd>
+
+Print list of RegEx matching and excluding commands with text argument.
+
+=item B<--show-config>
+
+Show values of configuration variables.
+
+=item B<--show-all>
+
+Combine all --show commands.
+
+NB For all --show commands, no C<old.tex> or C<new.tex> file needs to be specified, and no
+differencing takes place.
+
+=back
+
+=head2 Other configuration options:
+
+=over 4
+
+=item B<--allow-spaces>
+
+Allow spaces between bracketed or braced arguments to commands.  Note
+that this option might have undesirable side effects (unrelated scope
+might get lumpeded with preceding commands) so should only be used if the
+default produces erroneous results.  (Default requires arguments to
+directly follow each other without intervening spaces).
+
+=item B<--math-markup=level>
+
+Determine granularity of markup in displayed math environments:
+Possible values for level are (both numerical and text labels are acceptable):
+
+C<off> or C<0>: suppress markup for math environments.  Deleted equations will not
+appear in diff file. This mode can be used if all the other modes
+cause invalid latex code.
+
+C<whole> or C<1>: Differencing on the level of whole equations. Even trivial changes
+to equations cause the whole equation to be marked changed.  This
+mode can be used if processing in coarse or fine mode results in
+invalid latex code.
+
+C<coarse> or C<2>: Detect changes within equations marked up with a coarse
+granularity; changes in equation type (e.g.displaymath to equation)
+appear as a change to the complete equation. This mode is recommended
+for situations where the content and order of some equations are still
+being changed. [Default]
+
+C<fine> or C<3>: Detect small change in equations and mark up at fine granularity.
+This mode is most suitable, if only minor changes to equations are
+expected, e.g. correction of typos.
+
+=item B<--graphics-markup=level>
+
+ Change highlight style for graphics embedded with C<\includegraphics> commands.
+
+Possible values for level:
+
+C<none>, C<off> or C<0>: no highlighting for figures
+
+C<new-only> or C<1>: surround newly added or changed figures with a blue frame [Default if graphicx package loaded]
+
+C<both> or C<2>:     highlight new figures with a blue frame and show deleted figures at reduced
+scale, and crossed out with a red diagonal cross. Use configuration
+variable SCALEDELGRAPHICS to set size of deleted figures.
+
+Note that changes to the optional parameters will make the figure appear as changed
+to latexdiff, and this figure will thus be highlighted.
+
+In some circumstances "Misplaced \noalign" error can occur if there are certain types
+of changes in tables. In this case please use C<--graphics-markup=none> as a
+work-around.
+
+
+=item B<--no-del>
+
+Suppress deleted text from the diff. It is similar in effect to the BOLD style,
+but the deleted text ist not just invisible in the output, it is also not included in the diff text file.
+This can be more robust than just making it invisible.
+
+=item B<--disable-citation-markup> or B<--disable-auto-mbox>
+
+Suppress citation markup and markup of other vulnerable commands in styles
+using ulem (UNDERLINE,FONTSTRIKE, CULINECHBAR)
+(the two options are identical and are simply aliases)
+
+=item B<--enable-citation-markup> or B<--enforce-auto-mbox>
+
+Protect citation commands and other vulnerable commands in changed sections
+with C<\mbox> command, i.e. use default behaviour for ulem package for other packages
+(the two options are identical and are simply aliases)
+
+=back
+
+=head2 Miscellaneous
+
+=over 4
+
+=item B<--verbose> or B<-V>
+
+Output various status information to stderr during processing.
+Default is to work silently.
+
+=item B<--driver=type>
+
+Choose driver for changebar package (only relevant for styles using
+   changebar: CCHANGEBAR CFONTCHBAR CULINECHBAR CHANGEBAR). Possible
+drivers are listed in changebar manual, e.g. pdftex,dvips,dvitops
+  [Default: pdftex]
+
+=item B<--ignore-warnings>
+
+Suppress warnings about inconsistencies in length between input and
+parsed strings and missing characters.  These warning messages are
+often related to non-standard latex or latex constructions with a
+syntax unknown to C<latexdiff> but the resulting difference argument
+is often fully functional anyway, particularly if the non-standard
+latex only occurs in parts of the text which have not changed.
+
+=item B<--label=label> or
+B<-L label>
+
+Sets the labels used to describe the old and new files.  The first use
+of this option sets the label describing the old file and the second
+use of the option sets the label for the new file, i.e. set both
+labels like this C<-L labelold -L labelnew>.
+[Default: use the filename and modification dates for the label]
+
+=item B<--no-label>
+
+Suppress inclusion of old and new file names as comment in output file
+
+=item B<--visible-label>
+
+Include old and new filenames (or labels set with C<--label> option) as
+visible output.
+
+=item B<--flatten>
+
+Replace C<\input> and C<\include> commands within body by the content
+of the files in their argument.  If C<\includeonly> is present in the
+preamble, only those files are expanded into the document. However,
+no recursion is done, i.e. C<\input> and C<\include> commands within
+included sections are not expanded.  The included files are assumed to
+ be located in the same directories as the old and new master files,
+respectively, making it possible to organise files into old and new directories.
+--flatten is applied recursively, so inputted files can contain further
+C<\input> statements.  Also handles files included by the import package
+(C<\import> and C<\subimport>), and C<\subfile> command.
+
+Use of this option might result in prohibitive processing times for
+larger documents, and the resulting difference document
+no longer reflects the structure of the input documents.
+
+=item B<--filter-script=filterscript>
+
+Run files through this filterscript (full path preferred) before processing.
+The filterscript must take STDIN input and output to STDOUT.
+When coupled with --flatten, each file will be run through the filter as it is brought in.
+
+=item B<--ignore-filter-stderr>
+
+When running with --filter-script, STDERR from the script may cause readability issues.
+Turn this flag on to ignore STDERR from the filter script.
+
+
+
+=item B<--help> or
+B<-h>
+
+Show help text
+
+=item B<--version>
+
+Show version number
+
+=back
+
+
+=head2 Internal options
+
+These options are mostly for automated use by latexdiff-vc. They can be used directly, but the API should be considered less stable than for the other options.
+
+=over 4
+
+=item B<--no-links>
+
+Suppress generation of hyperreferences, used for minimal diffs (option --only-changes of latexdiff-vc)
+
+=back
+
+
+=head2 Predefined styles
+
+=head2 Major types
+
+The major type determine the markup of plain text and some selected latex commands outside floats by defining the markup commands C<\DIFadd{...}> and C<\DIFdel{...}> .
+
+=over 10
+
+=item C<UNDERLINE>
+
+Added text is wavy-underlined and blue, discarded text is struck out and red
+(Requires color and ulem packages).  Overstriking does not work in displayed math equations such that deleted parts of equation are underlined, not struck out (this is a shortcoming inherent to the ulem package).
+
+=item C<LUAUNDERLINE>
+
+Added text is underlined and blue, discarded text is struck out and red
+(Requires lua-ul package + LuaLaTeX).
+
+=item C<CTRADITIONAL>
+
+Added text is blue and set in sans-serif, and a red footnote is created for each discarded
+piece of text. (Requires color package)
+
+=item C<TRADITIONAL>
+
+Like C<CTRADITIONAL> but without the use of color.
+
+=item C<CFONT>
+
+Added text is blue and set in sans-serif, and discarded text is red and very small size.
+
+=item C<FONTSTRIKE>
+
+Added tex is set in sans-serif, discarded text small and struck out
+
+=item C<CCHANGEBAR>
+
+Added text is blue, and discarded text is red.  Additionally, the changed text is marked with a bar in the margin (Requires color and changebar packages).
+
+=item C<CFONTCHBAR>
+
+Like C<CFONT> but with additional changebars (Requires color and changebar packages).
+
+=item C<CULINECHBAR>
+
+Like C<UNDERLINE> but with additional changebars (Requires color, ulem and changebar packages).
+
+=item C<CHANGEBAR>
+
+No mark up of text, but mark margins with changebars (Requires changebar package).
+
+=item C<INVISIBLE>
+
+No visible markup (but generic markup commands will still be inserted.
+
+=item C<BOLD>
+
+Added text is set in bold face, discarded is not shown. (also see --no-del option for another possibility to hide deleted text)
+
+=item C<PDFCOMMENT>
+
+The pdfcomment package is used to underline new text, and mark deletions with a PDF comment. Note that this markup might appear differently or not at all based on the pdf viewer used. The viewer with best support for pdf markup is probably acroread. This style is only recommended if the number of differences is small.
+
+=back
+
+=head2 Subtypes
+
+The subtype defines the commands that are inserted at the begin and end of added or discarded blocks, irrespectively of whether these blocks contain text or commands (Defined commands: C<\DIFaddbegin, \DIFaddend, \DIFdelbegin, \DIFdelend>)
+
+=over 10
+
+=item C<SAFE>
+
+No additional markup (Recommended choice)
+
+=item C<MARGIN>
+
+Mark beginning and end of changed blocks with symbols in the margin nearby (using
+the standard C<\marginpar> command - note that this sometimes moves somewhat
+from the intended position.
+
+=item C<COLOR>
+
+An alternative way of marking added passages in blue, and deleted ones in red.
+(It is recommeneded to use instead the main types to effect colored markup,
+although in some cases coloring with dvipscol can be more complete, for example
+with citation commands).
+
+=item C<DVIPSCOL>
+
+An alternative way of marking added passages in blue, and deleted ones in red. Note
+that C<DVIPSCOL> only works with the dvips converter, e.g. not pdflatex.
+(it is recommeneded to use instead the main types to effect colored markup,
+although in some cases coloring with dvipscol can be more complete).
+
+
+=item C<ZLABEL>
+
+can be used to highlight only changed pages, but requires post-processing. It is recommend to not call this option manually but use C<latexdiff-vc> with C<--only-changes> option. Alternatively, use the script given within preamble of diff files made using this style.
+
+=item C<ONLYCHANGEDPAGE>
+
+also highlights changed pages, without the need for post-processing, but might not work reliably if
+there is floating material (figures, tables).
+
+=item C<LABEL>
+
+is similar to C<ZLABEL>, but does not need the zref package and works less reliably (deprecated).
+
+=back
+
+=head2 Float Types
+
+Some of the markup used in the main text might cause problems when used within
+floats (e.g. figures or tables).  For this reason alternative versions of all
+markup commands are used within floats. The float type defines these alternative commands.
+
+=over 10
+
+=item C<FLOATSAFE>
+
+Use identical markup for text as in the main body, but set all commands marking the begin and end of changed blocks to null-commands.  You have to choose this float type if your subtype is C<MARGIN> as C<\marginpar> does not work properly within floats.
+
+=item C<TRADITIONALSAFE>
+
+Mark additions the same way as in the main text.  Deleted environments are marked by angular brackets \[ and \] and the deleted text is set in scriptscript size. This float type should always be used with the C<TRADITIONAL> and  C<CTRADITIONAL> markup types as the \footnote command does not work properly in floating environments.
+
+=item C<IDENTICAL>
+
+Make no difference between the main text and floats.
+
+=back
+
+
+=head2 Configuration Variables
+
+=over 10
+
+=item C<ARRENV>
+
+If a match to C<ARRENV> is found within an inline math environment within a deleted or added block, then the inlined math
+is surrounded by C<\mbox{>...C<}>.  This is necessary as underlining does not work within inlined array environments.
+
+[ Default: C<ARRENV>=S<C<(?:array|[pbvBV]matrix)> >
+
+=item C<COUNTERCMD>
+
+If a command in a deleted block which is also in the textcmd list matches C<COUNTERCMD> then an
+additional command C<\addtocounter{>F<cntcmd>C<}{-1}>, where F<cntcmd> is the matching command, is appended in the diff file such that the numbering in the diff file remains synchronized with the
+numbering in the new file.
+
+[ Default: C<COUNTERCMD>=C<(?:footnote|part|section|subsection> ...
+
+C<|subsubsection|paragraph|subparagraph)>  ]
+
+=item C<CUSTOMDIFCMD>
+
+This option is for advanced users and allows definition of special versions of commands, which do not work as safe commands.
+
+Commands in C<CUSTOMDIFCMD> that occur in added or deleted blocks will be given an ADD or DEL prefix.
+The prefixed versions of the command must be defined in the preamble, either by putting them
+in the preamble of at least the new file, or by creating a custom preamble file (Option --preamble).
+For example the command C<\blindtext> (from package blindtext) does not interact well with underlining, so that
+for the standard markup type, it is not satisfactory to define it as a safe command. Instead, a customised versions
+without underlining can be defined in the preamble:
+
+C<\newcommand{\DELblindtext}{{\color{red}\blindtext}}>
+
+C<\newcommand{\ADDblindtext}{{\color{blue}\blindtext}}>
+
+and then latexdiff should be invoked with the option C<-c CUSTOMDIFCMD=blindtext>.
+
+[ Default: none ]
+
+=item C<FLOATENV>
+
+Environments whose name matches the regular expression in C<FLOATENV> are
+considered floats.  Within these environments, the I<latexdiff> markup commands
+are replaced by their FL variaties.
+
+[ Default: S<C<(?:figure|table|plate)[\w\d*@]*> >]
+
+=item C<ITEMCMD>
+
+Commands representing new item line with list environments.
+
+[ Default: \C<item> ]
+
+=item C<LISTENV>
+
+Environments whose name matches the regular expression in C<LISTENV> are list environments.
+
+[ Default: S<C<(?:itemize|enumerate|description)> >]
+
+=item C<MATHENV>,C<MATHREPL>
+
+If both \begin and \end for a math environment (environment name matching C<MATHENV> or \[ and \])
+are within the same deleted block, they are replaced by a \begin and \end commands for C<MATHREPL>
+rather than being commented out.
+
+[ Default: C<MATHENV>=S<C<(?:displaymath|equation)> >, C<MATHREPL>=S<C<displaymath> >]
+
+=item C<MINWORDSBLOCK>
+
+Minimum number of tokens required to form an independent block. This value is
+used in the algorithm to detect changes of complete blocks by merging identical text parts of less than C<MINWORDSBLOCK> to the preceding added and discarded parts.
+
+[ Default: 3 ]
+
+=item C<PICTUREENV>
+
+Within environments whose name matches the regular expression in C<PICTUREENV>
+all latexdiff markup is removed (in pathologic cases this might lead to
+inconsistent markup but this situation should be rare).
+
+[ Default: S<C<(?:picture|DIFnomarkup)[\w\d*@]*> >]
+
+=item C<SCALEDELGRAPHICS>
+
+If C<--graphics-markup=both> is chosen, C<SCALEDELGRAPHICS> is the factor, by which deleted figures will be scaled (i.e. 0.5 implies they are shown at half linear size).
+
+[ Default: 0.5 ]
+
+=item C<VERBATIMENV>
+
+RegEx describing environments like verbatim, whose contents should be taken verbatim. The content of these environments will not be processed in any way:
+deleted content is commented out, new content is not marked up
+
+[ Default:  S<C<comment> > ]
+
+=item C<VERBATIMLINEENV>
+
+RegEx describing environments like verbatim, whose contents should be taken verbatim. The content of environments described by VERBATIMLINEENV are compared in
+line mode, and changes are marked up using the listings package. The markup style is set based on the chosen mains markup type (Option -t), or on an analysis
+of the preamble.
+Note that "listings.sty" must be installed. If this file is not found the fallback solution is to
+treat VERBATIMLINEENV environments treated exactly the same way as VERBATIMENV environments.
+
+[ Default:  S<C<(?:verbatim[*]?|lstlisting> > ]
+
+=back
+
+=head1 COMMON PROBLEMS AND FAQ
+
+=over 10
+
+=item Changed citations result in overfull boxes
+
+There is an incompatibility between the C<ulem> package, which C<latexdiff> uses for underlining and striking out in the UNDERLINE style,
+the default style, and the way citations are generated. In order to be able to mark up citations properly, they are enclosed with an C<\mbox>
+command. As mboxes cannot be broken across lines, this procedure frequently results in overfull boxes, possibly obscuring the content as it extends beyond the right margin.  The same occurs for some other packages (e.g., siunitx). If this is a problem, you have several possibilities.
+
+1. Use C<CFONT> type markup (option C<-t CFONT>): If this markup is chosen, then changed citations are no longer marked up
+with the wavy line (additions) or struck out (deletions), but are still highlighted in the appropriate color, and deleted text is shown with a different font. Other styles not using the C<ulem> package will also work.
+
+2. Choose option C<--disable-citation-markup> which turns off the marking up of citations: deleted citations are no longer shown, and
+added citations are shown without markup. (This was the default behaviour of latexdiff at versions 0.6 and older).
+For custom packages you can define the commands which need to be protected by C<\mbox> with C<--append-mboxsafecmd> and C<--excludemboxsafecmd> options
+(submit your lists of command as feature request at github page to set the default behaviour of future versions, see section 6)
+
+3. If you are using luatex >= 1.12.0 you can use option LUAUNDERLINE that uses lua-ul instead of ulem for underlining, which does not have this problem (experimental feature).
+
+=item Changes in complicated mathematical equations result in latex processing errors.
+
+Try option C<--math-markup=whole>.   If even that fails, you can turn off mark up for equations with C<--math-markup=off>.
+
+=item Deleted parts in equations are not struck out but underlined.
+
+This is a limitation of the ulem package that implements the strike-out. If you use the amsmath package, then the strike-out command is redefined in such a way that deleted passages are wrapped with C<\text> command; adding C<\usepackage{amsmath}> to your preamble will trigger this behaviour. (If amsmath is not included directly, but imported by another package, latexdiff will not be able to detect its availability; in this case you can give latexdiff a hint with option C<--packages=amsmath>.
+
+
+=item How can I just show the pages where changes had been made?
+
+Use options C<--s ZLABEL>  (some postprocessing required) or C<-s ONLYCHANGEDPAGE>. C<latexdiff-vc --ps|--pdf> with C<--only-changes> option takes care of
+the post-processing for you (requires zref package to be installed).
+
+=item The character encoding breaks when running latexdiff from powershell.
+
+This problem is not limited to C<latexdiff> and has to do with the default settings of C<powershell> in Windows. It is recommended to use C<cmd> instead.
+
+=back
+
+=head1 BUGS
+
+=over 10
+
+=item Option allow-spaces not implemented entirely consistently. It breaks
+the rules that number and type of white space does not matter, as
+different numbers of inter-argument spaces are treated as significant.
+
+=back
+
+Please submit bug reports using the issue tracker of the github repository page I<https://github.com/ftilmann/latexdiff.git>,
+or send them to I<tilmann -- AT -- gfz-potsdam.de>.  Include the version number of I<latexdiff>
+(from comments at the top of the source or use B<--version>).  If you come across latex
+files that are error-free and conform to the specifications set out
+above, and whose differencing still does not result in error-free
+latex, please send me those files, ideally edited to only contain the
+offending passage as long as that still reproduces the problem. If your
+file relies on non-standard class files, you must include those.  I will not
+look at examples where I have trouble to latex the original files.
+
+=head1 SEE ALSO
+
+L<latexrevise>, L<latexdiff-vc>
+
+=head1 PORTABILITY
+
+I<latexdiff> does not make use of external commands and thus should run
+on any platform  supporting Perl 5.6 or higher.  If files with encodings
+other than ASCII or UTF-8 are processed, Perl 5.8 or higher is required.
+
+The standard version of I<latexdiff> requires installation of the Perl package
+C<Algorithm::Diff> (available from I<www.cpan.org> -
+I<http://search.cpan.org/~nedkonz/Algorithm-Diff-1.15>) but a stand-alone
+version, I<latexdiff-so>, which has this package inlined, is available, too.
+I<latexdiff-fast> requires the I<diff> command to be present.
+
+=head1 AUTHOR
+
+Version 1.3.4
+Copyright (C) 2004-2024 Frederik Tilmann
+
+This program is free software; you can redistribute it and/or modify
+it under the terms of the GNU General Public License Version 3
+
+Contributors of fixes and additions: V. Kuhlmann, J. Paisley, N. Becker, T. Doerges, K. Huebner,
+T. Connors, Sebastian Gouezel and many others.
+Thanks to the many people who sent in bug reports, feature suggestions, and other feedback.
+
+=cut
+
+__END__
+%%BEGIN SAFE COMMANDS
+% Regex matching commands which can safely be in the
+% argument of a \DIFadd or \DIFdel command (leave out the \)
+arabic
+dashbox
+emph
+fbox
+framebox
+hspace\*?
+math.*
+makebox
+mbox
+pageref
+ref
+eqref
+symbol
+raisebox
+rule
+text.*
+shortstack
+usebox
+dag
+ddag
+copyright
+pounds
+S
+P
+oe
+OE
+ae
+AE
+aa
+AA
+o
+O
+l
+L
+frac
+ss
+sqrt
+ldots
+cdots
+vdots
+ddots
+alpha
+beta
+gamma
+delta
+epsilon
+varepsilon
+zeta
+eta
+theta
+vartheta
+iota
+kappa
+lambda
+mu
+nu
+xi
+pi
+varpi
+rho
+varrho
+sigma
+varsigma
+tau
+upsilon
+phi
+varphi
+chi
+psi
+omega
+Gamma
+Delta
+Theta
+Lambda
+Xi
+Pi
+Sigma
+Upsilon
+Phi
+Psi
+Omega
+ps
+mp
+times
+div
+ast
+star
+circ
+bullet
+cdot
+cap
+cup
+uplus
+sqcap
+vee
+wedge
+setminus
+wr
+diamond
+(?:big)?triangle.*
+lhd
+rhd
+unlhd
+unrhd
+oplus
+ominus
+otimes
+oslash
+odot
+bigcirc
+d?dagger
+amalg
+leq
+prec
+preceq
+ll
+(?:sq)?su[bp]set(?:eq)?
+in
+vdash
+geq
+succ(?:eq)?
+gg
+ni
+dashv
+equiv
+sim(?:eq)?
+asymp
+approx
+cong
+neq
+doteq
+propto
+models
+perp
+mid
+parallel
+bowtie
+Join
+smile
+frown
+.*arrow
+(?:long)?mapsto
+.*harpoon.*
+leadsto
+aleph
+hbar
+imath
+jmath
+ell
+wp
+Re
+Im
+mho
+prime
+emptyset
+nabla
+surd
+top
+bot
+angle
+forall
+exists
+neg
+flat
+natural
+sharp
+backslash
+partial
+infty
+Box
+Diamond
+triangle
+clubsuit
+diamondsuit
+heartsuit
+spadesuit
+sum
+prod
+coprod
+int
+oint
+big(?:sq)?c[au]p
+bigvee
+bigwedge
+bigodot
+bigotimes
+bigoplus
+biguplus
+(?:arc)?(?:cos|sin|tan|cot)h?
+csc
+arg
+deg
+det
+dim
+exp
+gcd
+hom
+inf
+ker
+lg
+lim
+liminf
+limsup
+ln
+log
+max
+min
+Pr
+sec
+sup
+quad
+qquad
+bibfield
+bibinfo
+[Hclbkdruvt]
+[`'^"~=.]
+_
+AMPERSAND
+(SUPER|SUB)SCRIPTNB
+(SUPER|SUB)SCRIPT
+SQRT
+SQRTNB
+PERCENTAGE
+DOLLAR
+%%END SAFE COMMANDS
+
+%%BEGIN TEXT COMMANDS
+% Regex matching commands with a text argument (leave out the \)
+addcontents.*
+cc
+closing
+chapter
+dashbox
+emph
+encl
+fbox
+framebox
+footnote
+footnotetext
+framebox
+href
+intertext
+part
+(sub){0,2}section\*?
+(sub)?paragraph\*?
+makebox
+mbox
+opening
+parbox
+raisebox
+savebox
+sbox
+shortintertext
+shortstack
+sidenote
+signature
+text.*
+value
+underline
+sqrt
+(SUPER|SUB)SCRIPT
+%%END TEXT COMMANDS
+
+%%BEGIN CONTEXT1 COMMANDS
+% Regex matching commands with a text argument (leave out the \), which will fail out of context. These commands behave like text commands, except when they occur in a deleted section, where they are disabled, but their argument is shown as deleted text.
+caption
+subcaption
+multicolumn
+%%END CONTEXT1 COMMANDS
+
+%%BEGIN CONTEXT2 COMMANDS
+% Regex matching commands with a text argument (leave out the \), which will fail out of context.  As corresponding commands for context1.  The only difference is that context2 commands are completely disabled in deleted sections, including their arguments.
+title
+author
+date
+institute
+%%END CONTEXT2 COMMANDS
+
+%% CONFIGURATION variabe defaults
+%%BEGIN LISTENV CONFIG
+itemize
+description
+enumerate
+%%END LISTENV CONFIG
+
+%%BEGIN FLOATENV CONFIG
+figure[\w\d*@]*
+table[\w\d*@]*
+plate[\w\d*@]*
+%%END FLOATENV CONFIG
+
+%%BEGIN PICTUREENV CONFIG
+picture[\w\d*@]*
+tikzpicture[\w\d*@]*
+DIFnomarkup
+%%END PICTUREENV CONFIG
+
+%%BEGIN MATHENV CONFIG
+equation[*]?
+displaymath
+math
+DOLLARDOLLAR
+%%END MATHENV CONFIG
+
+%%BEGIN MATHARRENV CONFIG
+eqnarray[*]?
+align[*]?
+alignat[*]?
+gather[*]?
+multline[*]?
+flalign[*]?
+%%END MATHARRENV CONFIG
+
+%%BEGIN ARRENV CONFIG
+aligned
+gathered
+multlined
+array
+[pbvBV]?matrix
+smallmatrix
+cases
+%%END ARRENV CONFIG
+# split
+
+
+%%BEGIN COUNTERCMD CONFIG
+footnote
+part
+chapter
+section
+subsection
+subsubsection
+paragraph
+subparagraph
+%%END COUNTERCMD CONFIG
+
+%%BEGIN VERBATIMENV CONFIG
+comment
+%%END VERBATIMENV CONFIG
+
+%%BEGIN VERBATIMLINEENV CONFIG
+lstlisting
+verbatim[*]?
+%%END VERBATIMLINEENV CONFIG
+
+%%BEGIN CUSTOMDIFCMD CONFIG
+%%END CUSTOMDIFCMD CONFIG
+
+%%% TYPES (Commands for highlighting changed blocks)
+
+%DIF UNDERLINE PREAMBLE
+\RequirePackage[normalem]{ulem}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{{\protect\color{blue}\uwave{#1}}}
+\providecommand{\DIFdel}[1]{{\protect\color{red}\sout{#1}}}
+%DIF END UNDERLINE PREAMBLE
+
+%DIF LUAUNDERLINE PREAMBLE
+\RequirePackage{lua-ul}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{{\protect\color{blue}\underLine{#1}}}
+\providecommand{\DIFdel}[1]{{\protect\color{red}\strikeThrough{#1}}}
+%DIF END LUAUNDERLINE PREAMBLE
+
+%DIF CTRADITIONAL PREAMBLE
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\RequirePackage[stable]{footmisc}
+\DeclareOldFontCommand{\sf}{\normalfont\sffamily}{\mathsf}
+\providecommand{\DIFadd}[1]{{\protect\color{blue} \sf #1}}
+\providecommand{\DIFdel}[1]{{\protect\color{red} [..\footnote{removed: #1} ]}}
+%DIF END CTRADITIONAL PREAMBLE
+
+%DIF TRADITIONAL PREAMBLE
+\RequirePackage[stable]{footmisc}
+\DeclareOldFontCommand{\sf}{\normalfont\sffamily}{\mathsf}
+\providecommand{\DIFadd}[1]{{\sf #1}}
+\providecommand{\DIFdel}[1]{{[..\footnote{removed: #1} ]}}
+%DIF END TRADITIONAL PREAMBLE
+
+%DIF CFONT PREAMBLE
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\DeclareOldFontCommand{\sf}{\normalfont\sffamily}{\mathsf}
+\providecommand{\DIFadd}[1]{{\protect\color{blue} \sf #1}}
+\providecommand{\DIFdel}[1]{{\protect\color{red} \scriptsize #1}}
+%DIF END CFONT PREAMBLE
+
+%DIF FONTSTRIKE PREAMBLE
+\RequirePackage[normalem]{ulem}
+\DeclareOldFontCommand{\sf}{\normalfont\sffamily}{\mathsf}
+\providecommand{\DIFadd}[1]{{\sf #1}}
+\providecommand{\DIFdel}[1]{{\footnotesize \sout{#1}}}
+%DIF END FONTSTRIKE PREAMBLE
+
+%DIF CCHANGEBAR PREAMBLE
+\RequirePackage[pdftex]{changebar}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{\protect\cbstart{\protect\color{blue}#1}\protect\cbend}
+\providecommand{\DIFdel}[1]{\protect\cbdelete{\protect\color{red}#1}\protect\cbdelete}
+%DIF END CCHANGEBAR PREAMBLE
+
+%DIF CFONTCHBAR PREAMBLE
+\RequirePackage[pdftex]{changebar}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{\protect\cbstart{\protect\color{blue}\sf #1}\protect\cbend}
+\providecommand{\DIFdel}[1]{\protect\cbdelete{\protect\color{red}\scriptsize #1}\protect\cbdelete}
+%DIF END CFONTCHBAR PREAMBLE
+
+%DIF CULINECHBAR PREAMBLE
+\RequirePackage[normalem]{ulem}
+\RequirePackage[pdftex]{changebar}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{\protect\cbstart{\protect\color{blue}\uwave{#1}}\protect\cbend}
+\providecommand{\DIFdel}[1]{\protect\cbdelete{\protect\color{red}\sout{#1}}\protect\cbdelete}
+%DIF END CULINECHBAR PREAMBLE
+
+%DIF CHANGEBAR PREAMBLE
+\RequirePackage[pdftex]{changebar}
+\providecommand{\DIFadd}[1]{\protect\cbstart{#1}\protect\cbend}
+\providecommand{\DIFdel}[1]{\protect\cbdelete}
+%DIF END CHANGEBAR PREAMBLE
+
+%DIF INVISIBLE PREAMBLE
+\providecommand{\DIFadd}[1]{#1}
+\providecommand{\DIFdel}[1]{}
+%DIF END INVISIBLE PREAMBLE
+
+%DIF BOLD PREAMBLE
+\DeclareOldFontCommand{\bf}{\normalfont\bfseries}{\mathbf}
+\providecommand{\DIFadd}[1]{{\bf #1}}
+\providecommand{\DIFdel}[1]{}
+%DIF END BOLD PREAMBLE
+
+%DIF PDFCOMMENT PREAMBLE
+\RequirePackage{pdfcomment} %DIF PREAMBLE
+\providecommand{\DIFadd}[1]{\pdfmarkupcomment[author=ADD:,markup=Underline]{#1}{}}
+\providecommand{\DIFdel}[1]{\pdfcomment[icon=Insert,author=DEL:,hspace=12pt]{#1}}
+%DIF END PDFCOMMENT PREAMBLE
+
+%% SUBTYPES (Markers for beginning and end of changed blocks)
+
+%DIF SAFE PREAMBLE
+\providecommand{\DIFaddbegin}{}
+\providecommand{\DIFaddend}{}
+\providecommand{\DIFdelbegin}{}
+\providecommand{\DIFdelend}{}
+\providecommand{\DIFmodbegin}{}
+\providecommand{\DIFmodend}{}
+%DIF END SAFE PREAMBLE
+
+%DIF MARGIN PREAMBLE
+\providecommand{\DIFaddbegin}{\protect\marginpar{a[}}
+\providecommand{\DIFaddend}{\protect\marginpar{]}}
+\providecommand{\DIFdelbegin}{\protect\marginpar{d[}}
+\providecommand{\DIFdelend}{\protect\marginpar{]}}
+\providecommand{\DIFmodbegin}{\protect\marginpar{m[}}
+\providecommand{\DIFmodend}{\protect\marginpar{]}}
+%DIF END MARGIN PREAMBLE
+
+%DIF DVIPSCOL PREAMBLE
+%Note: only works with dvips converter
+\RequirePackage{color}
+\RequirePackage{dvipscol}
+\providecommand{\DIFaddbegin}{\protect\nogroupcolor{blue}}
+\providecommand{\DIFaddend}{\protect\nogroupcolor{black}}
+\providecommand{\DIFdelbegin}{\protect\nogroupcolor{red}}
+\providecommand{\DIFdelend}{\protect\nogroupcolor{black}}
+\providecommand{\DIFmodbegin}{}
+\providecommand{\DIFmodend}{}
+%DIF END DVIPSCOL PREAMBLE
+
+%DIF COLOR PREAMBLE
+\RequirePackage{color}
+\providecommand{\DIFaddbegin}{\protect\color{blue}}
+\providecommand{\DIFaddend}{\protect\color{black}}
+\providecommand{\DIFdelbegin}{\protect\color{red}}
+\providecommand{\DIFdelend}{\protect\color{black}}
+\providecommand{\DIFmodbegin}{}
+\providecommand{\DIFmodend}{}
+%DIF END COLOR PREAMBLE
+
+%DIF LABEL PREAMBLE
+% To show only pages with changes (pdf) (external program pdftk needs to be installed)
+% (only works for simple documents with non-repeated page numbers, otherwise use ZLABEL)
+% pdflatex diff.tex
+% pdflatex diff.tex
+%pdftk diff.pdf cat \
+%`perl -lne '\
+% if (m/\\newlabel{DIFchg[b](\d*)}{{.*}{(.*)}}/) { $start{$1}=$2; print $2}\
+% if (m/\\newlabel{DIFchg[e](\d*)}{{.*}{(.*)}}/) { \
+%      if (defined($start{$1})) { \
+%         for ($j=$start{$1}; $j<=$2; $j++) {print "$j";}\
+%      } else { \
+%         print "$2"\
+%      }\
+% }' diff.aux \
+% | uniq \
+% | tr  \\n ' '` \
+% output diff-changedpages.pdf
+% To show only pages with changes (dvips/dvipdf)
+% dvips -pp `\
+% [ put here the perl script from above]
+% | uniq | tr -s \\n ','`
+\typeout{Check comments in preamble of output for instructions how to show only pages where changes have been made}
+\newcount\DIFcounterb
+\global\DIFcounterb 0\relax
+\newcount\DIFcountere
+\global\DIFcountere 0\relax
+\providecommand{\DIFaddbegin}{\global\advance\DIFcounterb 1\relax\label{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFaddend}{\global\advance\DIFcountere 1\relax\label{DIFchge\the\DIFcountere}}
+\providecommand{\DIFdelbegin}{\global\advance\DIFcounterb 1\relax\label{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFdelend}{\global\advance\DIFcountere 1\relax\label{DIFchge\the\DIFcountere}}
+\providecommand{\DIFmodbegin}{\global\advance\DIFcounterb 1\relax\label{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFmodend}{\global\advance\DIFcountere 1\relax\label{DIFchge\the\DIFcountere}}
+%DIF END LABEL PREAMBLE
+
+%DIF ZLABEL PREAMBLE
+% To show only pages with changes (pdf) (external program pdftk needs to be installed)
+% (uses zref for reference to absolute page numbers)
+% pdflatex diff.tex
+% pdflatex diff.tex
+%pdftk diff.pdf cat \
+%`perl -lne 'if (m/\\zref\@newlabel{DIFchgb(\d*)}{.*\\abspage{(\d*)}}/ ) { $start{$1}=$2; print $2 } \
+%  if (m/\\zref\@newlabel{DIFchge(\d*)}{.*\\abspage{(\d*)}}/) { \
+%      if (defined($start{$1})) { \
+%         for ($j=$start{$1}; $j<=$2; $j++) {print "$j";}\
+%      } else { \
+%         print "$2"\
+%      }\
+% }' diff.aux \
+% | uniq \
+% | tr  \\n ' '` \
+% output diff-changedpages.pdf
+% To show only pages with changes (dvips/dvipdf)
+% latex diff.tex
+% latex diff.tex
+% dvips -pp `perl -lne 'if (m/\\newlabel{DIFchg[be]\d*}{{.*}{(.*)}}/) { print $1 }' diff.aux | uniq | tr -s \\n ','` diff.dvi
+\typeout{Check comments in preamble of output for instructions how to show only pages where changes have been made}
+\usepackage[user,abspage]{zref}
+\newcount\DIFcounterb
+\global\DIFcounterb 0\relax
+\newcount\DIFcountere
+\global\DIFcountere 0\relax
+\providecommand{\DIFaddbegin}{\global\advance\DIFcounterb 1\relax\zlabel{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFaddend}{\global\advance\DIFcountere 1\relax\zlabel{DIFchge\the\DIFcountere}}
+\providecommand{\DIFdelbegin}{\global\advance\DIFcounterb 1\relax\zlabel{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFdelend}{\global\advance\DIFcountere 1\relax\zlabel{DIFchge\the\DIFcountere}}
+\providecommand{\DIFmodbegin}{\global\advance\DIFcounterb 1\relax\zlabel{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFmodend}{\global\advance\DIFcountere 1\relax\zlabel{DIFchge\the\DIFcountere}}
+%DIF END ZLABEL PREAMBLE
+
+%DIF ONLYCHANGEDPAGE PREAMBLE
+\RequirePackage{atbegshi}
+\RequirePackage{etoolbox}
+\RequirePackage{zref}
+% redefine label command to write immediately to aux file - page references will be lost
+\makeatletter \let\oldlabel\label% Store \label
+\renewcommand{\label}[1]{% Update \label to write to the .aux immediately
+\zref@wrapper@immediate{\oldlabel{#1}}}
+\makeatother
+\newbool{DIFkeeppage}
+\newbool{DIFchange}
+\boolfalse{DIFkeeppage}
+\boolfalse{DIFchange}
+\AtBeginShipout{%
+  \ifbool{DIFkeeppage}
+        {\global\boolfalse{DIFkeeppage}}  % True DIFkeeppage
+         {\ifbool{DIFchange}{\global\boolfalse{DIFkeeppage}}{\global\boolfalse{DIFkeeppage}\AtBeginShipoutDiscard}} % False DIFkeeppage
+}
+\providecommand{\DIFaddbegin}{\global\booltrue{DIFkeeppage}\global\booltrue{DIFchange}}
+\providecommand{\DIFaddend}{\global\booltrue{DIFkeeppage}\global\boolfalse{DIFchange}}
+\providecommand{\DIFdelbegin}{\global\booltrue{DIFkeeppage}\global\booltrue{DIFchange}}
+\providecommand{\DIFdelend}{\global\booltrue{DIFkeeppage}\global\boolfalse{DIFchange}}
+\providecommand{\DIFmodbegin}{\global\booltrue{DIFkeeppage}\global\booltrue{DIFchange}}
+\providecommand{\DIFmodend}{\global\booltrue{DIFkeeppage}\global\boolfalse{DIFchange}}
+%DIF END ONLYCHANGEDPAGE PREAMBLE
+
+%% FLOAT TYPES
+
+%DIF FLOATSAFE PREAMBLE
+\providecommand{\DIFaddFL}[1]{\DIFadd{#1}}
+\providecommand{\DIFdelFL}[1]{\DIFdel{#1}}
+\providecommand{\DIFaddbeginFL}{}
+\providecommand{\DIFaddendFL}{}
+\providecommand{\DIFdelbeginFL}{}
+\providecommand{\DIFdelendFL}{}
+%DIF END FLOATSAFE PREAMBLE
+
+%DIF IDENTICAL PREAMBLE
+\providecommand{\DIFaddFL}[1]{\DIFadd{#1}}
+\providecommand{\DIFdelFL}[1]{\DIFdel{#1}}
+\providecommand{\DIFaddbeginFL}{\DIFaddbegin}
+\providecommand{\DIFaddendFL}{\DIFaddend}
+\providecommand{\DIFdelbeginFL}{\DIFdelbegin}
+\providecommand{\DIFdelendFL}{\DIFdelend}
+%DIF END IDENTICAL PREAMBLE
+
+%DIF TRADITIONALSAFE PREAMBLE
+% procidecommand color to make this work for TRADITIONAL and CTRADITIONAL
+\providecommand{\color}[1]{}
+\providecommand{\DIFaddFL}[1]{\DIFadd{#1}}
+\providecommand{\DIFdel}[1]{{\protect\color{red}[..{\scriptsize {removed: #1}} ]}}
+\providecommand{\DIFaddbeginFL}{}
+\providecommand{\DIFaddendFL}{}
+\providecommand{\DIFdelbeginFL}{}
+\providecommand{\DIFdelendFL}{}
+%DIF END TRADITIONALSAFE PREAMBLE
+
+% see:
+%  http://tex.stackexchange.com/questions/47351/can-i-redefine-a-command-to-contain-itself
+
+%DIF HIGHLIGHTGRAPHICS PREAMBLE
+\RequirePackage{settobox}
+\RequirePackage{letltxmacro}
+\newsavebox{\DIFdelgraphicsbox}
+\newlength{\DIFdelgraphicswidth}
+\newlength{\DIFdelgraphicsheight}
+% store original definition of \includegraphics
+\LetLtxMacro{\DIFOincludegraphics}{\includegraphics}
+\newcommand{\DIFaddincludegraphics}[2][]{{\color{blue}\fbox{\DIFOincludegraphics[#1]{#2}}}}
+\newcommand{\DIFdelincludegraphics}[2][]{%
+\sbox{\DIFdelgraphicsbox}{\DIFOincludegraphics[#1]{#2}}%
+\settoboxwidth{\DIFdelgraphicswidth}{\DIFdelgraphicsbox}
+\settoboxtotalheight{\DIFdelgraphicsheight}{\DIFdelgraphicsbox}
+\scalebox{\DIFscaledelfig}{%
+\parbox[b]{\DIFdelgraphicswidth}{\usebox{\DIFdelgraphicsbox}\\[-\baselineskip] \rule{\DIFdelgraphicswidth}{0em}}\llap{\resizebox{\DIFdelgraphicswidth}{\DIFdelgraphicsheight}{%
+\setlength{\unitlength}{\DIFdelgraphicswidth}%
+\begin{picture}(1,1)%
+\thicklines\linethickness{2pt}
+{\color[rgb]{1,0,0}\put(0,0){\framebox(1,1){}}}%
+{\color[rgb]{1,0,0}\put(0,0){\line( 1,1){1}}}%
+{\color[rgb]{1,0,0}\put(0,1){\line(1,-1){1}}}%
+\end{picture}%
+}\hspace*{3pt}}}
+}
+\LetLtxMacro{\DIFOaddbegin}{\DIFaddbegin}
+\LetLtxMacro{\DIFOaddend}{\DIFaddend}
+\LetLtxMacro{\DIFOdelbegin}{\DIFdelbegin}
+\LetLtxMacro{\DIFOdelend}{\DIFdelend}
+\DeclareRobustCommand{\DIFaddbegin}{\DIFOaddbegin \let\includegraphics\DIFaddincludegraphics}
+\DeclareRobustCommand{\DIFaddend}{\DIFOaddend \let\includegraphics\DIFOincludegraphics}
+\DeclareRobustCommand{\DIFdelbegin}{\DIFOdelbegin \let\includegraphics\DIFdelincludegraphics}
+\DeclareRobustCommand{\DIFdelend}{\DIFOaddend \let\includegraphics\DIFOincludegraphics}
+\LetLtxMacro{\DIFOaddbeginFL}{\DIFaddbeginFL}
+\LetLtxMacro{\DIFOaddendFL}{\DIFaddendFL}
+\LetLtxMacro{\DIFOdelbeginFL}{\DIFdelbeginFL}
+\LetLtxMacro{\DIFOdelendFL}{\DIFdelendFL}
+\DeclareRobustCommand{\DIFaddbeginFL}{\DIFOaddbeginFL \let\includegraphics\DIFaddincludegraphics}
+\DeclareRobustCommand{\DIFaddendFL}{\DIFOaddendFL \let\includegraphics\DIFOincludegraphics}
+\DeclareRobustCommand{\DIFdelbeginFL}{\DIFOdelbeginFL \let\includegraphics\DIFdelincludegraphics}
+\DeclareRobustCommand{\DIFdelendFL}{\DIFOaddendFL \let\includegraphics\DIFOincludegraphics}
+%DIF END HIGHLIGHTGRAPHICS PREAMBLE
+
+%% SPECIAL PACKAGE PREAMBLE COMMANDS
+
+%% Redefine strike out command to wrap in text box if amsmath is used and markup style with ulem is used
+%DIF AMSMATHULEM PREAMBLE
+\makeatletter
+\let\sout@orig\sout
+\renewcommand{\sout}[1]{\ifmmode\text{\sout@orig{\ensuremath{#1}}}\else\sout@orig{#1}\fi}
+\makeatother
+%DIF END AMSMATHULEM PREAMBLE
+
+
+% Standard \DIFadd and \DIFdel are redefined as \DIFaddtex and \DIFdeltex
+% when hyperref package is included.
+%DIF HYPERREF PREAMBLE
+\providecommand{\DIFadd}[1]{\texorpdfstring{\DIFaddtex{#1}}{#1}}
+\providecommand{\DIFdel}[1]{\texorpdfstring{\DIFdeltex{#1}}{}}
+%DIF END HYPERREF PREAMBLE
+
+%DIF LISTINGS PREAMBLE
+\RequirePackage{listings}
+\lstdefinelanguage{DIFcode}{
+  % note that the definitions in the following two lines are overwritten dependent on the markup type selected %DIFCODE TEMPLATE
+  morecomment=[il]{\%DIF\ <\ },          %DIFCODE TEMPLATE
+  moredelim=[il][\bfseries]{\%DIF\ >\ }  %DIFCODE TEMPLATE
+}
+\lstdefinestyle{DIFverbatimstyle}{
+	language=DIFcode,
+	basicstyle=\ttfamily,
+	columns=fullflexible,
+	keepspaces=true
+}
+\lstnewenvironment{DIFverbatim}{\lstset{style=DIFverbatimstyle}}{}
+\lstnewenvironment{DIFverbatim*}{\lstset{style=DIFverbatimstyle,showspaces=true}}{}
+%DIF END LISTINGS PREAMBLE
+
+%DIF COLORLISTINGS PREAMBLE
+\RequirePackage{listings}
+\RequirePackage{color}
+\lstdefinelanguage{DIFcode}{
+  % note that the definitions in the following two lines are overwritten dependent on the markup type selected %DIFCODE TEMPLATE
+  morecomment=[il]{\%DIF\ <\ },          %DIFCODE TEMPLATE
+  moredelim=[il][\bfseries]{\%DIF\ >\ }  %DIFCODE TEMPLATE
+}
+\lstdefinestyle{DIFverbatimstyle}{
+	language=DIFcode,
+	basicstyle=\ttfamily,
+	columns=fullflexible,
+	keepspaces=true
+}
+\lstnewenvironment{DIFverbatim}{\lstset{style=DIFverbatimstyle}}{}
+\lstnewenvironment{DIFverbatim*}{\lstset{style=DIFverbatimstyle,showspaces=true}}{}
+%DIF END COLORLISTINGS PREAMBLE
+
+%DIF DIFCODE_UNDERLINE
+  moredelim=[il][\color{red}\sout]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\uwave]{\%DIF\ >\ }
+%DIF END DIFCODE_UNDERLINE
+
+%DIF DIFCODE_CTRADITIONAL
+  moredelim=[il][\color{red}\scriptsize]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_CTRADITIONAL
+
+%DIF DIFCODE_TRADITIONAL
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il][\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_TRADITIONAL
+
+%DIF DIFCODE_CFONT
+  moredelim=[il][\color{red}\scriptsize]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_CFONT
+
+%DIF DIFCODE_FONTSTRIKE
+  moredelim=[il][\scriptsize \sout]{\%DIF\ <\ },
+  moredelim=[il][\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_FONTSTRIKE
+
+%DIF DIFCODE_INVISIBLE
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il]{\%DIF\ >\ }
+%DIF END DIFCODE_INVISIBLE
+
+%DIF DIFCODE_CHANGEBAR
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il]{\%DIF\ >\ }
+%DIF END DIFCODE_CHANGEBAR
+
+%DIF DIFCODE_CCHANGEBAR
+  moredelim=[il][\color{red}]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}]{\%DIF\ >\ }
+%DIF END DIFCODE_CCHANGEBAR
+
+%DIF DIFCODE_CULINECHBAR
+  moredelim=[il][\color{red}\sout]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\uwave]{\%DIF\ >\ }
+%DIF END DIFCODE_CULINECHBAR
+
+%DIF DIFCODE_CFONTCHBAR
+  moredelim=[il][\color{red}\scriptsize]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_CFONTCHBAR
+
+%DIF DIFCODE_BOLD
+  % unfortunately \bfseries cannot be combined with ttfamily without extra packages
+  % also morecomment=[il] is broken as of v1.5b of listings at least
+  % workaround: plot in white with tiny font
+  % morecomment=[il]{\%DIF\ <\ },
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il][\sffamily\bfseries]{\%DIF\ >\ }
+%DIF END DIFCODE_BOLD
+
+%DIF DIFCODE_PDFCOMMENT
+
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il][\sffamily\bfseries]{\%DIF\ >\ }
+%DIF END DIFCODE_PDFCOMMENT
+
diff --git a/scripts/latexdiff-so.pl b/scripts/latexdiff-so.pl
new file mode 100644
index 0000000..9ab8e41
--- /dev/null
+++ b/scripts/latexdiff-so.pl
@@ -0,0 +1,5927 @@
+#!/usr/bin/env perl
+##!/usr/bin/perl -w
+# latexdiff - differences two latex files on the word level
+#             and produces a latex file with the differences marked up.
+#
+#   Copyright (C) 2004-22  F J Tilmann (tilmann@gfz-potsdam.de)
+#
+# Repository/issue tracker:   https://github.com/ftilmann/latexdiff
+# CTAN page:          http://www.ctan.org/pkg/latexdiff
+#
+#    This program is free software: you can redistribute it and/or modify
+#    it under the terms of the GNU General Public License as published by
+#    the Free Software Foundation, either version 3 of the License, or
+#    (at your option) any later version.
+#
+#    This program is distributed in the hope that it will be useful,
+#    but WITHOUT ANY WARRANTY; without even the implied warranty of
+#    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+#    GNU General Public License for more details.
+#    You should have received a copy of the GNU General Public License
+#    along with this program.  If not, see <http://www.gnu.org/licenses/>.
+#
+# Detailed usage information at the end of the file
+#
+#
+# Note references to issue numbers are for the github repository of latexdiff: https://github.com/ftilmann/latexdiff
+#
+# Version 1.3.4:
+# New features:
+#  - Option to use lua-ul instead of ulem (for use with LuaLaTeX) (fix #17, #60, #188, #255, #270)
+# Enhancement
+#  - If amsmath is detected as one of the included (based on idea in PR #263 contributed by github user xlucn
+# Bug fix
+#  - sometimes the introduction of aux commands to show deleted lists or descriptions leaves in place empty auxiliary
+#    list environments, which would cause error messages (though skipping messages would result in a correct output pdf).
+#    These are now removed.
+#  - add more mboxsafecmd and safecmd commands for SIunitx to stay compatible with newer versionsof SIunitx (PR #283, fixing issue #282, contributed by github user joe6302413)
+#  - File added via --preamble option is no longer assumed to be ASCII, but read either with encoding as defined by --encoding option or using the encoding of the LOCALE (fixes issue #285 )
+#  - multicolumn argument is now treated as text
+#  - when tikz-dependency package is used, \& is no longer a safe command as it has special meaning inside dependency environment.  The fix is a little of a hack as really it should only be considered unsafe within dependency environment (fixes (mostly) issue #303 )
+#  - listings package had trouble with non-ASCII chars. The encoding is now set (thanks to github user anka-213 for finding this). Fixes #304
+#
+# Version 1.3.3:
+# New features:
+#  - Option --no-del to remove all deleted text (merge contributed by github user tdegeus PR #252, fixing issue #66
+#
+# Bug fixes:
+#  - Abbreviations involving punctuations within them. They need special treatment because otherwise in some
+#    circumstances the gnoring of white space differences in conjunction with merging according to MINWORDSBLOCK rule
+#    could turn 'i.e.' into 'i.\PAR e.' (see https://github.com/ftilmann/latexdiff/issues/269). A few abbreviations
+#    are now hard-coded and treated as atomic:
+#    English: i.e., e.g.   Deutsch: z.B.
+#    (fixes issue #269)
+#  - In WHOLE and COARSE math modes, now properly treat math environments with arguments such as \alignat. Fixes #251
+#  - For FINE math mode, multiple improvments to the processing work flow yield more robust outcomes. In particular, changes
+#    to the equation type, e.g. \begin{displaymath} -> \begin{equation}  without modifications  now usually no longer result
+#    in errors. (Partially) fixes issues #235  and #244.
+#  - When encountering deleted math array environments such as align or eqnarray, rather than replacing them with  a
+#    fixed replacement environment (e.g. align* or eqnarray*), an asterisk is now added to the original command, which
+#    in amsmath (and with eqnarray) will result in the same environment but without line numbers. Config variable MATHARRREPL
+#    is therefore (nearly) redundant, and a depracation warning is given when it is set. Reference to MATHARRREPL are have
+#    been removed from the manual (there is one exception, when it's still being used: sometimes latexdiff can figure out
+#    that there is a deleted array environment, but does not know which one. In this case, MATHARRREPL is still being used
+#    to encapsulate these parts of the source, and therefore it is still set internally.  But this is a quite rare situation).
+#    Fixes issue #216
+#  - Unlike 'array' environment, 'split' (amsmath) does not work in argument of \DIFadd or \DIFdl in UNDERLINE modes; therefore remove it from ARRENV configuration variable.
+#     Exclude \begin and \end in math environments in COARSE and WHOLE modes. Fixes #258. Fixes #109
+#  - --flatten now works for empty files. Fixes issue #242
+#  - improved processing of Chinese and Japanese texts in that splitting is done based on characters. Thanks to LuXu (Oliver Lew) in git for working this out. Fixes #229, fixes #145
+
+
+
+# Version 1.3.2
+# API adaptions:
+#  - latexdiff now completes with exit code 0 after --help or --version command (see issue #248)
+# New features / feature extensions
+#  - extend CUSTOMDIFCMD related postprocessing to deal properly with multiline commands, or a sequence of several commands in the same line (see github issue #204)
+#  - Support for additional macros from import package (\import, \inputfrom, \includefrom, \subimport,\subinputfrom, \subincludefrom). Provided by janniklasrose in PR #243 (fixes #239)
+#  - replace default driver dvips->pdftex
+# Bug fixes:
+#  - fix issue #206 affecting proper markup of text commands which are not also safe cmd's and have multiple arguments
+#  - fix issue #210 by adding \eqref (amsmath package) to the list of safe commands
+#  - fix bug reported in issue #168 mangled verbatim line environment
+#  - fix bug reported in issue #218 by replacing \hspace{0pt} after \mbox{..} auxiliary commands with \hskip0pt.
+#  - more ways to process \frac correctly with atomic arguments (committed by julianuu PR #246
+#  - fix a bug in biblatex mode, which prevented proper processing of modified \textcite (see: https://tex.stackexchange.com/questions/555157/latexdiff-and-biblatex-citation-commands)
+#  - -h string fix: add -driver option
+#
+# Version 1.3.1.1
+#  - remove spurious \n to fix error: Unknown regexp modifier "/n" at .../latexdiff line 1974, near "=~ " (see github issue #201)
+#
+# Version 1.3.1
+#    Bug fixes:
+#      - remove some uninitialised variable $2 warnings in string substitution in flatten function in case included file is not found
+#      - add minimal postprocessing to diff processing of preamble commands (replace \RIGHTBRACE by \} )
+#      - pre-processing: replace (contributed) routine take_comments_and_enter_from_frac() with take_comments_and_newline_from_frac(), which does the same thing
+#        (remove whitespace characters and comments between the argument of \frac commands) in an easier and more robust way. In addition, it
+#        will replace commands like \frac12 with \frac{1}{2} as pre-processing step.   Fixes issue #184
+#      - add "intertext" to list of unsafe math commands @UNSAFEMATHCMD . Fixes issue #179
+#      - provide citation command patterns for biblatex and protect them with mbox'es. Fixes issue #199
+#      - hardcode number of parameters for \href and \url commands to allow spaces between commands and arguments even if --allow-spaces option is not used (this
+#        is needed because some bibliography styles add such in-command-sequence spaces)  Fixes issues: #178 #198
+#      - bibitem is now kept even in deleted blocks such that deleted references show up properly (this implies that the actual numbers in numerical referencing schemes will change)
+#        (this is implemented by introducing a new class of commands KEEPCMD , which are kept as is in deleted environments (no effect in added environments). Currently
+#        \bibitem   is hardwired to be the only member of this class  (fixes issue #194, #174)
+#    Features:
+#      - add some special processing for revtex bibliography commands, so that the spaces between bibliography commands \bibfield and \bibinfo and their arguments are ignored.
+#         (fixes issue #194, should fix #174)
+#
+# Version 1.3.0 (7 October 2018):
+#    - treat options to \documentclass as potential package names (some packages allow implicit loading of or imply selected packages
+#    - improved pattern matching: now allows nested angular brackets, and is no longer confused by escaped curly braces
+#    - improved pattern matching in COARSE mode: occasionally, the closing bracket or some other elements would be matched in an 'unnatural' way due to another sequence being more minimal in the computational sense, sometimes even causing errors due to tokens moving in or out of the scope of math environments. This is now discouraged by adding internal \DIFANCHOR commands (which are removed again in post-processing) (fixes issues reported via email by li_ruomeng .
+#    - verbatim and lstlisting environments are marked-up with line-by-line in a similar style to non-verbatim text (requires the listing package to be installed)
+#       (see new configuration variable VERBATIMLINEENV) (several issues and pull requests by jprotze)
+#    - --flatten: now supports \verbatiminput and \lstlistinput
+#     - --flatten: if file is not found, do not fail, simply warn and leave command unexpanded (inspired by issue #112).  Don't warn if file name contains #[0-9] as it is then most likely an argument within a command definition rather than an actual file (applies to \input, \subfile, \include commands)
+#     - added to textcmds: \intertext
+#    - new config variable CUSTOMDIFCMD to allow defining special versions of  commands  in added or deleted blocks (Pull request by github user jprotze)
+#    - added option -no-links (mostly for use by latexdiff-vc in only-changes modes) (Pull request by github user jprotze)
+#    - new option --filter-script to run both input through a pre-processing script (PR jasonmccsmith  #167)
+#      new option --no-filter-stderr to hide stderr output from filter-script (potentially dangerous, as this might hide malfunctioning of filter scripts)
+#    - --flatten now can deal with imports made using the import package {PR jasonmccsmith #173)
+#   Bug fixes:
+#    - pattern matching of \verb and \lstinline commands had an error which meant they would trigger on commands beginning with \verb.
+#    - In description environments, mark up item descriptions by effectively reating the insides of item commannds as text commands (fixes #161)
+#
+#
+# Version 1.2.1 (22 June 2017)
+#    - add "DeclareOldFontCommand" to styles using \bf or \sf old style font commands (fixies issue #92 )
+#    - improved markup: process lstinline commands in listings package correctly
+#      for styles using colour, \verb and \lstinline arguments are marked up with colour (blue for added, red for deleted)
+#    - bug fix: protecting inline math expressions for mbox did not work as intended (see stack exchange question: http://tex.stackexchange.com/questions/359412/compiling-the-latexdiff-when-adding-a-subscript-before-a-pmatrix-environment-cau)
+#    - bug fix: when deleted \item commands are followed immediately by unsafe commands, they were not restored properly
+#      (thanks to J. Protze for pull request) (pull request #89)
+#    - treat lstlisting and comment as equivalent to verbatim environment
+#      make environments that are treated like verbatim environments configurable (config variable VERBATIMENV)
+#      treat lstinlne as equivalent to verb command
+#      partially addresses issue #38
+#    - refactoring: set default configuration variables in a hash, and those that correspond to lists
+#    - feature: option --add-to-config used to amend configuration variables, which are regex pattern lists
+#    - bug fix: deleted figures when endfloat package is activated
+#    - bug fix: alignat environment now always processed correctly (fix issues #65)
+#    - bug fix: avoid processing of commands as potential files in routine init_regex_arr (fix issue #70 )
+#    - minimal feature enhancement: treat '@' as allowed character in commands (strictly speaking requires prior \makeatletter statement, but always assuming it to be
+#       @       a letter if it is part of a command name will usually lead to the correct behaviour (see http://tex.stackexchange.com/questions/346651/latexdiff-and-let)
+#    - new feature/bug fix: --flatten option \endinput in included files now respected but only if \endinput stands right at the beginning of the line (issue #77)
+#    - bug fix: flatten would incorrectly attempt to process commented out \include commands (from discussion in issue #77 )
+#    - introduce an invisible space (\hspace{0pt} after \mbox{..} auxiliary commands (not in math mode), to allow line breaks between added and deleted citations (change should not cause adverse behaviour otherwise)
+#
+# Version 1.2.0:
+#    - highlight new and deleted figures
+#    - bug fix in title mark-up. Previously deleted commands in title (such as \title, \author or \date) were marked up erroneously
+#    - (minor) bug fixes in new 1.1.1 features: disabled label was commented out twice, additional spaces were introduced before list environment begin and end commands
+#    - depracation fix: left brace in RegEx now needs to be escaped
+#    - add type PDFCOMMENT based on issue #49 submitted by github user peci1 (Martin Pecka)
+#    - make utf8 the default encoding
+#
+# Version 1.1.1
+#    - patch mhchem: allow ce in equations
+#    - flatten now also expands \input etc. in the preamble (but not \usepackage!)
+#    - Better support for Japanese ( contributed by github user kshramt )
+#    - prevent duplicated verbatim hashes (patch contributed by github user therussianjig, issue #36)
+#    - disable deleted label commands (fixes issue #31)
+#    - introduce post-processing to reinstate most deleted environments and all needed item commands (fixes issue #1)
+#
+# Version 1.1.0
+#    - treat diacritics (\",\', etc) as safe commands
+#    - treat \_ and \& correctly as safe commands, even if used without spacing to the next word
+#    - Add a BOLD markup type that sets added text in bold face (Contribution by Victor Zabalza via pull request )
+#    - add append-mboxsafecmd list option to be able to specify special safe commands which need to be surrounded by mbox to avoid breaking (mostly this is needed with ulem package)
+#    - support for siunitx and cleveref packages: protect \SI command in siunitx package and \cref,\Cref{range}{*} in cleveref packages (thanks to Stefan Pinnow for testing)
+#    - experimental support for chemformula, mhchem packages: define \ch and \ce in packages as safe (but not \ch,\cee in equation array environments) - these unfortunately will not be marked up (thanks to Stefan Pinnow for testing)
+#    - bug fix: packages identified correctly even if \usepackage command options extend over several lines (previously \usepackage command needed to be fully contained in one line)
+#    - new subtype ONLYCHANGEDPAGE outputs only changed pages (might not work well for floating material)
+#    - new subtype ZLABEL operates similarly to LABEL but uses absolute page numbers (needs zref package)
+#    - undocumented option --debug/--nodebug to override default setting for debug mode (Default: 0 for release version, 1: for development version
+#
+# Version 1.0.4
+#    - introduce list UNSAFEMATHCMD, which holds list of commands which cannot be marked up with \DIFadd or \DIFdel commands  (only relevant for WHOLE and COARSE math markup modes)
+#    - new subtype LABEL which gives each change a label. This can later be used to only display pages where changes
+#      have been made (instructions for that are put as comments into the diff'ed file) inspired by answer on http://tex.stackexchange.com/questions/166049/invisible-markers-in-pdfs-using-pdflatex
+#    - Configuration variables take into accout some commands from additional packages:
+#      tikzpicture environment now treated as PICTUREENV, and \smallmatrix in ARRENV (amsmath)
+#    - --flatten: support for \subfile command (subfiles package)  (in response to http://tex.stackexchange.com/questions/167620/latexdiff-with-subfiles )
+#    - --flatten: \bibliography commands expand if corresponding bbl file present
+#    - angled bracket optional commands now parsed correctly (patch #3570) submitted by Dave Kleinschmidt (thanks)
+#    - \RequirePackage now treated as synonym of \usepackage with respect to setting packages
+#    - special rules for apacite package (redefine citation commands)
+#    - recognise /dev/null as 'file-like' arguments for --preamble and --config options
+#    - fix units package incompatibility with ulem for text maths statements $ ..$ (thanks to Stuart Prescott for reporting this)
+#    - amsmath environment cases treated correctly (Bug fix #19029) (thanks to Jalar)
+#    - {,} in comments no longer confuse latexdiff (Bug fix #19146)
+#    - \% in one-letter sub/Superscripts was not converted correctly
+#
+# Version 1.0.3
+#    - fix bug in add_safe_commands that made latexdiff hang on DeclareMathOperator
+#      command in preamble
+#    - \(..\) inline math expressions were not parsed correctly, if they contained a linebreak
+#    - applied patch contributed by tomflannaghan via Berlios: [ Patch #3431 ] Adds correct handling of \left< and \right>
+#    - \$ is treated correctly as a literal dollar sign (thanks to Reed Cartwright and Joshua Miller for reporting this bug
+#      and sketching out the solution)
+#    - \^ and \_ are correctly interpreted as accent and underlined space, respectively, not as superscript of subscript
+#      (thanks to Wail Yahyaoui for pointing out this bug)
+#
+# Version 1.0.1 - treat \big,\bigg etc. equivalently to \left and
+#              \right - include starred version in MATHENV - apply
+#            - flatten recursively and --flatten expansion is now
+#              aware of comments (thanks to Tim Connors for patch)
+#            - Change to post-processing for more reliability for
+#              deleted math environments
+#            - On linux systems, recognise  and remove DOS style newlines
+#            - Provide markup for some special preamble commands (\title,
+#              \author,\date,
+#            - configurable by setting context2cmd
+#            - for styles using ulem package, remove \emph and \text.. from list of
+#              safe commands in order to allow linebreaks within the
+#              highlighted sections.
+#            - for ulem style, now show citations by enclosing them in \mbox commands.
+#              This unfortunately implies linebreaks within citations no longer function,
+#              so this functionality can be turned off (Option --disable-citation-markup).
+#              With --enable-citation-markup, the mbox markup is forced for other styles)
+#            - new substyle COLOR.  This is particularly useful for marking up citations
+#              and some special post-processing is implemented to retain cite
+#              commands in deleted blocks.
+#            - four different levels of math-markup
+#            - Option --driver for choosing driver for modes employing changebar package
+#            - accept \\* as valid command (and other commands of form \.*). Also accept
+#              \<nl> (backslashed newline)
+#            - some typo fixes, include commands defined in preamble as safe commands
+#              (Sebastian Gouezel)
+#            - include compared filenames as comments as line 2 and 3 of
+#              the preamble (can be modified with option --label, and suppressed with
+#              --no-label), option --visible-label to show files in generated pdf or dvi
+#              at the beginning of main document
+#
+# Version 0.5  A number of minor improvements based on feedback
+#              Deleted blocks are now shown before added blocks
+#              Package specific processing
+#
+# Version 0.43 unreleased typo in list of styles at the end
+#              Add protect to all \cbstart, \cbend commands
+#              More robust substitution of deleted math commands
+#
+# Version 0.42 November 06  Bug fixes only
+#
+# Version 0.4   March 06 option for fast differencing using UNIX diff command, several minor bug fixes (\par bug, improved highlighting of textcmds)
+#
+# Version 0.3   August 05 improved parsing of displayed math, --allow-spaces
+#               option, several minor bug fixes
+#
+# Version 0.25  October 04 Fix bug with deleted equations, add math mode commands to safecmd, add | to allowed interpunctuation signs
+# Version 0.2   September 04 extension to utf-8 and variable encodings
+# Version 0.1   August 04    First public release
+
+# Inserted block for stand-alone version replaces
+# use Algorithm::Diff qw(traverse_sequences);
+# in standard version
+# The following BEGIN block contains a verbatim copy of
+# Ned Konz' Algorithm::Diff package version 1.15 except
+# that all POD documentation has been stripped out.
+# I encourage you to download and install the Algorithm::Diff
+# package and use the standard latexdiff version instead
+# (current distribution available from http://search.cpan.org/~nedkonz/Algorithm-Diff-1.15
+#  the most recent version can be found via  http://search.cpan.org/search?module=Algorithm::Diff )
+# Please note the LICENCE for Algorithm::Diff :
+#   "Copyright (c) 2000-2002 Ned Konz.  All rights reserved.
+#    This program is free software;
+#    you can redistribute it and/or modify it under the same terms
+#    as Perl itself."
+# The stand-alone version of latexdiff is provided as a convenience
+# for latex users with no knowledge of PERL who do not wish to install
+# additional packages to be able to use latexdiff.  If you believe
+# the inlining of Algorithm::Diff violates its license please contact
+# me and I will modify the latexdiff distribution accordingly.
+# Frederik Tilmann (tilmann@esc.cam.ac.uk)
+BEGIN { 
+package Algorithm::Diff;
+use strict;
+use vars qw($VERSION @EXPORT_OK @ISA @EXPORT);
+use integer;    # see below in _replaceNextLargerWith() for mod to make
+                # if you don't use this
+require Exporter;
+@ISA       = qw(Exporter);
+@EXPORT    = qw();
+@EXPORT_OK = qw(LCS diff traverse_sequences traverse_balanced sdiff);
+$VERSION = sprintf('%d.%02d so', (q$Revision: 1.15 $ =~ /\d+/g));
+
+# McIlroy-Hunt diff algorithm
+# Adapted from the Smalltalk code of Mario I. Wolczko, <mario@wolczko.com>
+# by Ned Konz, perl@bike-nomad.com
+
+
+# Create a hash that maps each element of $aCollection to the set of positions
+# it occupies in $aCollection, restricted to the elements within the range of
+# indexes specified by $start and $end.
+# The fourth parameter is a subroutine reference that will be called to
+# generate a string to use as a key.
+# Additional parameters, if any, will be passed to this subroutine.
+#
+# my $hashRef = _withPositionsOfInInterval( \@array, $start, $end, $keyGen );
+
+sub _withPositionsOfInInterval
+{
+	my $aCollection = shift;    # array ref
+	my $start       = shift;
+	my $end         = shift;
+	my $keyGen      = shift;
+	my %d;
+	my $index;
+	for ( $index = $start ; $index <= $end ; $index++ )
+	{
+		my $element = $aCollection->[$index];
+		my $key = &$keyGen( $element, @_ );
+		if ( exists( $d{$key} ) )
+		{
+			unshift ( @{ $d{$key} }, $index );
+		}
+		else
+		{
+			$d{$key} = [$index];
+		}
+	}
+	return wantarray ? %d : \%d;
+}
+
+# Find the place at which aValue would normally be inserted into the array. If
+# that place is already occupied by aValue, do nothing, and return undef. If
+# the place does not exist (i.e., it is off the end of the array), add it to
+# the end, otherwise replace the element at that point with aValue.
+# It is assumed that the array's values are numeric.
+# This is where the bulk (75%) of the time is spent in this module, so try to
+# make it fast!
+
+sub _replaceNextLargerWith
+{
+	my ( $array, $aValue, $high ) = @_;
+	$high ||= $#$array;
+
+	# off the end?
+	if ( $high == -1 || $aValue > $array->[-1] )
+	{
+		push ( @$array, $aValue );
+		return $high + 1;
+	}
+
+	# binary search for insertion point...
+	my $low = 0;
+	my $index;
+	my $found;
+	while ( $low <= $high )
+	{
+		$index = ( $high + $low ) / 2;
+
+		#		$index = int(( $high + $low ) / 2);		# without 'use integer'
+		$found = $array->[$index];
+
+		if ( $aValue == $found )
+		{
+			return undef;
+		}
+		elsif ( $aValue > $found )
+		{
+			$low = $index + 1;
+		}
+		else
+		{
+			$high = $index - 1;
+		}
+	}
+
+	# now insertion point is in $low.
+	$array->[$low] = $aValue;    # overwrite next larger
+	return $low;
+}
+
+# This method computes the longest common subsequence in $a and $b.
+
+# Result is array or ref, whose contents is such that
+# 	$a->[ $i ] == $b->[ $result[ $i ] ]
+# foreach $i in ( 0 .. $#result ) if $result[ $i ] is defined.
+
+# An additional argument may be passed; this is a hash or key generating
+# function that should return a string that uniquely identifies the given
+# element.  It should be the case that if the key is the same, the elements
+# will compare the same. If this parameter is undef or missing, the key
+# will be the element as a string.
+
+# By default, comparisons will use "eq" and elements will be turned into keys
+# using the default stringizing operator '""'.
+
+# Additional parameters, if any, will be passed to the key generation routine.
+
+sub _longestCommonSubsequence
+{
+	my $a      = shift;    # array ref
+	my $b      = shift;    # array ref
+	my $keyGen = shift;    # code ref
+	my $compare;           # code ref
+
+	# set up code refs
+	# Note that these are optimized.
+	if ( !defined($keyGen) )    # optimize for strings
+	{
+		$keyGen = sub { $_[0] };
+		$compare = sub { my ( $a, $b ) = @_; $a eq $b };
+	}
+	else
+	{
+		$compare = sub {
+			my $a = shift;
+			my $b = shift;
+			&$keyGen( $a, @_ ) eq &$keyGen( $b, @_ );
+		};
+	}
+
+	my ( $aStart, $aFinish, $bStart, $bFinish, $matchVector ) =
+	  ( 0, $#$a, 0, $#$b, [] );
+
+	# First we prune off any common elements at the beginning
+	while ( $aStart <= $aFinish
+		and $bStart <= $bFinish
+		and &$compare( $a->[$aStart], $b->[$bStart], @_ ) )
+	{
+		$matchVector->[ $aStart++ ] = $bStart++;
+	}
+
+	# now the end
+	while ( $aStart <= $aFinish
+		and $bStart <= $bFinish
+		and &$compare( $a->[$aFinish], $b->[$bFinish], @_ ) )
+	{
+		$matchVector->[ $aFinish-- ] = $bFinish--;
+	}
+
+	# Now compute the equivalence classes of positions of elements
+	my $bMatches =
+	  _withPositionsOfInInterval( $b, $bStart, $bFinish, $keyGen, @_ );
+	my $thresh = [];
+	my $links  = [];
+
+	my ( $i, $ai, $j, $k );
+	for ( $i = $aStart ; $i <= $aFinish ; $i++ )
+	{
+		$ai = &$keyGen( $a->[$i], @_ );
+		if ( exists( $bMatches->{$ai} ) )
+		{
+			$k = 0;
+			for $j ( @{ $bMatches->{$ai} } )
+			{
+
+				# optimization: most of the time this will be true
+				if ( $k and $thresh->[$k] > $j and $thresh->[ $k - 1 ] < $j )
+				{
+					$thresh->[$k] = $j;
+				}
+				else
+				{
+					$k = _replaceNextLargerWith( $thresh, $j, $k );
+				}
+
+				# oddly, it's faster to always test this (CPU cache?).
+				if ( defined($k) )
+				{
+					$links->[$k] =
+					  [ ( $k ? $links->[ $k - 1 ] : undef ), $i, $j ];
+				}
+			}
+		}
+	}
+
+	if (@$thresh)
+	{
+		for ( my $link = $links->[$#$thresh] ; $link ; $link = $link->[0] )
+		{
+			$matchVector->[ $link->[1] ] = $link->[2];
+		}
+	}
+
+	return wantarray ? @$matchVector : $matchVector;
+}
+
+sub traverse_sequences
+{
+	my $a                 = shift;                                  # array ref
+	my $b                 = shift;                                  # array ref
+	my $callbacks         = shift || {};
+	my $keyGen            = shift;
+	my $matchCallback     = $callbacks->{'MATCH'} || sub { };
+	my $discardACallback  = $callbacks->{'DISCARD_A'} || sub { };
+	my $finishedACallback = $callbacks->{'A_FINISHED'};
+	my $discardBCallback  = $callbacks->{'DISCARD_B'} || sub { };
+	my $finishedBCallback = $callbacks->{'B_FINISHED'};
+	my $matchVector = _longestCommonSubsequence( $a, $b, $keyGen, @_ );
+
+	# Process all the lines in @$matchVector
+	my $lastA = $#$a;
+	my $lastB = $#$b;
+	my $bi    = 0;
+	my $ai;
+
+	for ( $ai = 0 ; $ai <= $#$matchVector ; $ai++ )
+	{
+		my $bLine = $matchVector->[$ai];
+		if ( defined($bLine) )    # matched
+		{
+			&$discardBCallback( $ai, $bi++, @_ ) while $bi < $bLine;
+			&$matchCallback( $ai,    $bi++, @_ );
+		}
+		else
+		{
+			&$discardACallback( $ai, $bi, @_ );
+		}
+	}
+
+	# The last entry (if any) processed was a match.
+	# $ai and $bi point just past the last matching lines in their sequences.
+
+	while ( $ai <= $lastA or $bi <= $lastB )
+	{
+
+		# last A?
+		if ( $ai == $lastA + 1 and $bi <= $lastB )
+		{
+			if ( defined($finishedACallback) )
+			{
+				&$finishedACallback( $lastA, @_ );
+				$finishedACallback = undef;
+			}
+			else
+			{
+				&$discardBCallback( $ai, $bi++, @_ ) while $bi <= $lastB;
+			}
+		}
+
+		# last B?
+		if ( $bi == $lastB + 1 and $ai <= $lastA )
+		{
+			if ( defined($finishedBCallback) )
+			{
+				&$finishedBCallback( $lastB, @_ );
+				$finishedBCallback = undef;
+			}
+			else
+			{
+				&$discardACallback( $ai++, $bi, @_ ) while $ai <= $lastA;
+			}
+		}
+
+		&$discardACallback( $ai++, $bi, @_ ) if $ai <= $lastA;
+		&$discardBCallback( $ai, $bi++, @_ ) if $bi <= $lastB;
+	}
+
+	return 1;
+}
+
+sub traverse_balanced
+{
+	my $a                 = shift;                                  # array ref
+	my $b                 = shift;                                  # array ref
+	my $callbacks         = shift || {};
+	my $keyGen            = shift;
+	my $matchCallback     = $callbacks->{'MATCH'} || sub { };
+	my $discardACallback  = $callbacks->{'DISCARD_A'} || sub { };
+	my $discardBCallback  = $callbacks->{'DISCARD_B'} || sub { };
+	my $changeCallback    = $callbacks->{'CHANGE'};
+	my $matchVector = _longestCommonSubsequence( $a, $b, $keyGen, @_ );
+
+	# Process all the lines in match vector
+	my $lastA = $#$a;
+	my $lastB = $#$b;
+	my $bi    = 0;
+	my $ai    = 0;
+	my $ma    = -1;
+	my $mb;
+
+	while (1)
+	{
+
+		# Find next match indices $ma and $mb
+		do { $ma++ } while ( $ma <= $#$matchVector && !defined $matchVector->[$ma] );
+
+		last if $ma > $#$matchVector;    # end of matchVector?
+		$mb = $matchVector->[$ma];
+
+		# Proceed with discard a/b or change events until
+		# next match
+		while ( $ai < $ma || $bi < $mb )
+		{
+
+			if ( $ai < $ma && $bi < $mb )
+			{
+
+				# Change
+				if ( defined $changeCallback )
+				{
+					&$changeCallback( $ai++, $bi++, @_ );
+				}
+				else
+				{
+					&$discardACallback( $ai++, $bi, @_ );
+					&$discardBCallback( $ai, $bi++, @_ );
+				}
+			}
+			elsif ( $ai < $ma )
+			{
+				&$discardACallback( $ai++, $bi, @_ );
+			}
+			else
+			{
+
+				# $bi < $mb
+				&$discardBCallback( $ai, $bi++, @_ );
+			}
+		}
+
+		# Match
+		&$matchCallback( $ai++, $bi++, @_ );
+	}
+
+	while ( $ai <= $lastA || $bi <= $lastB )
+	{
+		if ( $ai <= $lastA && $bi <= $lastB )
+		{
+
+			# Change
+			if ( defined $changeCallback )
+			{
+				&$changeCallback( $ai++, $bi++, @_ );
+			}
+			else
+			{
+				&$discardACallback( $ai++, $bi, @_ );
+				&$discardBCallback( $ai, $bi++, @_ );
+			}
+		}
+		elsif ( $ai <= $lastA )
+		{
+			&$discardACallback( $ai++, $bi, @_ );
+		}
+		else
+		{
+
+			# $bi <= $lastB
+			&$discardBCallback( $ai, $bi++, @_ );
+		}
+	}
+
+	return 1;
+}
+
+sub LCS
+{
+	my $a = shift;                                           # array ref
+	my $matchVector = _longestCommonSubsequence( $a, @_ );
+	my @retval;
+	my $i;
+	for ( $i = 0 ; $i <= $#$matchVector ; $i++ )
+	{
+		if ( defined( $matchVector->[$i] ) )
+		{
+			push ( @retval, $a->[$i] );
+		}
+	}
+	return wantarray ? @retval : \@retval;
+}
+
+sub diff
+{
+	my $a      = shift;    # array ref
+	my $b      = shift;    # array ref
+	my $retval = [];
+	my $hunk   = [];
+	my $discard = sub { push ( @$hunk, [ '-', $_[0], $a->[ $_[0] ] ] ) };
+	my $add = sub { push ( @$hunk, [ '+', $_[1], $b->[ $_[1] ] ] ) };
+	my $match = sub { push ( @$retval, $hunk ) if scalar(@$hunk); $hunk = [] };
+	traverse_sequences( $a, $b,
+		{ MATCH => $match, DISCARD_A => $discard, DISCARD_B => $add }, @_ );
+	&$match();
+	return wantarray ? @$retval : $retval;
+}
+
+sub sdiff
+{
+	my $a      = shift;    # array ref
+	my $b      = shift;    # array ref
+	my $retval = [];
+	my $discard = sub { push ( @$retval, [ '-', $a->[ $_[0] ], "" ] ) };
+	my $add = sub { push ( @$retval, [ '+', "", $b->[ $_[1] ] ] ) };
+	my $change = sub {
+		push ( @$retval, [ 'c', $a->[ $_[0] ], $b->[ $_[1] ] ] );
+	};
+	my $match = sub {
+		push ( @$retval, [ 'u', $a->[ $_[0] ], $b->[ $_[1] ] ] );
+	};
+	traverse_balanced(
+		$a,
+		$b,
+		{
+			MATCH     => $match,
+			DISCARD_A => $discard,
+			DISCARD_B => $add,
+			CHANGE    => $change,
+		},
+		@_
+	);
+	return wantarray ? @$retval : $retval;
+}
+
+1;
+}
+import Algorithm::Diff qw(traverse_sequences);
+# End of inserted block for stand-alone version
+
+use Getopt::Long ;
+use strict ;
+use warnings;
+use utf8 ;
+
+use File::Spec ;
+
+
+my ($algodiffversion)=split(/ /,$Algorithm::Diff::VERSION);
+
+
+my ($versionstring)=<<EOF ;
+This is LATEXDIFF 1.3.4 (Algorithm::Diff $Algorithm::Diff::VERSION, Perl $^V)
+  (c) 2004-2024 F J Tilmann
+EOF
+
+# Hash with defaults for configuration variables. These marked undef have default values constructed from list defined in the DATA block
+# (under tag CONFIG)
+my %CONFIG=(
+   MINWORDSBLOCK => 3, # minimum number of tokens to form an independent block
+                        # shorter identical blocks will be merged to the previous word
+   SCALEDELGRAPHICS => 0.5, # factor with which deleted figures will be scaled down (i.e. 0.5 implies they are shown at half linear size)
+                             # this is only used for --dgraphics-markup=BOTH option
+   FLOATENV => undef ,   # Environments in which FL variants of defined commands are used
+   PICTUREENV => undef ,   # Environments in which all change markup is removed
+   MATHENV => undef ,           # Environments turning on display math mode (code also knows about \[ and \])
+   MATHREPL => 'displaymath',  # Environment introducing deleted maths blocks
+   MATHARRENV => undef ,           # Environments turning on eqnarray math mode
+   MATHARRREPL => 'eqnarray*',  # Environment introducing deleted maths blocks (note that now the starred varieties are being used, so this is only used to replace MATHMODE environments (where original environment is unknown)
+   ARRENV => undef , # Environments making arrays in math mode.  The underlining style does not cope well with those - as a result in-text math environments are surrounded by \mbox{ } if any of these commands is used in an inline math block
+   COUNTERCMD => undef,
+                                        # COUNTERCMD textcmds which are associated with a counter
+                                        # If any of these commands occur in a deleted block
+                                        # they will be followed by an \addtocounter{...}{-1}
+                                        # for the associated counter such that the overall numbers
+                                        # should be the same as in the new file
+   LISTENV => undef ,  # list making environments - they will generally be kept
+   VERBATIMENV => undef,      # Environments whose content should be treated as verbatim text and not be touched
+   VERBATIMLINEENV => undef,  # Environments whose content should be treated as verbatim text and processed in line diff mode
+   CUSTOMDIFCMD => undef,# Custom dif command. Is defined in the document as a \DELcommand and \ADDcommand version to be replaced by the diff
+   ITEMCMD => 'item'                    # command marking item in a list environment
+);
+# Configuration variables: these have to be visible from the subroutines
+my ($ARRENV,
+    $COUNTERCMD,
+    $FLOATENV,
+    $ITEMCMD,
+    $LISTENV,
+    $MATHARRENV,
+    $MATHARRREPL,
+    $MATHENV,
+    $MATHREPL,
+    $MINWORDSBLOCK,
+    $PICTUREENV,
+    $SCALEDELGRAPHICS,
+    $VERBATIMENV,
+    $VERBATIMLINEENV,
+    $CUSTOMDIFCMD
+    );
+
+# my $MINWORDSBLOCK=3; # minimum number of tokens to form an independent block
+#                      # shorter identical blocks will be merged to the previous word
+# my $SCALEDELGRAPHICS=0.5; # factor with which deleted figures will be scaled down (i.e. 0.5 implies they are shown at half linear size)
+#                       # this is only used for --graphics-markup=BOTH option
+# my $FLOATENV='(?:figure|table|plate)[\w\d*@]*' ;   # Environments in which FL variants of defined commands are used
+# my $PICTUREENV='(?:picture|tikzpicture|DIFnomarkup)[\w\d*@]*' ;   # Environments in which all change markup is removed
+# my $MATHENV='(?:equation[*]?|displaymath|DOLLARDOLLAR)[*]?' ;           # Environments turning on display math mode (code also knows about \[ and \])
+# my $MATHREPL='displaymath';  # Environment introducing deleted maths blocks
+# my $MATHARRENV='(?:eqnarray|align|alignat|gather|multline|flalign)[*]?' ;           # Environments turning on eqnarray math mode
+# my $MATHARRREPL='eqnarray*';  # Environment introducing deleted maths blocks
+# my $ARRENV='(?:aligned|gathered|multlined|array|[pbvBV]?matrix|smallmatrix|cases|split)'; # Environments making arrays in math mode.  The underlining style does not cope well with those - as a result in-text math environments are surrounded by \mbox{ } if any of these commands is used in an inline math block
+# my $COUNTERCMD='(?:footnote|part|chapter|section|subsection|subsubsection|paragraph|subparagraph)';  # textcmds which are associated with a counter
+#                                         # If any of these commands occur in a deleted block
+#                                         # they will be succeeded by an \addtocounter{...}{-1}
+#                                         # for the associated counter such that the overall numbers
+#                                         # should be the same as in the new file
+# my $LISTENV='(?:itemize|description|enumerate)'; # list making environments - they will generally be kept
+# my $ITEMCMD='item';   # command marking item in a list environment
+
+
+
+my $LABELCMD='(?:label)';                # matching commands are disabled within deleted blocks - mostly useful for maths mode, as otherwise it would be fine to just not add those to SAFECMDLIST
+my @UNSAFEMATHCMD=('qedhere','intertext','begin','end');           # Commands which are definitely unsafe for marking up in math mode (amsmath qedhere only tested to not work with UNDERLINE markup) (only affects WHOLE and COARSE math markup modes). Note that unlike text mode (or FINE math mode0 deleted unsafe commands are not deleted but simply taken outside \DIFdel
+my $MBOXINLINEMATH=0; # if set to 1 then surround marked-up inline maths expression with \mbox ( to get around compatibility
+                      # problems between some maths packages and ulem package
+
+
+# Markup strings
+# If at all possible, do not change these as parts of the program
+# depend on the actual name (particularly post-processing)
+# At the very least adapt subroutine postprocess to new tokens.
+my $ADDMARKOPEN='\DIFaddbegin ';   # Token to mark begin of appended text
+my $ADDMARKCLOSE='\DIFaddend ';   # Token to mark end of appended text
+my $ADDOPEN='\DIFadd{';  # To mark begin of added text passage
+my $ADDCLOSE='}';        # To mark end of added text passage
+my $ADDCOMMENT='DIF > ';   # To mark added comment line
+my $DELMARKOPEN='\DIFdelbegin ';   # Token to mark begin of deleted text
+my $DELMARKCLOSE='\DIFdelend ';   # Token to mark end of deleted text
+my $DELOPEN='\DIFdel{';  # To mark begin of deleted text passage
+my $DELCLOSE='}';        # To mark end of deleted text passage
+my $DELCMDOPEN='%DIFDELCMD < ';  # To mark begin of deleted commands (must begin with %, i.e., be a comment
+my $DELCMDCLOSE="%%%\n";    # To mark end of deleted commands (must end with a new line)
+my $AUXCMD='%DIFAUXCMD' ; #  follows auxiliary commands put in by latexdiff to make difference file legal
+                          # auxiliary commands must be on a line of their own
+                          # Note that for verbatim environment openings the %DIFAUXCMD cannot be placed in
+                          # the same line as this would mean they are shown
+                          # so the special form "%DIFAUXCMD NEXT" is used to indicate that the next line
+                          # is an auxiliary command
+                          # Similarly "%DIFAUXCMD LAST" would indicate the auxiliary command is in previous line (not currently used)
+my $DELCOMMENT='DIF < ';   # To mark deleted comment line
+my $VERBCOMMENT='DIFVRB ';  # to mark lines which are within a verbatim environment
+
+# main local variables:
+my @TEXTCMDLIST=();  # array containing patterns of commands with text arguments
+my @TEXTCMDEXCL=();  # array containing patterns of commands without text arguments (if a pattern
+                     # matches both TEXTCMDLIST and TEXTCMDEXCL it is excluded)
+my @CONTEXT1CMDLIST=();  # array containing patterns of commands with text arguments (subset of text commands),
+                         # but which cause confusion if used out of context (e.g. \caption).
+                         # In deleted passages, the command will be disabled but its argument is marked up
+                         # Otherwise they behave exactly like TEXTCMD's
+my @CONTEXT1CMDEXCL=();  # exclude list for above, but always empty
+my @CONTEXT2CMDLIST=();  # array containing patterns of commands with text arguments, but which fail or cause confusion
+                         # if used out of context (e.g. \title). They and their arguments will be disabled in deleted
+                         # passages
+my @CONTEXT2CMDEXCL=();  # exclude list for above, but always empty
+my @MATHTEXTCMDLIST=();  # treat like textcmd.  If a textcmd is in deleted or added block, just wrap the
+                         # whole content with \DIFadd or \DIFdel irrespective of content.  This functionality
+                         # is useful for pseudo commands \MATHBLOCK.. into which math environments are being
+                         # transformed
+my @MATHTEXTCMDEXCL=();  #
+
+# Note I need to declare this with "our" instead of "my" because later in the code I have to "local"ise these
+our @SAFECMDLIST=();  # array containing patterns of safe commands (which do not break when in the argument of DIFadd or DIFDEL)
+our @SAFECMDEXCL=();
+my @MBOXCMDLIST=();   # patterns for commands which are in principle safe but which need to be surrounded by an \mbox
+my @MBOXCMDEXCL=();           # all the patterns in MBOXCMDLIST will be appended to SAFECMDLIST
+
+my @KEEPCMDLIST=( qr/^bibitem$/ );   # patterns for commands which should not be deleted in nominally delete text passages
+my @KEEPCMDEXCL=();
+
+my ($i,$j,$l);
+my ($old,$new);
+my ($line,$key);
+my (@dumlist);
+my ($newpreamble,$oldpreamble);
+my (@newpreamble,@oldpreamble,@diffpreamble,@diffbody);
+my ($latexdiffpreamble);
+my ($oldbody, $newbody, $diffbo);
+my ($oldpost, $newpost);
+my ($diffall);
+# Option names
+my ($type,$subtype,$floattype,$config,$preamblefile,$encoding,$nolabel,$visiblelabel,
+    $filterscript,$ignorefilterstderr,
+    $showpreamble,$showsafe,$showtext,$showconfig,$showall,
+    $replacesafe,$appendsafe,$excludesafe,
+    $replacetext,$appendtext,$excludetext,
+    $replacecontext1,$appendcontext1,
+    $replacecontext2,$appendcontext2,
+    $help,$verbose,$driver,$version,$ignorewarnings,
+    $onlyadditions,
+    $enablecitmark,$disablecitmark,$allowspaces,$flatten,$nolinks,$debug,$earlylatexdiffpreamble);  ###$disablemathmark,
+my ($mboxsafe);
+# MNEMNONICS for mathmarkup
+my $mathmarkup;
+use constant {
+  OFF => 0,
+  WHOLE => 1,
+  COARSE => 2,
+  FINE => 3
+};
+# MNEMNONICS for graphicsmarkup
+my $graphicsmarkup;
+use constant {
+  NONE => 0,
+  NEWONLY => 1,
+  BOTH => 2
+};
+
+my ($mboxcmd);
+
+my (@configlist,@addtoconfiglist,@labels,
+    @appendsafelist,@excludesafelist,
+    @appendmboxsafelist,@excludemboxsafelist,
+    @appendtextlist,@excludetextlist,
+    @appendcontext1list,@appendcontext2list,
+    @packagelist);
+my ($assign,@config);
+# Hash where keys corresponds to the names of  all included packages (including the documentclass as another package
+# the optional arguments to the package are the values of the hash elements
+my ($pkg,%packages);
+
+# Defaults
+$mathmarkup=COARSE;
+$verbose=0;
+$onlyadditions=0;
+# output debug and intermediate files, set to 0 in final distribution
+$debug=0;
+# insert preamble directly after documentclass - experimental feature, set to 0 in final distribution
+# Note that this failed with mini example (or other files, where packages used in latexdiff preamble
+# are called again with incompatible options in preamble of resulting file)
+$earlylatexdiffpreamble=0;
+
+
+# define character properties
+sub IsNonAsciiPunct { return <<'END'    # Unicode punctuation but excluding ASCII punctuation
++utf8::IsPunct
+-utf8::IsASCII
+END
+}
+sub IsNonAsciiS { return <<'END'       # Unicode symbol but excluding ASCII
++utf8::IsS
+-utf8::IsASCII
+END
+}
+
+
+my %verbhash;
+
+Getopt::Long::Configure('bundling');
+GetOptions('type|t=s' => \$type,
+	   'subtype|s=s' => \$subtype,
+	   'floattype|f=s' => \$floattype,
+	   'config|c=s' => \@configlist,
+	   'add-to-config=s' => \@addtoconfiglist,
+	   'preamble|p=s' => \$preamblefile,
+	   'encoding|e=s' => \$encoding,
+	   'label|L=s' => \@labels,
+	   'no-label' => \$nolabel,
+	   'visible-label' => \$visiblelabel,
+	   'exclude-safecmd|A=s' => \@excludesafelist,
+	   'replace-safecmd=s' => \$replacesafe,
+	   'append-safecmd|a=s' => \@appendsafelist,
+	   'exclude-textcmd|X=s' => \@excludetextlist,
+	   'replace-textcmd=s' => \$replacetext,
+	   'append-textcmd|x=s' => \@appendtextlist,
+	   'replace-context1cmd=s' => \$replacecontext1,
+	   'append-context1cmd=s' => \@appendcontext1list,
+	   'replace-context2cmd=s' => \$replacecontext2,
+	   'append-context2cmd=s' => \@appendcontext2list,
+	   'exclude-mboxsafecmd=s' => \@excludemboxsafelist,
+	   'append-mboxsafecmd=s' => \@appendmboxsafelist,
+	   'show-preamble' => \$showpreamble,
+	   'show-safecmd' => \$showsafe,
+	   'show-textcmd' => \$showtext,
+	   'show-config' => \$showconfig,
+	   'show-all' => \$showall,
+           'packages=s' => \@packagelist,
+	   'allow-spaces' => \$allowspaces,
+           'math-markup=s' => \$mathmarkup,
+           'graphics-markup=s' => \$graphicsmarkup,
+           'enable-citation-markup|enforce-auto-mbox' => \$enablecitmark,
+           'disable-citation-markup|disable-auto-mbox' => \$disablecitmark,
+	   'verbose|V' => \$verbose,
+	   'ignore-warnings' => \$ignorewarnings,
+	   'driver=s'=> \$driver,
+	   'flatten' => \$flatten,
+	   'filter-script=s' => \$filterscript,
+	   'ignore-filter-stderr' => \$ignorefilterstderr,
+	   'no-links' => \$nolinks,
+	   'no-del' => \$onlyadditions,
+	   'version' => \$version,
+	   'help|h' => \$help,
+	   'debug!' => \$debug ) or die "Use latexdiff -h to get help.\n" ;
+
+if ( $help ) {
+  usage() ;
+}
+
+
+if ( $version ) {
+  print STDERR $versionstring ;
+  exit 0;
+}
+
+print STDERR $versionstring if $verbose;
+
+if (defined($showall)){
+  $showpreamble=$showsafe=$showtext=$showconfig=1;
+}
+# Default types
+$type='UNDERLINE' unless defined($type);
+$subtype='SAFE' unless defined($subtype);
+# set floattype to IDENTICAL for LABEL and ONLYCHANGEDPAGE subtype, unless it has been set explicitly on the command line
+$floattype=($subtype eq 'LABEL' || $subtype eq 'ONLYCHANGEDPAGE') ? 'IDENTICAL' : 'FLOATSAFE' unless defined($floattype);
+if ( $subtype eq 'LABEL' ) {
+  print STDERR "Note that LABEL subtype is deprecated. If possible, use ZLABEL instead (requires zref package)";
+}
+
+if (defined($mathmarkup)) {
+  $mathmarkup=~tr/a-z/A-Z/;
+  if ( $mathmarkup eq 'OFF' ){
+    $mathmarkup=OFF;
+  } elsif ( $mathmarkup eq 'WHOLE' ){
+    $mathmarkup=WHOLE;
+  } elsif ( $mathmarkup eq 'COARSE' ){
+    $mathmarkup=COARSE;
+  } elsif ( $mathmarkup eq 'FINE' ){
+    $mathmarkup=FINE;
+  } elsif ( $mathmarkup !~ m/^[0123]$/ ) {
+    die "latexdiff Illegal value: ($mathmarkup)  for option--math-markup. Possible values: OFF,WHOLE,COARSE,FINE,0-3\n";
+  }
+  # else use numerical value
+}
+
+# Give filterscript a default empty string
+$filterscript="" unless defined($filterscript);
+
+# setting extra preamble commands
+if (defined($preamblefile)) {
+  $latexdiffpreamble=join "\n",(extrapream($preamblefile),"");
+} else {
+  $latexdiffpreamble=join "\n",(extrapream($type,$subtype,$floattype),"");
+}
+
+if ( defined($driver) ) {
+  # for changebar only
+  $latexdiffpreamble=~s/\[pdftex\]/[$driver]/sg;
+}
+# setting up @SAFECMDLIST and @SAFECMDEXCL
+if (defined($replacesafe)) {
+  init_regex_arr_ext(\@SAFECMDLIST,$replacesafe);
+} else {
+  init_regex_arr_data(\@SAFECMDLIST, "SAFE COMMANDS");
+}
+foreach $appendsafe ( @appendsafelist ) {
+  init_regex_arr_ext(\@SAFECMDLIST, $appendsafe);
+}
+foreach $excludesafe ( @excludesafelist ) {
+  init_regex_arr_ext(\@SAFECMDEXCL, $excludesafe);
+}
+# setting up @MBOXCMDLIST and @MBOXCMDEXCL
+foreach $mboxsafe ( @appendmboxsafelist ) {
+  init_regex_arr_ext(\@MBOXCMDLIST, $mboxsafe);
+}
+foreach $mboxsafe ( @excludemboxsafelist ) {
+  init_regex_arr_ext(\@MBOXCMDEXCL, $mboxsafe);
+}
+
+
+
+# setting up @TEXTCMDLIST and @TEXTCMDEXCL
+if (defined($replacetext)) {
+  init_regex_arr_ext(\@TEXTCMDLIST,$replacetext);
+} else {
+  init_regex_arr_data(\@TEXTCMDLIST, "TEXT COMMANDS");
+}
+foreach $appendtext ( @appendtextlist ) {
+  init_regex_arr_ext(\@TEXTCMDLIST, $appendtext);
+}
+foreach $excludetext ( @excludetextlist ) {
+  init_regex_arr_ext(\@TEXTCMDEXCL, $excludetext);
+}
+
+
+# setting up @CONTEXT1CMDLIST ( @CONTEXT1CMDEXCL exist but is always empty )
+if (defined($replacecontext1)) {
+  init_regex_arr_ext(\@CONTEXT1CMDLIST,$replacecontext1);
+} else {
+  init_regex_arr_data(\@CONTEXT1CMDLIST, "CONTEXT1 COMMANDS");
+}
+foreach $appendcontext1 ( @appendcontext1list ) {
+  init_regex_arr_ext(\@CONTEXT1CMDLIST, $appendcontext1);
+}
+
+
+# setting up @CONTEXT2CMDLIST ( @CONTEXT2CMDEXCL exist but is always empty )
+if (defined($replacecontext2)) {
+  init_regex_arr_ext(\@CONTEXT2CMDLIST,$replacecontext2);
+} else {
+  init_regex_arr_data(\@CONTEXT2CMDLIST, "CONTEXT2 COMMANDS");
+}
+foreach $appendcontext2 ( @appendcontext2list ) {
+  init_regex_arr_ext(\@CONTEXT2CMDLIST, $appendcontext2);
+}
+
+# setting configuration variables
+@config=();
+foreach $config ( @configlist ) {
+  if (-f $config || lc $config eq '/dev/null' ) {
+    open(FILE,$config) or die ("Couldn't open configuration file $config: $!");
+    while (<FILE>) {
+      chomp;
+      next if /^\s*#/ || /^\s*%/ || /^\s*$/ ;
+      push (@config,$_);
+    }
+    close(FILE);
+  }
+  else {
+#    foreach ( split(",",$config) ) {
+#      push @config,$_;
+#    }
+     push @config,split(",",$config)
+  }
+}
+foreach $assign ( @config ) {
+  $assign=~ m/\s*(\w*)\s*=\s*(\S*)\s*$/ or die "Illegal assignment $assign in configuration list (must be variable=value)";
+  exists $CONFIG{$1} or die "Unknown configuration variable $1.";
+  $CONFIG{$1}=$2;
+}
+
+my @addtoconfig=();
+foreach $config ( @addtoconfiglist ) {
+  if (-f $config || lc $config eq '/dev/null' ) {
+    open(FILE,$config) or die ("Couldn't open addd-to-config file $config: $!");
+    while (<FILE>) {
+      chomp;
+      next if /^\s*#/ || /^\s*%/ || /^\s*$/ ;
+      push (@addtoconfig,$_);
+    }
+    close(FILE);
+  }
+  else {
+#    foreach ( split(",",$config) ) {
+#      push @addtoconfig,$_;
+#    }
+     push @addtoconfig,split(",",$config)
+  }
+}
+
+# initialise default lists from DATA
+# for those configuration variables, which have not been set explicitly, initiate from list in document
+foreach $key ( keys(%CONFIG) ) {
+  if (!defined $CONFIG{$key}) {
+    @dumlist=();
+    init_regex_arr_data(\@dumlist,"$key CONFIG");
+    $CONFIG{$key}=join(";",@dumlist)
+  }
+}
+
+
+foreach $assign ( @addtoconfig ) {
+  ###print STDERR "assign:|$assign|\n";
+  $assign=~ m/\s*(\w*)\s*=\s*(\S*)\s*$/ or die "Illegal assignment $assign in configuration list (must be variable=value)";
+  exists $CONFIG{$1} or die "Unknown configuration variable $1.";
+  $CONFIG{$1}.=";$2";
+}
+
+# Map from hash to variables (we do this to have more concise code later, change from comma-separated list)
+foreach  ( keys(%CONFIG) ) {
+  if ( $_ eq "MINWORDSBLOCK" ) { $MINWORDSBLOCK = $CONFIG{$_}; }
+  elsif ( $_ eq "FLOATENV" ) { $FLOATENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "ITEMCMD" ) { $ITEMCMD = $CONFIG{$_} ; }
+  elsif ( $_ eq "LISTENV" ) { $LISTENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "PICTUREENV" ) { $PICTUREENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "MATHENV" ) { $MATHENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "MATHREPL" ) { $MATHREPL = $CONFIG{$_} ; }
+  elsif ( $_ eq "MATHARRENV" ) { $MATHARRENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "MATHARRREPL" ) { $MATHARRREPL = $CONFIG{$_} ;
+				  print STDERR "WARNING: Setting MATHARRREPL is depracated. Generally deleted math array environments will be set to their starred varieties and the setting of MATHARREPL is ignored.\n\n" unless $MATHARRREPL =~ /eqnarray\*/ ;
+				}
+  elsif ( $_ eq "ARRENV" ) { $ARRENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "VERBATIMENV" ) { $VERBATIMENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "VERBATIMLINEENV" ) { $VERBATIMLINEENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "CUSTOMDIFCMD" ) { $CUSTOMDIFCMD = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "COUNTERCMD" ) { $COUNTERCMD = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "SCALEDELGRAPHICS" ) { $SCALEDELGRAPHICS = $CONFIG{$_} ; }
+  else { die "Unknown configuration variable $_.";}
+}
+
+if ( $mathmarkup == COARSE || $mathmarkup == WHOLE ) {
+  push(@MATHTEXTCMDLIST,qr/^MATHBLOCK(?:$MATHENV|$MATHARRENV|SQUAREBRACKET)$/);
+}
+
+
+
+
+
+foreach $pkg ( @packagelist ) {
+  map { $packages{$_}="" } split(/,/,$pkg) ;
+}
+
+
+if ($showconfig || $showtext || $showsafe || $showpreamble) {
+
+  show_configuration();
+  exit 0;
+}
+
+
+if ( @ARGV != 2 ) {
+  print STDERR "2 and only 2 non-option arguments required.  Write latexdiff -h to get help\n";
+  exit(2);
+}
+
+# Are extra spaces between command arguments permissible?
+my $extraspace;
+if ($allowspaces) {
+  $extraspace='\s*';
+} else {
+  $extraspace='';
+}
+
+# append context lists to text lists (as text property is implied)
+push @TEXTCMDLIST, @CONTEXT1CMDLIST;
+push @TEXTCMDLIST, @CONTEXT2CMDLIST;
+
+push @TEXTCMDLIST, @MATHTEXTCMDLIST if $mathmarkup==COARSE;
+
+# internal additions to SAFECMDLIST
+push(@SAFECMDLIST, qr/^QLEFTBRACE$/, qr/^QRIGHTBRACE$/);
+
+
+# Patterns. These are used by some of the subroutines, too
+# I can only define them down here because value of extraspace depends on an option
+
+  my $pat0 = '(?:[^{}])*';
+  my $pat_n = $pat0;
+# if you get "undefined control sequence MATHBLOCKmath" error, increase the maximum value in this loop
+  for (my $i_pat = 0; $i_pat < 20; ++$i_pat){
+    $pat_n = '(?:[^{}]|\{'.$pat_n.'\}|\\\\\{|\\\\\})*';
+    # Actually within the text body, quoted braces are replaced in pre-processing. The only place where
+    # the last part of the pattern matters is when processing the arguments of context2cmds in the preamble
+    # and these contain a \{ or \} combination, probably rare.
+    # It should thus be fine to use the simpler version below.
+    ###  $pat_n = '(?:[^{}]|\{'.$pat_n.'\})*';
+  }
+
+  my $brat0 = '(?:[^\[\]]|\\\[|\\\])*';
+  my $brat_n = $brat0;
+  for (my $i_pat = 0; $i_pat < 4; ++$i_pat){
+    $brat_n = '(?:[^\[\]]|\['.$brat_n.'\]|\\\[|\\\])*';
+    ###  $brat_n = '(?:[^\[\]]|\['.$brat_n.'\])*';   # Version not taking into account escaped \[ and \]
+  }
+  my $abrat0 = '(?:[^<>])*';
+
+
+# variable definitions are in order that they are matched
+  my $and = '&';
+  my $quotemarks = '(?:\'\')|(?:\`\`)';
+  # some common abbreviations involving punctuations within them. They need special treatment because otherwise in some
+  # circumstances the gnoring of white space differences in conjunction with merging according to MINWORDSBLOCK rule
+  # could turn 'i.e.' into 'i.\PAR e.' (see https://github.com/ftilmann/latexdiff/issues/269)
+  # English: i.e., e.g.   Deutsch: z.B.
+  my $abbreviation='(?:i\. ?e\.|e\. ?g\.|z\. ?B\.)' ;
+  my $number='-?\d*\.\d*';
+
+  # word: sequence of letters or accents followed by letter
+  my $word_cj='\p{Han}|\p{InHiragana}|\p{InKatakana}';
+  my $word='(?:' . $word_cj . '|(?:(?:[-\w\d*]|\\\\[\"\'\`~^][A-Za-z\*])(?!(?:' . $word_cj . ')))+)';
+
+  # quoted underscore - this needs special treatment as perl treats _ as a letter (\w) but latex does not
+  # such that a\_b would otherwise be interpreted as a{\_}b by latex but a{\_b} by latexdiff
+  my $quotedunderscore='\\\\_';
+  # Handle tex \def macro: \def\MAKRONAME#1[#2]#3{DEFINITION}
+  my $defseq='\\\\def\\\\[\w\d@\*]+(?:#\d+|\[#\d+\])+(?:\{'. $pat_n . '\})?';
+  my $cmdleftright='\\\\(?:left|right|[Bb]igg?[lrm]?|middle)\s*(?:[<>()\[\]|\.]|\\\\(?:[|{}]|\w+))';
+
+  # for selected commands, the number of arguments is known, and we can therefore allow spaces between command and its argument
+  # Note that it is still expected that the arguments are blocks marked by parentheses rather than single characters, and that intervening comments will inhibit the association
+  my $predefinedcmdoptseq12='\\\\(?:href|bibfield|bibinfo)\s*(?:\['.$brat_n.'\])?\s*(?:\{'. $pat_n . '\}\s*){2}';  # Commands with one optional and two non-optional arguments
+  my $predefinedcmdoptseq01='\\\\(?:url|BibitemShut)\s*\s*(?:\{'. $pat_n . '\}\s*){1}';  # Commands with one non-optional argument
+  # \bibitem in revtex styles appears to be always followed by \BibItemOpen. We bind \BibItemOpen to the bibitem (if present) in order to prevent the comparison algorithm to interpret the \BibItemOpen as an identical part of the sequence; this interpretation can lead to added and removed entries to the reference list to become mixed.
+  my $predefinedbibitem='\\\\(?:bibitem)\s*(?:\['.$brat_n.'\])?\s*(?:\{'. $pat_n . '\})(?:%?\s*\\\\BibitemOpen)?';  # Commands with one optional and one non-optional arguments
+  my $predefinedcmdoptseq='(?:'.$predefinedcmdoptseq12.'|'.$predefinedcmdoptseq01.'|'.$predefinedbibitem.')';
+
+  # standard $cmdoptseq (default: no intrevening spaces, controlled by extraspcae) - a final open parentheses is merged to the commend if it exists to deal properly with multi-argument text command
+  my $coords= '[\-.,\s\d]*';
+  my $cmdoptseq='\\\\[\w\d@\*]+'.$extraspace.'(?:(?:<'.$abrat0.'>|\['.$brat_n.'\]|\{'. $pat_n . '\}|\(' . $coords .'\))'.$extraspace.')*\{?';
+
+  # inline math $....$ or \(..\)
+  ### the commented out version is simpler but for some reason cannot cope with newline (in spite of s option) - need to include \newline explicitly
+  ###  my $math='\$(?:[^$]|\\\$)*?\$|\\\\[(].*?\\\\[)]';
+  my $math='\$(?:[^$]|\\\$)*?\$|\\\\[(](?:.|\n)*?\\\\[)]';
+  ### test version (this seems to give the same results as version above)
+  ## the current maths command cannot cope with newline within the math expression
+  ### my $math='\$(?:[^$]|\\\$)*?\$|\\[(].*?\\[)]';
+  ###  my $math='\$(?:[^$]|\\\$)*\$';
+
+  my $backslashnl='\\\\\n';
+  my $oneletcmd='\\\\.\*?(?:\['.$brat_n.'\]|\{'. $pat_n . '\})*';
+  my $comment='%[^\n]*\n';
+  my $punct='[0.,\/\'\`:;\"\?\(\)\[\]!~\p{IsNonAsciiPunct}\p{IsNonAsciiS}]';
+  my $mathpunct='[+=<>\-\|]';
+
+  # Assembled pattern
+  my $pat=qr/(?:\A\s*)?(?:${abbreviation}|${and}|${quotemarks}|${number}|${word}|$quotedunderscore|${defseq}|$cmdleftright|${predefinedcmdoptseq}|${cmdoptseq}|${math}|${backslashnl}|${oneletcmd}|${comment}|${punct}|${mathpunct}|\{|\})\s*/ ;
+
+
+
+
+# now we are done setting up and can start working
+my ($oldfile, $newfile) = @ARGV;
+# check for existence of input files
+if ( ! -e $oldfile ) {
+  die "Input file $oldfile does not exist";
+}
+if ( ! -e $newfile ) {
+  die "Input file $newfile does not exist";
+}
+
+
+# set the labels to be included into the file
+# first find out which file name is longer for correct alignment
+my ($diff,$oldlabel_n_spaces,$newlabel_n_spaces);
+$oldlabel_n_spaces = 0;
+$newlabel_n_spaces = 0;
+$diff = length($newfile) - length($oldfile);
+if ($diff > 0) {
+  $oldlabel_n_spaces = $diff;
+}
+if ($diff < 0) {
+  $newlabel_n_spaces = abs($diff);
+}
+
+my ($oldtime,$newtime,$oldlabel,$newlabel);
+if (defined($labels[0])) {
+  $oldlabel=$labels[0] ;
+} else {
+  $oldtime=localtime((stat($oldfile))[9]);
+  $oldlabel="$oldfile   " . " "x($oldlabel_n_spaces) . $oldtime;
+}
+if (defined($labels[1])) {
+  $newlabel=$labels[1] ;
+} else {
+  $newtime=localtime((stat($newfile))[9]);
+  $newlabel="$newfile   " . " "x($newlabel_n_spaces) . $newtime;
+}
+
+$encoding=guess_encoding($newfile) unless defined($encoding);
+
+$encoding = "utf8" if $encoding =~ m/^utf8/i ;
+print STDERR "Encoding $encoding\n" if $verbose;
+if (lc($encoding) eq "utf8" ) {
+  binmode(STDOUT, ":utf8");
+  binmode(STDERR, ":utf8");
+}
+
+# filter($text)
+# Runs $text through the script provided in $filterscript argument, if set
+# If not set, just returns $text unchanged.
+# If flatten was set, defer filtering to flatten.  flatten will run the filter
+# on all incoming text prior to its own processing.
+# If flatten was not set, filter each of old and new once (see just below this def)
+sub filter {
+  my ($text)=@_;
+  my ($textout,$pid);
+  if ($filterscript ne "") {
+    print STDERR "Passing " . length($text) . " chars to filter script " . $filterscript . "\n" if $verbose;
+
+    if ($ignorefilterstderr) {
+        # If we need to capture and bury STDERR, use the Open3 version, and close CHLD_ERR below.
+        use IPC::Open3;
+        # We consume STDERR from the process, and hide it
+        $pid = open3(\*CHLD_IN, \*CHLD_OUT, \*CHLD_ERR, $filterscript) or die "open3() failed $!";
+    }
+    else {
+        # Capture STDOUT and use as our new $text.  Allow STDERR to go to console.
+        use IPC::Open2;
+        $pid = open2(\*CHLD_OUT, \*CHLD_IN, $filterscript) or die "open2() failed $!";
+    }
+    # Send in $text
+    print CHLD_IN $text."\n";  # Adding a newline just to make sure there is one.
+    close CHLD_IN;
+    # Wait for output and gather it up
+    while (<CHLD_OUT>) {
+      $textout = $textout.$_;
+    }
+    if ($ignorefilterstderr) {
+        close CHLD_ERR;       # Enable only if Open3 used above
+    }
+    # On the off chance a very long running and/or frequently called script is used.
+    waitpid( $pid, 0 );
+    $text = $textout;
+    print STDERR "Received " . length($text) . " chars after filtering\n" if $verbose;
+    print STDERR $text if $verbose;
+  }
+  return $text;
+}
+
+$old=read_file_with_encoding($oldfile,$encoding);
+$new=read_file_with_encoding($newfile,$encoding);
+
+if (not defined($flatten)) {
+    $old=filter($old);
+    $new=filter($new);
+}
+
+
+
+
+# reset time
+exetime(1);
+($oldpreamble,$oldbody,$oldpost)=splitdoc($old,'\\\\begin\{document\}','\\\\end\{document\}');
+
+($newpreamble,$newbody,$newpost)=splitdoc($new,'\\\\begin\{document\}','\\\\end\{document\}');
+
+
+if ($flatten) {
+  $oldbody=flatten($oldbody,$oldpreamble,File::Spec->rel2abs($oldfile),$encoding);
+  $newbody=flatten($newbody,$newpreamble,File::Spec->rel2abs($newfile),$encoding);
+  # flatten preamble
+  $oldpreamble=flatten($oldpreamble,$oldpreamble,File::Spec->rel2abs($oldfile),$encoding);
+  $newpreamble=flatten($newpreamble,$newpreamble,File::Spec->rel2abs($newfile),$encoding);
+
+}
+
+
+
+
+my @auxlines;
+
+# boolean variab
+my ($ulem)=0;
+
+if ( length $oldpreamble && length $newpreamble ) {
+  # pre-process preamble by looking for commands used in \maketitle (title, author, date etc commands)
+  # and marking up content with latexdiff markup
+  @auxlines=preprocess_preamble($oldpreamble,$newpreamble);
+
+  @oldpreamble = split /\n/, $oldpreamble;
+  @newpreamble = split /\n/, $newpreamble;
+
+  # If a command is defined in the preamble of the new file, and only uses safe commands, then it can be considered to be safe) (contribution S. Gouezel)
+  # Base this assessment on the new preamble
+  add_safe_commands($newpreamble);
+
+  # get a list of packages from preamble if not predefined
+  %packages=list_packages($newpreamble) unless %packages;
+  if ( %packages && $debug ) { my $key ; foreach $key (keys %packages) { print STDERR "DEBUG \\usepackage[",$packages{$key},"]{",$key,"}\n" ;} }
+}
+
+# have to return to all processing to properly add preamble additions based on packages found
+if (defined($graphicsmarkup)) {
+  $graphicsmarkup=~tr/a-z/A-Z/;
+  if ( $graphicsmarkup eq 'OFF' or $graphicsmarkup eq 'NONE' ) {
+    $graphicsmarkup=NONE;
+  } elsif ( $graphicsmarkup eq 'NEWONLY' or $graphicsmarkup eq 'NEW-ONLY' ) {
+    $graphicsmarkup=NEWONLY;
+  } elsif ( $graphicsmarkup eq 'BOTH' ) {
+    $graphicsmarkup=BOTH;
+  } elsif ( $graphicsmarkup !~ m/^[012]$/ ) {
+    die "latexdiff Illegal value: ($graphicsmarkup)  for option --highlight-graphics. Possible values: OFF,WHOLE,COARSE,FINE,0-2\n";
+  }
+  # else use numerical value
+} else {
+  # Default: no explicit setting in menu
+  if ( defined $packages{"graphicx"} or defined $packages{"graphics"} ) {
+    $graphicsmarkup=NEWONLY;
+  } else {
+    $graphicsmarkup=NONE;
+  }
+}
+
+if (defined $packages{"hyperref"} ) {
+  # deleted lines should not generate or appear in link names:
+  print STDERR "hyperref package detected.\n" if $verbose ;
+  $latexdiffpreamble =~ s/\{\\DIFadd\}/{\\DIFaddtex}/g;
+  $latexdiffpreamble =~ s/\{\\DIFdel\}/{\\DIFdeltex}/g;
+  $latexdiffpreamble .= join "\n",(extrapream("HYPERREF"),"");
+  if($nolinks){
+    $latexdiffpreamble .= "\n\\hypersetup{bookmarks=false}";
+  }
+  ###    $latexdiffpreamble .= '%DIF PREAMBLE EXTENSION ADDED BY LATEXDIFF FOR HYPERREF PACKAGE' . "\n";
+  ###    $latexdiffpreamble .= '\providecommand{\DIFadd}[1]{\texorpdfstring{\DIFaddtex{#1}}{#1}}' . "\n";
+  ###    $latexdiffpreamble .= '\providecommand{\DIFdel}[1]{\texorpdfstring{\DIFdeltex{#1}}{}}' . "\n";
+  ###    $latexdiffpreamble .= '%DIF END PREAMBLE EXTENSION ADDED BY LATEXDIFF FOR HYPERREF PACKAGE' . "\n";
+}
+
+# add commands for figure highlighting to preamble
+if ($graphicsmarkup != NONE ) {
+  my @matches;
+  # Check if \DIFaddbeginFL definition calls \DIFaddbegin - if so we will issue an error message that graphics highlighting is
+  # is not compatible with this.
+  # (A more elegant solution would be to suppress the redefinitions of the \DIFaddbeginFL etc commands, but for this narrow use case
+  #  I currently don't see this as an efficient use of time)
+  ### The foreach loop does not make sense here. I don't know why I put this in -  (F Tilmann)
+  ###foreach my $cmd ( "DIFaddbegin","DIFaddend","DIFdelbegin","DIFdelend" ) {
+  @matches=( $latexdiffpreamble =~ m/command\{\\DIFaddbeginFL}\{($pat_n)}/sg ) ;
+  # we look at the last one of the list to take into account possible redefinition but almost always matches should have exactly one element
+  if ( $matches[$#matches] =~ m/\\DIFaddbegin/ ) {
+    die "Cannot combine graphics markup with float styles defining \\DIFaddbeginFL in terms of \\DIFaddbegin. Use --graphics-markup=none option or choose a different float style.";
+    exit 10;
+  }
+  ###}
+  $latexdiffpreamble .= join "\n",("\\newcommand{\\DIFscaledelfig}{$SCALEDELGRAPHICS}",extrapream("HIGHLIGHTGRAPHICS"),"");
+
+  # only change required for highlighting both is to declare \includegraphics safe, as preamble already contains commands for deleted environment
+  if ( $graphicsmarkup == BOTH ) {
+    init_regex_arr_list(\@SAFECMDLIST,'includegraphics');
+  }
+}
+
+$ulem = ($latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{ulem\}/ || defined $packages{"ulem"});
+
+# If amsmath is defined and $ulem is used for markup, redefine the \sout command to also work (mostly) in math mode
+# See stack exchange https://tex.stackexchange.com/questions/20609/strikeout-in-math-mode/308647#308647 based on comment by Taylor Raine
+if ( defined($packages{'amsmath'}) and $ulem ) {
+    $latexdiffpreamble .= join "\n",(extrapream('AMSMATHULEM'),"");
+}
+
+# If listings is being used or can be found in the latexdiff search path, add to the preamble auxiliary code to enable line-by-line markup
+if ( defined($packages{"listings"}) or `kpsewhich listings.sty` ne "" ) {
+  my @listingpreamble=extrapream("LISTINGS");
+  if ($latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/ ) {
+    @listingpreamble=extrapream("COLORLISTINGS");
+  }
+  my @listingDIFcode=();
+  my $replaced;
+  # note that in case user supplies preamblefile the type might not reflect well the actual markup style
+  @listingDIFcode=extrapream("-nofail","DIFCODE_" . $type) unless defined($preamblefile);
+  if (!(@listingDIFcode)) {
+    # if listingDIFcode is empty try to guess a suitable one from the preamble
+    if ($latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/ and $ulem ) {
+      @listingDIFcode=extrapream("DIFCODE_UNDERLINE");
+    } elsif ( $latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/ ) {
+      # only colour used
+      @listingDIFcode=extrapream("DIFCODE_CFONT");
+    } else {
+      # fall-back solution
+      @listingDIFcode=extrapream("DIFCODE_BOLD");
+    }
+  }
+  # add configuration so that listings work with utf-8
+  push @listingpreamble, '\lstset{extendedchars=\true,inputencoding='.$encoding."}\n";
+
+
+  # now splice it in
+  $replaced=0;
+  ###print STDERR "DEBUG: listingDIFcode: ",join("\n",@listingDIFcode),"|||\n" if $debug;
+
+  @listingpreamble=grep {
+    # only replace if this has not been done already (use short-circuit property of and)
+    if (!$replaced and $_ =~ s/^.*%DIFCODE TEMPLATE.*$/join("\n",@listingDIFcode)/e ) {
+      ###print STDERR "DEBUG: Replaced text $_\n" if $debug;
+      $replaced=1;
+      1;
+    } else {
+      # return false for those lines matching %DIFCODE TEMPLATE (so that they are not included in output)
+      not m/%DIFCODE TEMPLATE/;
+    }
+  } @listingpreamble;
+  ###  print STDERR "DEBUG: listingpreamble @listingpreamble\n";
+  $latexdiffpreamble .= join "\n",(@listingpreamble,"");
+} else {
+  print STDERR "WARNING: listings package not detected. Disabling mark-up in verbatim environments \n" ;
+  # if listings does not exist disable line-by-line markup and treat all verbatim environments as opaque
+  $VERBATIMENV = liststringtoregex($CONFIG{VERBATIMENV}.";".$CONFIG{VERBATIMLINEENV});
+  $VERBATIMLINEENV = "";
+}
+
+# adding begin and end marker lines to preamble
+$latexdiffpreamble = "%DIF PREAMBLE EXTENSION ADDED BY LATEXDIFF\n" . $ latexdiffpreamble . "%DIF END PREAMBLE EXTENSION ADDED BY LATEXDIFF\n";
+
+# and return to preamble specific processing
+if ( length $oldpreamble && length $newpreamble ) {
+  print STDERR "Differencing preamble.\n" if $verbose;
+
+  # insert dummy first line such that line count begins with line 1 (rather than perl's line 0) - just so that line numbers inserted by linediff are correct
+  unshift @newpreamble,'';
+  unshift @oldpreamble,'';
+  @diffpreamble = linediff(\@oldpreamble, \@newpreamble);
+  # remove dummy line again
+  shift @diffpreamble;
+  # add filenames, modification time and latexdiff mark
+  defined($nolabel) or splice @diffpreamble,1,0,
+      "%DIF LATEXDIFF DIFFERENCE FILE",
+      ,"%DIF DEL $oldlabel",
+      "%DIF ADD $newlabel";
+  if ( @auxlines ) {
+    push @diffpreamble,"%DIF DELETED TITLE COMMANDS FOR MARKUP";
+    push @diffpreamble,join("\n",@auxlines);
+  }
+  if ( $earlylatexdiffpreamble) {
+    # insert latexdiff command directly after documentclass at beginning of preamble
+    # note that grep is only run for its side effect
+    ( grep { s/^([^%]*\\documentclass.*)$/$1$latexdiffpreamble/ } @diffpreamble )==1 or die "Could not find documentclass statement in preamble";
+  } else {
+    # insert latexdiff commands at the end of preamble (default behaviour)
+    push @diffpreamble,$latexdiffpreamble;
+  }
+  push @diffpreamble,'\begin{document}';
+  if (defined $packages{"hyperref"} && $nolinks) {
+    push @diffpreamble, '\begin{NoHyper}';
+  }
+}
+elsif ( !length $oldpreamble && !length $newpreamble ) {
+  @diffpreamble=();
+} else {
+  print STDERR "Either both texts must have preamble or neither text must have the preamble.\n";
+  exit(2);
+}
+
+# Special: treat all cite commands as safe except in UNDERLINE and FONTSTRIKE mode
+# (there is a conflict between citation and ulem package, see
+# package documentation)
+# Use post-processing
+# and $packages{"apacite"}!~/natbibpapa/
+
+
+if (defined $packages{"units"}  && $ulem ) {
+  # protect inlined maths environments by surrounding with an \mbox
+  # this is done to get around an incompatibility between the ulem and units package
+  # where spaces in the argument to underlined or crossed-out \unit commands cause an error message
+  print STDERR "units package detected at the same time as style using ulem.\n" if $verbose ;
+  $MBOXINLINEMATH=1;
+}
+
+if (defined $packages{"siunitx"} ) {
+  # protect SI command by surrounding them with an \mbox
+  # this is done to get around an incompatibility between the ulem and siunitx package
+  print STDERR "siunitx package detected.\n" if $verbose ;
+  my $mboxcmds='SI,ang,numlist,numrange,SIlist,SIrange,qty,qtylist,qtyproduct,qtyrange,complexqty';
+  init_regex_arr_list(\@SAFECMDLIST,'num,si,numproduct,unit,complexnum');
+  if ( $enablecitmark || ( $ulem  && ! $disablecitmark )) {
+    init_regex_arr_list(\@MBOXCMDLIST,$mboxcmds);
+  } else {
+    init_regex_arr_list(\@SAFECMDLIST,$mboxcmds);
+  }
+}
+
+if (defined $packages{"cleveref"} ) {
+  # protect selected command by surrounding them with an \mbox
+  # this is done to get around an incompatibility between ulem and cleveref package
+  print STDERR "cleveref package detected.\n" if $verbose ;
+  my $mboxcmds='[Cc]ref(?:range)?\*?,labelcref,(?:lc)?name[cC]refs?' ;
+  if ( $enablecitmark || ( $ulem  && ! $disablecitmark )) {
+    init_regex_arr_list(\@MBOXCMDLIST,$mboxcmds);
+  } else {
+    init_regex_arr_list(\@SAFECMDLIST,$mboxcmds);
+  }
+}
+
+if (defined $packages{"glossaries"} ) {
+  # protect selected command by surrounding them with an \mbox
+  # this is done to get around an incompatibility between ulem and glossaries package
+  print STDERR "glossaries package detected.\n" if $verbose ;
+  my $mboxcmds='[gG][lL][sS](?:|pl|disp|link|first|firstplural|desc|user[iv][iv]?[iv]?),[aA][cC][rR](?:long|longpl|full|fullpl),[aA][cC][lfp]?[lfp]?';
+  init_regex_arr_list(\@SAFECMDLIST,'[gG][lL][sS](?:(?:entry)?(?:text|plural|name|symbol)|displaynumberlist|entryfirst|entryfirstplural|entrydesc|entrydescplural|entrysymbolplural|entryuser[iv][iv]?[iv]?|entrynumberlist|entrydisplaynumberlist|entrylong|entrylongpl|entryshort|entryshortpl|entryfull|entryfullpl),[gG]lossentry(?:name|desc|symbol),[aA][cC][rR](?:short|shortpl),[aA]csp?');
+  if ( $enablecitmark || ( $ulem  && ! $disablecitmark )) {
+    init_regex_arr_list(\@MBOXCMDLIST,$mboxcmds);
+  } else {
+    init_regex_arr_list(\@SAFECMDLIST,$mboxcmds);
+  }
+}
+
+if (defined $packages{"chemformula"} or defined $packages{"chemmacros"} ) {
+  print STDERR "chemformula package detected.\n" if $verbose ;
+  init_regex_arr_list(\@SAFECMDLIST,'ch');
+  push(@UNSAFEMATHCMD,'ch');
+  # The next command would be needed to allow highlighting the interior of \ch commands in math environments
+  # but the redefinitions in chemformula are too deep to make this viable
+  # push(@MATHTEXTCMDLIST,'ch');
+}
+
+if (defined $packages{"mhchem"} ) {
+  print STDERR "mhchem package detected.\n" if $verbose ;
+  init_regex_arr_list(\@SAFECMDLIST,'ce');
+  push(@UNSAFEMATHCMD,'ce','cee');
+  # The next command would be needed to allow highlighting the interior of \cee commands in math environments
+  # but the redefinitions in chemformula are too deep to make this viable
+  # push(@MATHTEXTCMDLIST,'cee');
+}
+
+if ( defined $packages{"tikz-dependency"} ) {
+  init_regex_arr_ext(\@SAFECMDEXCL, 'AMPERSAND');
+}
+
+
+my ( $citpat);
+
+if ( defined $packages{"apacite"}  ) {
+  print STDERR "apacite package detected.\n" if $verbose ;
+  $citpat='(?:mask)?(?:full|short|no)?cite(?:A|author|year|meta)?(?:NP)?';
+} elsif (defined $packages{"biblatex"}) {
+  print STDERR "biblatex package detected.\n" if $verbose ;
+  $citpat='(?:[cC]ites?|(?:[pP]aren|foot|[Tt]ext|[sS]mart|super)cites?\*?|footnotecitetex)';
+  push(@TEXTCMDEXCL, qr/^textcite$/);
+} else {
+  # citation command pattern for all other citation schemes
+  $citpat='(?:cite\w*|nocite)';
+};
+
+if ( ! $ulem ) {
+  # modes not using ulem: citation is safe
+  push (@SAFECMDLIST, $citpat);
+} else {
+  ### Experimental: disable text and emph commands
+  push(@SAFECMDEXCL, qr/^emph$/, qr/^text..$/);
+  # replace \cite{..} by \mbox{\cite{..}} in added or deleted blocks in post-processing
+  push(@MBOXCMDLIST,$citpat) unless $disablecitmark;
+  if ( uc($subtype) eq "COLOR" or uc($subtype) eq "DVIPSCOL" ) {
+    # remove \cite command again from list of safe commands
+    pop @MBOXCMDLIST;
+    # deleted cite commands
+  }
+}
+push(@MBOXCMDLIST,$citpat) if $enablecitmark ;
+
+
+if (defined $packages{"amsmath"}  or defined $packages{"amsart"} or defined $packages{"amsbook"} ) {
+  print STDERR "amsmath package detected.\n" if $verbose ;
+  $MATHARRREPL='align*';
+}
+
+# add commands in MBOXCMDLIST to SAFECMDLIST
+foreach $mboxcmd ( @MBOXCMDLIST ) {
+  init_regex_arr_list(\@SAFECMDLIST, $mboxcmd);
+}
+
+# check if \label is in SAFECMDLIST, and if yes replace "label" in $LABELCMD by something that never matches (we hope!)
+if ( iscmd("label",\@SAFECMDLIST,\@SAFECMDEXCL) ) {
+  $LABELCMD=~ s/label/NEVERMATCHLABEL/;
+}
+
+
+
+print STDERR "Preprocessing body.  " if $verbose;
+preprocess($oldbody,$newbody);
+writedebugfile($oldbody,'old-preprocess');
+writedebugfile($newbody,'new-preprocess');
+
+# run difference algorithm
+@diffbody=bodydiff($oldbody, $newbody);
+$diffbo=join("",@diffbody);
+writedebugfile($diffbo,"bodydiff");
+
+print STDERR "(",exetime()," s)\n","Postprocessing body. \n" if $verbose;
+postprocess($diffbo);
+$diffall =join("\n",@diffpreamble) ;
+# add visible labels
+if (defined($visiblelabel)) {
+  # Give information right after \begin{document} (or at the beginning of the text for files without preamble
+  ### if \date command is used, add information to \date argument, otherwise give right after \begin{document}
+  ###  $diffall=~s/(\\date$extraspace(?:\[$brat0\])?$extraspace)\{($pat_n)\}/$1\{$2 \\ LATEXDIFF comparison \\ Old: $oldlabel \\ New: $newlabel \}/  or
+  $diffbo = "\\begin{verbatim}LATEXDIFF comparison\nOld: $oldlabel\nNew: $newlabel\\end{verbatim}\n$diffbo"   ;
+}
+
+$diffall .= "$diffbo" ;
+if (defined $packages{"hyperref"} && $nolinks) {
+  $diffall .= "\\end{NoHyper}\n";
+}
+$diffall .= "\\end{document}$newpost" if length $newpreamble ;
+if ( lc($encoding) ne "utf8" && lc($encoding) ne "ascii" ) {
+  print STDERR "Encoding output file to $encoding\n" if $verbose;
+  $diffall=Encode::encode($encoding,$diffall);
+  binmode STDOUT;
+}
+print $diffall;
+
+
+print STDERR "(",exetime()," s)\n","Done.\n" if $verbose;
+
+
+# liststringtoregex(liststring)
+# expands string with semi-colon separated list into a regular expression corresponding
+# matching any of the elements
+sub liststringtoregex {
+  my ($liststring)=@_;
+  my @elements=grep /\S/,split(";",$liststring);
+  if ( @elements) {
+    return('(?:(?:' . join(')|(?:',@elements) .'))');
+  } else {
+    return "";
+  }
+}
+
+# show_configuration
+# note that this is not encapsulated but uses variables from the main program
+# It is provided for convenience because in the future it is planned to allow output
+# to be modified based on what packages are read etc - this works only if the input files are actually read
+# whether or not additional files are provided
+sub show_configuration {
+  if ($showpreamble) {
+    print "\nPreamble commands:\n";
+    print $latexdiffpreamble ;
+  }
+
+  if ($showsafe) {
+    print "\nsafecmd: Commands safe within scope of $ADDOPEN $ADDCLOSE and $DELOPEN $DELCLOSE (unless excluded):\n";
+    print_regex_arr(@SAFECMDLIST);
+    print "\nsafecmd-exlude: Commands not safe within scope of $ADDOPEN $ADDCLOSE and $DELOPEN $DELCLOSE :\n";
+    print_regex_arr(@SAFECMDEXCL);
+    print "\nmboxsafecmd:  Commands safe only if they are surrounded by \\mbox command:\n";
+    print_regex_arr(@MBOXCMDLIST);
+    print "\nnmboxsafecmd: Commands not safe:\n";
+    print_regex_arr(@MBOXCMDEXCL);
+  }
+
+  if ($showtext) {
+    print "\nCommands with last argument textual (unless excluded) and safe in every context:\n";
+    print_regex_arr(@TEXTCMDLIST);
+    print "\nContext1 commands (last argument textual, command will be disabled in deleted passages, last argument will be shown as plain text):\n";
+    print_regex_arr(@CONTEXT1CMDLIST);
+    print "\nContext2 commands (last argument textual, command and its argument will be disabled in deleted passages):\n";
+    print_regex_arr(@CONTEXT2CMDLIST);
+    print "\nExclude list of Commands with last argument not textual (overrides patterns above):\n";
+    print_regex_arr(@TEXTCMDEXCL);
+  }
+
+
+  if ($showconfig) {
+    print "Configuration variables:\n";
+    print "ARRENV=$ARRENV\n";
+    print "COUNTERCMD=$COUNTERCMD\n";
+    print "FLOATENV=$FLOATENV\n";
+    print "ITEMCMD=$ITEMCMD\n";
+    print "LISTENV=$LISTENV\n";
+    print "MATHARRENV=$MATHARRENV\n";
+#    print "MATHARRREPL=$MATHARRREPL\n";     # this is not deprecated and thus no longer shown
+    print "MATHENV=$MATHENV\n";
+    print "MATHREPL=$MATHREPL\n";
+    print "MINWORDSBLOCK=$MINWORDSBLOCK\n";
+    print "PICTUREENV=$PICTUREENV\n";
+    print "SCALEDELGRAPHICS=$SCALEDELGRAPHICS\n";
+    print "VERBATIMENV=$VERBATIMENV\n";
+    print "VERBATIMLINEENV=$VERBATIMLINEENV\n";
+    print "CUSTOMDIFCMD=$CUSTOMDIFCMD\n";
+  }
+}
+
+
+
+## guess_encoding(filename)
+## reads the first 20 lines of filename and looks for call of inputenc package
+## if found, return the option of this package (encoding), otherwise return utf8
+sub guess_encoding {
+  my ($filename)=@_;
+  my ($i,$enc);
+  open (FH, $filename) or die("Couldn't open $filename: $!");
+  $i=0;
+  while (<FH>) {
+    next if /^\s*%/;    # skip comment lines
+    if (m/\\usepackage\[(\w*?)\]\{inputenc\}/) {
+      close(FH);
+      return($1);
+    }
+    last if (++$i > 20 ); # scan at most 20 non-comment lines
+  }
+  close(FH);
+  ### return("ascii");
+  return("utf8");
+}
+
+
+sub read_file_with_encoding {
+  my ($output);
+  my ($filename, $encoding) = @_;
+
+  if (lc($encoding) eq "utf8" ) {
+    open (FILE, "<:utf8",$filename) or die("Couldn't open $filename: $!");
+    local $/ ; # locally set record operator to undefined, ie. enable whole-file mode
+    $output=<FILE>;
+  } elsif ( lc($encoding) eq "ascii") {
+    open (FILE, $filename) or die("Couldn't open $filename: $!");
+    local $/ ; # locally set record operator to undefined, ie. enable whole-file mode
+    $output=<FILE>;
+  } else {
+    require Encode;
+    open (FILE, "<",$filename) or die("Couldn't open $filename: $!");
+    local $/ ; # locally set record operator to undefined, ie. enable whole-file mode
+    $output=<FILE>;
+    print STDERR "Converting $filename from $encoding to utf8\n" if $verbose;
+    $output=Encode::decode($encoding,$output);
+  }
+  close FILE;
+  if ($^O eq "linux" ) {
+    $output =~ s/\r\n/\n/g ;
+  }
+  return $output;
+}
+
+## %packages=list_packages(@preamble)
+## scans the arguments for \documentclass,\RequirePackage and \usepackage statements and constructs a hash
+## whose keys are the included packages, and whose values are the associated optional arguments
+#sub list_packages {
+#  my (@preamble)=@_;
+#  my %packages=();
+#  foreach $line ( @preamble ) {
+#    # get rid of comments
+#    $line=~s/(?<!\\)%.*$// ;
+#    if ( $line =~ m/\\(?:documentclass|usepackage|RequirePackage)(?:\[(.+?)\])?\{(.*?)\}/ ) {
+##      print STDERR "Found something: |$line|\n" if $debug;
+#      if (defined($1)) {
+#	$packages{$2}=$1;
+#      } else {
+#	$packages{$2}="";
+#      }
+#    }
+#  }
+#  return (%packages);
+#}
+
+
+# %packages=list_packages($preamble)
+# scans the arguments for \documentclass,\RequirePackage and \usepackage statements and constructs a hash
+# whose keys are the included packages, and whose values are the associated optional arguments
+# if argument of \usepackage or \RequirePackage is comma separated list, treat as different packages
+sub list_packages {
+  my ($preamble)=@_;
+  my %packages=();
+  my $pkg;
+
+  # remove comments
+  $preamble=~s/(?<!\\)%.*$//mg ;
+
+  while ( $preamble =~  m/\\(?:documentclass|usepackage|RequirePackage)(?:\[($brat_n)\])?\{(.*?)\}/gs ) {
+    if (defined($1)) {
+      foreach $pkg ( split /,/,$2 ) {
+	$packages{$pkg}=$1;
+      }
+    } else {
+      foreach $pkg ( split /,/,$2 ) {
+	$packages{$pkg}="";
+      }
+    }
+  }
+
+  # sometimes, class options are defined in such a way that they imply the loading and/or presence of a package
+  # so we also treat all class options as 'packages.
+  if ( $preamble =~  m/\\documentclass\s*\[($brat_n)\]\s*\{.*?\}/s ) {
+     foreach $pkg ( split /,/,$1 ) {
+       $pkg =~ s/\s//g ;   # remove space and newline characters
+       $packages{$pkg}="" unless exists($packages{$pkg});
+     }
+  }
+  return (%packages);
+}
+
+# Subroutine add_safe_commands modified from version provided by S. Gouezel
+# add_safe_commands($preamble)
+# scans the argument for \newcommand and \DeclareMathOperator,
+# and adds the created commands which are clearly safe to @SAFECMDLIST
+sub add_safe_commands {
+  my ($preamble)=@_;
+
+  # get rid of comments
+  $preamble=~s/(?<!\\)%.*$//mg ;
+
+  my $to_test = "";
+  # test for \DeclareMathOperator{\foo}{myoperator}
+  while ( $preamble =~ m/\DeclareMathOperator\s*\*?\{\\(\w*?)\}/osg) {
+    $to_test=$1;
+    if ($to_test ne "" and not iscmd($to_test,\@SAFECMDLIST,\@SAFECMDEXCL) and not iscmd($to_test, \@SAFECMDEXCL, [])) {
+      # one should add $to_test to the list of safe commands.
+      init_regex_arr_list(\@SAFECMDLIST, $to_test);
+      print STDERR "Adding $to_test to the list of safe commands\n" if $verbose;
+    }
+  }
+
+  while ( $preamble =~ m/\\(?:new|renew|provide)command\s*{\\(\w*)\}(?:|\[\d*\])\s*\{(${pat_n})\}/osg ) {
+    my $maybe_to_test  = $1;
+    my $should_be_safe = $2;
+    print STDERR "DEBUG Checking new command: maybe_to_test, should_be_safe: $1 $2\n" if $debug;
+    my $success = 0;
+    # skip custom diff commands
+    next if ($maybe_to_test =~ m/^(?:ADD|DEL)?${CUSTOMDIFCMD}$/);
+    ###print STDERR "DEBUG: really test it. \n";
+    # test if all latex commands inside it are safe
+    $success = 1;
+    if ($should_be_safe =~ m/\\\\/) {
+      $success = 0;
+    } else {
+      while ($should_be_safe =~ m/\\(\w+)/g) {
+	###	  print STDERR "DEBUG: Testing command $1 " if $debug;
+	$success = 0 unless iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL); ### or $1 eq "";
+	###        print STDERR " success=$success\n" if $debug;
+      }
+    }
+    ###      }
+    if ($success) {
+      $to_test = $maybe_to_test;
+      if (  not iscmd($to_test,\@SAFECMDLIST,\@SAFECMDEXCL) and not iscmd($to_test, \@SAFECMDEXCL, [])) {
+	#        # one should add $to_test to the list of safe commands.
+	init_regex_arr_list(\@SAFECMDLIST, $to_test);
+	print STDERR "Adding $to_test to the list of safe commands\n" if $verbose;
+      }
+    }
+  }
+}
+
+
+# helper function for flatten
+# remove \endinput at beginning of line and everything
+# following it, # if \endinput is not at the beginning of
+# the line, nothing will be removed. It is assumed that
+# this case is most common when \endinput is part of a
+# conditional clause.  The file will only be processed
+# correctly if the conditional is always false,
+# i.e. \endinput # not actually reached
+sub remove_endinput {
+  # s/// operates on default input
+  $_[0] =~ s/^\\endinput.*\Z//ms ;
+  return($_[0]);
+}
+
+# flatten($text,$preamble,$filename,$encoding)
+# expands \input and \include commands within text
+# expands \bibliography command with corresponding bbl file if available
+# expands \subfile command (from subfiles package - not part of standard text distribution)
+# expands \import etc commands (from import package - not part of standard text distribution)
+# preamble is scanned for includeonly commands
+# encoding is the encoding
+sub flatten {
+  my ($text,$preamble,$filename,$encoding)=@_;
+  my ($includeonly,$dirname,$fname,$newpage,$fullfile,$filecontent,$replacement,$begline,$inputcmd,$bblfile,$subfile,$command,$verbenv,$verboptions,$ignore,$fileonly);
+  my ($subpreamble,$subbody,$subpost);
+  my ($subdir,$subdirfull,$importfilepath);
+  require File::Basename ;
+  ###  require File::Spec ;    # now this is needed even if flatten option not given
+  $filename = File::Spec->rel2abs( $filename ) ;
+  ($ignore, $dirname, $fileonly) = File::Spec->splitpath($filename) ;
+  $bblfile = $filename;
+  $bblfile=~s/\.tex$//;
+  $bblfile.=".bbl";
+
+  if ( ($includeonly) = ($preamble =~ m/\\includeonly\{(.*?)\}/ ) ) {
+    $includeonly =~ s/,/|/g;
+  } else {
+    $includeonly = '.*?';
+  }
+
+  print STDERR "DEBUG: includeonly $includeonly\n" if $debug;
+
+
+  # Run through filter, to let filterscript have a pass if it was set
+  $text = filter($text);
+
+  # Recursively replace \\import, \\subimport, and related import commands
+  $text =~ s/(^(?:[^%\n]|\\%)*)(\\(sub)?(?:import|inputfrom|includefrom))\{(.*?)\}(?:[\s]*)\{(.*?)\}/{
+          #  (--------1-------)(--(=3=)-------------2-------------------)  (-4-)             (-5-)
+          # $1 is begline
+          # $2 is the import macro name
+          # $3 is (optional) prefix "sub"
+          # $4 is directory
+          # $5 is filename
+          $begline = (defined($1)? $1 : "");
+          $subdir = $4;
+          $fname = $5;
+          $fname .= ".tex" unless $fname =~ m|\.\w{3,4}$|;
+          print STDERR "DEBUG begline:", $begline, "\n" if $debug;
+          print STDERR "DEBUG", (defined($3)? "subimport_file:" : "import_file:"), $subdir, "\n" if $debug;
+          print STDERR "DEBUG file:", $fname, "\n" if $debug;
+
+          # subimport appends $subdir to the current $dirname.  import replaces it with an absolute path.
+          $subdirfull = (defined($3) ? File::Spec->catdir($dirname,$subdir) : $subdir);
+
+          $importfilepath = File::Spec->catfile($subdirfull, $fname);
+
+          print STDERR "importing importfilepath:", $importfilepath,"\n" if $verbose;
+          if ( -f $importfilepath ) {
+              # If file exists, replace input or include command with expanded input
+              #TODO: need remove_endinput & newpage similar to other replacements inside flatten
+              $replacement=flatten(read_file_with_encoding($importfilepath, $encoding), $preamble,$importfilepath,$encoding);
+          } else {
+              # if file does not exist, do not expand include or input command (do not warn if fname contains #[0-9] as it is then likely part of a command definition
+              # and is not meant to be expanded directly
+              print STDERR "WARNING: Could not find included file ",$importfilepath,". I will continue but not expand |$2|\n";
+              $replacement = $2;
+              $replacement .= "{$subdir}{$fname} % Processed";
+          }
+          "$begline$replacement";
+  }/exgm;
+
+  # recursively replace \\input and \\include files
+  $text =~ s/(^(?:[^%\n]|\\%)*)(\\input\{(.*?)\}|\\include\{(${includeonly}(?:\.tex)?)\})/{
+	    $begline=(defined($1)? $1 : "") ;
+	    $inputcmd=$2;
+	    $fname = $3 if defined($3) ;
+	    $fname = $4 if defined($4) ;
+            $newpage=(defined($4)? " \\newpage " : "") ;
+            #      # add tex extension unless there is a three or four letter extension already
+            $fname .= ".tex" unless $fname =~ m|\.\w{3,4}$|;
+            $fullfile = File::Spec->catfile($dirname,$fname);
+            print STDERR "DEBUG Beg of line match |$1|\n" if defined($1) && $debug ;
+            print STDERR "Include file $fname\n" if $verbose;
+            print STDERR "DEBUG looking for file ",$fullfile, "\n" if $debug;
+            # content of file becomes replacement value (use recursion), add \newpage if the command was include
+            if ( -f $fullfile ) {
+	      # If file exists, replace input or include command with expanded input
+	      $replacement=flatten(read_file_with_encoding($fullfile, $encoding), $preamble,$filename,$encoding);
+	      $replacement = remove_endinput($replacement);
+	      # \include always starts a new page; use explicit \newpage command to simulate this
+	    } else {
+	      # if file does not exist, do not expand include or input command (do not warn if fname contains #[0-9] as it is then likely part of a command definition
+              # and is not meant to be expanded directly
+	      print STDERR "WARNING: Could not find included file ",$fullfile,". I will continue but not expand |$inputcmd|\n" unless $fname =~ m(#[0-9]) ;
+	      $replacement = $inputcmd ;   # i.e. just the original command again -> make no change file does not exist
+	      $newpage="";
+	    }
+	    "$begline$newpage$replacement$newpage";
+          }/exgm;
+
+  # replace bibliography with bbl file if it exists
+  $text=~s/(^(?:[^%\n]|\\%)*)\\bibliography\{(.*?)\}/{
+           if ( -f $bblfile ){
+	     $replacement=read_file_with_encoding(File::Spec->catfile($bblfile), $encoding);
+	   } else {
+	     warn "Bibliography file $bblfile cannot be found. No flattening of \\bibliography done. Run bibtex on old and new files first";
+	     $replacement="\\bibliography{$2}";
+	   }
+	   $begline=(defined($1)? $1 : "") ;
+	   "$begline$replacement";
+  }/exgm;
+
+  # replace subfile with contents (subfile package)
+  $text=~s/(^(?:[^%\n]|\\%)*)\\subfile\{(.*?)\}/{
+           $begline=(defined($1)? $1 : "") ;
+     	   $fname = $2;
+           #      # add tex extension unless there is a three or four letter extension already
+           $fname .= ".tex" unless $fname =~ m|\.\w{3,4}|;
+           print STDERR "Include file as subfile $fname\n" if $verbose;
+           # content of file becomes replacement value (use recursion)
+           # now strip away everything outside and including \begin{document} and \end{document} pair#
+	   #             # note: no checking for comments is made
+           $fullfile=File::Spec->catfile($dirname,$fname);
+           if ( -f $fullfile) {
+	     # if file exists, expand \subfile command by contents of file
+	     $subfile=read_file_with_encoding($fullfile,$encoding) or die "Could not open included subfile ",$fullfile,": $!";
+	     ($subpreamble,$subbody,$subpost)=splitdoc($subfile,'\\\\begin\{document\}','\\\\end\{document\}');
+	     ###           $subfile=~s|^.*\\begin{document}||s;
+	     ###           $subfile=~s|\\end{document}.*$||s;
+	     $replacement=flatten($subbody, $preamble,$fullfile,$encoding);
+	     ### $replacement = remove_endinput($replacement);
+	   } else {
+	      # if file does not exist, do not expand subfile
+	      print STDERR "WARNING: Could not find subfile ",$fullfile,". I will continue but not expand |$2|\n" unless $fname =~ m(#[0-9]) ;
+	      $replacement = "\\subfile\{$2\}" ;   # i.e. just the original command again -> make no change file does not exist
+	    }
+
+	   "$begline$replacement";
+  }/exgm;
+
+  # replace \verbatiminput and \lstlistinginput
+  $text=~s/(^(?:[^%\n]|\\%)*)\\(verbatiminput\*?|lstinputlisting)$extraspace(\[$brat_n\])?$extraspace\{(.*?)\}/{
+     $begline=(defined($1)? $1 : "") ;
+     $command = $2 ;
+     $fname = $4 ;
+     $verboptions = defined($3)? $3 : "" ;
+     if ($command eq 'verbatiminput' ) {
+       $verbenv = "verbatim" ;
+     } elsif ($command eq 'verbatiminput*' ) {
+       $verbenv = "verbatim*" ;
+     } elsif ($command eq 'lstinputlisting' ) {
+       $verbenv = "lstlisting" ;
+     } else {
+       die "Internal errorL Unexpected verbatim input type $command.\n";
+     }
+     print STDERR "DEBUG Beg of line match |$begline|\n" if $debug ;
+     print STDERR "Include file $fname  verbatim\n" if $verbose;
+     print STDERR "DEBUG looking for file ",File::Spec->catfile($dirname,$fname), "\n" if $debug;
+     # content of file becomes replacement value (do not use recursion), add \newpage if the command was include
+     ###$replacement=read_file_with_encoding(File::Spec->catfile($dirname,$fname), $encoding) or die "Couldn't find file ",File::Spec->catfile($dirname,$fname),": $!";
+     $replacement=read_file_with_encoding(File::Spec->catfile($dirname,$fname), $encoding); # (cannot on apparent failure as this triggers for empty fie. Original:  or die "Couldn't find file ",File::Spec->catfile($dirname,$fname),": $!";
+     # Add a new line if it not already there (note that the matching operator needs to use different delimiters, as we are still inside an outer scope that takes precedence
+     $replacement .= "\n" unless  $replacement =~ m(\n$)  ;
+     "$begline\\begin{$verbenv}$verboptions\n$replacement\\end{$verbenv}\n";
+    }/exgm;
+
+  return($text);
+}
+
+
+# print_regex_arr(@arr)
+# prints regex array without x-ism expansion put in by pearl to stdout
+sub print_regex_arr {
+  my $dumstring;
+  $dumstring = join(" ",@_);     # PERL generates string (?-xism:^ref$) for quoted refex ^ref$
+  $dumstring =~ s/\(\?-xism:\^(.*?)\$\)/$1/g;   # remove string and ^,$ marks before output
+  print $dumstring,"\n";
+}
+
+
+# @lines=extrapream($type,...)
+# reads line from appendix or external file
+# (end of file after __END__ token)
+# if $type is a filename, it will read the file instead of reading from the appendix
+# otherwise it will screen appendix for line "%DIF $TYPE" and copy everything up to line
+# '%DIF END $TYPE' (where $TYPE is upcased version of $type)
+# extrapream('-nofail',$type) will---instead of failing---simply return nothing if
+# it does not find the matching line in appendix (do not use -nofail option with multiple types!)
+sub extrapream {
+  my @types=@_;
+  my ($type,$arg);
+  my $nofail=0;
+  ###my @retval=("%DIF PREAMBLE EXTENSION ADDED BY LATEXDIFF") ;
+  my @retval=();
+  my ($copy);
+
+  foreach $arg ( @types ) {
+    if ( $arg eq '-nofail' ) {
+      $nofail=1;
+      next;
+    }
+    $type=$arg;
+    $copy=0;
+    if ( -f $type || lc $type eq '/dev/null' ) {
+      print STDERR "Reading preamble file $type\n" if $verbose ;
+      open (FILE,$type) or die "Cannot open preamble file $type: $!";
+      if (defined($encoding)) {
+	binmode(FILE,":encoding($encoding)");
+      } else {
+	require Encode::Locale;
+	binmode(FILE,":encoding(locale)");
+      }
+      while (<FILE>) {
+	chomp ;
+	if ( $_ =~ m/%DIF PREAMBLE/ ) {
+	  push (@retval,"$_");
+	} else {
+	  push (@retval,"$_ %DIF PREAMBLE");
+	}
+      }
+    } else {    # not (-f $type)
+       $type=uc($type);   # upcase argument
+       print STDERR "Preamble Internal Type $type\n" if $verbose;
+       # save filehandle position (before first read this points to line after __END__)
+       # but seek DATA,0,0 resets it to the beginning of the file
+       # see https://stackoverflow.com/questions/4459601/how-can-i-use-data-twice
+       my $data_start = tell DATA;
+       while (<DATA>) {
+	 if ( m/^%DIF $type/ ) {
+	   $copy=1;
+	 } elsif ( m/^%DIF END $type/ ) {
+	   last;
+	 }
+	 chomp;
+	 push (@retval,"$_ %DIF PREAMBLE") if $copy;
+       }
+       if ( $copy == 0 ) {
+	 unless ($nofail) {
+	   print STDERR "\nPreamble style $type not implemented.\n";
+	   print STDERR "Write latexdiff -h to get help with available styles\n";
+	   exit(2);
+	 }
+       }
+       seek DATA,$data_start,0;    # rewind DATA handle to beginning of data record
+     }
+  }
+  ###push (@retval,"%DIF END PREAMBLE EXTENSION ADDED BY LATEXDIFF")  ;
+  return @retval;
+}
+
+
+# ($part1,$part2,$part3)=splitdoc($text,$word1,$word2)
+# splits $text into 3 parts at $word1 and $word2.
+# if neither $word1 nor $word2 exist, $part1 and $part3 are empty, $part2 is $text
+# If only $word1 or $word2 exist but not the other, output an error message.
+
+# NB this version avoids $` and $' for performance reason although it only makes a tiny difference
+# (in one test gain a tenth of a second for a 30s run)
+sub splitdoc {
+  my ($text,$word1,$word2)=@_;
+  my ($part1,$part2,$part3)=("","","");
+  my ($rest,$pos);
+
+  if ( $text =~ m/(^[^%]*)($word1)/mg ) {
+    $pos=pos $text;
+    $part1=substr($text,0,$pos-length($2));
+    $rest=substr($text,$pos);
+    if ( $rest =~ m/(^[^%]*)($word2)/mg ) {
+      $pos=pos $rest;
+      $part2=substr($rest,0,$pos-length($2));
+      $part3=substr($rest,$pos);
+    }
+    else {
+      die "$word1 and $word2 not in the correct order or not present as a pair." ;
+    }
+  } else {
+    $part2=$text;
+    die "$word2 present but not $word1." if ( $text =~ m/(^[^%]*)$word2/ms );
+  }
+  return ($part1,$part2,$part3);
+}
+
+
+
+
+
+# bodydiff($old,$new)
+sub bodydiff {
+  my ($oldwords, $newwords) = @_;
+  my @retwords;
+
+  print STDERR "(",exetime()," s)\n","Splitting into latex tokens \n" if $verbose;
+  print STDERR "Parsing $oldfile \n" if $verbose;
+  my @oldwords = splitlatex($oldwords);
+  print STDERR "Parsing $newfile \n" if $verbose;
+  my @newwords = splitlatex($newwords);
+
+  if ( $debug ) {
+    open(TOKENOLD,">","latexdiff.debug.tokenold");
+    print TOKENOLD join("***\n",@oldwords);
+    close(TOKENOLD);
+    open(TOKENNEW,">","latexdiff.debug.tokennew");
+    print TOKENNEW join("***\n",@newwords);
+    close(TOKENNEW);
+  }
+
+  print STDERR "(",exetime()," s)\n","Pass 1: Expanding text commands and merging isolated identities with changed blocks  " if $verbose;
+  pass1(\@oldwords, \@newwords);
+
+
+  print STDERR "(",exetime()," s)\n","Pass 2: inserting DIF tokens and mark up.  " if $verbose;
+  if ( $debug ) {
+    open(TOKENOLD,">","latexdiff.debug.tokenold2");
+    print TOKENOLD join("***\n",@oldwords);
+    close(TOKENOLD);
+    open(TOKENNEW,">","latexdiff.debug.tokennew2");
+    print TOKENNEW join("***\n",@newwords);
+    close(TOKENNEW);
+  }
+
+  @retwords=pass2(\@oldwords, \@newwords);
+
+  return(@retwords);
+}
+
+
+
+
+# @words=splitlatex($string)
+# split string according to latex rules
+# Each element of words is either
+# a word (including trailing spaces and punctuation)
+# a latex command
+# if there is white space in the beginning return that as first token
+sub splitlatex {
+  my ($inputstring) = @_ ;
+  my $string=$inputstring ;
+  # if input is empty, return empty list
+  length($string)>0 or return ();
+  $string=~s/^(\s*)//s;
+  my $leadin=$1;
+  length($string)>0 or return ($leadin);
+
+  my @retval=($string =~ m/$pat/osg);
+
+  if (length($string) != length(join("",@retval))) {
+    print STDERR "\nWARNING: Inconsistency in length of input string and parsed string:\n     This often indicates faulty or non-standard latex code.\n     In many cases you can ignore this and the following warning messages.\n Note that character numbers in the following are counted beginning after \\begin{document} and are only approximate." unless $ignorewarnings;
+    print STDERR "DEBUG Original length ",length($string),"  Parsed length ",length(join("",@retval)),"\n" if $debug;
+    print STDERR "DEBUG Input string:  |$string|\n" if (length($string)<500) && $debug;
+    print STDERR "DEBUG Token parsing: |",join("+",@retval),"|\n" if (length($string)<500) && $debug ;
+    @retval=();
+    # slow way only do this if other m//sg method fails
+    my $last = 0;
+    while ( $string =~ m/$pat/osg ) {
+      my $match=$&;
+      if ($last + length $& != pos $string  ) {
+	my $pos=pos($string);
+	my $offset=30<$last ? 30 : $last;
+	my $dum=substr($string,$last-$offset,$pos-$last+2*$offset);
+	my $dum1=$dum;
+	my $cnt=$#retval;
+	my $i;
+	$dum1 =~ s/\n/ /g;
+	unless ($ignorewarnings) {
+	  print STDERR "\n$dum1\n";
+	  print STDERR " " x 30,"^" x ($pos-$last)," " x 30,"\n";
+	  print STDERR "Missing characters near word " . (scalar @retval) . " character index: " . $last . "-" .  pos($string) . " Length: " . length($match) . " Match: |$match| (expected match marked above).\n";
+	}
+	  # put in missing characters `by hand'
+	push (@retval, substr($dum,$offset,$pos-$last-length($match)));
+#       Note: there seems to be a bug in substr with utf8 that made the following line output substr which were too long,
+#             using dum instead appears to work
+#	push (@retval, substr($string,$last, pos($string)-$last-length($match)));
+      }
+      push (@retval, $match);
+      $last=pos $string;
+    }
+
+  }
+
+  unshift(@retval,$leadin) if (length($leadin)>0);
+  return @retval;
+}
+
+
+# pass1( \@seq1,\@seq2)
+# Look for differences between seq1 and seq2.
+# Where an common-subsequence block is flanked by deleted or appended blocks,
+# and is shorter than $MINWORDSBLOCK words it is appended
+# to the last deleted or appended word.  If the block contains tokens other than words
+# or punctuation it is not merged.
+# Deleted or appended block consisting of words and safe commands only are
+# also merged, to prevent break-up in pass2 (after previous isolated words have been removed)
+# If there are commands with textual arguments (e.g. \caption) both in corresponding
+# appended and deleted blocks split them such that the command and opening bracket
+# are one token, then the rest is split up following standard rules, and the closing
+# bracket is a separate token, ie. turn
+# "\caption{This is a textual argument}" into
+# ("\caption{","This ","is ","a ","textual ","argument","}")
+# No return value.  Destructively changes sequences
+sub pass1 {
+  my $seq1 = shift ;
+  my $seq2 = shift ;
+
+  my $len1 = scalar @$seq1;
+  my $len2 = scalar @$seq2;
+  my $wpat=qr/^(?:[a-zA-Z.,'`:;?()!]*)[\s~]*$/;   #'
+
+  my ($last1,$last2)=(-1,-1) ;
+  my $cnt=0;
+  my $block=[];
+  my $addblock=[];
+  my $delblock=[];
+  my $todo=[];
+  my $instruction=[];
+  my $i;
+  my (@delmid,@addmid,@dummy);
+
+  my ($addcmds,$delcmds,$matchindex);
+  my ($addtextblocks,$deltextblocks);
+  my ($addtokcnt,$deltokcnt,$mattokcnt)=(0,0,0);
+  my ($addblkcnt,$delblkcnt,$matblkcnt)=(0,0,0);
+
+  my $adddiscard = sub {
+                      if ($cnt > 0 ) {
+			$matblkcnt++;
+			# just after an unchanged block
+#			print STDERR "Unchanged block $cnt, $last1,$last2 \n";
+                        if ($cnt < $MINWORDSBLOCK
+			    && $cnt==scalar (
+				     grep { /^$wpat/ || ( /^\\((?:[`'^"~=.]|[\w\d@*]+))((?:\[$brat_n\]|\{$pat_n\})*)/o
+							   && iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL)
+							   && scalar(@dummy=split(" ",$2))<3 ) }
+					     @$block) )  {
+			  # merge identical blocks shorter than $MINWORDSBLOCK
+			  # and only containing ordinary words
+			  # with preceding different word
+			  # We cannot carry out this merging immediately as this
+			  # would change the index numbers of seq1 and seq2 and confuse
+			  # the algorithm, instead we store in @$todo where we have to merge
+                          push(@$todo, [ $last1,$last2,$cnt,@$block ]);
+			}
+			$block = [];
+			$cnt=0; $last1=-1; $last2=-1;
+		      }
+		    };
+  my $discard=sub { $deltokcnt++;
+                    &$adddiscard; #($_[0],$_[1]);
+		    push(@$delblock,[ $seq1->[$_[0]],$_[0] ]);
+		    $last1=$_[0] };
+
+  my $add =   sub { $addtokcnt++;
+                    &$adddiscard; #($_[0],$_[1]);
+		    push(@$addblock,[ $seq2->[$_[1]],$_[1] ]);
+		    $last2=$_[1] };
+
+  my $match = sub { $mattokcnt++;
+                    if ($cnt==0) {   # first word of matching sequence after changed sequence or at beginning of word sequence
+		      $deltextblocks = extracttextblocks($delblock);
+		      $delblkcnt++ if scalar @$delblock;
+		      $addtextblocks = extracttextblocks($addblock);
+		      $addblkcnt++ if scalar @$addblock;
+
+		      # make a list of all TEXTCMDLIST commands in deleted and added blocks
+		      $delcmds = extractcommands($delblock);
+      		      $addcmds = extractcommands($addblock);
+		      # now find those text commands, which are found in both deleted and added blocks, and expand them
+		      # keygen(third argument of _longestCommonSubsequence) implies to sort on command (0th elements of $addcmd elements)
+		      # the calling format for longestCommonSubsequence has changed between versions of
+		      # Algorithm::Diff so we need to check which one we are using
+		      if ( $algodiffversion  > 1.15 ) {
+			### Algorithm::Diff 1.19
+			$matchindex=Algorithm::Diff::_longestCommonSubsequence($delcmds,$addcmds, 0, sub { $_[0]->[0] } );
+		      } else {
+			### Algorithm::Diff 1.15
+			$matchindex=Algorithm::Diff::_longestCommonSubsequence($delcmds,$addcmds, sub { $_[0]->[0] } );
+		      }
+
+		      for ($i=0 ; $i<=$#$matchindex ; $i++) {
+			if (defined($matchindex->[$i])){
+			  $j=$matchindex->[$i];
+			  @delmid=splitlatex($delcmds->[$i][3]);
+			  @addmid=splitlatex($addcmds->[$j][3]);
+			  while (scalar(@$deltextblocks)  && $deltextblocks->[0][0]<$delcmds->[$i][1]) {
+			    my ($index,$block,$cnt)=@{ shift(@$deltextblocks) };
+			    push(@$todo, [$index,-1,$cnt,@$block]);
+			  }
+			  push(@$todo, [ $delcmds->[$i][1],-1,-1,$delcmds->[$i][2],@delmid,$delcmds->[$i][4]]);
+
+			  while (scalar(@$addtextblocks) && $addtextblocks->[0][0]<$addcmds->[$j][1]) {
+			    my ($index,$block,$cnt)=@{ shift(@$addtextblocks) };
+			    push(@$todo, [-1,$index,$cnt,@$block]);
+			  }
+			  push(@$todo, [ -1,$addcmds->[$j][1],-1,$addcmds->[$j][2],@addmid,$addcmds->[$j][4]]);
+			}
+		      }
+		      # mop up remaining textblocks
+		      while (scalar(@$deltextblocks)) {
+			my ($index,$block,$cnt)=@{ shift(@$deltextblocks) } ;
+			push(@$todo, [$index,-1,$cnt,@$block]);
+		      }
+		      while (scalar(@$addtextblocks)) {
+			my ($index,$block,$cnt)=@{ shift(@$addtextblocks) };
+			push(@$todo, [-1,$index,$cnt,@$block]);
+		      }
+
+		      $addblock=[];
+		      $delblock=[];
+		    }
+		    push(@$block,$seq2->[$_[1]]);
+		    $cnt++  };
+
+  my $keyfunc = sub { join("  ",split(" ",shift())) };
+
+  traverse_sequences($seq1,$seq2, { MATCH=>$match, DISCARD_A=>$discard, DISCARD_B=>$add }, $keyfunc );
+
+
+  # now carry out the merging/splitting.  Refer to elements relative from
+  # the end (with negative indices) as these offsets don't change before the instruction is executed
+  # cnt>0: merged small unchanged groups with previous changed blocks
+  # cnt==-1: split textual commands into components
+  foreach $instruction ( @$todo) {
+    ($last1,$last2,$cnt,@$block)=@$instruction ;
+    if ($cnt>=0) {
+      splice(@$seq1,$last1-$len1,1+$cnt,join("",$seq1->[$last1-$len1],@$block)) if $last1>=0;
+      splice(@$seq2,$last2-$len2,1+$cnt,join("",$seq2->[$last2-$len2],@$block)) if $last2>=0;
+    } else {
+      splice(@$seq1,$last1-$len1,1,@$block) if $last1>=0;
+      splice(@$seq2,$last2-$len2,1,@$block) if $last2>=0;
+    }
+  }
+
+  if ($verbose) {
+    print STDERR "\n";
+    print STDERR "  $mattokcnt matching  tokens in $matblkcnt blocks.\n";
+    print STDERR "  $deltokcnt discarded tokens in $delblkcnt blocks.\n";
+    print STDERR "  $addtokcnt appended  tokens in $addblkcnt blocks.\n";
+  }
+}
+
+
+# extracttextblocks(\@blockindex)
+# $blockindex has the following format
+# [ [ token1, index1 ], [token2, index2],.. ]
+# where index refers to the index in the original old or new word sequence
+# Returns: reference to an array of the form
+# [[ $index, $textblock, $cnt ], ..
+# where $index index of block to be merged
+#       $textblock contains all the words to be merged with the word at $index (but does not contain this word)
+#       $cnt   is length of block
+#
+# requires: iscmd
+#
+sub extracttextblocks {
+  my $block=shift;
+  my ($i,$token,$index);
+  my $textblock=[];
+  my $last=-1;
+  my $wpat=qr/^(?:[a-zA-Z.,'`:;?()!]*)[\s~]*$/;  #'
+  my $retval=[];
+
+  # we redefine locally $extraspace (shadowing the global definition) to capture command sequences with intervening spaces no matter what the global setting
+  # this is done so we can capture those commands with a predefined number of arguments without having to introduce them again explicitly here
+  my $extraspace='\s*';
+
+  for ($i=0;$i< scalar @$block;$i++) {
+    ($token,$index)=@{ $block->[$i] };
+    # store pure text blocks
+    if ($token =~ /$wpat/ ||  ( $token =~/^\\((?:[`'^"~=.]|[\w\d@\*]+))((?:${extraspace}\[$brat_n\]${extraspace}|${extraspace}\{$pat_n\})*)/
+				&& iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL)
+				&& !iscmd($1,\@TEXTCMDLIST,\@TEXTCMDEXCL))) {
+      # we have text or a command which can be treated as text
+      if ($last<0) {
+	# new pure-text block
+	$last=$index;
+      } else {
+	# add to pure-text block
+	push(@$textblock, $token);
+      }
+    } else {
+      # it is not text
+      if (scalar(@$textblock)) {
+	push(@$retval,[ $last, $textblock, scalar(@$textblock) ]);
+      }
+      $textblock=[];
+      $last=-1;
+    }
+  }
+  # finish processing a possibly unfinished block before returning
+  if (scalar(@$textblock)) {
+    push(@$retval,[ $last, $textblock, scalar(@$textblock) ]);
+  }
+  return($retval)
+}
+
+
+
+# extractcommands( \@blockindex )
+# $blockindex has the following format
+# [ [ token1, index1 ], [token2, index2],.. ]
+# where index refers to the index in the original old or new word sequence
+# Returns: reference to an array of the form
+# [ [ "\cmd1", index, "\cmd1[optarg]{arg1}{", "arg2" ,"} " ],..
+# where index is just taken from input array
+# command must have a textual argument as last argument
+#
+# requires: iscmd
+#
+sub extractcommands {
+  my $block=shift;
+  my ($i,$token,$index,$cmd,$open,$mid,$closing);
+  my $retval=[];
+
+  # we redefine locally $extraspace (shadowing the global definition) to capture command sequences with intervening spaces no matter what the global setting
+  # this is done so we can capture those commands with a predefined number of arguments without having to introduce them again explicitly here
+  my $extraspace='\s*';
+
+  for ($i=0;$i< scalar @$block;$i++) {
+    ($token,$index)=@{ $block->[$i] };
+    # check if token is an alphanumeric command sequence with at least one non-optional argument
+    # \cmd[...]{...}{last argument}
+    # Capturing in the following results in these associations
+    # $1: \cmd[...]{...}{
+    # $2: \cmd
+    # $3: last argument
+    # $4: }  + trailing spaces
+    if ( ( $token =~ m/^(\\([\w\d\*]+)(?:${extraspace}\[$brat_n\]|${extraspace}\{$pat_n\})*${extraspace}\{)($pat_n)(\}\s*)$/so )
+	 && iscmd($2,\@TEXTCMDLIST,\@TEXTCMDEXCL) ) {
+      print STDERR "DEBUG EXTRACTCOMMANDS Match |$1|$2|$3|$4|$index \n" if $debug;
+      #      push(@$retval,[ $2,$index,$1,$3,$4 ]);
+      ($cmd,$open,$mid,$closing) = ($2,$1,$3,$4) ;
+      $closing =~ s/\}/\\RIGHTBRACE/ ;
+      push(@$retval,[ $cmd,$index,$open,$mid,$closing ]);
+    }
+  }
+  return $retval;
+}
+
+# iscmd($cmd,\@regexarray,\@regexexcl) checks
+# return 1 if $cmd matches any of the patterns in the
+# array $@regexarray, and none of the patterns in \@regexexcl, otherwise return 0
+sub iscmd {
+  my ($cmd,$regexar,$regexexcl)=@_;
+  my ($ret)=0;
+  ### print STDERR "DEBUG: iscmd($cmd)=" if $debug;
+  foreach $pat ( @$regexar ) {
+    if ( $cmd =~ m/^${pat}$/ ) {
+      $ret=1 ;
+      last;
+    }
+  }
+  ### print STDERR "0\n" if ($debug && !$ret) ;
+  return 0 unless $ret;
+  foreach $pat ( @$regexexcl ) {
+    ### print STDERR "0\n" if ( $debug && $cmd =~ m/^${pat}$/) ;
+    return 0 if ( $cmd =~ m/^${pat}$/ );
+  }
+  ### print STDERR "1\n" if $debug;
+  return 1;
+}
+
+
+# pass2( \@seq1,\@seq2)
+# Look for differences between seq1 and seq2.
+# Mark begin and end of deleted and appended sequences with tags $DELOPEN and $DELCLOSE
+# and $ADDOPEN and $ADDCLOSE, respectively, however exclude { } & and all comands, unless
+# they match an element of the whitelist (SAFECMD)
+# For words in TEXTCMD but not in SAFECMD, enclose interior with $ADDOPEN and $ADDCLOSE brackets
+# Deleted comment lines are marked with %DIF <
+# Added comment lines are marked with %DIF >
+sub pass2 {
+  my $seq1 = shift ;
+  my $seq2 = shift ;
+
+  my ($addtokcnt,$deltokcnt,$mattokcnt)=(0,0,0);
+  my ($addblkcnt,$delblkcnt,$matblkcnt)=(0,0,0);
+
+  my $retval = [];
+  my $delhunk   = [];
+  my $addhunk   = [];
+
+  my $discard = sub { $deltokcnt++;
+                      push ( @$delhunk, $seq1->[$_[0]]) };
+
+  my $add = sub { $addtokcnt++;
+                  push ( @$addhunk, $seq2->[$_[1]]) };
+
+  my $match = sub { $mattokcnt++;
+		    if ( scalar @$delhunk ) {
+                      $delblkcnt++;
+		      # mark up changes, but comment out commands
+                      push @$retval,marktags($DELMARKOPEN,$DELMARKCLOSE,$DELOPEN,$DELCLOSE,$DELCMDOPEN,$DELCMDCLOSE,$DELCOMMENT,$delhunk);
+		      $delhunk = [];
+		    }
+                    if ( scalar @$addhunk ) {
+                      $addblkcnt++;
+                      # we mark up changes, but simply quote commands
+                      push @$retval,marktags($ADDMARKOPEN,$ADDMARKCLOSE,$ADDOPEN,$ADDCLOSE,"","",$ADDCOMMENT,$addhunk);
+		      $addhunk = [];
+		    }
+		    push(@$retval,$seq2->[$_[1]]) };
+
+  my $keyfunc = sub { join("  ",split(" ",shift())) };
+
+  traverse_sequences($seq1,$seq2, { MATCH=>$match, DISCARD_A=>$discard, DISCARD_B=>$add }, $keyfunc );
+  # clear up unprocessed hunks
+  push @$retval,marktags($DELMARKOPEN,$DELMARKCLOSE,$DELOPEN,$DELCLOSE,$DELCMDOPEN,$DELCMDCLOSE,$DELCOMMENT,$delhunk) if scalar @$delhunk;
+  push @$retval,marktags($ADDMARKOPEN,$ADDMARKCLOSE,$ADDOPEN,$ADDCLOSE,"","",$ADDCOMMENT,$addhunk) if scalar @$addhunk;
+
+
+  if ($verbose) {
+    print STDERR "\n";
+    print STDERR "  $mattokcnt matching  tokens. \n";
+    print STDERR "  $deltokcnt discarded tokens in $delblkcnt blocks.\n";
+    print STDERR "  $addtokcnt appended  tokens in $addblkcnt blocks.\n";
+  }
+  return(@$retval);
+}
+
+# marktags($openmark,$closemark,$open,$close,$opencmd,$closecmd,$comment,\@block)
+# returns ($openmark,$open,$block,$close,$closemark) if @block contains no commands (except white-listed ones),
+# braces, ampersands, or comments
+# mark comments with $comment
+# exclude all other exceptions from scope of open, close like this
+# ($openmark, $open,...,$close, $opencmd,command, command,$closecmd, $open, ..., $close, $closemark)
+# If $opencmd begins with "%" marktags assumes it is operating on a deleted block, otherwise on an added block
+sub marktags {
+  my ($openmark,$closemark,$open,$close,$opencmd,$closecmd,$comment,$block)=@_;
+  my $word;
+  my (@argtext);
+  my $retval=[];
+  my $noncomment=0;
+  my $cmd=-1;    # -1 at beginning 0: last token written is a ordinary word
+                 # 1: last token written is a command
+                # for keeping track whether we are just in a command sequence or in a word sequence
+  my $cmdcomment= ($opencmd =~ m/^%/);  # Flag to indicate whether opencmd is a comment (i.e. if we intend to simply comment out changed commands)
+  my ($command,$commandword,$closingbracket) ; # temporary variables needed below to remember sub-pattern matches
+
+
+
+# split this block to split sequences joined in pass1
+  ### print STDERR "DEBUG: marktags before splitlatex blocksplit ",join("|",@$block),"\n" if $debug;
+  @$block=splitlatex(join "",@$block);
+  ### print STDERR "DEBUG: marktags $openmark,$closemark,$open,$close,$opencmd,$closecmd,$comment\n" if $debug;
+  print STDERR "DEBUG: after splitlatex ",join("|",@$block),"\n" if $debug;
+
+  # we redefine locally $extraspace (shadowing the global definition) to capture command sequences with intervening spaces no matter what the global setting
+  # this is done so we can capture those commands with a predefined number of arguments without having to introduce them again explicitly here
+  my $extraspace_mt='\s*';
+
+
+  foreach (@$block) {
+    $word=$_;
+    if ( $word =~ s/^%/%$comment/ ) {
+      # a comment
+      if ($cmd==1) {
+	push (@$retval,$closecmd) ;
+	$cmd=-1;
+      }
+      push (@$retval,$word);
+      next;
+    }
+    if ( $word =~ m/^\s*$/ ) {
+      ### print STDERR "DEBUG MARKTAGS: whitespace detected |$word| cmdcom |$cmdcomment| |$opencmd|\n" if $debug;
+      # a sequence of white-space characters - this should only ever happen for the first element of block.
+      # in deleted block, omit, otherwise just copy it in
+      if ( ! $cmdcomment) {   # ignore in deleted blocks
+	push(@$retval,$word);
+      }
+      next;
+    }
+    if (! $noncomment) {
+      push (@$retval,$openmark);
+      $noncomment=1;
+    }
+    # negative lookahead pattern (?!) in second clause is put in to avoid matching \( .. \) patterns
+    # also note that second pattern will match \\
+    if (  $word =~ /^[&{}\[\]]/ || ( $word =~ /^\\(?!\()(\\|[`'^"~=.]|[\w*@]+)/ &&  !iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL)) ) {
+      ###    if ( $word =~ /^[&{}\[\]]/ || ( $word =~ /^\\([\w*@\\% ]+)/ && !iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL)) ) {
+      # word is a command or other significant token (not in SAFECMDLIST)
+	## same conditions as in subroutine extractcommand:
+	# check if token is an alphanumeric command sequence with at least one non-optional argument
+	# \cmd[...]{...}{last argument}
+	# Capturing in the following results in these associations
+	# $1: \cmd[...]{...}{
+	# $2: cmd
+	# $3: last argument
+	# $4: }  + trailing spaces
+	### pre-0.3    if ( ( $token =~ m/^(\\([\w\d\*]+)(?:\[$brat0\]|\{$pat_n\})*\{)($pat_n)(\}\s*)$/so )
+      if ( ( $word =~ m/^(\\([\w\d\*]+)(?:${extraspace_mt}\[$brat_n\]|${extraspace_mt}\{$pat_n\})*${extraspace_mt}\{)($pat_n)(\}\s*)$/so )
+	   && (iscmd($2,\@TEXTCMDLIST,\@TEXTCMDEXCL)|| iscmd($2,\@MATHTEXTCMDLIST,\@MATHTEXTCMDEXCL))
+           && ( !$cmdcomment || !iscmd($2,\@CONTEXT2CMDLIST, \@CONTEXT2CMDEXCL) )  ) {
+	# Condition 1: word is a command? - if yes, $1,$2,.. will be set as above
+        # Condition 2: word is a text command - we mark up the interior of the word. There is a separate check for MATHTEXTCMDLIST
+        #              because for $mathmarkup=WHOLE, the commands should not be split in pass1 (ie. math mode commands are not in
+        #              TEXTCMDLIST, but the interior of MATHTEXT commnds should be highlighted in both deleted and added blocks
+        # Condition 3: But if we are in a deleted block ($cmdcomment=1) and
+        #            $2 (the command) is in context2, just treat it as an ordinary command (i.e. comment it open with $opencmd)
+        # Because we do not want to disable this command
+	# here we do not use $opencmd and $closecmd($opencmd is empty)
+	print STDERR "DEBUG: Detected text |$word| but not safe command \$2: $2 \$3: $3\n." if $debug;
+	if ($cmd==1) {
+	  push (@$retval,$closecmd) ;
+	} elsif ($cmd==0) {
+	  push (@$retval,$close) ;
+	}
+        $command=$1; $commandword=$2; $closingbracket=$4;
+	@argtext=splitlatex($3);   # split textual argument into tokens
+	# and mark it up (but we do not need openmark and closemark)
+        # insert command with initial arguments, marked-up final argument, and closing bracket
+	if ( $cmdcomment && iscmd($commandword,\@CONTEXT1CMDLIST, \@CONTEXT1CMDEXCL) ) {
+	  # context1cmd in a deleted environment; delete command itself but keep last argument, marked up
+	  push (@$retval,$opencmd);
+	  $command =~ s/\n/\n${opencmd}/sg ; # repeat opencmd at the beginning of each line
+	  # argument, note that the additional comment character is included
+          # to suppress linebreak after opening parentheses, which is important
+          # for latexrevise
+          push (@$retval,$command,"%\n{$AUXCMD\n",marktags("","",$open,$close,$opencmd,$closecmd,$comment,\@argtext),$closingbracket);
+        } elsif ( iscmd($commandword,,\@MATHTEXTCMDLIST, \@MATHTEXTCMDEXCL) ) {
+	  # MATHBLOCK pseudo command: consider all commands safe, except & and \\, \begin and \end and a few package sprcific one (look at UNSAFEMATHCMD definition)
+	  # Keep these commands even in deleted blocks, hence set $opencmd and $closecmd (5th and 6th argument of marktags) to
+	  # ""
+	  local @SAFECMDLIST=(".*");
+	  local @SAFECMDEXCL=('\\','\\\\',@UNSAFEMATHCMD);
+	  push(@$retval,$command,marktags("","",$open,$close,"","",$comment,\@argtext)#@argtext
+                       ,$closingbracket);
+        } else {
+	  # normal textcmd or context1cmd in an added block
+	  push (@$retval,$command,marktags("","",$open,$close,$opencmd,$closecmd,$comment,\@argtext),$closingbracket);
+	}
+	push (@$retval,$AUXCMD,"\n") if $cmdcomment ;
+	$cmd=-1 ;
+      } elsif ( $cmdcomment &&
+	       ( $word =~ m/^(\\([\w\d\*]+)(?:${extraspace_mt}\[$brat_n\]|${extraspace_mt}\{$pat_n\})*${extraspace_mt}\{)($pat_n)(\}\s*)/so )
+	       && iscmd($2,\@KEEPCMDLIST, \@KEEPCMDEXCL) ) {
+	# 'keepcmd' in a deleted environment: keep  the command as is
+	push (@$retval,$close) if $cmd==0 ;
+	push (@$retval,$word);
+	$cmd=-1;  # pretend we are at the beginning of a sequence because we do not want to add an additional $closecmd or $close before the next token, no matter what it is
+      } else {
+	# ordinary command
+	push (@$retval,$opencmd) if $cmd==-1 ;
+	push (@$retval,$close,$opencmd) if $cmd==0 ;
+	$word =~ s/\n/\n${opencmd}/sg if $cmdcomment ;   # if opencmd is a comment, repeat this at the beginning of every line
+        ### print STDERR "MARKTAGS: Add command |$word|\n";
+	push (@$retval,$word);
+	$cmd=1;
+      }
+    } else {
+      ###print STDERR "DEBUG MARKTAGS is an ordinary word or SAFECMD command \n" if $debug;
+      # just an ordinary word or command in SAFECMD
+      push (@$retval,$open) if $cmd==-1 ;
+      push (@$retval,$closecmd,$open) if $cmd==1 ;
+      ###TODO:  check here if it is a command in MBOXCMD list, and surround it with \mbox{...}
+      ### $word =~ /^\\(?!\()(\\|[`'^"~=.]|[\w*@]+)/ &&  iscmd($1,\@MBOXCMDLIST,\@MBOXCMDEXCL))
+      ### but actually this check has been carried out already so can simply check if word begins with backslash
+      if ( $word =~ /^\\(?!\()(\\|[`'^"~=.]|[\w*@]+)(.*?)(\s*)$/s &&  iscmd($1,\@MBOXCMDLIST,\@MBOXCMDEXCL)) {
+	# $word is a safe command in MBOXCMDLIST
+	###print STDERR "DEBUG Mboxsafecmd detected:$word:\n" if $debug ;
+	push(@$retval,"\\mbox{$AUXCMD\n\\" . $1 . $2 . $3 ."}\\hskip0pt$AUXCMD\n" );
+      } else {
+	# $word is a normal word or a safe command (not in MBOXCMDLIST)
+	push (@$retval,$word);
+      }
+      $cmd=0;
+    }
+  }
+  push (@$retval,$close) if $cmd==0;
+  push (@$retval,$closecmd) if $cmd==1;
+
+  push (@$retval,$closemark) if ($noncomment);
+  return @$retval;
+}
+
+#used in preprocess
+sub take_comments_and_newline_from_frac() {
+  # some special magic for common usage of frac, which does not conform to the latexdiff requirements but can be made to fit
+  # note that this is a rare exception to the general rule that the new tex can be reconstructed from the diff file
+
+  # regex that matches space and comment characters
+  my $space = qr/\s|%[^\n]*?/;
+  # \frac {abc} -> \frac{abc}
+  # \frac1 -> \frac{1}
+  # \frac a -> \frac{a}
+  # \frac \lambda -> \frac{\lambda}
+  s/\\frac(?|${space}+\{($pat_n)\}|${space}*(\d)|${space}+(\w)|${space}*(\\[a-zA-Z]+))/\\frac\{$1\}/g;
+  # same as above for the second argument of frac
+  s/\\frac(\{$pat_n\})(?|${space}*\{($pat_n)\}|${space}*(\d)|${space}+(\w)|${space}*(\\[a-zA-Z]+))/\\frac$1\{$2\}/g;
+}
+
+# preprocess($string, ..)
+# carry out the following pre-processing steps for all arguments:
+# 1. Remove leading white-space
+#    Change \{ to \QLEFTBRACE and \} to \QRIGHTBRACE and \& to \AMPERSAND
+# #.   Change {,},\frac in comments to \CLEFTBRACE, \CRIGHTBRACE, \CFRAC
+# 2. mark all first empty line (in block of several) with \PAR tokens
+# 3. Convert all '\%' into '\PERCENTAGE ' and all '\$' into \DOLLAR to make parsing regular expressions easier
+# 4. Convert all \verb|some verbatim text| commands (where | can be an arbitrary character)
+#    into \verb{hash}  (also lstinline)
+# 5. Convert \begin{verbatim} some verbatim text \end{verbatim} into \verbatim{hash}  (not only verbatim, all patterns matching VERBATIMENV)
+# 6. Convert _n into \SUBSCRIPTNB{n} and _{nnn} into \SUBSCRIPT{nn}
+# 7. Convert ^n into \SUPERSCRIPTNB{n} and ^{nnn} into \SUPERSCRIPT{nn}
+# 8. a. Convert $$ $$ into \begin{DOLLARDOLLAR} \end{DOLLARDOLLAR}
+#    b. Convert \[ \] into \begin{SQUAREBRACKET} \end{SQUAREBRACKET}
+# 9. Convert all picture environmentent (\begin{PICTUREENV} .. \end{PICTUREENV} \PICTUREBLOCKenv
+#     For math-mode COARSE,WHOLE or NONE option -convert all \begin{MATH} .. \end{MATH}
+#    into \MATHBLOCKmath{...} commands, where MATH/math is any valid math environment
+
+# 10. Add final token STOP to the very end.  This is put in because the algorithm works better if the last token is identical.  This is removed again in postprocessing.
+#
+# NB: step 6 and 7 is likely to  convert some "_" inappropriately, e.g. in file
+#     names or labels but it does not matter because they are converted back in the postprocessing step
+# Returns: leading white space removed in step 1
+sub preprocess {
+  for (@_) {
+
+
+    # change in \verb and similar commands - note that I introduce an extra space here so that the
+    #       already hashed variants do not trigger again
+    # transform \lstinline{...}
+#    s/\\lstinline(\[$brat0\])?(\{(?:.*?)\})/"\\DIFlstinline". $1 ."{". tohash(\%verbhash,"$2") ."}"/esg;
+#    s/\\lstinline(\[$brat0\])?((\S).*?\2)/"\\DIFlstinline". $1 ."{". tohash(\%verbhash,"$2") ."}"/esg;
+    s/\\lstinline((?:\[$brat_n\])?)(\{(?:.*?)\})/"\\DIFlstinline". $1 ."{". tohash(\%verbhash,"$2") ."}"/esg;
+    s/\\lstinline((?:\[$brat_n\])?)(([^\s\w]).*?\3)/"\\DIFlstinline". $1 ."{". tohash(\%verbhash,"$2") ."}"/esg;
+    s/\\(verb\*?|lstinline)([^\s\w])(.*?)\2/"\\DIF${1}{". tohash(\%verbhash,"${2}${3}${2}") ."}"/esg;
+
+    #    Change \{ to \QLEFTBRACE, \} to \QRIGHTBRACE, and \& to \AMPERSAND
+    s/(?<!\\)\\\{/\\QLEFTBRACE /sg;
+    s/(?<!\\)\\\}/\\QRIGHTBRACE /sg;
+    s/(?<!\\)\\&/\\AMPERSAND /sg;
+# replace {,}, \frac in comments with \\CLEFTBRACE,\\CRIGHTBRACE, CFRAC to protect from special treatment
+    1 while s/((?<!\\)%.*)\{(.*)$/$1\\CLEFTBRACE $2/mg ;
+    1 while s/((?<!\\)%.*)\}(.*)$/$1\\CRIGHTBRACE $2/mg ;
+    1 while s/((?<!\\)%.*)\\frac(.*)$/$1\\CFRAC $2/mg ;
+
+    s/(?<!\\)\\%/\\PERCENTAGE /g ;  # (?<! is negative lookbehind assertion to prevent \\% from being converted
+    s/(?<!\\)\\\$/\\DOLLAR /g ;  # (?<! is negative lookbehind assertion to prevent \\$ from being converted
+    s/\\begin\{($VERBATIMENV)\}(.*?)\\end\{\1\}/"\\${1}{". tohash(\%verbhash,"${2}") . "}"/esg;
+    s/\\begin\{($VERBATIMLINEENV)\}(.*?)\\end\{\1\}/"\\begin{$1}". linecomment($2) . "\\end{$1}"/esg;
+
+    # mark all first empty line (in block of several) with \PAR tokens
+    s/\n(\s*?)\n((?:\s*\n)*)/\n$1\\PAR\n$2/g ;
+    # Convert _n or _\cmd into \SUBSCRIPTNB{n} or \SUBSCRIPTNB{\cmd} and _{nnn} into \SUBSCRIPT{nn}
+    1 while s/(?<!\\)_(\s*([^{\\\s]|\\\w+))/\\SUBSCRIPTNB{$1}/g ;
+    1 while s/(?<!\\)_(\s*{($pat_n)})/\\SUBSCRIPT$1/g ;
+    # Convert ^n into \SUPERSCRIPTNB{n} and ^{nnn} into \SUPERSCRIPT{nn}
+    1 while s/(?<!\\)\^(\s*([^{\\\s]|\\\w+))/\\SUPERSCRIPTNB{$1}/g ;
+    1 while s/(?<!\\)\^(\s*{($pat_n)})/\\SUPERSCRIPT$1/g ;
+    # Convert  \sqrt{n} into \SQRT{n}  and  \sqrt nn into SQRTNB{nn}
+    1 while s/(?<!\\)\\sqrt(\s*([^{\\\s]|\\\w+))/\\SQRTNB{$1}/g ;
+    1 while s/(?<!\\)\\sqrt(\s*{($pat_n)})/\\SQRT$1/g ;
+    # Convert $$ $$ into \begin{DOLLARDOLLAR} \end{DOLLARDOLLAR}
+    s/\$\$(.*?)\$\$/\\begin{DOLLARDOLLAR}$1\\end{DOLLARDOLLAR}/sg;
+    # Convert \[ \] into \begin{SQUAREBRACKET} \end{SQUAREBRACKET}
+    s/(?<!\\)\\\[/\\begin{SQUAREBRACKET}/sg;
+    s/\\\]/\\end{SQUAREBRACKET}/sg;
+    # Convert all picture environmentent (\begin{PICTUREENV} .. \end{PICTUREENV} \PICTUREBLOCKenv
+    s/\\begin\{($PICTUREENV)}(.*?)\\end\{\1}/\\PICTUREBLOCK$1\{$2\}/sg;
+    #    For math-mode COARSE,WHOLE or NONE option -convert all \begin{MATH} .. \end{MATH}
+    #    into \MATHBLOCKMATH{...} commands, where MATH is any valid math environment
+    #    Also convert all array environments into ARRAYBLOCK environments.
+    #    Where these environments have arguments they become optional arguments to the MATHBLOCK command enclosed in < > brackets
+    #    Example: \begin{alignat}{3} ... \end{alignat} will turn into \MATHBLOCKalignat[{3}]{ ... }
+
+    if ( $mathmarkup != FINE ) {
+      # DIFANCHORARRB and DIFANCHORARRE, DIFANCHORMATHB and DIFANCHORMATHE markers are inserted here to encourage the matching algorithm
+      # to always match up the closing brace. Otherwise sometimes one ends up with a situation where
+      # the closing brace is deleted and added at another point. The deleted closing brace is then
+      # prevented by a %DIFDELCMD, leading to material leaking in or out of the math environment.
+      # The anchors are removed in post-processing again. (note that they are simple text to cause least amount of complications
+      # Admittedly, this is something of a hack and will not always work. If it does not, then one needs to
+      # resort to WHOLE or FINE, or NONE math mode processing.
+      s/\\begin\{($ARRENV)}(.*?)\\end\{\1}/\\ARRAYBLOCK$1\{$2\\DIFANCHORARRB \}\\DIFANCHORARRE /sg;
+
+      take_comments_and_newline_from_frac();
+
+      # Convert Math environments with arguments
+      s/\\begin\{($MATHENV|$MATHARRENV|SQUAREBRACKET)\}((?:\[$brat_n\])|(?:\{$pat_n\}))+(.*?)\\end\{\1\}/\\MATHBLOCK$1\[$2\]\{$3\\DIFANCHORMATHB \}\\DIFANCHORMATHE /sg;
+      # Convert Math environments without arguments
+      s/\\begin\{($MATHENV|$MATHARRENV|SQUAREBRACKET)\}(.*?)\\end\{\1\}/\\MATHBLOCK$1\{$2\\DIFANCHORMATHB \}\\DIFANCHORMATHE /sg;
+    }
+
+    # add final token " STOP"
+    $_ .= " STOP"
+  }
+}
+
+
+# $expanded=linecomment($string)
+#preface all lines with verbatim marker (usually DIFVRB)
+sub linecomment {
+  my @verbatimlines=split("\n",$_[0]);
+  # the first line needs special treatment - we do want to retain optional arguments as is but wrap the remainder also with VERBCOMMENT
+  ### print STDERR "DEBUG: before verbatimlines[0] = ",$verbatimlines[0],"\n";
+  $verbatimlines[0]=~s/^((?:\s*\[$brat_n\])?\s*)([^\s\[].*)/ defined($2) ? ( "$1\%$VERBCOMMENT$2" ) : ( $1 )/e;
+  ### print STDERR "DEBUG: after  verbatimlines[0] = ",$verbatimlines[0],"\n";
+  return(join("\n%$VERBCOMMENT",@verbatimlines)."\n");
+}
+
+# $simple=reverselinecomment($env $string)
+# remove DIFVRB comments but leave changed lines marked
+sub reverselinecomment {
+  my ($environment, $verbatimtext)=@_;
+  ### print STDERR "DEBUG REVERSELINECOMMENT input: $environment,|$verbatimtext|\n" if $debug;
+  # remove markup added by latexdiff
+  # (this should occur only if the type of verbatim environment was changed)
+  # (note that this destroys some information in old file)
+  #  in theory I could save it by moving it out of the verbatim environment
+  #  but this requires more bookkeeping and is probably not necessary)
+  $verbatimtext =~ s/\\DIFaddbegin //g;
+  $verbatimtext =~ s/\\DIFaddend //g;
+  $verbatimtext =~ s/\\DIFdelbegin //g;
+  $verbatimtext =~ s/\\DIFdelend //g;
+  $verbatimtext =~ s/$DELCMDOPEN.*//g;
+
+  # remove DIFVRB mark
+  $verbatimtext=~ s/%$VERBCOMMENT//g;
+
+  # remove part of the markup in changed lines
+  # if any of these substitution was made, then there was at least
+  # one changed line, and we have to extend the style
+  if ( $verbatimtext=~ s/$VERBCOMMENT//g ) {
+    # in the next line we add ~alsolanguage~ modifier, but also deletes the rest of the line after the optional argument, as lstlisting commands gets sometimes
+    # very confused by what is there   (and othertimes seems to ignore this anyway)
+    unless ( $verbatimtext =~ s/^(\s*)\[($brat_n)\](.*)\n/$1\[$2,alsolanguage=DIFcode\]\n/ ) {
+      if ( $verbatimtext =~ m/^\s*\n/ ) {
+	$verbatimtext = "[alsolanguage=DIFcode]" . $verbatimtext;
+      } else {
+	$verbatimtext = "[alsolanguage=DIFcode]\n" . $verbatimtext;
+      }
+    }
+    # There is a bug in listings package (at least v1.5b) for empty comments where the actual comment command is not made invisible
+    # I therefore have to introduce an artificial '-' character at the end of empty added or deleted lines
+    $verbatimtext =~ s/($DELCOMMENT\s*)$/$1-/mg;
+    $verbatimtext = "\\DIFmodbegin\n\\begin{${environment}}${verbatimtext}\\end{${environment}}\n\\DIFmodend"
+  } else {
+    $verbatimtext = "\\begin{${environment}}${verbatimtext}\\end{${environment}}"
+  }
+  ### print STDERR "DEBUG REVERSELINECOMMENT output: |$verbatimtext|\n" if $debug;
+  return($verbatimtext);
+}
+
+
+#hashstring=tohash(\%hash,$string)
+# creates a hash value based on string and stores in %hash
+sub tohash {
+  my ($hash,$string)=@_;
+  my (@arr,$val);
+  my ($sum,$i)=(0,1);
+  my ($hstr);
+
+  @arr=unpack('c*',$string);
+
+  while (1) {
+    foreach $val (@arr) {
+      $sum += $i*$val;
+      $i++;
+    }
+    $hstr= "$sum";
+    last unless (defined($hash->{$hstr}) && $string ne $hash->{$hstr});
+    # else found a duplicate HASH need to repeat for a higher hash value
+  }
+  $hash->{$hstr}=$string;
+  ###  print STDERR "Hash:$hstr: Content:$string:\n";
+  return($hstr);
+}
+
+#string=fromhash(\%hash,$fromstring)
+# restores string value stored in hash
+#string=fromhash(\%hash,$fromstring,$prependstring)
+# additionally begins each line with prependstring
+sub fromhash {
+  my ($hash,$hstr)=($_[0],$_[1]);
+  my $retstr=$hash->{$hstr};
+  if ( $#_ >= 2) {
+    $retstr =~ s/^/$_[2]/mg;
+  }
+  return $retstr;
+}
+
+# stripdelcmpopen(string)
+# return string with $DELCMDOPEN removed
+sub stripdelcmdopen {
+  my ($str) = $_[0];
+  $str =~ s/${DELCMDOPEN}//mg;
+  return $str;
+}
+
+# writedebugfile(string, label)
+# if $debug set writes <string> to file latexdiff.debug.<label>
+# otherwise do nothing
+sub writedebugfile {
+  my ($string,$label)=@_;
+  if ( $debug ) {
+    open(RAWDIFF,">","latexdiff.debug." . $label);
+    print RAWDIFF $string;
+    close(RAWDIFF);
+  }
+}
+
+
+# postprocess($string, ..)
+# carry out the following post-processing steps for all arguments:
+# * Remove STOP token from the end
+# * Replace \RIGHTBRACE by }
+# * If option --no-del is set delete all deleted blocks
+# 1. Check all deleted blocks:
+#    a.where a deleted block contains a matching \begin and
+#      \end environment (these will be disabled by a %DIFDELCMD statements), for selected environments enable
+#      these commands again (such that for example displayed math in a deleted equation
+#      is properly within math mode.  For math mode environments replace numbered equation
+#      environments with their display only variety (so that equation numbers in new file and
+#      diff file are identical).  Where the correct type of math environment cannot be determined
+#      use a place holder MATHMODE
+#    b.where one of the commands matching $COUNTERCMD is used as a DIFAUXCMD, add a statement
+#      subtracting one from the respective counter to keep numbering consistent with new file
+#    Replace all MATHMODE environment commands by the correct environment to achieve matching
+#    pairs
+#    c. Convert MATHBLOCKmath commands to their uncounted numbers (e.g. convert equation -> displaymath
+#       (environments defined in $MATHENV will be replaced by $MATHREPL, and  environments in $MATHARRENV
+#       will be replaced by their equivalent with a * appended, e.g. align -> align*
+#    d. If in-line math mode contains array environment, enclose the whole environment in \mbox'es
+#    d. place \cite commands in mbox'es (for UNDERLINE style)
+#
+#   For added blocks:
+#    c. If in-line math mode contains array environment, enclose the whole environment in \mbox'es
+#    d. place \cite commands in mbox'es (for UNDERLINE style)
+#
+# 2.   If math-mode COARSE,WHOLE or NONE option set: Convert \MATHBLOCKmath{..} commands back to environments
+#
+#      Convert all PICTUREblock{..} commands back to the appropriate environments
+# 3. Convert DIFadd, DIFdel, DIFaddbegin , ... into FL varieties
+#    within floats (currently recognised float environments: plate,table,figure
+#    plus starred varieties).
+# 4. Remove empty %DIFDELCMD < lines
+# 4. Convert \begin{SQUAREBRACKET} \end{SQUAREBRACKET} into \[ \]
+#    Convert \begin{DOLLARDOLLAR} \end{DOLLARDOLLAR} into $$ $$
+# 5. Convert  \SUPERSCRIPTNB{n} into ^n  and  \SUPERSCRIPT{nn} into ^{nnn}
+# 6. Convert  \SUBSCRIPTNB{n} into _n  and  \SUBCRIPT{nn} into _{nnn}
+# 7. Expand hashes of verb and verbatim environments
+# 8. Convert '\PERCENTAGE ' back into '\%' and '\DOLLAR ' into '\$'
+# 9.. remove all \PAR tokens
+# 10.  package specific processing:  endfloat: make sure \begin{figure} and \end{figure} are always
+#      on a line by themselves, similarly for table environment
+#  4, undo renaming of the \begin, \end,{,}  in comments
+#    Change \QLEFTBRACE, \QRIGHTBRACE,\AMPERSAND to \{,\},\&
+#
+# Note have to manually synchronize substitution commands below and
+# DIF.. command names in the header
+sub postprocess {
+  my ($begin,$len,$float,$delblock,$addblock); ### $cnt
+  # second level blocks
+  my ($begin2,$len2,$eqarrayblock,$eqblock,$mathblock); ### $cnt2
+
+  my (@textparts,@newtextparts,@liststack,$listtype,$listlast);
+
+  my (@itemargs, $itemarg);
+
+
+  for (@_) {
+    # change $'s in comments to something harmless
+    1 while s/(%.*)\$/$1DOLLARDIF/mg ;
+
+    # Remove final STOP token
+    s/ STOP$//;
+    # Replace \RIGHTBRACE in comments by \MBLOCKRIGHTBRACE
+    # the only way to get these is as %DIFDELCMD < \RIGHTBRACE construction
+    # This essentially marks closing right braces of MATHBLOCK environments, which did not get matched
+    # up. This case should be rare, so I just leave this in the diff file output. Not really elegant
+    # but can still be dealt with later if it results in problems.
+    s/%DIFDELCMD < \\RIGHTBRACE/%DIFDELCMD < \\MBLOCKRIGHTBRACE/g ;
+    # Replace \RIGHTBRACE by }
+    s/\\RIGHTBRACE/}/g;
+
+    # Optional: Remove deleted block entirely
+    if ($onlyadditions) {
+        s/\\DIFdelbegin.*?\\DIFdelend//sg ;
+    }
+
+    # Check all deleted blocks: where a deleted block contains a matching \begin and
+    #    \end environment (these will be disabled by a %DIFDELCMD statements), enable
+    #    these commands again (such that for example displayed math in a deleted equation
+    #    is properly within math mode).  For math mode environments replace numbered equation
+    #    environments with their display only variety (so that equation numbers in new file and
+    #    diff file are identical)
+    writedebugfile($_,'postprocess1');
+
+
+    while ( m/\\DIFdelbegin.*?\\DIFdelend/sg ) {
+      # special processing within each deleted block
+      ###    while ( m/\\DIFdelbegin.*?\\DIFdelend/sg ) {
+      ###      print STDERR "DEBUG Match delblock \n||||$&||||\n at ",pos,"\n";
+      my $arrenv;    # dummy variable only needed within this block
+      my @symbols;   # another temporary variable
+      $len=length($&);
+      $begin=pos($_) - $len;
+      $delblock=$&;
+      ###   A much simpler method for math replacement might follow this strategy (can recycle part of the commands below for following
+      ###   this strategy:
+      ###   1. a Insert aux commands \begin{MATHMODE} or \end{MATHMODE} for all deleted commands opening or closing displayed math mode
+      ###      b Insert aux commands \begin{MATHARRMODE} or \end{MATHARRMODE} for all deleted commands opening or closing math array mode
+      ###   2  Replace MATHMODE and MATHARRMODE by correct pairing if appropriate partner  math command is found in text
+      ###   3  a Replace remaining \begin{MATHMODE}...\end{MATHMODE} pairs with \begin{$MATHREPL}..\end{$MATHREPL}
+      ###      b Replace remaining \begin{MATHARRMODE}...\end{MATHARRMODE} pairs with \begin{$MATHREPL}..\end{$MATHREPL}
+      ###   4  Delete all aux command math mode pairs which have simply comments or empty lines between them
+      ###   As written this won't actually work!
+
+
+      ###   Most general case: allow all included environments
+      ###      $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{(\w*\*?)\}\s*?\n)(.*?)(\%DIFDELCMD < \s*\\end\{\2\})/$1\\begin{$2}$AUXCMD\n$3\n\\end{$2}$AUXCMD\n$4/sg;
+      ### (.*?[^\n]?)\n? construct is necessary to avoid empty lines in math mode, which result in
+      ### an error
+      # displayed math environments
+      ###0.5:     $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{((?:$MATHENV)|SQUAREBRACKET)\}\s*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$MATHREPL}$AUXCMD\n$1$3\n\\end{$MATHREPL}$AUXCMD\n$4/sg;
+
+      if ($mathmarkup == FINE ) {
+        # if block contains both beginning and and markers for an explicit math environment, restore opening and closing commands
+	$delblock=~ s/(\%DIFDELCMD < \s*\\begin\{((?:$MATHENV)|SQUAREBRACKET)\}.*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$MATHREPL}$AUXCMD\n$1$3\n\\end{$MATHREPL}$AUXCMD\n$4/sg;
+	# also transform the opposite pair \end{displaymath} .. \begin{displaymath} but we have to be careful not to interfere with the results of the transformation in the line directly above
+	### pre-0.42 obsolete version which did not work on eqnarray test      $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\end\{($MATHENV)\}\s*?\n)(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\begin\{\2\})/$1\\end{$MATHREPL}$AUXCMD\n$3\n\\begin{$MATHREPL}$AUXCMD\n$4/sg;
+	###0.5:      $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\end\{((?:$MATHENV)|SQUAREBRACKET)\}\s*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\begin\{\2\})/\\end{MATHMODE}$AUXCMD\n$1$3\n\\begin{MATHMODE}$AUXCMD\n$4/sg;
+	$delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\end\{((?:$MATHENV)|SQUAREBRACKET)\}.*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\begin\{\2\})/\\end\{MATHMODE\}$AUXCMD\n$1$3\n\\begin\{MATHMODE\}$AUXCMD\n$4/sg;
+
+        # now look for unpaired %DIFDELCMD < \begin{MATHENV}; if found add \begin{$MATHREPL} and insert \end{$MATHREPL}
+        # just before end of block; again we use look-behind assertion to avoid matching constructions which have already been converted
+        if ($delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\begin\{((?:$MATHENV)|SQUAREBRACKET)\}\s*?(?:$DELCMDCLOSE|\n))/$1\\begin{$MATHREPL}$AUXCMD\n/sg ) {
+	  ### print STDERR "BINGO: begin block: \nBefore: |" . substr($_,$begin,$len) . "|\n" if $debug ;
+	  $delblock =~ s/(\\DIFdelend$)/\\end{$MATHREPL}$AUXCMD\n$1/s ;
+	  ### print STDERR "After: |" . $delblock . "|\n\n" if $debug ;
+        }
+        # now look for unpaired %DIFDELCMD < \end{MATHENV}; if found add \end{MATHMODE} and insert \begin{MATHMODE}
+        # just before end of block; again we use look-behind assertion to avoid matching constructions which have already been converted
+        if ($delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\end\{((?:$MATHENV)|SQUAREBRACKET)\}\s*?(?:$DELCMDCLOSE|\n))/$1\\end\{MATHMODE\}$AUXCMD\n/sg ) {
+	  ### print STDERR "BINGO: end block:\nBefore: |" . substr($_,$begin,$len) . "|\n" if $debug;
+	  $delblock =~ s/(\\DIFdelend$)/\\begin\{MATHMODE\}$AUXCMD\n$1/s ;
+	  ### print STDERR "After: |" . $delblock . "|\n\n" if $debug ;
+	}
+
+	### pre-0.42      # same as above for special case \[.\] (latex abbreviation for displaymath)
+        ### pre-0.42      $delblock=~ s/(\%DIFDELCMD < \s*\\\[\s*?\n())(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\\])/$1\\\[$AUXCMD\n$3\n\\\]$AUXCMD\n$4/sg;
+        ### pre-0.42      $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\\]\s*?\n())(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\\[)/$1\\\]$AUXCMD\n$3\n\\\[$AUXCMD\n$4/sg;
+        # equation array environment
+        ###pre-0.3      $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($MATHARRENV)\}\s*?\n)(.*?)(\%DIFDELCMD < \s*\\end\{\2\})/$1\\begin{$MATHARRREPL}$AUXCMD\n$3\n\\end{$MATHARRREPL}$AUXCMD\n$4/sg;
+        ###0.5      $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($MATHARRENV)\}\s*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$MATHARRREPL}$AUXCMD\n$1$3\n\\end{$MATHARRREPL}$AUXCMD\n$4/sg;
+	###1.3.2         $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($MATHARRENV)\}.*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$MATHARRREPL}$AUXCMD\n$1$3\n\\end{$MATHARRREPL}$AUXCMD\n$4/sg;
+	# Example Input: %DIFDELCMD < \begin{alignat}{2} \n\exp(ix)&=&\cos(x)+i\sin(x)\n%DIFDECMD < \end{alignat}
+        #                < $1                <$2   >><$3>  < $4                      ><$5                       >
+	# Example Output:\begin{alignat*}{2}%DIFAUXCMD\n\%DIFDELCMD\begin{align}\n\exp(ix)&=&\cos(x)+i\sin(x)\n\end{align*}%DIFAUXCMD\n%DIFDELCMD > \end{alignat*}
+        $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($MATHARRENV)\}(.*?)(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$2*}$3$AUXCMD\n$1$4\n\\end{$2*}$AUXCMD\n$5/sg;
+        ###  pre-0.42 obsolete version which did not work on eqnarray test     $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\end\{($MATHARRENV)\}\s*?\n)(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\begin\{\2\})/$1\\end{$MATHARRREPL}$AUXCMD\n$3\n\\begin{$MATHARRREPL}$AUXCMD\n$4/sg;
+        $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\end\{($MATHARRENV)\}\s*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\begin\{\2\})/\\end{MATHMODE}$AUXCMD\n$1$3\n\\begin{MATHMODE}$AUXCMD\n$4/sg;
+        ### print STDERR "STEP1: |" . $delblock . "|\n\n" if $debug ;
+
+        # now look for unpaired %DIFDELCMD < \begin{MATHARRENV}; if found add \begin{$MATHARRREPL} and insert \end{$MATHARRREPL}
+        # just before end of block; again we use look-behind assertion to avoid matching constructions which have already been converted
+        if ($delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\begin\{($MATHARRENV)\}(.*?)(?:$DELCMDCLOSE|\n))/$1\\begin{$2*}$3$AUXCMD\n/sg ) {
+	  $arrenv=$2;
+	  $delblock =~ s/(\\DIFdelend$)/\\end{$arrenv*}$AUXCMD\n$1/s ;
+        }
+        ### print STDERR "STEP2: |" . $delblock . "|\n\n" if $debug ;
+
+        # now look for unpaired %DIFDELCMD < \end{MATHENV}; if found add \end{MATHMODE} and insert \begin{MATHMODE}
+        # just before end of block; again we use look-behind assertion to avoid matching constructions which have already been converted
+        if ($delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\end\{($MATHARRENV)\}\s*?(?:$DELCMDCLOSE|\n))/$1\\end{MATHMODE}$AUXCMD\n/sg ) {
+	  $delblock =~ s/(\\DIFdelend$)/\\begin{MATHMODE}$AUXCMD\n$1/s ;
+        }
+
+	# parse $delblock for deleted and reinstated eqnarray* environments - within those reinstate \\ and & commands
+	###      while ( $delblock =~ m/\\begin{$MATHARRREPL}$AUXCMD\n.*?\n\\end{$MATHARRREPL}$AUXCMD\n/sg ) {
+        ### while ( $delblock =~ m/\\begin\Q{$MATHARRREPL}$AUXCMD\E\n.*?\n\\end\Q{$MATHARRREPL}$AUXCMD\E\n/sg ) {
+	while ( $delblock =~ m/\\begin\{$MATHARRENV\*}[^\n]*?$AUXCMD\n.*?\n\\end\{$MATHARRENV\*\}$AUXCMD\n/sg ) {
+	  ###	      print STDERR "DEBUG Match eqarrayblock $& at ",pos,"\n";
+	  $len2=length($&);
+	  $begin2=pos($delblock) - $len2;
+	  $eqarrayblock=$&;
+	  # reinstate deleted & and \\ commands
+	  ### $eqarrayblock=~ s/(\%DIFDELCMD < \s*(\&|\\\\)\s*?(?:$DELCMDCLOSE|\n))/$1$2$AUXCMD\n/sg ;
+	  $eqarrayblock    =~ s/(\%DIFDELCMD < (.*?(?:\&|\\\\).*)(?:$DELCMDCLOSE|\n))/
+        	  # The pattern captures comments with at least one of & or \\
+                  @symbols= split(m@((?:&|\\\\)\s*)@,$2);   #  extract & and \\ and other material from sequence
+                  @symbols= grep ( m@&|\\\\\s*@, @symbols); #  select & and \\ (and subsequent spaces)
+                  "$1@symbols$AUXCMD\n"
+            /eg ;
+	  ### print STDERR "Modified block:|$eqarrayblock|\n" if  $debug;
+	  # splice in modified block
+	  substr($delblock,$begin2,$len2)=$eqarrayblock;
+	  pos($delblock) = $begin2 + length($eqarrayblock);
+	}
+
+      } elsif ( $mathmarkup == COARSE || $mathmarkup == WHOLE ) {
+	#       Convert MATHBLOCKmath commands to their uncounted numbers (e.g. convert equation -> displaymath
+	#       (environments defined in $MATHENV will be replaced by $MATHREPL, and  environments in $MATHARRENV
+	#       will be replaced by their starred variety
+	$delblock=~ s/\\MATHBLOCK($MATHENV)((?:\[$brat_n\])?)\{($pat_n)\}/\\MATHBLOCK$MATHREPL$2\{$3\}/sg;
+	$delblock=~ s/\\MATHBLOCK($MATHARRENV)((?:\[$brat_n\])?)\{($pat_n)\}/\\MATHBLOCK$1\*$2\{$3\}/sg;
+      }
+
+      # Replacing math array environments with their starred varieties in deleted blocks as implemented now causes double asterisks
+      # in situation where there had been starred already, e.g. \begin{alignat*} is turned into \begin{alignat**}
+      # The following command seeks to undo this double-starring. It is a little bit of a hack because it relies on the fact that **
+      # double-starred math array environment does not occur naturally
+      $delblock =~ s/($MATHARRENV)(?<=\*)\*/$1/sg;
+
+      print STDERR "DELBLOCK after maths processing: |" . $delblock . "|\n\n" if $debug ;
+
+
+      # Reinstate completely deleted list environments. note that items within the
+      # environment will still be commented out.  They will be restored later
+      $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($LISTENV)\}\s*?(?:\n|$DELCMDCLOSE))(.*?)(\%DIFDELCMD < \s*\\end\{\2\})/{
+															###   # block within the search; replacement environment
+															###   "$1\\begin{$2}$AUXCMD\n". restore_item_commands($3). "\n\\end{$2}$AUXCMD\n$4";
+															"$1\\begin{$2}$AUXCMD\n$3\n\\end{$2}$AUXCMD\n$4";
+														       }/esg;
+      ###      $delblock=~ s/\\begin\{$MATHENV}$AUXCMD/\\begin{$MATHREPL}$AUXCMD/g;
+      ###      $delblock=~ s/\\end\{$MATHENV}$AUXCMD/\\end{$MATHREPL}$AUXCMD/g;
+      ###      $delblock=~ s/\\begin\{$MATHARRENV}$AUXCMD/\\begin{$MATHARRREPL}$AUXCMD/g;
+      ###      $delblock=~ s/\\end\{$MATHARRENV}$AUXCMD/\\end{$MATHARRREPL}$AUXCMD/g;
+
+      #    b.where one of the commands matching $COUNTERCMD is used as a DIFAUXCMD, add a statement
+      #      subtracting one from the respective counter to keep numbering consistent with new file
+      $delblock=~ s/\\($COUNTERCMD)((?:${extraspace}\[$brat_n\]${extraspace}|${extraspace}\{$pat_n\})*\s*${AUXCMD}\n)/\\$1$2\\addtocounter{$1}{-1}${AUXCMD}\n/sg ;
+
+      #    bb. disable active labels within deleted blocks (i.e. those not commented out) (as these are not safe commands, this should normally only
+      #        happen within deleted maths blocks
+      ###      $delblock=~ s/(?<!$DELCMDOPEN)(\\$LABELCMD(?:${extraspace})\{(?:[^{}])*\}[\t ]*)\n?/${DELCMDOPEN}$1${DELCMDCLOSE}/smg ;
+      ###      previous line caused trouble as by issue #90 I might need to modify this
+      $delblock=~ s/^([^%]*)(\\$LABELCMD(?:${extraspace})\{(?:[^{}])*\}[\t ]*)\n?/$1${DELCMDOPEN}$2${DELCMDCLOSE}/smg ;
+      ###      print STDERR "<<<$delblock>>>\n" if $debug;
+
+
+      #     c. If in-line math mode contains array environment, enclose the whole environment in \mbox'es
+      while ( $delblock =~ m/($math)(\s*)/sg ) {
+##		      print STDERR "DEBUG Delblock Match math $& at ",pos,"\n";
+	$len2=length($&);
+	$begin2=pos($delblock) - $len2;
+	$mathblock="%\n\\mbox{$AUXCMD\n$1\n}$AUXCMD\n";
+        next unless ( $mathblock =~ /ARRAYBLOCK/ or $mathblock =~ m/\{$ARRENV\}/ );
+	substr($delblock,$begin2,$len2)=$mathblock;
+	pos($delblock) = $begin2 + length($mathblock);
+      }
+      ###      if ($CITE2CMD) {
+      ######   ${extraspace}(?:\[$brat0\]${extraspace}){0,2}\{$pat_n\}))  .*?%%%\n
+      ###	$delblock=~s/($DELCMDOPEN\s*\\($CITE2CMD)(.*)$DELCMDCLOSE)/
+      ###	  # Replacement code
+      ###	  {my ($aux,$all);
+      ###	   $aux=$all=$1;
+      ###	   $aux=~s#\n?($DELCMDOPEN|$DELCMDCLOSE)##g;
+      ###	   $all."$aux$AUXCMD\n";}/sge;
+      ###      }
+      ###      # or protect \cite commands with \mbox
+      ###      if ($CITECMD) {
+      ######	$delblock=~s/(\\($CITECMD)${extraspace}(?:\[$brat0\]${extraspace}){0,2}\{$pat_n\})(\s*)/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/msg ;
+      ###	$delblock=~s/(\\($CITECMD)${extraspace}(?:<$abrat0>${extraspace})?(?:\[$brat0\]${extraspace}){0,2}\{$pat_n\})(\s*)/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/msg ;
+      ###      }
+      # if MBOXINLINEMATH is set, protect inlined math environments with an extra mbox
+
+
+      if ( $MBOXINLINEMATH ) {
+	# note additional \newline after command is omitted from output if right at the end of deleted block (otherwise a spurious empty line is generated)
+	$delblock=~s/($math)(?:[\s\n]*)?/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/sg;
+      }
+      ###if ( defined($packages{"listings"} and $latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/))   {
+      ###  #     change included verbatim environments
+      ###  $delblock =~ s/\\DIFverb\{/\\DIFDIFdelverb\{/g;
+      ###  $delblock =~ s/\\DIFlstinline/\\DIFDIFdellstinline/g;
+      ###}
+      # Mark deleted verbatim commands
+      $delblock =~ s/(${DELCMDOPEN}\\DIF((?:verb\*?|lstinline(?:\[$brat_n\])?)\{([-\d]*?)\}\s*).*)$/%\n\\DIFDIFdel$2${AUXCMD}\n$1/gm;
+      if ( $CUSTOMDIFCMD ) {
+        ###$delblock =~ s/(${DELCMDOPEN}.*)\\($CUSTOMDIFCMD)((?:\[${brat_n}\])*?(?:\s*\{${pat_n}\})*)/"$1${DELCMDCLOSE}\\DEL$2". stripdelcmdopen($3) ." ${DELCMDOPEN}"/egms;
+        ###$delblock =~ s/(${DELCMDOPEN}.*)\\($CUSTOMDIFCMD)/$1${DELCMDCLOSE}\\DEL$2/gm;
+        # ($1 ? "${DELCMDOPEN}$1${DELCMDCLOSE}":"") : only add the DELCMDOPEN / DELCMDCLOSE pair if there are actually any commands in between, otherwise this is redundant
+	$delblock    =~ s/${DELCMDOPEN}(.*?)\\($CUSTOMDIFCMD)((?:\[${brat_n}\])*?(?:\s*\{${pat_n}\})*)/($1 ? "${DELCMDOPEN}$1${DELCMDCLOSE}":"") ."\\DEL$2". stripdelcmdopen($3)/egs;
+        # if there is a sequence of several commands in the same row only the first will be converted due to the need to be connected to the DELCMDOPEN. To mop these up, just add the DEL to the front of any remaining cmd's in the deleted block
+        $delblock =~ s/\\($CUSTOMDIFCMD)/\\DEL$1/g;     # this will also convert comments but I guess it does not matter much
+      }
+
+      #     splice in modified delblock
+      substr($_,$begin,$len)=$delblock;
+      pos = $begin + length($delblock);
+    }
+
+    ### print STDERR "<<<$_>>>\n" if $debug;
+
+
+    # make the array modification in added blocks
+    while ( m/\\DIFaddbegin.*?\\DIFaddend/sg ) {
+      $len=length($&);
+      $begin=pos($_) - $len;
+      $addblock=$&;
+      while ( $addblock =~ m/($math)(\s*)/sg ) {
+	$len2=length($&);
+	$begin2=pos($addblock) - $len2;
+	$mathblock="%\n\\mbox{$AUXCMD\n$1\n}$AUXCMD\n";
+        next unless ( $mathblock =~ /ARRAYBLOCK/ or $mathblock =~ m/\{$ARRENV\}/) ;
+	substr($addblock,$begin2,$len2)=$mathblock;
+	pos($addblock) = $begin2 + length($mathblock);
+      }
+      # if MBOXINLINEMATH is set, protect inlined math environments with an extra mbox
+      if ( $MBOXINLINEMATH ) {
+	##$addblock=~s/($math)/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/sg;
+	$addblock=~s/($math)(?:[\s\n]*)?/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/sg;
+      }
+      ###if ( defined($packages{"listings"} and $latexdiffpreamble =~ /\\RequirePackage(?:\[$brat0\])?\{color\}/))   {
+	# mark added verbatim commands
+      $addblock =~ s/\\DIFverb/\\DIFDIFaddverb/g;
+      $addblock =~ s/\\DIFlstinline/\\DIFDIFaddlstinline/g;
+      if( $CUSTOMDIFCMD ) {
+        $addblock =~ s/\\($CUSTOMDIFCMD)/\\ADD$1/g;     # this will also convert comments but I guess it does not matter much
+      }
+      # markup the optional arguments of \item
+      $addblock =~ s/(\\$ITEMCMD$extraspace(?:<$abrat0>)?$extraspace)\[($brat_n)\]/
+	@itemargs=splitlatex(substr($2,0,length($2)));
+        $itemarg="[".join("",marktags("","",$ADDOPEN,$ADDCLOSE,"","",$ADDCOMMENT,\@itemargs))."]";
+      "$1$itemarg"/sge;   # old substitution: $1\[$ADDOPEN$2$ADDCLOSE\]
+      ###}
+#     splice in modified addblock
+      substr($_,$begin,$len)=$addblock;
+      pos = $begin + length($addblock);
+    }
+
+    # Go through whole text, and by counting list environment commands, find out when we are within a list environment.
+    # Within those restore deleted \item commands
+    @textparts=split /(?<!$DELCMDOPEN)(\\(?:begin|end)\{$LISTENV\})/ ;
+    @liststack=();
+    @newtextparts=map {
+      ### print STDERR ":::::::: $_\n";
+      if ( ($listtype) = m/^\\begin\{($LISTENV)\}$/ ) {
+	print STDERR "DEBUG: postprocess \\begin{$listtype}\n" if $debug;
+	push @liststack,$listtype;
+      } elsif ( ($listtype) = m/^\\end\{($LISTENV)\}$/ ) {
+ 	print STDERR "DEBUG: postprocess \\end{$listtype}\n" if $debug;
+	if (scalar  @liststack > 0) {
+ 	  $listlast=pop(@liststack);
+ 	  ($listtype eq $listlast) or warn "Invalid nesting of list environments: $listlast environment closed by \\end{$listtype}.";
+ 	} else {
+ 	  warn "WARNING: Invalid nesting of list environments: \\end{$listtype} encountered without matching \\begin{$listtype}.\n";
+ 	}
+      } else {
+	print STDERR "DEBUG: postprocess \@liststack=(",join(",",@liststack),")\n" if $debug;
+	if (scalar  @liststack > 0 ) {
+	  # we are within a list environment and should replace all item commands
+	  $_=restore_item_commands($_);
+	}
+	# else: we are outside a list environment and do not need to do anything
+      }
+      $_ } @textparts;     # end of map command
+    # replace the main text with the modified version
+    $_= join("",@newtextparts);
+
+
+
+
+    # Replace MATHMODE environments from step 1a above by the correct Math environment and remove unncessary pairings
+
+    if ( $mathmarkup == FINE ) {
+      # look for AUXCMD math-mode pairs which have only comments (or empty lines between them), and remove the added commands
+      # \begin{..} ... \end{..} pairs
+      s/\\begin\{((?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET|MATHMODE)\}$AUXCMD\n((?:\s*%.[^\n]*\n)*)\\end\{\1\}$AUXCMD\n/$2/sg;
+      # \end{..} .... \begin{..} pairs
+      s/\\end\{((?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET|MATHMODE)\}$AUXCMD\n((?:\s*%.[^\n]*\n)*)\\begin\{\1\}$AUXCMD\n/$2/sg;
+
+      writedebugfile($_,'postprocess15');
+      # The next line is complicated.  The negative look-ahead insertion makes sure that no \end{$MATHENV} (or other mathematical
+      # environments) are between the \begin{$MATHENV} and \end{MATHMODE} commands. This is necessary as the minimal matching
+      # is not globally minimal but only 'locally' (matching is beginning from the left side of the string)
+      # [NB: Do not be tempted to prettify the expression with /x modified. It seems this is applied after strings are expanded so will ignore spaces in strings
+      1 while s/(?<!$DELCMDOPEN)\\begin\{((?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET)}((?:${DELCMDOPEN}[^\n]*|.(?!(?:\\end\{(?:(?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET)}|\\begin\{MATHMODE})))*?)\\end\{MATHMODE}/\\begin{$1}$2\\end{$1}/s;
+      writedebugfile($_,'postprocess16');
+      1 while s/\\begin\{MATHMODE}((?:.(?!\\end\{MATHMODE}))*?)(?<!$DELCMDOPEN)\\end\{((?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET)}/\\begin{$2}$1\\end{$2}/s;
+      # convert remaining \begin{MATHMODE} \end{MATHMODE} (and not containing & or \\ )into MATHREPL environments
+      s/\\begin\{MATHMODE\}((?:(.(?!(?<!\\)\&|\\\\))*)?)\\end\{MATHMODE\}/\\begin{$MATHREPL}$1\\end{$MATHREPL}/sg;
+      # others into MATHARRREPL
+      s/\\begin\{MATHMODE\}(.*?)\\end\{MATHMODE\}/\\begin{$MATHARRREPL}$1\\end{$MATHARRREPL}/sg;
+      # Final cleanup of all equation environments: Sometimes I end up with %DIFDELCMD < \PAR lines, which I need to make sure do not get expanded into
+      # actual empty lines within equation environments (later part)
+      while (  m/\\begin\{((?:$MATHENV|$MATHARRENV|SQUAREBRACKET)\*?)}.*?(?<!\%DIFDELCMD < )\\end\{\1}/sg ) {
+	$len=length($&);
+	$begin=pos($_) - $len;
+	$eqblock=$&;
+	$eqblock =~ s/(%DIFDELCMD < )([^\n]*?)\\PAR\n/$1$2\n$1\n/sg;
+	#     splice in modified delblock
+	print STDERR "DEBUG EQBLOCK: |$eqblock|\n\n" if $debug;
+	substr($_,$begin,$len)=$eqblock;
+	pos = $begin + length($eqblock);
+      }
+    } else {
+      #   math modes OFF,WHOLE,COARSE: Convert \MATHBLOCKmath{..} commands back to environments
+      # with optionl argument to MATHBLOCK, e.g \MATHBLOCKalignat[{3}]{ ...}
+      s/\\MATHBLOCK($MATHENV|$MATHARRENV\*?|SQUAREBRACKET)\[($brat0)\]\{($pat_n)\}/\\begin{$1}$2$3\\end{$1}/sg;
+      # without optional argument e.g \MATHBLOCKalign{ ...}
+      s/\\MATHBLOCK($MATHENV|$MATHARRENV\*?|SQUAREBRACKET)\{($pat_n)\}/\\begin{$1}$2\\end{$1}/sg;
+      # convert ARRAYBLOCK.. commands back to environments
+      s/\\ARRAYBLOCK($ARRENV)\{($pat_n)\}/\\begin{$1}$2\\end{$1}/sg;
+      # get rid of the DIFANCHOR markers, first the delete comments, then everywhere
+      s/%DIFDELCMD < \\DIFANCHOR(?:MATH|ARR)[BE] (?:\n%DIFDELCMD < )?%%%\n//g ;
+      s/\\DIFANCHOR(?:MATH|ARR)[BE] //g;
+    }
+    writedebugfile($_,'postprocess2');
+
+    #  Convert all PICTUREblock{..} commands back to the appropriate environments
+    s/\\PICTUREBLOCK($PICTUREENV)\{($pat_n)\}/\\begin{$1}$2\\end{$1}/sg;
+#0.5:    # Remove all mark up within picture environments
+#     while ( m/\\begin\{($PICTUREENV)\}.*?\\end\{\1\}/sg ) {
+#       $cnt=0;
+#       $len=length($&);
+#       $begin=pos($_) - $len;
+#       $float=$&;
+#       $float =~ s/\\DIFaddbegin //g;
+#       $float =~ s/\\DIFaddend //g;
+#       $float =~ s/\\DIFadd\{($pat_n)\}/$1/g;
+#       $float =~ s/\\DIFdelbegin //g;
+#       $float =~ s/\\DIFdelend //g;
+#       $float =~ s/\\DIFdel\{($pat_n)\}//g;
+#       $float =~ s/$DELCMDOPEN.*//g;
+#       substr($_,$begin,$len)=$float;
+#       pos = $begin + length($float);
+#     }
+    # Convert DIFadd, DIFdel, DIFFaddbegin , ... into  varieties
+    #    within floats (currently recognised float environments: plate,table,figure
+    #    plus starred varieties).
+    while ( m/\\begin\{($FLOATENV)\}.*?\\end\{\1\}/sg ) {
+      $len=length($&);
+      $begin=pos($_) - $len;
+      $float=$&;
+      $float =~ s/\\DIFaddbegin /\\DIFaddbeginFL /g;
+      $float =~ s/\\DIFaddend /\\DIFaddendFL /g;
+      $float =~ s/\\DIFadd\{/\\DIFaddFL{/g;
+      $float =~ s/\\DIFdelbegin /\\DIFdelbeginFL /g;
+      $float =~ s/\\DIFdelend /\\DIFdelendFL /g;
+      $float =~ s/\\DIFdel\{/\\DIFdelFL{/g;
+      substr($_,$begin,$len)=$float;
+      pos = $begin + length($float);
+    }
+    ### former location of undo renaming of \begin and \end in comments
+
+    # remove empty DIFCMD < lines
+    s/^\Q${DELCMDOPEN}\E\n//msg;
+
+    # Expand hashes of verb and verbatim environments
+    s/${DELCMDOPEN}\\($VERBATIMENV)\{([-\d]*?)\}/"${DELCMDOPEN}\\begin{${1}}".fromhash(\%verbhash,$2,$DELCMDOPEN)."${DELCMDOPEN}\\end{${1}}"/esg;
+    # revert changes to verbatim environments for line diffs (and add code to mark up changes) (note negative look behind assertions to not leak out of DIFDELCMD comments)
+    # Example:
+    # < \begin{verbatim}
+    # < %DIF < DIFVRB old verbatim line
+    # < %DIF > DIFVRB new verbatim line
+    # < \end{verbatim}
+    # ---
+    # > \DIFmodbegin
+    # > \begin{verbatim}[alsolanguage=DIFcode]
+    # > %DIF < old verbatim line
+    # > %DIF > new verbatim line
+    # > \end{verbatim}
+    # > \DIFmodend
+    s/(?<!$DELCMDOPEN)\\begin\{($VERBATIMLINEENV)\}(.*?)(?<!$DELCMDOPEN)\\end\{\1\}/"". reverselinecomment($1, $2) .""/esg;
+#    # we do the same for deleted environments but additionally reinstate the framing commands
+#   s/$DELCMDOPEN\\begin\{($VERBATIMLINEENV)\}$extraspace(?:\[$brat0\])?$DELCMDCLOSE(.*?)$DELCMDOPEN\\end\{\1\}$DELCMDCLOSE/"\\begin{$1}". reverselinecomment($2) . "\\end{$1}"/esg;
+##    s/$DELCMDOPEN\\begin\{($VERBATIMLINEENV)\}($extraspace(?:\[$brat0\])?\s*)(?:\n|$DELCMDOPEN)*$DELCMDCLOSE((?:\%$DELCOMMENT$VERBCOMMENT.*?\n)*)($DELCMDOPEN\\end\{\1\}(?:\n|\s|$DELCMDOPEN)*$DELCMDCLOSE)/"SUBSTITUTION: \\begin{$1}$2 INTERIOR: |$3| END: |$4|"/esg;
+     s/ # Deleted \begin command of verbatim environment (Captures $1: whole deleted command, $2: environment, $3: optional arguments with white space
+          (\Q$DELCMDOPEN\E\\begin\{($VERBATIMLINEENV)\}(\Q$extraspace\E(?:\[$brat_n\])?\s*)(?:\n|\Q$DELCMDOPEN\E)*\Q$DELCMDCLOSE\E)
+        # Interior of deleted verbatim environment should consist entirely of delete DIFVRB comments, i.e. match only lines beginning with % DIF < DIFVRB
+        #   Captures: $4: all lines combined
+          ((?:\%\Q$DELCOMMENT$VERBCOMMENT\E[^\n]*?\n)*)
+        # Deleted \end command of verbatim environment. Note that the type is forced to match the opening. Captures: $5: Whole deleted environment  (previous way this line was written: (\Q$DELCMDOPEN\E\\end\{\2\}(?:\n|\s|\Q$DELCMDOPEN\E)*\Q$DELCMDCLOSE\E)
+          (\Q$DELCMDOPEN\E\\end\{\2\})
+      / # Substitution part
+            $1                   # Leave expression as is
+            . "$AUXCMD NEXT\n"   # Mark the following line as an auxiliary command
+            . ""    # reinstate the original environment without options
+            . reverselinecomment($2, "$3$4")   # modify the body to change the markup; reverselinecomment parses for options
+            . " $AUXCMD\n"  # close the auxiliary environment
+            . $5               # and again leave the original deleted closing environment as is
+      /esgx;  # Modifiers of substitution command
+    writedebugfile($_,'postprocess2');
+    # where changes have occurred in verbatim environment, change verbatim to DIFverbatim to allow mark-up
+    # (I use the presence of optional paramater to verbatim environment as the marker - normal verbatim
+    # environment does not take optional arguments)
+    s/(?<!$DELCMDOPEN)\\begin\{(verbatim[*]?)\}(\[$brat_n\].*?)\\end\{\1\}/\\begin{DIF$1}$2\\end{DIF$1}/sg;
+
+    s/\\($VERBATIMENV)\{([-\d]*?)\}/"\\begin{${1}}".fromhash(\%verbhash,$2)."\\end{${1}}"/esg;
+
+    # remove all \PAR tokens (taking care to properly keep commented out PAR's
+    # from introducing uncommented newlines - next line)
+    s/(%DIF < )([^\n]*?)\\PAR\n/$1$2\n$1\n/sg;
+    # convert PAR commands which are on a line by themselves
+    s/\n(\s*?)\\PAR\n/\n\n/sg;
+    # convert remaining PAR commands (which are preceded by non-white space characters, usually "}" ($ADDCLOSE)
+    s/\\PAR\n/\n\n/sg;
+
+    #  package specific processing:
+    if ( defined($packages{"endfloat"})) {
+      #endfloat: make sure \begin{figure} and \end{figure} are always
+      #      on a line by themselves, similarly for table environment
+      print STDERR "endfloat package detected.\n" if $verbose ;
+      # eliminate whitespace before and after
+      s/^(\s*)(\\(?:end|begin)\{(?:figure|table)\})(\s*?)$/$2/mg;
+      # split lines with remaining characters before float environment conmmand
+      s/^([^%]+)(\\(?:begin|end)\{(?:figure|table)\})/$1\n$2/mg;
+      # split lines with remaining characters after float environment conmmand
+      s/^((?:[^%]+)\\(?:begin|end)\{(?:figure|table)\}(?:\[[a-zA-Z]+\])?)(.+)((?:%.*)?)$/$1\n$2$3/mg;
+    }
+
+    # Remove empty auxiliary LISTENV  (sometmes these are generated if list environment occurs within the argument of a deleted comment)
+    #   (slightly hacky but I could not see an easy way to see if in argument of another command when this extra markup is added)
+    s/\\begin\{($LISTENV)\}$AUXCMD\n((?:\s*\%[^\n]*\n)*\n?)\\end\{\1\}$AUXCMD\n/$2\n/msg ;
+
+    # Convert '\PERCENTAGE ' back into '\%' (the final question mark catches a special situation where due to a latter pre-processing step the ' ' becomes separated
+    s/\\PERCENTAGE ?/\\%/g;
+    # Convert '\DOLLAR ' back into '\$'
+    s/\\DOLLAR /\\\$/g;
+
+    # undo renaming of the \begin and \end,{,}  and dollars in comments
+
+    # although we only renamed $ in comments to DOLLARDIFF, we might have lost the % in unchanged verbatim blocks, so rename all
+    s/DOLLARDIF/\$/g;
+#   Convert \begin{SQUAREBRACKET} \end{SQUAREBRACKET} into \[ \]
+    s/\\end\{SQUAREBRACKET\}/\\\]/sg;
+    s/\\begin\{SQUAREBRACKET\}/\\\[/sg;
+# 4. Convert \begin{DOLLARDOLLAR} \end{DOLLARDOLLAR} into $$ $$
+    s/\\begin\{DOLLARDOLLAR\}(.*?)\\end\{DOLLARDOLLAR\}/\$\$$1\$\$/sg;
+# 5. Convert  \SUPERSCRIPTNB{n} into ^n  and  \SUPERSCRIPT{nn} into ^{nnn}
+    1 while s/\\SUPERSCRIPT(\s*\{($pat_n)\})/^$1/g ;
+    1 while s/\\SUPERSCRIPTNB\{(\s*$pat0)\}/^$1/g ;
+    # Convert  \SUBSCRIPNB{n} into _n  and  \SUBCRIPT{nn} into _{nnn}
+    1 while s/\\SUBSCRIPT(\s*\{($pat_n)\})/_$1/g ;
+    1 while s/\\SUBSCRIPTNB\{(\s*$pat0)\}/_$1/g ;
+    # Convert  \SQRT{n} into \sqrt{n}  and  \SQRTNB{nn} into \sqrt nn
+    1 while s/\\SQRT(\s*\{($pat_n)\})/\\sqrt$1/g ;
+    1 while s/\\SQRTNB\{(\s*$pat0)\}/\\sqrt$1/g ;
+
+    1 while s/(%.*)\\CFRAC (.*)$/$1\\frac$2/mg ;
+    1 while s/(%.*)\\CRIGHTBRACE (.*)$/$1\}$2/mg ;
+    1 while s/(%.*)\\CLEFTBRACE (.*)$/$1\{$2/mg ;
+
+
+#    Change \QLEFTBRACE, \QRIGHTBRACE to \{,\}
+    s/\\QLEFTBRACE /\\\{/sg;
+    s/\\QRIGHTBRACE /\\\}/sg;
+    s/\\AMPERSAND /\\&/sg;
+    # Highlight added inline verbatim commands if possible
+    if ( $latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/ )   {
+      # wrap added verb commands with color commands
+      s/\\DIFDIFadd((?:verb\*?|lstinline(?:\[$brat_n\])?)\{[-\d]*?\}[\s\n]*)/\{\\color{blue}$AUXCMD\n\\DIF$1%\n\}$AUXCMD\n/sg;
+      s/\\DIFDIFdel((?:verb\*?|lstinline(?:\[$brat_n\])?)\{[-\d]*?\}[\s\n]*$AUXCMD)/\{\\color{red}${AUXCMD}\n\\DIF$1\n\}${AUXCMD}/sg;
+    } else {
+      # currently if colour markup is not used just remove the added mark
+      s/\\DIFDIFadd(verb\*?|lstinline)/\\DIF$1/sg;
+      s/\\DIFDIFdel((?:verb\*?|lstinline(?:\[$brat_n\])?)\{[-\d]*?\}[\s\n]*$AUXCMD\n)//sg;
+    }
+    # expand \verb and friends inline arguments
+    s/\\DIF((?:DIFadd|DIFdel)?(?:verb\*?|lstinline(?:\[$brat_n\])?))\{([-\d]*?)\}/"\\${1}". fromhash(\%verbhash,$2)/esg;
+    # add basicstyle color{blue} to added lstinline commands
+    # finally add the comment to the ones not having an optional argument before
+    ###s/\\DIFaddlstinline(?!\[)/\\lstinline\n[basicstyle=\\color{blue}]$AUXCMD\n/g;
+
+  }
+  return;
+}
+
+# $out = restore_item_commands($listenviron)
+# short helper function for post-process, which restores deleted \item commands in its argument (as DIFAUXCMDs)
+sub restore_item_commands {
+  my ($string)=@_ ;
+  my ($itemarg,@itemargs);
+  $string =~ s/(\%DIFDELCMD < \s*(\\$ITEMCMD$extraspace)((?:<$abrat0>)?$extraspace)((?:\[$brat_n\])?)\s*((?:${cmdoptseq}\s*?)*)(?:\n|$DELCMDCLOSE))/
+     # if \item has an []argument, then mark up the argument as deleted)
+     if (length($4)>0) {
+       # use substr to exclude square brackets at end points
+       @itemargs=splitlatex(substr($4,1,length($4)-2));
+       $itemarg="[".join("",marktags("","",$DELOPEN,$DELCLOSE,$DELCMDOPEN,$DELCMDCLOSE,$DELCOMMENT,\@itemargs))."]";
+     } else {
+       $itemarg="";
+     }
+     "$1$2$3$itemarg$AUXCMD\n";  ###.((length($5)>0) ? "%DIFDELCMD $5 $DELCMDCLOSE\n" : "")
+     /sge;
+  return($string);
+}
+
+
+# @auxlines=preprocess_preamble($oldpreamble,$newpreamble);
+  # pre-process preamble by looking for commands used in \maketitle (title, author, date etc commands)
+  # the list of commands is defined in CONTEXT2CMD
+  # if found then use a bodydiff to mark up content, and replace the corresponding commands
+  # in both preambles by marked up version to 'fool' the linediff (such that only body is marked up).
+  # A special case are e.g. author commands being added (or removed)
+  # 1. If commands are added, then the entire content is marked up as new, but also the lines are marked as new in the linediff
+  # 2. If commands are removed, then the linediff will mark the line as deleted.  The program returns
+  #    with $auxlines a text to be appended at the end of the preamble, which shows the respective fields as deleted
+sub preprocess_preamble {
+  my ($oldpreambleref,$newpreambleref)=(\$_[0],\$_[1]) ;
+  my @auxlines=();
+  # Remember to use $$oldpreambleref to refer to oldpreamble
+  my ($titlecmd,$titlecmdpat);
+  my (@oldtitlecommands,@newtitlecommands );
+  my  %oldhash  = ();
+  my  %newhash  = ();
+  my ($line,$cmd,$optarg,$arg,$optargnew,$optargold,$optargdiff,$argold,$argnew,$argdiff,$auxline);
+
+  my $warnmsgdetail = <<EOF ;
+     This should not occur for standard styles, but can occur for some specific styles, document classes,
+     e.g. journal house styles.
+     Workaround: Use --replace-context2cmd option to specifically set those commands, which are not repeated.
+EOF
+
+  # resuse context2cmdlist to define these commands to  look out for in preamble
+  $titlecmd = "(?:".join("|",@CONTEXT2CMDLIST).")";
+  # as context2cmdlist is stored as regex, e.g. ((?-xism:^title$), we need to remove ^- fo
+  # resue in a more complex regex
+  $titlecmd =~ s/[\$\^]//g;
+  # make sure to not match on comment lines:
+  $titlecmdpat=qr/^(?:[^%\n]|\\%)*(\\($titlecmd)$extraspace(?:\[($brat_n)\])?(?:\{($pat_n)\}))/ms;
+  ###print STDERR "DEBUG:",$titlecmdpat,"\n";
+  @oldtitlecommands= ( $$oldpreambleref =~ m/$titlecmdpat/g );
+  @newtitlecommands= ( $$newpreambleref =~ m/$titlecmdpat/g );
+
+
+  while ( @oldtitlecommands ) {
+    $line=shift @oldtitlecommands;
+    $cmd=shift @oldtitlecommands;
+    $optarg=shift @oldtitlecommands;
+    $arg=shift @oldtitlecommands;
+
+    if ( defined($oldhash{$cmd})) {
+      warn "WARNING: $cmd is used twice in preamble of old file. Reverting to pure line diff mode for preamble.\n";
+      print STDERR $warnmsgdetail;
+      return;
+    }
+    $oldhash{$cmd}=[ $line, $optarg, $arg ];
+  }
+  while ( @newtitlecommands ) {
+    $line=shift @newtitlecommands;
+    $cmd=shift @newtitlecommands;
+    $optarg=shift @newtitlecommands;
+    $arg=shift @newtitlecommands;
+
+    if ( defined($newhash{$cmd})) {
+      warn "$cmd is used twice in preamble of new file. Reverting to pure line diff mode for preamble.\n";
+      print STDERR $warnmsgdetail;
+      return;
+    }
+    $newhash{$cmd}=[ $line, $optarg, $arg ];
+  }
+  foreach $cmd ( keys %newhash ) {
+    if ( defined($newhash{$cmd}->[1])) {
+       $optargnew=$newhash{$cmd}->[1];
+    } else {
+      $optargnew="";
+    }
+    if ( defined($oldhash{$cmd}->[1])) {
+       $optargold=$oldhash{$cmd}->[1];
+    } else {
+      $optargold="";
+    }
+
+    if ( defined($oldhash{$cmd}->[2]) ) {
+      $argold=$oldhash{$cmd}->[2];
+    } else {
+      $argold="";
+    }
+    $argnew=$newhash{$cmd}->[2];
+    $argdiff="{" . join("",bodydiff($argold,$argnew)) ."}";
+    # Replace \RIGHTBRACE by }
+    $argdiff =~ s/\\RIGHTBRACE/}/g;
+
+    if ( length $optargnew ) {
+      $optargdiff="[".join("",bodydiff($optargold,$optargnew))."]" ;
+      $optargdiff =~ s/\\DIFaddbegin /\\DIFaddbeginFL /g;
+      $optargdiff =~ s/\\DIFaddend /\\DIFaddendFL /g;
+      $optargdiff =~ s/\\DIFadd\{/\\DIFaddFL{/g;
+      $optargdiff =~ s/\\DIFdelbegin /\\DIFdelbeginFL /g;
+      $optargdiff =~ s/\\DIFdelend /\\DIFdelendFL /g;
+      $optargdiff =~ s/\\DIFdel\{/\\DIFdelFL{/g;
+    } else {
+      $optargdiff="";
+    }
+    ### print STDERR "DEBUG s/\\Q$newhash{$cmd}->[0]\\E/\\$cmd$optargdiff$argdiff/s\n";
+    # Note: \Q and \E force literal interpretation of what it between them but allow
+    #      variable interpolation, such that e.g. \title matches just that and not TAB-itle
+    $$newpreambleref=~s/\Q$newhash{$cmd}->[0]\E/\\$cmd$optargdiff$argdiff/s;
+    # replace this in old preamble if necessary
+    if ( defined($oldhash{$cmd}->[0])) {
+      $$oldpreambleref=~s/\Q$oldhash{$cmd}->[0]\E/\\$cmd$optargdiff$argdiff/s ;
+    }
+    ### print STDERR "DEBUG NEW PRE ".$$newpreambleref."\n";
+  }
+
+  foreach $cmd ( keys %oldhash ) {
+    # if this has already been dealt with above can just skip
+    next if defined($newhash{$cmd}) ;
+    $argold=$oldhash{$cmd}->[2];
+    $argdiff="{" . join("",bodydiff($argold,"")) ."}";
+    if ( defined($oldhash{$cmd}->[1])) {
+      $optargold=$oldhash{$cmd}->[1];
+      $optargdiff="[".join("",bodydiff($optargold,""))."]" ;
+      $optargdiff =~ s/\\DIFdelbegin /\\DIFdelbeginFL /g;
+      $optargdiff =~ s/\\DIFdelend /\\DIFdelendFL /g;
+      $optargdiff =~ s/\\DIFdel\{/\\DIFdelFL{/g;
+    } else {
+      $optargdiff="";
+    }
+    $auxline = "\\$cmd$optargdiff$argdiff";
+    $auxline =~s/$/$AUXCMD/sg;
+    push @auxlines,$auxline;
+  }
+  # add auxcmd comment to highlight added lines
+  return(@auxlines);
+}
+
+
+
+# @diffs=linediff(\@seq1, \@seq2)
+# mark up lines like this
+#%DIF mm-mmdnn
+#%< old deleted line(s)
+#%DIF -------
+#%DIF mmann-nn
+#new appended line %<
+#%DIF -------
+# Future extension: mark change explicitly
+# Assumes: traverse_sequence traverses deletions before insertions in changed sequences
+#          all line numbers relative to line 0 (first line of real file)
+sub linediff {
+  my $seq1 = shift ;
+  my $seq2 = shift ;
+
+  my $block = [];
+  my $retseq = [];
+  my @begin=('','',''); # dummy initialisation
+  my $instring ;
+
+  my $discard = sub { @begin=('d',$_[0],$_[1]) unless scalar @$block ;
+                      push(@$block, "%DIF < " . $seq1->[$_[0]]) };
+  my $add = sub { if (! scalar  @$block) {
+		    @begin=('a',$_[0],$_[1]) ;}
+		  elsif ( $begin[0] eq 'd' ) {
+                    $begin[0]='c'; $begin[2]=$_[1];
+		    push(@$block, "%DIF -------") }
+                  push(@$block,  $seq2->[$_[1]] . " %DIF > " ) };
+  my $match = sub { if ( scalar @$block ) {
+                      if ( $begin[0] eq 'd' && $begin[1]!=$_[0]-1) {
+			$instring = sprintf "%%DIF %d-%dd%d",$begin[1],$_[0]-1,$begin[2]; }
+		      elsif ( $begin[0] eq 'a' && $begin[2]!=$_[1]-1) {
+			$instring = sprintf "%%DIF %da%d-%d",$begin[1],$begin[2],$_[1]-1; }
+		      elsif ( $begin[0] eq 'c' ) {
+			$instring = sprintf "%%DIF %sc%s",
+			                     ($begin[1]==$_[0]-1) ? "$begin[1]" : $begin[1]."-".($_[0]-1)  ,
+			                     ($begin[2]==$_[1]-1) ? "$begin[2]" : $begin[2]."-".($_[1]-1)  ; }
+		      else {
+			$instring = sprintf "%%DIF %d%s%d",$begin[1],$begin[0],$begin[2]; }
+		      push @$retseq, $instring,@$block, "%DIF -------" ;
+		      $block = [];
+		    }
+		    push @$retseq, $seq2->[$_[1]]
+		  };
+  # key function: remove multiple spaces (such that insertion or deletion of redundant white space is not reported)
+  my $keyfunc = sub { join("  ",split(" ",shift())) };
+
+  traverse_sequences($seq1,$seq2, { MATCH=>$match, DISCARD_A=>$discard, DISCARD_B=>$add }, $keyfunc );
+  push @$retseq, @$block if scalar @$block;
+
+  return wantarray ? @$retseq : $retseq ;
+}
+
+
+
+# init_regex_arr_data(\@array,"TOKEN INIT")
+# scans DATA file handel for line "%% TOKEN INIT" line
+# then appends each line not beginning with % into array (as a quoted regex)
+# This is used for command lists and configuration variables, but the processing is slightly
+# different:
+# For lists, the regular expression is extended to include beginning (^) and end ($) markers, to require full-string matching
+# For configuration variables (and all others), simply an unadorned list is copied
+sub init_regex_arr_data {
+  my ($arr,$token)=@_;
+  my $copy=0;
+  my ($mode);
+  if ($token =~ m/COMMANDS/ ) {
+    $mode=0;  # Reading command list
+  } else {
+    $mode=1;  # Reading configuration variables
+  }
+
+  while (<DATA>) {
+    if ( m/^%%BEGIN $token\s*$/ ) {
+      $copy=1;
+      next;
+    } elsif ( m/^%%END $token\s*$/ )  {
+      last; }
+    chomp;
+    if ( $mode==0 ) {
+#      print STDERR "DEBUG init_regex_arr_data regex >$_<\n" if ($debug && $copy);
+      push (@$arr,qr/^$_$/) if ( $copy && !/^%/ ) ;
+    } elsif ($mode==1) {
+      push (@$arr,"$_") if ( $copy && !/^%/ ) ;
+    }
+  }
+  seek DATA,0,0;    # rewind DATA handle to file begin
+}
+
+
+# init_regex_arr_ext(\@array,$arg)
+# appends array with regular expressions.
+# if arg is a file name, then read in list of regular expressions from that file
+# (one expression per line)
+# Otherwise treat arg as a comma separated list of regular expressions
+sub init_regex_arr_ext {
+  my ($arr,$arg)=@_;
+  if ( -f $arg ) {
+    init_regex_arr_file($arr,$arg);
+  } else {
+    init_regex_arr_list($arr,$arg);
+  }
+}
+
+# init_regex_arr_file(\@array,$fname)
+# appends array with regular expressions.
+# Read in list of regular expressions from $fname
+# (one expression per line)
+sub init_regex_arr_file {
+  my ($arr,$fname)=@_;
+  open(FILE,"$fname") or die ("Couldn't open $fname: $!");
+  while (<FILE>) {
+    chomp;
+    next if /^\s*#/ || /^\s*%/ || /^\s*$/ ;
+    push (@$arr,qr/^$_$/);
+  }
+  close(FILE);
+}
+
+# init_regex_arr_list(\@array,$arg)
+# appends array with regular expressions.
+# read from comma separated list of regular expressions ($arg)
+sub init_regex_arr_list {
+  my ($arr,$arg)=@_;
+  my $regex;
+  ###    print STDERR "DEBUG init_regex_arr_list arg >$arg<\n" if $debug;
+  foreach $regex (split(qr/(?<!\\),/,$arg)) {
+    $regex =~ s/\\,/,/g;
+    print STDERR "DEBUG init_regex_arr_list regex >$regex<\n" if $debug;
+    push (@$arr,qr/^$regex$/);
+  }
+}
+
+
+#exetime() returns time since last execution of this command
+#exetime(1) resets this time
+my $lasttime=-1;   # global variable for persistence
+sub exetime {
+  my $reset=0;
+  my $retval;
+  if ((scalar @_) >=1) {
+    $reset=shift;
+  }
+  if ($reset) {
+    $lasttime=times();
+  }
+  else {
+    $retval=times()-$lasttime;
+    $lasttime=$lasttime+$retval;
+    return($retval);
+  }
+}
+
+
+sub usage {
+  print STDERR <<"EOF";
+Usage: $0 [options] old.tex new.tex > diff.tex
+
+Compares two latex files and writes tex code to stdout, which has the same format as new.tex but
+has all changes relative to old.tex marked up or commented. Note that old.tex and new.tex need to
+be real files (not pipes or similar) as they are opened twice.
+
+--type=markupstyle
+-t markupstyle         Add code to preamble for selected markup style
+                       Available styles: UNDERLINE CTRADITIONAL TRADITIONAL CFONT FONTSTRIKE INVISIBLE
+                                         CHANGEBAR CCHANGEBAR CULINECHBAR CFONTCHBAR BOLD PDFCOMMENT
+                                         LUAUNDERLINE
+                       [ Default: UNDERLINE ]
+
+--subtype=markstyle
+-s markstyle           Add code to preamble for selected style for bracketing
+                       commands (e.g. to mark changes in  margin)
+                       Available styles: SAFE MARGIN DVIPSCOL COLOR ZLABEL ONLYCHANGEDPAGE (LABEL)*
+                       [ Default: SAFE ]
+                       * LABEL subtype is deprecated
+
+--floattype=markstyle
+-f markstyle           Add code to preamble for selected style which
+                       replace standard marking and markup commands within floats
+                       (e.g., marginal remarks cause an error within floats
+                       so marginal marking can be disabled thus)
+                       Available styles: FLOATSAFE IDENTICAL
+                       [ Default: FLOATSAFE ]
+
+--encoding=enc
+-e enc                 Specify encoding of old.tex and new.tex. Typical encodings are
+                       ascii, utf8, latin1, latin9.  A list of available encodings can be
+                       obtained by executing
+                       perl -MEncode -e 'print join ("\\n",Encode->encodings( ":all" )) ;'
+                       [Default encoding is utf8 unless the first few lines of the preamble contain
+                       an invocation "\\usepackage[..]{inputenc} in which case the
+                       encoding chosen by this command is asssumed. Note that ASCII (standard
+                       latex) is a subset of utf8]
+
+--preamble=file
+-p file                Insert file at end of preamble instead of auto-generating
+                       preamble.  The preamble must define the following commands
+                       \\DIFaddbegin,\\DIFaddend,\\DIFadd{..},
+                       \\DIFdelbegin,\\DIFdelend,\\DIFdel{..},
+                       and varieties for use within floats
+                       \\DIFaddbeginFL,\\DIFaddendFL,\\DIFaddFL{..},
+                       \\DIFdelbeginFL,\\DIFdelendFL,\\DIFdelFL{..}
+                       (If this option is set -t, -s, and -f options
+                       are ignored.)
+
+--exclude-safecmd=exclude-file
+--exclude-safecmd="cmd1,cmd2,..."
+-A exclude-file
+--replace-safecmd=replace-file
+--append-safecmd=append-file
+--append-safecmd="cmd1,cmd2,..."
+-a append-file         Exclude from, replace or append to the list of regex
+                       matching commands which are safe to use within the
+                       scope of a \\DIFadd or \\DIFdel command.  The file must contain
+                       one Perl-RegEx per line (Comment lines beginning with # or % are
+                       ignored). A literal comma within the comma-separated list must be
+                       escaped thus "\\,",   Note that the RegEx needs to match the whole of
+                       the token, i.e., /^regex\$/ is implied and that the initial
+                       "\\" of the command is not included. The --exclude-safecmd
+                       and --append-safecmd options can be combined with the --replace-safecmd
+                       option and can be used repeatedly to add cumulatively to the lists.
+
+--exclude-textcmd=exclude-file
+--exclude-textcmd="cmd1,cmd2,..."
+-X exclude-file
+--replace-textcmd=replace-file
+--append-textcmd=append-file
+--append-textcmd="cmd1,cmd2,..."
+-x append-file         Exclude from, replace or append to the list of regex
+                       matching commands whose last argument is text.  See
+                       entry for --exclude-safecmd directly above for further details.
+
+--replace-context1cmd=replace-file
+--append-context1cmd=append-file
+--append-context1cmd="cmd1,cmd2,..."
+                       Replace or append to the list of regex matching commands
+                       whose last argument is text but which require a particular
+                       context to work, e.g. \\caption will only work within a figure
+                       or table.  These commands behave like text commands, except when
+                       they occur in a deleted section, when they are disabled, but their
+                       argument is shown as deleted text.
+
+--replace-context2cmd=replace-file
+--append-context2cmd=append-file
+--append-context2cmd="cmd1,cmd2,..."
+                       As corresponding commands for context1.  The only difference is that
+                       context2 commands are completely disabled in deleted sections, including
+                       their arguments.
+                       context2 commands are also the only commands in the preamble, whose argument will
+                       be processed in word-by-word mode (which only works, if they occur no more than
+		       once in the preamble).
+
+--exclude-mboxsafecmd=exclude-file
+--exclude-mboxsafecmd="cmd1,cmd2,..."
+--append-mboxsafecmd=append-file
+--append-mboxsafecmd="cmd1,cmd2,..."
+                       Define safe commands, which additionally need to be protected by encapsulating
+                       in an \\mbox{..}. This is sometimes needed to get around incompatibilities
+                       between external packages and the ulem package, which is  used for highlighting
+                       in the default style UNDERLINE as well as CULINECHBAR CFONTSTRIKE
+
+
+
+--config var1=val1,var2=val2,...
+-c var1=val1,..        Set configuration variables.
+-c configfile           Available variables:
+                          ARRENV (RegEx)
+                          COUNTERCMD (RegEx)
+                          FLOATENV (RegEx)
+                          ITEMCMD (RegEx)
+                          LISTENV (RegEx)
+                          MATHARRENV (RegEx)
+                          MATHENV (RegEx)
+                          MATHREPL (String)
+                          MINWORDSBLOCK (Integer)
+                          PICTUREENV (RegEx)
+                          SCALEDELGRAPHICS (Float)
+                          VERBATIMENV (RegEx)
+                          VERBATIMLINEENV (RegEx)
+                          CUSTOMDIFCMD (RegEx)
+                       This option can be repeated.
+
+--add-to-config  varenv1=pattern1,varenv2=pattern2
+                       For configuration variables containing a regular expression (essentially those ending
+                       in ENV, and COUNTERCMD) this provides an alternative way to modify the configuration
+                       variables. Instead of setting the complete pattern, with this option it is possible to add an
+                       alternative pattern. varenv must be one of the variables listed above that take a regular
+                       expression as argument, and pattern is any regular expression (which might need to be
+                       protected from the shell by quotation). Several patterns can be added at once by using semi-colons
+                       to separate them, e.g. --add-to-config "LISTENV=myitemize;myenumerate,COUNTERCMD=endnote"
+
+--packages=pkg1,pkg2,..
+                       Tell latexdiff that .tex file is processed with the packages in list
+                       loaded.  This is normally not necessary if the .tex file includes the
+                       preamble, as the preamble is automatically scanned for \\usepackage commands.
+                       Use of the --packages option disables automatic scanning, so if for any
+                       reason package specific parsing needs to be switched off, use --packages=none.
+                       The following packages trigger special behaviour:
+                       endfloat hyperref amsmath apacite siunitx cleveref glossaries mhchem chemformula/chemmacros
+                       biblatex
+                       [ Default: scan the preamble for \\usepackage commands to determine
+                         loaded packages.]
+
+--show-preamble        Print generated or included preamble commands to stdout.
+
+--show-safecmd         Print list of regex matching and excluding safe commands.
+
+--show-textcmd         Print list of regex matching and excluding commands with text argument.
+
+--show-config          Show values of configuration variables
+
+--show-all             Show all of the above
+
+   NB For all --show commands, no old.tex or new.tex file needs to be given, and no
+      differencing takes place.
+
+Other configuration options:
+
+--allow-spaces         Allow spaces between bracketed or braced arguments to commands
+                       [Default requires arguments to directly follow each other without
+                                intervening spaces]
+
+--math-markup=level    Determine granularity of markup in displayed math environments:
+                      Possible values for level are (both numerical and text labels are acceptable):
+                      off or 0: suppress markup for math environments.  Deleted equations will not
+                               appear in diff file. This mode can be used if all the other modes
+                               cause invalid latex code.
+                      whole or 1: Differencing on the level of whole equations. Even trivial changes
+                               to equations cause the whole equation to be marked changed.  This
+                               mode can be used if processing in coarse or fine mode results in
+                               invalid latex code.
+                      coarse or 2: Detect changes within equations marked up with a coarse
+                               granularity; changes in equation type (e.g.displaymath to equation)
+                               appear as a change to the complete equation. This mode is recommended
+                               for situations where the content and order of some equations are still
+                               being changed. [Default]
+                      fine or 3: Detect small change in equations and mark up and fine granularity.
+                               This mode is most suitable, if only minor changes to equations are
+                               expected, e.g. correction of typos.
+
+--graphics-markup=level   Change highlight style for graphics embedded with \\includegraphics commands
+                      Possible values for level:
+                      none,off or 0: no highlighting for figures
+                      new-only or 1: surround newly added or changed figures with a blue frame [Default]
+                      both or 2:     highlight new figures with a blue frame and show deleted figures
+                                at reduced scale, and crossed out with a red diagonal cross. Use configuration
+                                variable SCALEDELGRAPHICS to set size of deleted figures.
+                      Note that changes to the optional parameters will make the figure appear as changed
+                      to latexdiff, and this figure will thus be highlighted.
+                      In some circumstances "Misplaced \\noalign" errors can occur if there are certain types
+                      of changes in tables. In this case please use option --graphics-markup=none as a
+                      work-around.
+
+--no-del               Suppress deleted text from the diff. It is similar in effect to the BOLD style,
+                       but the deleted text ist not just invisible in the output, it is also not included in the
+                       diff text file. This can be more robust than just making it invisible.
+
+--disable-citation-markup
+--disable-auto-mbox    Suppress citation markup and markup of other vulnerable commands in styles
+                       using ulem (UNDERLINE,FONTSTRIKE, CULINECHBAR)
+                       (the two options are identical and are simply aliases)
+
+--enable-citation-markup
+--enforce-auto-mbox    Protect citation commands and other vulnerable commands in changed sections
+                       with \\mbox command, i.e. use default behaviour for ulem package for other packages
+                       (the two options are identical and are simply aliases)
+
+
+
+Miscellaneous options
+
+--label=label
+-L label               Sets the labels used to describe the old and new files.  The first use
+                       of this option sets the label describing the old file and the second
+                       use of the option sets the label for the new file.
+                       [Default: use the filename and modification dates for the label]
+
+--no-label             Suppress inclusion of old and new file names as comment in output file
+
+--visible-label         Include old and new filenames (or labels set with --label option) as
+                       visible output
+
+--flatten              Replace \\input and \\include commands within body by the content
+                       of the files in their argument.  If \\includeonly is present in the
+                       preamble, only those files are expanded into the document. However,
+                       no recursion is done, i.e. \\input and \\include commands within
+                       included sections are not expanded.  The included files are assumed to
+                       be located in the same directories as the old and new master files,
+                       respectively, making it possible to organise files into old and new directories.
+                       --flatten is applied recursively, so inputted files can contain further
+                       \\input statements.  Also handles files included by the import package
+                       (\\import and \\subimport), and \\subfile command.
+
+--filter-script=filterscript    Run files through this filterscript (full path preferred) before processing.
+                       The filterscript must take STDIN input and output to STDOUT.
+                       When coupled with --flatten, each file will be run through the filter as it is brought in.
+
+--ignore-filter-stderr When running with --filter-script, STDERR from the script may cause readability issues.
+                       Turn this flag on to ignore STDERR from the filter script.
+
+--driver=type          Choose driver for changebar package (only relevant for styles using
+                       changebar: CCHANGEBAR CFONTCHBAR CULINECHBAR CHANGEBAR). Possible
+                       drivers are listed in changebar manual, e.g. pdftex,dvips,dvitops
+                       [Default: pdftex]
+
+--help
+-h                     Show this help text.
+
+--ignore-warnings      Suppress warnings about inconsistencies in length between input
+                       and parsed strings and missing characters.
+
+--verbose
+-V                     Output various status information to stderr during processing.
+                       Default is to work silently.
+
+--version              Show version number.
+
+Internal options:
+These options are mostly for automated use by latexdiff-vc. They can be used directly, but
+the API should be considered less stable than for the other options.
+
+--no-links             Suppress generation of hyperreferences, used for minimal diffs
+                       (option --only-changes of latexdiff-vc).
+EOF
+  exit 0;
+}
+
+=head1 NAME
+
+latexdiff - determine and markup differences between two latex files
+
+=head1 SYNOPSIS
+
+B<latexdiff> [ B<OPTIONS> ] F<old.tex> F<new.tex> > F<diff.tex>
+
+=head1 DESCRIPTION
+
+Briefly, I<latexdiff> is a utility program to aid in the management of
+revisions of latex documents. It compares two valid latex files, here
+called C<old.tex> and C<new.tex>, finds significant differences
+between them (i.e., ignoring the number of white spaces and position
+of line breaks), and adds special commands to highlight the
+differences.  Where visual highlighting is not possible, e.g. for changes
+in the formatting, the differences are
+nevertheless marked up in the source. Note that old.tex and new.tex need to
+be real files (not pipes or similar) as they are opened twice (unless C<--encoding> option is used)
+
+The program treats the preamble differently from the main document.
+Differences between the preambles are found using line-based
+differencing (similarly to the Unix diff command, but ignoring white
+spaces).  A comment, "S<C<%DIF E<gt>>>" is appended to each added line, i.e. a
+line present in C<new.tex> but not in C<old.tex>.  Discarded lines
+ are deactivated by prepending "S<C<%DIF E<lt>>>". Changed blocks are preceded  by
+comment lines giving information about line numbers in the original files.  Where there are insignificant
+differences, the resulting file C<diff.tex> will be similar to
+C<new.tex>.  At the end of the preamble, the definitions for I<latexdiff> markup commands are inserted.
+In differencing the main body of the text, I<latexdiff> attempts to
+satisfy the following guidelines (in order of priority):
+
+=over 3
+
+=item 1
+
+If both C<old.tex> and C<new.tex> are valid LaTeX, then the resulting
+C<diff.tex> should also be valid LateX. (NB If a few plain TeX commands
+are used within C<old.tex> or C<new.tex> then C<diff.tex> is not
+guaranteed to work but usually will).
+
+=item 2
+
+Significant differences are determined on the level of
+individual words. All significant differences, including differences
+between comments should be clearly marked in the resulting source code
+C<diff.tex>.
+
+=item 3
+
+If a changed passage contains text or text-producing commands, then
+running C<diff.tex> through LateX should produce output where added
+and discarded passages are highlighted.
+
+=item 4
+
+Where there are insignificant differences, e.g. in the positioning of
+line breaks, C<diff.tex> should follow the formatting of C<new.tex>
+
+=back
+
+For differencing the same algorithm as I<diff> is used but words
+instead of lines are compared.  An attempt is made to recognize
+blocks which are completely changed such that they can be marked up as a unit.
+Comments are differenced line by line
+but the number of spaces within comments is ignored. Commands including
+all their arguments are generally compared as one unit, i.e., no mark-up
+is inserted into the arguments of commands.  However, for a selected
+number of commands (for example, C<\caption> and all sectioning
+commands) the last argument is known to be text. This text is
+split into words and differenced just as ordinary text (use options to
+show and change the list of text commands, see below). As the
+algorithm has no detailed knowledge of LaTeX, it assumes all pairs of
+curly braces immediately following a command (i.e. a sequence of
+letters beginning with a backslash) are arguments for that command.
+As a restriction to condition 1 above it is thus necessary to surround
+all arguments with curly braces, and to not insert
+extraneous spaces.  For example, write
+
+  \section{\textem{This is an emphasized section title}}
+
+and not
+
+  \section {\textem{This is an emphasized section title}}
+
+or
+
+  \section\textem{This is an emphasized section title}
+
+even though all varieties are the same to LaTeX (but see
+B<--allow-spaces> option which allows the second variety).
+
+For environments whose content does not conform to standard LaTeX or
+where graphical markup does not make sense all markup commands can be
+removed by setting the PICTUREENV configuration variable, set by
+default to C<picture> and C<DIFnomarkup> environments; see B<--config>
+option).  The latter environment (C<DIFnomarkup>) can be used to
+protect parts of the latex file where the markup results in illegal
+markup. You have to surround the offending passage in both the old and
+new file by C<\begin{DIFnomarkup}> and C<\end{DIFnomarkup}>. You must
+define the environment in the preambles of both old and new
+documents. I prefer to define it as a null-environment,
+
+C<\newenvironment{DIFnomarkup}{}{}>
+
+but the choice is yours.  Any markup within the environment will be
+removed, and generally everything within the environment will just be
+taken from the new file.
+
+It is also possible to difference files which do not have a preamble.
+ In this case, the file is processed in the main document
+mode, but the definitions of the markup commands are not inserted.
+
+All markup commands inserted by I<latexdiff> begin with "C<\DIF>".  Added
+blocks containing words, commands or comments which are in C<new.tex>
+but not in C<old.tex> are marked by C<\DIFaddbegin> and C<\DIFaddend>.
+Discarded blocks are marked by C<\DIFdelbegin> and C<\DIFdelend>.
+Within added blocks all text is highlighted with C<\DIFadd> like this:
+C<\DIFadd{Added text block}>
+Selected `safe' commands can be contained in these text blocks as well
+(use options to show and change the list of safe commands, see below).
+All other commands as well as braces "{" and "}" are never put within
+the scope of C<\DIFadd>.  Added comments are marked by prepending
+"S<C<%DIF E<gt> >>".
+
+Within deleted blocks text is highlighted with C<\DIFdel>.  Deleted
+comments are marked by prepending "S<C<%DIF E<lt> >>".  Non-safe command
+and curly braces within deleted blocks are commented out with
+"S<C<%DIFDELCMD E<lt> >>".
+
+
+
+=head1 OPTIONS
+
+=head2 Preamble
+
+The following options determine the visual markup style by adding the appropriate
+command definitions to the preamble. See the end of this section for a description of
+available styles.
+
+=over 4
+
+=item B<--type=markupstyle> or
+B<-t markupstyle>
+
+Add code to preamble for selected markup style. This option defines
+C<\DIFadd> and C<\DIFdel> commands.
+Available styles:
+
+C<UNDERLINE CTRADITIONAL TRADITIONAL CFONT FONTSTRIKE INVISIBLE
+CHANGEBAR CCHANGEBAR CULINECHBAR CFONTCHBAR BOLD PDFCOMMENT
+LUAUNDERLINE>
+
+[ Default: C<UNDERLINE> ]
+
+=item B<--subtype=markstyle> or
+B<-s markstyle>
+
+Add code to preamble for selected style for bracketing
+commands (e.g. to mark changes in  margin). This option defines
+C<\DIFaddbegin>, C<\DIFaddend>, C<\DIFdelbegin> and C<\DIFdelend> commands.
+Available styles: C<SAFE MARGIN COLOR DVIPSCOL  ZLABEL ONLYCHANGEDPAGE (LABEL)*>
+
+[ Default: C<SAFE> ]
+* Subtype C<LABEL> is deprecated
+
+=item B<--floattype=markstyle> or
+B<-f markstyle>
+
+Add code to preamble for selected style which
+replace standard marking and markup commands within floats
+(e.g., marginal remarks cause an error within floats
+so marginal marking can be disabled thus). This option defines all
+C<\DIF...FL> commands.
+Available styles: C<FLOATSAFE TRADITIONALSAFE IDENTICAL>
+
+[ Default: C<FLOATSAFE> ]
+
+=item B<--encoding=enc> or
+B<-e enc>
+
+Specify encoding of old.tex and new.tex. Typical encodings are
+C<ascii>, C<utf8>, C<latin1>, C<latin9>.  A list of available encodings can be
+obtained by executing
+
+C<perl -MEncode -e 'print join ("\n",Encode->encodings( ":all" )) ;' >
+
+If this option is used, then old.tex, new.tex are only opened once.
+[Default encoding is utf8 unless the first few lines of the preamble contain
+an invocation C<\usepackage[..]{inputenc}> in which case the
+encoding chosen by this command is asssumed. Note that ASCII (standard
+latex) is a subset of utf8]
+
+=item B<--preamble=file> or
+B<-p file>
+
+Insert file at end of preamble instead of generating
+preamble.  The preamble must define the following commands
+C<\DIFaddbegin, \DIFaddend, \DIFadd{..},
+\DIFdelbegin,\DIFdelend,\DIFdel{..},>
+and varieties for use within floats
+C<\DIFaddbeginFL, \DIFaddendFL, \DIFaddFL{..},
+\DIFdelbeginFL, \DIFdelendFL, \DIFdelFL{..}>
+(If this option is set B<-t>, B<-s>, and B<-f> options
+are ignored.)
+
+=item B<--packages=pkg1,pkg2,..>
+
+Tell latexdiff that .tex file is processed with the packages in list
+loaded.  This is normally not necessary if the .tex file includes the
+preamble, as the preamble is automatically scanned for C<\usepackage> commands.
+Use of the B<--packages> option disables automatic scanning, so if for any
+reason package specific parsing needs to be switched off, use B<--packages=none>.
+The following packages trigger special behaviour:
+
+=over 8
+
+=item C<endfloat>
+
+Ensure that C<\begin{figure}> and C<\end{figure}> always appear by themselves on a line.
+
+=item C<hyperref>
+
+Change name of C<\DIFadd> and C<\DIFdel> commands to C<\DIFaddtex> and C<\DIFdeltex> and
+define new C<\DIFadd> and C<\DIFdel> commands, which provide a wrapper for these commands,
+using them for the text but not for the link defining command (where any markup would cause
+errors).
+
+=item C<apacite>, C<biblatex>
+
+Redefine the commands recognised as citation commands.
+
+=item C<siunitx>
+
+Treat C<\SI> as equivalent to citation commands (i.e. protect with C<\mbox> if markup style uses ulem package.
+
+=item C<cleveref>
+
+Treat C<\cref,\Cref>, etc as equivalent to citation commands (i.e. protect with C<\mbox> if markup style uses ulem package.
+
+=item C<glossaries>
+
+Define most of the glossaries commands as safe, protecting them with \mbox'es where needed
+
+=item C<mhchem>
+
+Treat C<\ce> as a safe command, i.e. it will be highlighted (note that C<\cee> will not be highlighted in equations as this leads to processing errors)
+
+=item C<chemformula> or C<chemmacros>
+
+Treat C<\ch> as a safe command outside equations, i.e. it will be highlighted (note that C<\ch> will not be highlighted in equations as this leads to processing errors)
+
+
+=back
+
+[ Default: scan the preamble for C<\usepackage> commands to determine
+  loaded packages. ]
+
+
+
+=item B<--show-preamble>
+
+Print generated or included preamble commands to stdout.
+
+=back
+
+=head2 Configuration
+
+=over 4
+
+=item B<--exclude-safecmd=exclude-file> or
+B<-A exclude-file> or  B<--exclude-safecmd="cmd1,cmd2,...">
+
+=item B<--replace-safecmd=replace-file>
+
+=item B<--append-safecmd=append-file> or
+B<-a append-file> or B<--append-safecmd="cmd1,cmd2,...">
+
+Exclude from, replace or append to the list of regular expressions (RegEx)
+matching commands which are safe to use within the
+scope of a C<\DIFadd> or C<\DIFdel> command.  The file must contain
+one Perl-RegEx per line (Comment lines beginning with # or % are
+ignored).  Note that the RegEx needs to match the whole of
+the token, i.e., /^regex$/ is implied and that the initial
+"\" of the command is not included.
+The B<--exclude-safecmd> and B<--append-safecmd> options can be combined with the -B<--replace-safecmd>
+option and can be used repeatedly to add cumulatively to the lists.
+ B<--exclude-safecmd>
+and B<--append-safecmd> can also take a comma separated list as input. If a
+comma for one of the regex is required, escape it thus "\,". In most cases it
+will be necessary to protect the comma-separated list from the shell by putting
+it in quotation marks.
+
+=item B<--exclude-textcmd=exclude-file> or
+B<-X exclude-file> or B<--exclude-textcmd="cmd1,cmd2,...">
+
+=item B<--replace-textcmd=replace-file>
+
+=item B<--append-textcmd=append-file> or
+B<-x append-file> or B<--append-textcmd="cmd1,cmd2,...">
+
+Exclude from, replace or append to the list of regular expressions
+matching commands whose last argument is text.  See
+entry for B<--exclude-safecmd> directly above for further details.
+
+
+=item B<--replace-context1cmd=replace-file>
+
+=item B<--append-context1cmd=append-file> or
+
+=item B<--append-context1cmd="cmd1,cmd2,...">
+
+Replace or append to the list of regex matching commands
+whose last argument is text but which require a particular
+context to work, e.g. C<\caption> will only work within a figure
+or table.  These commands behave like text commands, except when
+they occur in a deleted section, when they are disabled, but their
+argument is shown as deleted text.
+
+=item B<--replace-context2cmd=replace-file>
+
+=item B<--append-context2cmd=append-file> or
+
+=item B<--append-context2cmd="cmd1,cmd2,...">
+
+As corresponding commands for context1.  The only difference is that
+context2 commands are completely disabled in deleted sections, including
+their arguments.
+
+context2 commands are also the only commands in the preamble, whose argument will be processed in
+word-by-word mode (which only works, if they occur no more than once in the preamble). The algorithm currently cannot cope with repeated context2 commands in the preamble, as they occur e.g. for the C<\author> argument in some journal styles (not in the standard styles, though
+If such a repetition is detected, the whole preamble will be processed in line-by-line mode. In such a case, use C<--replace-context2cmd> option to just select the commands, which should be processed and are not used repeatedly in the preamble.
+
+
+
+=item B<--exclude-mboxsafecmd=exclude-file> or B<--exclude-mboxsafecmd="cmd1,cmd2,...">
+
+=item B<--append-mboxsafecmd=append-file> or B<--append-mboxsafecmd="cmd1,cmd2,...">
+
+Define safe commands, which additionally need to be protected by encapsulating
+in an C<\mbox{..}>. This is sometimes needed to get around incompatibilities
+between external packages and the ulem package, which is  used for highlighting
+in the default style UNDERLINE as well as CULINECHBAR CFONTSTRIKE
+
+
+
+
+
+=item B<--config var1=val1,var2=val2,...> or B<-c var1=val1,..>
+
+=item B<-c configfile>
+
+Set configuration variables.  The option can be repeated to set different
+variables (as an alternative to the comma-separated list).
+Available variables (see below for further explanations):
+
+C<ARRENV> (RegEx)
+
+C<COUNTERCMD> (RegEx)
+
+C<CUSTODIFCMD> (RegEx)
+
+C<FLOATENV> (RegEx)
+
+C<ITEMCMD> (RegEx)
+
+C<LISTENV>  (RegEx)
+
+C<MATHARRENV> (RegEx)
+
+C<MATHENV> (RegEx)
+
+C<MATHREPL> (String)
+
+C<MINWORDSBLOCK> (Integer)
+
+C<PICTUREENV> (RegEx)
+
+C<SCALEDELGRAPHICS> (Float)
+
+
+=item B<--add-to-config varenv1=pattern1,varenv2=pattern2,...>
+
+For configuration variables, which are a regular expression (essentially those ending
+in ENV, COUNTERCMD and CUSTOMDIFCMD, see list above) this option provides an alternative way to modify the configuration
+variables. Instead of setting the complete pattern, with this option it is possible to add an
+alternative pattern. C<varenv> must be one of the variables listed above that take a regular
+expression as argument, and pattern is any regular expression (which might need to be
+protected from the shell by quotation). Several patterns can be added at once by using semi-colons
+to separate them, e.g. C<--add-to-config "LISTENV=myitemize;myenumerate,COUNTERCMD=endnote">
+
+=item B<--show-safecmd>
+
+Print list of RegEx matching and excluding safe commands.
+
+=item B<--show-textcmd>
+
+Print list of RegEx matching and excluding commands with text argument.
+
+=item B<--show-config>
+
+Show values of configuration variables.
+
+=item B<--show-all>
+
+Combine all --show commands.
+
+NB For all --show commands, no C<old.tex> or C<new.tex> file needs to be specified, and no
+differencing takes place.
+
+=back
+
+=head2 Other configuration options:
+
+=over 4
+
+=item B<--allow-spaces>
+
+Allow spaces between bracketed or braced arguments to commands.  Note
+that this option might have undesirable side effects (unrelated scope
+might get lumpeded with preceding commands) so should only be used if the
+default produces erroneous results.  (Default requires arguments to
+directly follow each other without intervening spaces).
+
+=item B<--math-markup=level>
+
+Determine granularity of markup in displayed math environments:
+Possible values for level are (both numerical and text labels are acceptable):
+
+C<off> or C<0>: suppress markup for math environments.  Deleted equations will not
+appear in diff file. This mode can be used if all the other modes
+cause invalid latex code.
+
+C<whole> or C<1>: Differencing on the level of whole equations. Even trivial changes
+to equations cause the whole equation to be marked changed.  This
+mode can be used if processing in coarse or fine mode results in
+invalid latex code.
+
+C<coarse> or C<2>: Detect changes within equations marked up with a coarse
+granularity; changes in equation type (e.g.displaymath to equation)
+appear as a change to the complete equation. This mode is recommended
+for situations where the content and order of some equations are still
+being changed. [Default]
+
+C<fine> or C<3>: Detect small change in equations and mark up at fine granularity.
+This mode is most suitable, if only minor changes to equations are
+expected, e.g. correction of typos.
+
+=item B<--graphics-markup=level>
+
+ Change highlight style for graphics embedded with C<\includegraphics> commands.
+
+Possible values for level:
+
+C<none>, C<off> or C<0>: no highlighting for figures
+
+C<new-only> or C<1>: surround newly added or changed figures with a blue frame [Default if graphicx package loaded]
+
+C<both> or C<2>:     highlight new figures with a blue frame and show deleted figures at reduced
+scale, and crossed out with a red diagonal cross. Use configuration
+variable SCALEDELGRAPHICS to set size of deleted figures.
+
+Note that changes to the optional parameters will make the figure appear as changed
+to latexdiff, and this figure will thus be highlighted.
+
+In some circumstances "Misplaced \noalign" error can occur if there are certain types
+of changes in tables. In this case please use C<--graphics-markup=none> as a
+work-around.
+
+
+=item B<--no-del>
+
+Suppress deleted text from the diff. It is similar in effect to the BOLD style,
+but the deleted text ist not just invisible in the output, it is also not included in the diff text file.
+This can be more robust than just making it invisible.
+
+=item B<--disable-citation-markup> or B<--disable-auto-mbox>
+
+Suppress citation markup and markup of other vulnerable commands in styles
+using ulem (UNDERLINE,FONTSTRIKE, CULINECHBAR)
+(the two options are identical and are simply aliases)
+
+=item B<--enable-citation-markup> or B<--enforce-auto-mbox>
+
+Protect citation commands and other vulnerable commands in changed sections
+with C<\mbox> command, i.e. use default behaviour for ulem package for other packages
+(the two options are identical and are simply aliases)
+
+=back
+
+=head2 Miscellaneous
+
+=over 4
+
+=item B<--verbose> or B<-V>
+
+Output various status information to stderr during processing.
+Default is to work silently.
+
+=item B<--driver=type>
+
+Choose driver for changebar package (only relevant for styles using
+   changebar: CCHANGEBAR CFONTCHBAR CULINECHBAR CHANGEBAR). Possible
+drivers are listed in changebar manual, e.g. pdftex,dvips,dvitops
+  [Default: pdftex]
+
+=item B<--ignore-warnings>
+
+Suppress warnings about inconsistencies in length between input and
+parsed strings and missing characters.  These warning messages are
+often related to non-standard latex or latex constructions with a
+syntax unknown to C<latexdiff> but the resulting difference argument
+is often fully functional anyway, particularly if the non-standard
+latex only occurs in parts of the text which have not changed.
+
+=item B<--label=label> or
+B<-L label>
+
+Sets the labels used to describe the old and new files.  The first use
+of this option sets the label describing the old file and the second
+use of the option sets the label for the new file, i.e. set both
+labels like this C<-L labelold -L labelnew>.
+[Default: use the filename and modification dates for the label]
+
+=item B<--no-label>
+
+Suppress inclusion of old and new file names as comment in output file
+
+=item B<--visible-label>
+
+Include old and new filenames (or labels set with C<--label> option) as
+visible output.
+
+=item B<--flatten>
+
+Replace C<\input> and C<\include> commands within body by the content
+of the files in their argument.  If C<\includeonly> is present in the
+preamble, only those files are expanded into the document. However,
+no recursion is done, i.e. C<\input> and C<\include> commands within
+included sections are not expanded.  The included files are assumed to
+ be located in the same directories as the old and new master files,
+respectively, making it possible to organise files into old and new directories.
+--flatten is applied recursively, so inputted files can contain further
+C<\input> statements.  Also handles files included by the import package
+(C<\import> and C<\subimport>), and C<\subfile> command.
+
+Use of this option might result in prohibitive processing times for
+larger documents, and the resulting difference document
+no longer reflects the structure of the input documents.
+
+=item B<--filter-script=filterscript>
+
+Run files through this filterscript (full path preferred) before processing.
+The filterscript must take STDIN input and output to STDOUT.
+When coupled with --flatten, each file will be run through the filter as it is brought in.
+
+=item B<--ignore-filter-stderr>
+
+When running with --filter-script, STDERR from the script may cause readability issues.
+Turn this flag on to ignore STDERR from the filter script.
+
+
+
+=item B<--help> or
+B<-h>
+
+Show help text
+
+=item B<--version>
+
+Show version number
+
+=back
+
+
+=head2 Internal options
+
+These options are mostly for automated use by latexdiff-vc. They can be used directly, but the API should be considered less stable than for the other options.
+
+=over 4
+
+=item B<--no-links>
+
+Suppress generation of hyperreferences, used for minimal diffs (option --only-changes of latexdiff-vc)
+
+=back
+
+
+=head2 Predefined styles
+
+=head2 Major types
+
+The major type determine the markup of plain text and some selected latex commands outside floats by defining the markup commands C<\DIFadd{...}> and C<\DIFdel{...}> .
+
+=over 10
+
+=item C<UNDERLINE>
+
+Added text is wavy-underlined and blue, discarded text is struck out and red
+(Requires color and ulem packages).  Overstriking does not work in displayed math equations such that deleted parts of equation are underlined, not struck out (this is a shortcoming inherent to the ulem package).
+
+=item C<LUAUNDERLINE>
+
+Added text is underlined and blue, discarded text is struck out and red
+(Requires lua-ul package + LuaLaTeX).
+
+=item C<CTRADITIONAL>
+
+Added text is blue and set in sans-serif, and a red footnote is created for each discarded
+piece of text. (Requires color package)
+
+=item C<TRADITIONAL>
+
+Like C<CTRADITIONAL> but without the use of color.
+
+=item C<CFONT>
+
+Added text is blue and set in sans-serif, and discarded text is red and very small size.
+
+=item C<FONTSTRIKE>
+
+Added tex is set in sans-serif, discarded text small and struck out
+
+=item C<CCHANGEBAR>
+
+Added text is blue, and discarded text is red.  Additionally, the changed text is marked with a bar in the margin (Requires color and changebar packages).
+
+=item C<CFONTCHBAR>
+
+Like C<CFONT> but with additional changebars (Requires color and changebar packages).
+
+=item C<CULINECHBAR>
+
+Like C<UNDERLINE> but with additional changebars (Requires color, ulem and changebar packages).
+
+=item C<CHANGEBAR>
+
+No mark up of text, but mark margins with changebars (Requires changebar package).
+
+=item C<INVISIBLE>
+
+No visible markup (but generic markup commands will still be inserted.
+
+=item C<BOLD>
+
+Added text is set in bold face, discarded is not shown. (also see --no-del option for another possibility to hide deleted text)
+
+=item C<PDFCOMMENT>
+
+The pdfcomment package is used to underline new text, and mark deletions with a PDF comment. Note that this markup might appear differently or not at all based on the pdf viewer used. The viewer with best support for pdf markup is probably acroread. This style is only recommended if the number of differences is small.
+
+=back
+
+=head2 Subtypes
+
+The subtype defines the commands that are inserted at the begin and end of added or discarded blocks, irrespectively of whether these blocks contain text or commands (Defined commands: C<\DIFaddbegin, \DIFaddend, \DIFdelbegin, \DIFdelend>)
+
+=over 10
+
+=item C<SAFE>
+
+No additional markup (Recommended choice)
+
+=item C<MARGIN>
+
+Mark beginning and end of changed blocks with symbols in the margin nearby (using
+the standard C<\marginpar> command - note that this sometimes moves somewhat
+from the intended position.
+
+=item C<COLOR>
+
+An alternative way of marking added passages in blue, and deleted ones in red.
+(It is recommeneded to use instead the main types to effect colored markup,
+although in some cases coloring with dvipscol can be more complete, for example
+with citation commands).
+
+=item C<DVIPSCOL>
+
+An alternative way of marking added passages in blue, and deleted ones in red. Note
+that C<DVIPSCOL> only works with the dvips converter, e.g. not pdflatex.
+(it is recommeneded to use instead the main types to effect colored markup,
+although in some cases coloring with dvipscol can be more complete).
+
+
+=item C<ZLABEL>
+
+can be used to highlight only changed pages, but requires post-processing. It is recommend to not call this option manually but use C<latexdiff-vc> with C<--only-changes> option. Alternatively, use the script given within preamble of diff files made using this style.
+
+=item C<ONLYCHANGEDPAGE>
+
+also highlights changed pages, without the need for post-processing, but might not work reliably if
+there is floating material (figures, tables).
+
+=item C<LABEL>
+
+is similar to C<ZLABEL>, but does not need the zref package and works less reliably (deprecated).
+
+=back
+
+=head2 Float Types
+
+Some of the markup used in the main text might cause problems when used within
+floats (e.g. figures or tables).  For this reason alternative versions of all
+markup commands are used within floats. The float type defines these alternative commands.
+
+=over 10
+
+=item C<FLOATSAFE>
+
+Use identical markup for text as in the main body, but set all commands marking the begin and end of changed blocks to null-commands.  You have to choose this float type if your subtype is C<MARGIN> as C<\marginpar> does not work properly within floats.
+
+=item C<TRADITIONALSAFE>
+
+Mark additions the same way as in the main text.  Deleted environments are marked by angular brackets \[ and \] and the deleted text is set in scriptscript size. This float type should always be used with the C<TRADITIONAL> and  C<CTRADITIONAL> markup types as the \footnote command does not work properly in floating environments.
+
+=item C<IDENTICAL>
+
+Make no difference between the main text and floats.
+
+=back
+
+
+=head2 Configuration Variables
+
+=over 10
+
+=item C<ARRENV>
+
+If a match to C<ARRENV> is found within an inline math environment within a deleted or added block, then the inlined math
+is surrounded by C<\mbox{>...C<}>.  This is necessary as underlining does not work within inlined array environments.
+
+[ Default: C<ARRENV>=S<C<(?:array|[pbvBV]matrix)> >
+
+=item C<COUNTERCMD>
+
+If a command in a deleted block which is also in the textcmd list matches C<COUNTERCMD> then an
+additional command C<\addtocounter{>F<cntcmd>C<}{-1}>, where F<cntcmd> is the matching command, is appended in the diff file such that the numbering in the diff file remains synchronized with the
+numbering in the new file.
+
+[ Default: C<COUNTERCMD>=C<(?:footnote|part|section|subsection> ...
+
+C<|subsubsection|paragraph|subparagraph)>  ]
+
+=item C<CUSTOMDIFCMD>
+
+This option is for advanced users and allows definition of special versions of commands, which do not work as safe commands.
+
+Commands in C<CUSTOMDIFCMD> that occur in added or deleted blocks will be given an ADD or DEL prefix.
+The prefixed versions of the command must be defined in the preamble, either by putting them
+in the preamble of at least the new file, or by creating a custom preamble file (Option --preamble).
+For example the command C<\blindtext> (from package blindtext) does not interact well with underlining, so that
+for the standard markup type, it is not satisfactory to define it as a safe command. Instead, a customised versions
+without underlining can be defined in the preamble:
+
+C<\newcommand{\DELblindtext}{{\color{red}\blindtext}}>
+
+C<\newcommand{\ADDblindtext}{{\color{blue}\blindtext}}>
+
+and then latexdiff should be invoked with the option C<-c CUSTOMDIFCMD=blindtext>.
+
+[ Default: none ]
+
+=item C<FLOATENV>
+
+Environments whose name matches the regular expression in C<FLOATENV> are
+considered floats.  Within these environments, the I<latexdiff> markup commands
+are replaced by their FL variaties.
+
+[ Default: S<C<(?:figure|table|plate)[\w\d*@]*> >]
+
+=item C<ITEMCMD>
+
+Commands representing new item line with list environments.
+
+[ Default: \C<item> ]
+
+=item C<LISTENV>
+
+Environments whose name matches the regular expression in C<LISTENV> are list environments.
+
+[ Default: S<C<(?:itemize|enumerate|description)> >]
+
+=item C<MATHENV>,C<MATHREPL>
+
+If both \begin and \end for a math environment (environment name matching C<MATHENV> or \[ and \])
+are within the same deleted block, they are replaced by a \begin and \end commands for C<MATHREPL>
+rather than being commented out.
+
+[ Default: C<MATHENV>=S<C<(?:displaymath|equation)> >, C<MATHREPL>=S<C<displaymath> >]
+
+=item C<MINWORDSBLOCK>
+
+Minimum number of tokens required to form an independent block. This value is
+used in the algorithm to detect changes of complete blocks by merging identical text parts of less than C<MINWORDSBLOCK> to the preceding added and discarded parts.
+
+[ Default: 3 ]
+
+=item C<PICTUREENV>
+
+Within environments whose name matches the regular expression in C<PICTUREENV>
+all latexdiff markup is removed (in pathologic cases this might lead to
+inconsistent markup but this situation should be rare).
+
+[ Default: S<C<(?:picture|DIFnomarkup)[\w\d*@]*> >]
+
+=item C<SCALEDELGRAPHICS>
+
+If C<--graphics-markup=both> is chosen, C<SCALEDELGRAPHICS> is the factor, by which deleted figures will be scaled (i.e. 0.5 implies they are shown at half linear size).
+
+[ Default: 0.5 ]
+
+=item C<VERBATIMENV>
+
+RegEx describing environments like verbatim, whose contents should be taken verbatim. The content of these environments will not be processed in any way:
+deleted content is commented out, new content is not marked up
+
+[ Default:  S<C<comment> > ]
+
+=item C<VERBATIMLINEENV>
+
+RegEx describing environments like verbatim, whose contents should be taken verbatim. The content of environments described by VERBATIMLINEENV are compared in
+line mode, and changes are marked up using the listings package. The markup style is set based on the chosen mains markup type (Option -t), or on an analysis
+of the preamble.
+Note that "listings.sty" must be installed. If this file is not found the fallback solution is to
+treat VERBATIMLINEENV environments treated exactly the same way as VERBATIMENV environments.
+
+[ Default:  S<C<(?:verbatim[*]?|lstlisting> > ]
+
+=back
+
+=head1 COMMON PROBLEMS AND FAQ
+
+=over 10
+
+=item Changed citations result in overfull boxes
+
+There is an incompatibility between the C<ulem> package, which C<latexdiff> uses for underlining and striking out in the UNDERLINE style,
+the default style, and the way citations are generated. In order to be able to mark up citations properly, they are enclosed with an C<\mbox>
+command. As mboxes cannot be broken across lines, this procedure frequently results in overfull boxes, possibly obscuring the content as it extends beyond the right margin.  The same occurs for some other packages (e.g., siunitx). If this is a problem, you have several possibilities.
+
+1. Use C<CFONT> type markup (option C<-t CFONT>): If this markup is chosen, then changed citations are no longer marked up
+with the wavy line (additions) or struck out (deletions), but are still highlighted in the appropriate color, and deleted text is shown with a different font. Other styles not using the C<ulem> package will also work.
+
+2. Choose option C<--disable-citation-markup> which turns off the marking up of citations: deleted citations are no longer shown, and
+added citations are shown without markup. (This was the default behaviour of latexdiff at versions 0.6 and older).
+For custom packages you can define the commands which need to be protected by C<\mbox> with C<--append-mboxsafecmd> and C<--excludemboxsafecmd> options
+(submit your lists of command as feature request at github page to set the default behaviour of future versions, see section 6)
+
+3. If you are using luatex >= 1.12.0 you can use option LUAUNDERLINE that uses lua-ul instead of ulem for underlining, which does not have this problem (experimental feature).
+
+=item Changes in complicated mathematical equations result in latex processing errors.
+
+Try option C<--math-markup=whole>.   If even that fails, you can turn off mark up for equations with C<--math-markup=off>.
+
+=item Deleted parts in equations are not struck out but underlined.
+
+This is a limitation of the ulem package that implements the strike-out. If you use the amsmath package, then the strike-out command is redefined in such a way that deleted passages are wrapped with C<\text> command; adding C<\usepackage{amsmath}> to your preamble will trigger this behaviour. (If amsmath is not included directly, but imported by another package, latexdiff will not be able to detect its availability; in this case you can give latexdiff a hint with option C<--packages=amsmath>.
+
+
+=item How can I just show the pages where changes had been made?
+
+Use options C<--s ZLABEL>  (some postprocessing required) or C<-s ONLYCHANGEDPAGE>. C<latexdiff-vc --ps|--pdf> with C<--only-changes> option takes care of
+the post-processing for you (requires zref package to be installed).
+
+=item The character encoding breaks when running latexdiff from powershell.
+
+This problem is not limited to C<latexdiff> and has to do with the default settings of C<powershell> in Windows. It is recommended to use C<cmd> instead.
+
+=back
+
+=head1 BUGS
+
+=over 10
+
+=item Option allow-spaces not implemented entirely consistently. It breaks
+the rules that number and type of white space does not matter, as
+different numbers of inter-argument spaces are treated as significant.
+
+=back
+
+Please submit bug reports using the issue tracker of the github repository page I<https://github.com/ftilmann/latexdiff.git>,
+or send them to I<tilmann -- AT -- gfz-potsdam.de>.  Include the version number of I<latexdiff>
+(from comments at the top of the source or use B<--version>).  If you come across latex
+files that are error-free and conform to the specifications set out
+above, and whose differencing still does not result in error-free
+latex, please send me those files, ideally edited to only contain the
+offending passage as long as that still reproduces the problem. If your
+file relies on non-standard class files, you must include those.  I will not
+look at examples where I have trouble to latex the original files.
+
+=head1 SEE ALSO
+
+L<latexrevise>, L<latexdiff-vc>
+
+=head1 PORTABILITY
+
+I<latexdiff> does not make use of external commands and thus should run
+on any platform  supporting Perl 5.6 or higher.  If files with encodings
+other than ASCII or UTF-8 are processed, Perl 5.8 or higher is required.
+
+The standard version of I<latexdiff> requires installation of the Perl package
+C<Algorithm::Diff> (available from I<www.cpan.org> -
+I<http://search.cpan.org/~nedkonz/Algorithm-Diff-1.15>) but a stand-alone
+version, I<latexdiff-so>, which has this package inlined, is available, too.
+I<latexdiff-fast> requires the I<diff> command to be present.
+
+=head1 AUTHOR
+
+Version 1.3.4
+Copyright (C) 2004-2024 Frederik Tilmann
+
+This program is free software; you can redistribute it and/or modify
+it under the terms of the GNU General Public License Version 3
+
+Contributors of fixes and additions: V. Kuhlmann, J. Paisley, N. Becker, T. Doerges, K. Huebner,
+T. Connors, Sebastian Gouezel and many others.
+Thanks to the many people who sent in bug reports, feature suggestions, and other feedback.
+
+=cut
+
+__END__
+%%BEGIN SAFE COMMANDS
+% Regex matching commands which can safely be in the
+% argument of a \DIFadd or \DIFdel command (leave out the \)
+arabic
+dashbox
+emph
+fbox
+framebox
+hspace\*?
+math.*
+makebox
+mbox
+pageref
+ref
+eqref
+symbol
+raisebox
+rule
+text.*
+shortstack
+usebox
+dag
+ddag
+copyright
+pounds
+S
+P
+oe
+OE
+ae
+AE
+aa
+AA
+o
+O
+l
+L
+frac
+ss
+sqrt
+ldots
+cdots
+vdots
+ddots
+alpha
+beta
+gamma
+delta
+epsilon
+varepsilon
+zeta
+eta
+theta
+vartheta
+iota
+kappa
+lambda
+mu
+nu
+xi
+pi
+varpi
+rho
+varrho
+sigma
+varsigma
+tau
+upsilon
+phi
+varphi
+chi
+psi
+omega
+Gamma
+Delta
+Theta
+Lambda
+Xi
+Pi
+Sigma
+Upsilon
+Phi
+Psi
+Omega
+ps
+mp
+times
+div
+ast
+star
+circ
+bullet
+cdot
+cap
+cup
+uplus
+sqcap
+vee
+wedge
+setminus
+wr
+diamond
+(?:big)?triangle.*
+lhd
+rhd
+unlhd
+unrhd
+oplus
+ominus
+otimes
+oslash
+odot
+bigcirc
+d?dagger
+amalg
+leq
+prec
+preceq
+ll
+(?:sq)?su[bp]set(?:eq)?
+in
+vdash
+geq
+succ(?:eq)?
+gg
+ni
+dashv
+equiv
+sim(?:eq)?
+asymp
+approx
+cong
+neq
+doteq
+propto
+models
+perp
+mid
+parallel
+bowtie
+Join
+smile
+frown
+.*arrow
+(?:long)?mapsto
+.*harpoon.*
+leadsto
+aleph
+hbar
+imath
+jmath
+ell
+wp
+Re
+Im
+mho
+prime
+emptyset
+nabla
+surd
+top
+bot
+angle
+forall
+exists
+neg
+flat
+natural
+sharp
+backslash
+partial
+infty
+Box
+Diamond
+triangle
+clubsuit
+diamondsuit
+heartsuit
+spadesuit
+sum
+prod
+coprod
+int
+oint
+big(?:sq)?c[au]p
+bigvee
+bigwedge
+bigodot
+bigotimes
+bigoplus
+biguplus
+(?:arc)?(?:cos|sin|tan|cot)h?
+csc
+arg
+deg
+det
+dim
+exp
+gcd
+hom
+inf
+ker
+lg
+lim
+liminf
+limsup
+ln
+log
+max
+min
+Pr
+sec
+sup
+quad
+qquad
+bibfield
+bibinfo
+[Hclbkdruvt]
+[`'^"~=.]
+_
+AMPERSAND
+(SUPER|SUB)SCRIPTNB
+(SUPER|SUB)SCRIPT
+SQRT
+SQRTNB
+PERCENTAGE
+DOLLAR
+%%END SAFE COMMANDS
+
+%%BEGIN TEXT COMMANDS
+% Regex matching commands with a text argument (leave out the \)
+addcontents.*
+cc
+closing
+chapter
+dashbox
+emph
+encl
+fbox
+framebox
+footnote
+footnotetext
+framebox
+href
+intertext
+part
+(sub){0,2}section\*?
+(sub)?paragraph\*?
+makebox
+mbox
+opening
+parbox
+raisebox
+savebox
+sbox
+shortintertext
+shortstack
+sidenote
+signature
+text.*
+value
+underline
+sqrt
+(SUPER|SUB)SCRIPT
+%%END TEXT COMMANDS
+
+%%BEGIN CONTEXT1 COMMANDS
+% Regex matching commands with a text argument (leave out the \), which will fail out of context. These commands behave like text commands, except when they occur in a deleted section, where they are disabled, but their argument is shown as deleted text.
+caption
+subcaption
+multicolumn
+%%END CONTEXT1 COMMANDS
+
+%%BEGIN CONTEXT2 COMMANDS
+% Regex matching commands with a text argument (leave out the \), which will fail out of context.  As corresponding commands for context1.  The only difference is that context2 commands are completely disabled in deleted sections, including their arguments.
+title
+author
+date
+institute
+%%END CONTEXT2 COMMANDS
+
+%% CONFIGURATION variabe defaults
+%%BEGIN LISTENV CONFIG
+itemize
+description
+enumerate
+%%END LISTENV CONFIG
+
+%%BEGIN FLOATENV CONFIG
+figure[\w\d*@]*
+table[\w\d*@]*
+plate[\w\d*@]*
+%%END FLOATENV CONFIG
+
+%%BEGIN PICTUREENV CONFIG
+picture[\w\d*@]*
+tikzpicture[\w\d*@]*
+DIFnomarkup
+%%END PICTUREENV CONFIG
+
+%%BEGIN MATHENV CONFIG
+equation[*]?
+displaymath
+math
+DOLLARDOLLAR
+%%END MATHENV CONFIG
+
+%%BEGIN MATHARRENV CONFIG
+eqnarray[*]?
+align[*]?
+alignat[*]?
+gather[*]?
+multline[*]?
+flalign[*]?
+%%END MATHARRENV CONFIG
+
+%%BEGIN ARRENV CONFIG
+aligned
+gathered
+multlined
+array
+[pbvBV]?matrix
+smallmatrix
+cases
+%%END ARRENV CONFIG
+# split
+
+
+%%BEGIN COUNTERCMD CONFIG
+footnote
+part
+chapter
+section
+subsection
+subsubsection
+paragraph
+subparagraph
+%%END COUNTERCMD CONFIG
+
+%%BEGIN VERBATIMENV CONFIG
+comment
+%%END VERBATIMENV CONFIG
+
+%%BEGIN VERBATIMLINEENV CONFIG
+lstlisting
+verbatim[*]?
+%%END VERBATIMLINEENV CONFIG
+
+%%BEGIN CUSTOMDIFCMD CONFIG
+%%END CUSTOMDIFCMD CONFIG
+
+%%% TYPES (Commands for highlighting changed blocks)
+
+%DIF UNDERLINE PREAMBLE
+\RequirePackage[normalem]{ulem}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{{\protect\color{blue}\uwave{#1}}}
+\providecommand{\DIFdel}[1]{{\protect\color{red}\sout{#1}}}
+%DIF END UNDERLINE PREAMBLE
+
+%DIF LUAUNDERLINE PREAMBLE
+\RequirePackage{lua-ul}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{{\protect\color{blue}\underLine{#1}}}
+\providecommand{\DIFdel}[1]{{\protect\color{red}\strikeThrough{#1}}}
+%DIF END LUAUNDERLINE PREAMBLE
+
+%DIF CTRADITIONAL PREAMBLE
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\RequirePackage[stable]{footmisc}
+\DeclareOldFontCommand{\sf}{\normalfont\sffamily}{\mathsf}
+\providecommand{\DIFadd}[1]{{\protect\color{blue} \sf #1}}
+\providecommand{\DIFdel}[1]{{\protect\color{red} [..\footnote{removed: #1} ]}}
+%DIF END CTRADITIONAL PREAMBLE
+
+%DIF TRADITIONAL PREAMBLE
+\RequirePackage[stable]{footmisc}
+\DeclareOldFontCommand{\sf}{\normalfont\sffamily}{\mathsf}
+\providecommand{\DIFadd}[1]{{\sf #1}}
+\providecommand{\DIFdel}[1]{{[..\footnote{removed: #1} ]}}
+%DIF END TRADITIONAL PREAMBLE
+
+%DIF CFONT PREAMBLE
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\DeclareOldFontCommand{\sf}{\normalfont\sffamily}{\mathsf}
+\providecommand{\DIFadd}[1]{{\protect\color{blue} \sf #1}}
+\providecommand{\DIFdel}[1]{{\protect\color{red} \scriptsize #1}}
+%DIF END CFONT PREAMBLE
+
+%DIF FONTSTRIKE PREAMBLE
+\RequirePackage[normalem]{ulem}
+\DeclareOldFontCommand{\sf}{\normalfont\sffamily}{\mathsf}
+\providecommand{\DIFadd}[1]{{\sf #1}}
+\providecommand{\DIFdel}[1]{{\footnotesize \sout{#1}}}
+%DIF END FONTSTRIKE PREAMBLE
+
+%DIF CCHANGEBAR PREAMBLE
+\RequirePackage[pdftex]{changebar}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{\protect\cbstart{\protect\color{blue}#1}\protect\cbend}
+\providecommand{\DIFdel}[1]{\protect\cbdelete{\protect\color{red}#1}\protect\cbdelete}
+%DIF END CCHANGEBAR PREAMBLE
+
+%DIF CFONTCHBAR PREAMBLE
+\RequirePackage[pdftex]{changebar}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{\protect\cbstart{\protect\color{blue}\sf #1}\protect\cbend}
+\providecommand{\DIFdel}[1]{\protect\cbdelete{\protect\color{red}\scriptsize #1}\protect\cbdelete}
+%DIF END CFONTCHBAR PREAMBLE
+
+%DIF CULINECHBAR PREAMBLE
+\RequirePackage[normalem]{ulem}
+\RequirePackage[pdftex]{changebar}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{\protect\cbstart{\protect\color{blue}\uwave{#1}}\protect\cbend}
+\providecommand{\DIFdel}[1]{\protect\cbdelete{\protect\color{red}\sout{#1}}\protect\cbdelete}
+%DIF END CULINECHBAR PREAMBLE
+
+%DIF CHANGEBAR PREAMBLE
+\RequirePackage[pdftex]{changebar}
+\providecommand{\DIFadd}[1]{\protect\cbstart{#1}\protect\cbend}
+\providecommand{\DIFdel}[1]{\protect\cbdelete}
+%DIF END CHANGEBAR PREAMBLE
+
+%DIF INVISIBLE PREAMBLE
+\providecommand{\DIFadd}[1]{#1}
+\providecommand{\DIFdel}[1]{}
+%DIF END INVISIBLE PREAMBLE
+
+%DIF BOLD PREAMBLE
+\DeclareOldFontCommand{\bf}{\normalfont\bfseries}{\mathbf}
+\providecommand{\DIFadd}[1]{{\bf #1}}
+\providecommand{\DIFdel}[1]{}
+%DIF END BOLD PREAMBLE
+
+%DIF PDFCOMMENT PREAMBLE
+\RequirePackage{pdfcomment} %DIF PREAMBLE
+\providecommand{\DIFadd}[1]{\pdfmarkupcomment[author=ADD:,markup=Underline]{#1}{}}
+\providecommand{\DIFdel}[1]{\pdfcomment[icon=Insert,author=DEL:,hspace=12pt]{#1}}
+%DIF END PDFCOMMENT PREAMBLE
+
+%% SUBTYPES (Markers for beginning and end of changed blocks)
+
+%DIF SAFE PREAMBLE
+\providecommand{\DIFaddbegin}{}
+\providecommand{\DIFaddend}{}
+\providecommand{\DIFdelbegin}{}
+\providecommand{\DIFdelend}{}
+\providecommand{\DIFmodbegin}{}
+\providecommand{\DIFmodend}{}
+%DIF END SAFE PREAMBLE
+
+%DIF MARGIN PREAMBLE
+\providecommand{\DIFaddbegin}{\protect\marginpar{a[}}
+\providecommand{\DIFaddend}{\protect\marginpar{]}}
+\providecommand{\DIFdelbegin}{\protect\marginpar{d[}}
+\providecommand{\DIFdelend}{\protect\marginpar{]}}
+\providecommand{\DIFmodbegin}{\protect\marginpar{m[}}
+\providecommand{\DIFmodend}{\protect\marginpar{]}}
+%DIF END MARGIN PREAMBLE
+
+%DIF DVIPSCOL PREAMBLE
+%Note: only works with dvips converter
+\RequirePackage{color}
+\RequirePackage{dvipscol}
+\providecommand{\DIFaddbegin}{\protect\nogroupcolor{blue}}
+\providecommand{\DIFaddend}{\protect\nogroupcolor{black}}
+\providecommand{\DIFdelbegin}{\protect\nogroupcolor{red}}
+\providecommand{\DIFdelend}{\protect\nogroupcolor{black}}
+\providecommand{\DIFmodbegin}{}
+\providecommand{\DIFmodend}{}
+%DIF END DVIPSCOL PREAMBLE
+
+%DIF COLOR PREAMBLE
+\RequirePackage{color}
+\providecommand{\DIFaddbegin}{\protect\color{blue}}
+\providecommand{\DIFaddend}{\protect\color{black}}
+\providecommand{\DIFdelbegin}{\protect\color{red}}
+\providecommand{\DIFdelend}{\protect\color{black}}
+\providecommand{\DIFmodbegin}{}
+\providecommand{\DIFmodend}{}
+%DIF END COLOR PREAMBLE
+
+%DIF LABEL PREAMBLE
+% To show only pages with changes (pdf) (external program pdftk needs to be installed)
+% (only works for simple documents with non-repeated page numbers, otherwise use ZLABEL)
+% pdflatex diff.tex
+% pdflatex diff.tex
+%pdftk diff.pdf cat \
+%`perl -lne '\
+% if (m/\\newlabel{DIFchg[b](\d*)}{{.*}{(.*)}}/) { $start{$1}=$2; print $2}\
+% if (m/\\newlabel{DIFchg[e](\d*)}{{.*}{(.*)}}/) { \
+%      if (defined($start{$1})) { \
+%         for ($j=$start{$1}; $j<=$2; $j++) {print "$j";}\
+%      } else { \
+%         print "$2"\
+%      }\
+% }' diff.aux \
+% | uniq \
+% | tr  \\n ' '` \
+% output diff-changedpages.pdf
+% To show only pages with changes (dvips/dvipdf)
+% dvips -pp `\
+% [ put here the perl script from above]
+% | uniq | tr -s \\n ','`
+\typeout{Check comments in preamble of output for instructions how to show only pages where changes have been made}
+\newcount\DIFcounterb
+\global\DIFcounterb 0\relax
+\newcount\DIFcountere
+\global\DIFcountere 0\relax
+\providecommand{\DIFaddbegin}{\global\advance\DIFcounterb 1\relax\label{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFaddend}{\global\advance\DIFcountere 1\relax\label{DIFchge\the\DIFcountere}}
+\providecommand{\DIFdelbegin}{\global\advance\DIFcounterb 1\relax\label{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFdelend}{\global\advance\DIFcountere 1\relax\label{DIFchge\the\DIFcountere}}
+\providecommand{\DIFmodbegin}{\global\advance\DIFcounterb 1\relax\label{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFmodend}{\global\advance\DIFcountere 1\relax\label{DIFchge\the\DIFcountere}}
+%DIF END LABEL PREAMBLE
+
+%DIF ZLABEL PREAMBLE
+% To show only pages with changes (pdf) (external program pdftk needs to be installed)
+% (uses zref for reference to absolute page numbers)
+% pdflatex diff.tex
+% pdflatex diff.tex
+%pdftk diff.pdf cat \
+%`perl -lne 'if (m/\\zref\@newlabel{DIFchgb(\d*)}{.*\\abspage{(\d*)}}/ ) { $start{$1}=$2; print $2 } \
+%  if (m/\\zref\@newlabel{DIFchge(\d*)}{.*\\abspage{(\d*)}}/) { \
+%      if (defined($start{$1})) { \
+%         for ($j=$start{$1}; $j<=$2; $j++) {print "$j";}\
+%      } else { \
+%         print "$2"\
+%      }\
+% }' diff.aux \
+% | uniq \
+% | tr  \\n ' '` \
+% output diff-changedpages.pdf
+% To show only pages with changes (dvips/dvipdf)
+% latex diff.tex
+% latex diff.tex
+% dvips -pp `perl -lne 'if (m/\\newlabel{DIFchg[be]\d*}{{.*}{(.*)}}/) { print $1 }' diff.aux | uniq | tr -s \\n ','` diff.dvi
+\typeout{Check comments in preamble of output for instructions how to show only pages where changes have been made}
+\usepackage[user,abspage]{zref}
+\newcount\DIFcounterb
+\global\DIFcounterb 0\relax
+\newcount\DIFcountere
+\global\DIFcountere 0\relax
+\providecommand{\DIFaddbegin}{\global\advance\DIFcounterb 1\relax\zlabel{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFaddend}{\global\advance\DIFcountere 1\relax\zlabel{DIFchge\the\DIFcountere}}
+\providecommand{\DIFdelbegin}{\global\advance\DIFcounterb 1\relax\zlabel{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFdelend}{\global\advance\DIFcountere 1\relax\zlabel{DIFchge\the\DIFcountere}}
+\providecommand{\DIFmodbegin}{\global\advance\DIFcounterb 1\relax\zlabel{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFmodend}{\global\advance\DIFcountere 1\relax\zlabel{DIFchge\the\DIFcountere}}
+%DIF END ZLABEL PREAMBLE
+
+%DIF ONLYCHANGEDPAGE PREAMBLE
+\RequirePackage{atbegshi}
+\RequirePackage{etoolbox}
+\RequirePackage{zref}
+% redefine label command to write immediately to aux file - page references will be lost
+\makeatletter \let\oldlabel\label% Store \label
+\renewcommand{\label}[1]{% Update \label to write to the .aux immediately
+\zref@wrapper@immediate{\oldlabel{#1}}}
+\makeatother
+\newbool{DIFkeeppage}
+\newbool{DIFchange}
+\boolfalse{DIFkeeppage}
+\boolfalse{DIFchange}
+\AtBeginShipout{%
+  \ifbool{DIFkeeppage}
+        {\global\boolfalse{DIFkeeppage}}  % True DIFkeeppage
+         {\ifbool{DIFchange}{\global\boolfalse{DIFkeeppage}}{\global\boolfalse{DIFkeeppage}\AtBeginShipoutDiscard}} % False DIFkeeppage
+}
+\providecommand{\DIFaddbegin}{\global\booltrue{DIFkeeppage}\global\booltrue{DIFchange}}
+\providecommand{\DIFaddend}{\global\booltrue{DIFkeeppage}\global\boolfalse{DIFchange}}
+\providecommand{\DIFdelbegin}{\global\booltrue{DIFkeeppage}\global\booltrue{DIFchange}}
+\providecommand{\DIFdelend}{\global\booltrue{DIFkeeppage}\global\boolfalse{DIFchange}}
+\providecommand{\DIFmodbegin}{\global\booltrue{DIFkeeppage}\global\booltrue{DIFchange}}
+\providecommand{\DIFmodend}{\global\booltrue{DIFkeeppage}\global\boolfalse{DIFchange}}
+%DIF END ONLYCHANGEDPAGE PREAMBLE
+
+%% FLOAT TYPES
+
+%DIF FLOATSAFE PREAMBLE
+\providecommand{\DIFaddFL}[1]{\DIFadd{#1}}
+\providecommand{\DIFdelFL}[1]{\DIFdel{#1}}
+\providecommand{\DIFaddbeginFL}{}
+\providecommand{\DIFaddendFL}{}
+\providecommand{\DIFdelbeginFL}{}
+\providecommand{\DIFdelendFL}{}
+%DIF END FLOATSAFE PREAMBLE
+
+%DIF IDENTICAL PREAMBLE
+\providecommand{\DIFaddFL}[1]{\DIFadd{#1}}
+\providecommand{\DIFdelFL}[1]{\DIFdel{#1}}
+\providecommand{\DIFaddbeginFL}{\DIFaddbegin}
+\providecommand{\DIFaddendFL}{\DIFaddend}
+\providecommand{\DIFdelbeginFL}{\DIFdelbegin}
+\providecommand{\DIFdelendFL}{\DIFdelend}
+%DIF END IDENTICAL PREAMBLE
+
+%DIF TRADITIONALSAFE PREAMBLE
+% procidecommand color to make this work for TRADITIONAL and CTRADITIONAL
+\providecommand{\color}[1]{}
+\providecommand{\DIFaddFL}[1]{\DIFadd{#1}}
+\providecommand{\DIFdel}[1]{{\protect\color{red}[..{\scriptsize {removed: #1}} ]}}
+\providecommand{\DIFaddbeginFL}{}
+\providecommand{\DIFaddendFL}{}
+\providecommand{\DIFdelbeginFL}{}
+\providecommand{\DIFdelendFL}{}
+%DIF END TRADITIONALSAFE PREAMBLE
+
+% see:
+%  http://tex.stackexchange.com/questions/47351/can-i-redefine-a-command-to-contain-itself
+
+%DIF HIGHLIGHTGRAPHICS PREAMBLE
+\RequirePackage{settobox}
+\RequirePackage{letltxmacro}
+\newsavebox{\DIFdelgraphicsbox}
+\newlength{\DIFdelgraphicswidth}
+\newlength{\DIFdelgraphicsheight}
+% store original definition of \includegraphics
+\LetLtxMacro{\DIFOincludegraphics}{\includegraphics}
+\newcommand{\DIFaddincludegraphics}[2][]{{\color{blue}\fbox{\DIFOincludegraphics[#1]{#2}}}}
+\newcommand{\DIFdelincludegraphics}[2][]{%
+\sbox{\DIFdelgraphicsbox}{\DIFOincludegraphics[#1]{#2}}%
+\settoboxwidth{\DIFdelgraphicswidth}{\DIFdelgraphicsbox}
+\settoboxtotalheight{\DIFdelgraphicsheight}{\DIFdelgraphicsbox}
+\scalebox{\DIFscaledelfig}{%
+\parbox[b]{\DIFdelgraphicswidth}{\usebox{\DIFdelgraphicsbox}\\[-\baselineskip] \rule{\DIFdelgraphicswidth}{0em}}\llap{\resizebox{\DIFdelgraphicswidth}{\DIFdelgraphicsheight}{%
+\setlength{\unitlength}{\DIFdelgraphicswidth}%
+\begin{picture}(1,1)%
+\thicklines\linethickness{2pt}
+{\color[rgb]{1,0,0}\put(0,0){\framebox(1,1){}}}%
+{\color[rgb]{1,0,0}\put(0,0){\line( 1,1){1}}}%
+{\color[rgb]{1,0,0}\put(0,1){\line(1,-1){1}}}%
+\end{picture}%
+}\hspace*{3pt}}}
+}
+\LetLtxMacro{\DIFOaddbegin}{\DIFaddbegin}
+\LetLtxMacro{\DIFOaddend}{\DIFaddend}
+\LetLtxMacro{\DIFOdelbegin}{\DIFdelbegin}
+\LetLtxMacro{\DIFOdelend}{\DIFdelend}
+\DeclareRobustCommand{\DIFaddbegin}{\DIFOaddbegin \let\includegraphics\DIFaddincludegraphics}
+\DeclareRobustCommand{\DIFaddend}{\DIFOaddend \let\includegraphics\DIFOincludegraphics}
+\DeclareRobustCommand{\DIFdelbegin}{\DIFOdelbegin \let\includegraphics\DIFdelincludegraphics}
+\DeclareRobustCommand{\DIFdelend}{\DIFOaddend \let\includegraphics\DIFOincludegraphics}
+\LetLtxMacro{\DIFOaddbeginFL}{\DIFaddbeginFL}
+\LetLtxMacro{\DIFOaddendFL}{\DIFaddendFL}
+\LetLtxMacro{\DIFOdelbeginFL}{\DIFdelbeginFL}
+\LetLtxMacro{\DIFOdelendFL}{\DIFdelendFL}
+\DeclareRobustCommand{\DIFaddbeginFL}{\DIFOaddbeginFL \let\includegraphics\DIFaddincludegraphics}
+\DeclareRobustCommand{\DIFaddendFL}{\DIFOaddendFL \let\includegraphics\DIFOincludegraphics}
+\DeclareRobustCommand{\DIFdelbeginFL}{\DIFOdelbeginFL \let\includegraphics\DIFdelincludegraphics}
+\DeclareRobustCommand{\DIFdelendFL}{\DIFOaddendFL \let\includegraphics\DIFOincludegraphics}
+%DIF END HIGHLIGHTGRAPHICS PREAMBLE
+
+%% SPECIAL PACKAGE PREAMBLE COMMANDS
+
+%% Redefine strike out command to wrap in text box if amsmath is used and markup style with ulem is used
+%DIF AMSMATHULEM PREAMBLE
+\makeatletter
+\let\sout@orig\sout
+\renewcommand{\sout}[1]{\ifmmode\text{\sout@orig{\ensuremath{#1}}}\else\sout@orig{#1}\fi}
+\makeatother
+%DIF END AMSMATHULEM PREAMBLE
+
+
+% Standard \DIFadd and \DIFdel are redefined as \DIFaddtex and \DIFdeltex
+% when hyperref package is included.
+%DIF HYPERREF PREAMBLE
+\providecommand{\DIFadd}[1]{\texorpdfstring{\DIFaddtex{#1}}{#1}}
+\providecommand{\DIFdel}[1]{\texorpdfstring{\DIFdeltex{#1}}{}}
+%DIF END HYPERREF PREAMBLE
+
+%DIF LISTINGS PREAMBLE
+\RequirePackage{listings}
+\lstdefinelanguage{DIFcode}{
+  % note that the definitions in the following two lines are overwritten dependent on the markup type selected %DIFCODE TEMPLATE
+  morecomment=[il]{\%DIF\ <\ },          %DIFCODE TEMPLATE
+  moredelim=[il][\bfseries]{\%DIF\ >\ }  %DIFCODE TEMPLATE
+}
+\lstdefinestyle{DIFverbatimstyle}{
+	language=DIFcode,
+	basicstyle=\ttfamily,
+	columns=fullflexible,
+	keepspaces=true
+}
+\lstnewenvironment{DIFverbatim}{\lstset{style=DIFverbatimstyle}}{}
+\lstnewenvironment{DIFverbatim*}{\lstset{style=DIFverbatimstyle,showspaces=true}}{}
+%DIF END LISTINGS PREAMBLE
+
+%DIF COLORLISTINGS PREAMBLE
+\RequirePackage{listings}
+\RequirePackage{color}
+\lstdefinelanguage{DIFcode}{
+  % note that the definitions in the following two lines are overwritten dependent on the markup type selected %DIFCODE TEMPLATE
+  morecomment=[il]{\%DIF\ <\ },          %DIFCODE TEMPLATE
+  moredelim=[il][\bfseries]{\%DIF\ >\ }  %DIFCODE TEMPLATE
+}
+\lstdefinestyle{DIFverbatimstyle}{
+	language=DIFcode,
+	basicstyle=\ttfamily,
+	columns=fullflexible,
+	keepspaces=true
+}
+\lstnewenvironment{DIFverbatim}{\lstset{style=DIFverbatimstyle}}{}
+\lstnewenvironment{DIFverbatim*}{\lstset{style=DIFverbatimstyle,showspaces=true}}{}
+%DIF END COLORLISTINGS PREAMBLE
+
+%DIF DIFCODE_UNDERLINE
+  moredelim=[il][\color{red}\sout]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\uwave]{\%DIF\ >\ }
+%DIF END DIFCODE_UNDERLINE
+
+%DIF DIFCODE_CTRADITIONAL
+  moredelim=[il][\color{red}\scriptsize]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_CTRADITIONAL
+
+%DIF DIFCODE_TRADITIONAL
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il][\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_TRADITIONAL
+
+%DIF DIFCODE_CFONT
+  moredelim=[il][\color{red}\scriptsize]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_CFONT
+
+%DIF DIFCODE_FONTSTRIKE
+  moredelim=[il][\scriptsize \sout]{\%DIF\ <\ },
+  moredelim=[il][\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_FONTSTRIKE
+
+%DIF DIFCODE_INVISIBLE
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il]{\%DIF\ >\ }
+%DIF END DIFCODE_INVISIBLE
+
+%DIF DIFCODE_CHANGEBAR
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il]{\%DIF\ >\ }
+%DIF END DIFCODE_CHANGEBAR
+
+%DIF DIFCODE_CCHANGEBAR
+  moredelim=[il][\color{red}]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}]{\%DIF\ >\ }
+%DIF END DIFCODE_CCHANGEBAR
+
+%DIF DIFCODE_CULINECHBAR
+  moredelim=[il][\color{red}\sout]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\uwave]{\%DIF\ >\ }
+%DIF END DIFCODE_CULINECHBAR
+
+%DIF DIFCODE_CFONTCHBAR
+  moredelim=[il][\color{red}\scriptsize]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_CFONTCHBAR
+
+%DIF DIFCODE_BOLD
+  % unfortunately \bfseries cannot be combined with ttfamily without extra packages
+  % also morecomment=[il] is broken as of v1.5b of listings at least
+  % workaround: plot in white with tiny font
+  % morecomment=[il]{\%DIF\ <\ },
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il][\sffamily\bfseries]{\%DIF\ >\ }
+%DIF END DIFCODE_BOLD
+
+%DIF DIFCODE_PDFCOMMENT
+
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il][\sffamily\bfseries]{\%DIF\ >\ }
+%DIF END DIFCODE_PDFCOMMENT
+
diff --git a/scripts/latexdiff-vc.pl b/scripts/latexdiff-vc.pl
new file mode 100644
index 0000000..31760ca
--- /dev/null
+++ b/scripts/latexdiff-vc.pl
@@ -0,0 +1,913 @@
+#!/usr/bin/env perl 
+#
+# latexdiff-vc  - wrapper script for applying latexdiff to rcs managed files
+#                 and for automatised creation of postscript or pdf from difference file
+#
+#   Copyright (C) 2005-20  F J Tilmann (tilmann@gfz-potsdam.de)
+#
+# Repository:         https://github.com/ftilmann/latexdiff
+# CTAN page:          http://www.ctan.org/pkg/latexdiff
+#
+#
+#   Contributors: S Utcke, H Bruyninckx, C Junghans
+#
+#    This program is free software: you can redistribute it and/or modify
+#    it under the terms of the GNU General Public License as published by
+#    the Free Software Foundation, either version 3 of the License, or
+#    (at your option) any later version.
+#
+#    This program is distributed in the hope that it will be useful,
+#    but WITHOUT ANY WARRANTY; without even the implied warranty of
+#    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+#    GNU General Public License for more details.
+#    You should have received a copy of the GNU General Public License
+#    along with this program.  If not, see <http://www.gnu.org/licenses/>.
+#
+# Detailed usage information at the end of the file
+#
+# TODO/IDEAS: - option to call external pre-processing codes
+#
+# version 1.3.4:
+#    - Enhancement: if --flatten option is used, and no bbl file is included in the version management, then latexdiff-vc will try to run bibtex to generate the bbl file. Implements the suggestion of PR #127 (contributed by github user cousteaulecommandant )
+#    - Efficiency boost: use -draftmode for non-interactive preliminary runs of latex (idea from #
+#    - bug fix: minor fix for --only-changed that makes detection of changed pages more reliable [ PR #297 contributed by user nelijuc ]
+#
+# version 1.3.2:
+#    - bug fix: when setting config variables with the command lines it is now possible to use quotes to includes spaces in the value, e.g. -c LATEX="pdflatex --shell-escape"
+#    - bug fix: --only-changes is not compatible with graphics markup. --graphics-markup=none is now set automatically
+#    - when --pdf or --postscript is select, then replace tildes in version names with underscores in output file (e.g. diffHEAD_1 instead of diff HEAD~1. This is necessary because pdflatex strips ~n strings at the end of the filename, when generating output file names
+
+# version 1.3.1:
+#    - bug fix, import File::Path explicitly as otherwise mkpath is not found
+# version 1.3.0 ((7 October 2018)
+#    - option --only-changes with hyperref will suppress hyperrefs (pull request jprotze)_
+#    - option --only-changes now moves (rather than copies) file with only changes 
+# version 1.2.1 (22 June 2017):
+#    - update: use qpdf (instead of pdftk) to select particular pages of output (pull request #102 submitted by Tom Scogland via github). This was necessary because pdftk is being deprecated. pdftk is still used as a fall-back
+#    - bug fix: --hg option was not recognised (partially fixes github issue #93 )
+#    - wrap passed-through options to latexdiff in quotation marks (fix github issue #58 )
+#    - program names for latexdiff, latex, dvips, bibtex configurable (fixes issue #40)
+#
+# version 1.2.0:
+#    - deprecation fix: left brace in RegEx now needs to be escaped
+#
+# version 1.1.1:
+#    - better detection of RCS system 
+#    - undocumented option --debug/--nodebug to override default setting for debug mode (Default: 0)#
+#    - bug fix: --flatten option combined with --pdf caused confusion of old and new file
+#
+# version 1.1.0:
+#
+#      - with option --flatten and version control option, checkout the whole tree into a temporary directory
+#        (use svn info + checkout ; git archive )
+#       - add option --only-changes to output only changed pages (in combination with --ps or --pdf option)
+# version 1.0.4:
+#	- [ Patch #3486 contributed by cbarbu] add additional compilation for styles using chbar + avoid pdf/postscripts repeat in the code
+# version 1.0.3: Bug fix: replace use of system('cp...') with File::Copy::copy (Patch contributed by D. Bremner)
+#                 Quotes around system call file arguments to allow filenames with spaces (Patch contributed by ssteve)
+# version 1.0.2: - option --so to use latexdiff-so
+# version 1.0.1 (change version numbering to match that of latexdiff)
+#   - Option --fast to use latexdiff-fast, 
+#   - git support (thanks to Bjorn Magnus Mathisen, Santi Béjar, Pietro Battiston and Stefan Alfredson for patches) - UNTESTED
+# version 0.25:
+#   - bbl is allowed as alternative extension (instead of .tex)
+# version 0.26a
+#   - Bug fix: it copes now correctly with the possibility that there are no changes between current
+#         and archived version
+use Getopt::Long ;
+use Pod::Usage qw/pod2usage/ ;
+use File::Temp qw/tempdir/ ;
+use File::Basename qw/dirname fileparse/;
+use File::Copy;
+use File::Path;
+
+use strict ;
+use warnings ;
+
+my $versionstring=<<EOF ;
+This is LATEXDIFF-VC 1.3.4
+  (c) 2005-2024 F J Tilmann
+EOF
+
+# output debug and intermediate files, set to 0 in final distribution
+my $debug=0; 
+
+
+# Option names
+my ($version,$help,$fast,$showconfig,$so,$postscript,$pdf,$onlychanges,$flatten,$force,$dir,$cvs,$rcs,$svn,$git,$hg,
+    $run,$rundvi2,$diffcmd,$patchcmd,@revs,@configlist);
+# for passing through config options to latexdiff:
+my ($configlatexdiff,@config,$config,$assign);
+
+# Preset Variables
+#my $latexdiff="latexdiff"; # Program for making the comparison
+#my $latexcmd="latex"; # latex compiler if not further identified
+my $vc="";
+my $tempdir=tempdir(CLEANUP => 1);   # generate a temp dir, which will automatically be deleted at program exit
+#printf STDERR "DEBUG: tempdir $tempdir\n";
+
+# Variables
+my ($file1,$file2,$diff,$diffbase,$rootdir,$answer,$options,$infile,$append,$dirname,$cwd);
+my (@files,@ldoptions,@tmpfiles,@ptmpfiles,@difffiles,$extracomp); # ,
+
+
+# keep commands to be used in config hash
+my (%CFG)=( 
+   LATEX     => "latex",
+   LATEXDIFF => "latexdiff",
+   DVI2      => "dvips",
+   BIBTEX    => "bibtex"
+   );
+
+
+
+Getopt::Long::Configure('pass_through','bundling');
+
+GetOptions('revision|r:s' => \@revs,
+           'cvs' => \$cvs,
+           'rcs' => \$rcs,
+           'svn' => \$svn,
+           'git' => \$git,
+           'hg' => \$hg,
+           'dir|d:s' => \$dir,
+	   'run' => \$run,
+	   'dvi' => \$rundvi2,
+	   'fast' => \$fast,
+	   'so' => \$so,
+           'postscript|ps' => \$postscript,
+           'pdf' => \$pdf,
+           'force' => \$force,
+	   'only-changes' => \$onlychanges,
+	   'flatten:s' => \$flatten,
+	   'config|c=s' => \@configlist,
+	   'show-config' => \$showconfig,
+           'version' => \$version,
+	   'help|h' => \$help,
+	   'debug!' => \$debug);
+
+##print STDERR "DEBUG 1:revs($#revs): " . join(":",@revs) . "\n";
+##print STDERR "DEBUG 1:ARGV($#ARGV): " . join(":",@ARGV) . "\n";
+
+$extracomp = join(" ",grep(/BAR/,@ARGV)); # special latexdiff options requiring additional compilation
+
+if ( $help ) {
+  pod2usage(1) ;
+}
+
+if ( $version ) {
+  print STDERR $versionstring ;  
+  exit 0; 
+}
+
+if ( $so ) {
+  $CFG{LATEXDIFF}='latexdiff-so';
+}
+if ( $fast ) { 
+  die "Cannot specify more than one of --fast or --so " if ($so);
+  $CFG{LATEXDIFF}='latexdiff-fast';
+}
+
+if ( $cvs ) {
+  $vc="CVS";
+}
+if ( $rcs ) {
+  die "Cannot specify more than one of --cvs, --rcs, --svn --git or --hg." if ($vc);
+  $vc="RCS";
+}
+if ( $svn ) {
+  die "Cannot specify more than one of --cvs, --rcs, --svn --git or --hg." if ($vc);
+  $vc="SVN";
+}
+if ( $git ) {
+  die "Cannot specify more than one of --cvs, --rcs, --svn --git or --hg." if ($vc);
+  $vc="GIT";
+}
+if ( $hg ) {
+  die "Cannot specify more than one of --cvs, --rcs, --svn --git or --hg." if ($vc);
+  $vc="HG";
+}
+
+if ( $pdf ) {
+  $CFG{LATEX} = "pdflatex";
+  $run=1;
+} elsif ( $postscript ) {
+  $CFG{LATEX} = "latex";
+  $rundvi2=1;
+}
+
+if ( $rundvi2 ) {
+  $run=1;
+}
+
+# setting config variables from command line. Explicitly set configuration options take precedence over convenience options,
+# therefore this is executed last
+@config=();
+foreach $config ( @configlist ) {
+  if (-f $config || lc $config eq '/dev/null' ) {
+    open(FILE,$config) or die ("Couldn't open configuration file $config: $!");
+    while (<FILE>) {
+      chomp;
+      next if /^\s*#/ || /^\s*%/ || /^\s*$/ ;
+      push (@config,$_);
+    }
+    close(FILE);
+  }
+  else {
+#    foreach ( split(",",$config) ) {
+#      push @config,$_;
+#    }
+     push @config,split(",",$config)
+  }
+}
+
+$configlatexdiff="";
+foreach $assign ( @config ) {
+    $assign=~ m/\s*(\w*)\s*=\s*(.*?)\s*$/ 
+      or die "Illegal assignment |$assign| in configuration list (must be variable=value)";  
+  if ( defined($CFG{$1})) {
+    # known latexdiff-vc option
+    $CFG{$1}=$2;
+  } else {
+    # unknown option, assume it is a latexdiff configuration variable
+    $configlatexdiff .= "$1=$2,";
+  }
+}
+# delete last comma inv $configlatexdiff if present
+$configlatexdiff =~ s/,$// ;
+
+if ($showconfig) {
+  print "# Configuration variables (latexdiff-vc):\n";
+  foreach ( keys %CFG ) {
+    print "$_=$CFG{$_}\n";
+  }
+  exit 0;
+}
+
+
+# check whether the first file name or first passed-through option for latexdiff got misinterpreted as an option to an empty -r option
+if ( @revs && ( -f $revs[$#revs] || $revs[$#revs] =~ /^-/ ) ) {
+  push @ARGV,$revs[$#revs];
+  $revs[$#revs]="";
+}
+# check whether the first file name or first passed-through option for latexdiff got misinterpreted as an option to an empty -d option
+if ( defined($dir) && ( -f $dir || $dir =~ /^-/ ) ) {
+  push @ARGV,$dir;
+  $dir="";
+}
+# check whether the first file name or first passed-through option for latexdiff got misinterpreted as an option to an empty --flatten option
+
+if ( defined($flatten) && ( -f $flatten || $flatten =~ /^-/ ) ) {
+  unshift @ARGV,$flatten;
+  $flatten="";
+}
+
+print STDERR "DEBUG: PDF $pdf latexdiff-vc command line: ", join(" ",@ARGV), "\n" if $debug; 
+
+$file2=pop @ARGV;
+( defined($file2) && $file2 =~ /\.(tex|bbl|flt)$/ ) or pod2usage("Must specify at least one tex, bbl or flt file");
+
+if (! $vc && scalar(@revs)>0 ) {
+  # have to guess $vc
+  # check whether we have a special name
+  if ( $0 =~ /-cvs$/ ) {
+    $vc="CVS";
+  } elsif ( $0 =~ /-rcs$/ ) {
+    $vc="RCS";
+  } elsif ( $0 =~ /-svn$/ ) {
+    $vc="SVN";
+  } elsif ( $0 =~ /-git$/ ) {
+    $vc="GIT";
+  } elsif ( $0 =~ /-hg$/ ) {
+    $vc="HG";
+  } elsif ( -e "$file2,v" || -d "RCS" ) {
+    print STDERR "Guess you are using RCS ...\n";
+    $vc="RCS";
+  } elsif ( -d ".svn" ) {
+    print STDERR "Guess you are using SVN ...\n";
+    $vc="SVN";
+  } elsif ( -d ".git" ) {
+    print STDERR "Guess you are using GIT ...\n";
+    $vc="GIT";
+  } elsif ( -d ".hg" ) {
+    print STDERR "Guess you are using HG ...\n";
+    $vc="HG";
+  } elsif ( -e "CVSROOT" || defined($ENV{"CVSROOT"}) ) {
+    print STDERR "Guess you are using CVS ...\n";
+    $vc="CVS";
+  } else {
+    print STDERR "Cannot figure out version control system, so I default to CVS\n";
+    $vc="CVS";
+  }
+}
+
+if (defined($dir) && $dir=~/^\.\/?/ ) {
+  print STDERR "You wrote -dir=.  but you do not really like to do that, do you ?\n";
+  exit 10
+}
+
+if ( scalar(@revs)>0 ) {
+  if ( $vc eq "CVS" ) { 
+    $diffcmd  = "cvs diff -u -r"; 
+    $patchcmd = "patch -R -p0";
+  } elsif ( $vc eq "RCS" ) {
+    $diffcmd  = "rcsdiff -u -r";  
+    $patchcmd = "patch -R -p0";
+  } elsif ( $vc eq "SVN" ) {
+    $diffcmd  = "svn diff --internal-diff -r ";  
+    $patchcmd = "patch -R -p0";
+  } elsif ( $vc eq "GIT" ) {
+    $diffcmd  = "git diff -r --relative --no-prefix ";  
+    $patchcmd = "patch -R -p0";
+    # alternatively:
+    # $diffcmd  = "git diff ";
+    # $patchcmd = "patch -R -p1";
+  } elsif ( $vc eq "HG" ) {
+    $diffcmd  = "hg diff --root . -r";
+    $patchcmd = "patch -R -p1";
+  } else {
+    print STDERR "Unknown versioning system $vc \n";
+    exit 10;
+  }
+}
+
+
+# make file list (last arguments), initial arguments have to be passed to latexdiff
+# We assume an argument is a valid file rather than a latexdiff argument
+# if it has extension .tex, .bbl or .flt (output of flatex), and does not start with a - or --
+
+@files=($file2);
+while( $file1=pop @ARGV ) {
+  if ( $file1 =~ /\.(tex|bbl|flt)$/ && $file1 !~ /^-/) {
+    # $file1 looks like a valid file name and is prepended to file list
+    unshift @files, $file1 ;
+  } else {
+    # $file1 looks like an option for latexdiff, push it back to argument stack
+    unshift @ldoptions, $file1 ;
+  }
+}
+
+if ( length $configlatexdiff >0 ) {
+  push @ldoptions, "--config=$configlatexdiff";
+}
+
+if ( defined($flatten) ) {
+  push @ldoptions, "--flatten" ;
+}
+
+if ( defined($debug) && $debug ) {
+  push @ldoptions, "--debug" ;
+}
+
+# impose ZLABEL subtype if --only-changes option
+if ( $onlychanges ) {
+  push @ldoptions, "-s", "ZLABEL","-f","IDENTICAL","--no-links", "--graphics-markup=none" ;
+}
+
+if ( scalar(@revs) == 0 ) {
+  pod2usage("When -r option is not used, two .tex files (old and new) must be given on the command line") unless @files==2;
+  warn "Option -flatten should normally be combined with -r option (Results will probably not be as expected)" if defined($flatten);
+    #  compare two files
+  $file1=shift @files ;
+} else {
+  # revision control
+  if ( defined($flatten) ) {
+    pod2usage("Only one root file must be given  if --flatten option is used with version control") if  @files != 1;
+    $tempdir='.' if ( $flatten eq "keep-intermediate" );
+    print STDERR "flatten tempdir |$tempdir|\n" if $debug;
+    if ( $vc eq "SVN" ) {
+      my (@infoout)=`svn info`;
+      my (@urlline) = grep(/^URL:/, @infoout);
+      @urlline ==1 or die "Ambiguous output from svn info command."; 
+      ( $rootdir = $urlline[0] ) =~ s/URL:\s*// ;
+      chomp $rootdir;
+    } elsif ( $vc eq "GIT" || $vc eq "HG" ) {
+      # do nothing
+    } else {
+      die "--flatten option with revision control only possible with SVN, HG and GIT modes";
+    }
+  }
+}
+
+if ( scalar(@revs) == 2 ) {
+  $append = "-diff$revs[0]-$revs[1]";
+} elsif ( scalar(@revs) == 1 || $revs[0] ) {
+  $append = "-diff$revs[0]";
+} else {
+  $append = "-diff";
+}
+
+if ( defined($pdf) or defined($postscript) ) {
+  # when using options such as -r HEAD~1  the ~1 is ignored by pdflatex in generating output file names; this could
+  # be confusing and actually causes error to only-changes mode
+  $append =~ s/~/_/g ;
+}
+
+if ( defined ($dir) && ! $dir ) {
+  # bare -d option => choose directory name
+  ($dir=$append) =~ s/^-//;
+}
+
+if ( ($vc eq "SVN" || $vc eq "CVS") && scalar(@revs)) {
+  length($revs[$#revs]) > 0 or $revs[$#revs]="HEAD";
+  length($revs[0]) > 0 or $revs[0]="HEAD";
+}
+
+if ($vc eq "HG" && scalar(@revs)) {
+  length($revs[$#revs]) > 0 or $revs[$#revs]="tip";
+  length($revs[0]) > 0 or $revs[0]="tip";
+}
+
+print STDERR "DEBUG LDOptions($#ldoptions): " . join(":",@ldoptions) . "\n" if $debug;
+
+
+# cycle through all files
+@difffiles=();
+
+while ( $infile=$file2=shift @files ) {
+  print STDERR "Working on  $infile \n";
+  if ( scalar(@revs) == 1 ) {
+    if ( defined($flatten) ) {
+      my $olddir=$tempdir . "/latexdiff-vc-$revs[0]";
+      print STDERR "Checking out old dir into: $olddir (rev: $revs[0])\n";
+      checkout_dir($revs[0],$olddir,$infile);
+      $file1=$olddir ."/".$infile;
+
+      # generate bibliography for new file if 
+      if ( greptex('^[^%]*\\\\bibliography\\{',$infile) == 0)  {
+	my ($filebase,$filedir)=fileparse($infile,".tex");
+	if ( ! -e "$filedir$filebase.bbl" ) {
+	  system("$CFG{LATEX} -draftmode -interaction=batchmode '$infile'; $CFG{BIBTEX} '$filedir$filebase'")
+	      or die  "Something went wrong in executing:  $CFG{LATEX} -draftmode -interaction=batchmode '$infile'; $CFG{BIBTEX} '$filedir$filebase'" ;
+	}
+      }
+    } else {
+      ($file1=$infile) =~ s/\.(tex|bbl|flt)/-oldtmp-$$.$1/ ;
+      push @tmpfiles,$file1;
+      # compare file with previous version ($revs[0]="") or specified version
+      ### system("$diffcmd$revs[0] $infile| $patchcmd -o$file1") ;
+      if (system("$diffcmd$revs[0] \"$infile\" | $patchcmd -o\"$file1\"")==0  and -z $file1 ) {
+        # no differences detected, i.e. file is equal to current version
+	copy($infile,$file1) || die "copy($infile,$file1) failed: $!";
+      }
+    }
+  } elsif ( scalar(@revs) == 2 ) {
+    if ( defined($flatten) ) {
+      my $olddir=$tempdir . "/latexdiff-vc-$revs[0]";
+      print STDERR "Checking out old dir into: $olddir (rev: $revs[0])\n";
+      checkout_dir($revs[0],$olddir,$infile);
+      $file1=$olddir ."/".$infile;
+
+      my $newdir=$tempdir . "/latexdiff-vc-$revs[1]";
+      print STDERR "Checking out new dir into: $newdir\n";
+      checkout_dir($revs[1],$newdir,$infile);
+      $file2=$newdir ."/".$infile;
+    } else {
+      ($file1=$infile) =~ s/\.(tex|bbl|flt)/-oldtmp-$$.$1/ ;
+      $file2 =~ s/\.(tex|bbl|flt)/-newtmp-$$.$1/ ;
+      push @tmpfiles,$file2;
+      if (system("$diffcmd$revs[1] $infile | $patchcmd -o$file2")==0 and -z $file2 ) {
+	copy($infile,$file2) || die "copy($infile,$file2) failed: $!";
+      }
+      if (system("$diffcmd$revs[0] $infile | $patchcmd -o$file1")==0 and -z $file1 ) {
+	copy($infile,$file1) || die "copy($infile,$file1) failed: $!";
+      };
+    }
+  }
+
+  if ( -z $file1 || -z $file2) {
+    print STDERR "One or both of the files to compare are empty. Possibly something went wrong in the retrieval of older versions. Aborting ...\n" ;
+    exit(10);
+  }
+
+  # Get name of difference file
+  if ( defined($dir) ) {
+    $diff="$dir/$infile" ; 
+  } else {
+    ($diff=$infile) =~ s/\.(tex|bbl|flt)$/$append.$1/ ;
+  }
+
+  # make directories if needed
+  $dirname=dirname($diff) ;
+  system("mkdir -p $dirname") unless ( -e $dirname );
+
+  # Remaining options are passed to latexdiff
+  if (scalar(@ldoptions) > 0 ) {
+    $options = "\'" . join("\' \'",@ldoptions) . "\'";
+  } else {
+    $options = "";
+  }
+   
+  if ( -e $diff && ! $force ) {
+    print STDERR "OK to overwrite existing file $diff (y/n)? ";
+    $answer = <STDIN> ;
+    unless ($answer =~ /^y/i ) {
+      unlink @tmpfiles;
+      die "Abort ... " ;
+    }
+  }
+  
+  print STDERR "Running: $CFG{LATEXDIFF}  $options \"$file1\" \"$file2\" > \"$diff\"\n";
+  unless ( system("$CFG{LATEXDIFF} $options \"$file1\" \"$file2\" > \"$diff\"") == 0 ) { 
+    print STDERR  "Something went wrong in $CFG{LATEXDIFF}. Deleting $diff and abort\n" ; unlink $diff ; exit(5) 
+  };
+  print "Generated difference file $diff\n";
+   
+  if ( $run and !( scalar(@revs) && greptex( qr/\\document(?:class|style)/ , $diff,25) ) ) {
+    # save filename for later processing if postscript or pdf conversion is requested and either two-file mode was used (scalar(@revs)==0) or the diff file contains documentclass statement (ie. is a root document)
+    push @difffiles, $diff ;
+  }
+  
+  unlink @tmpfiles;
+}
+
+foreach $diff ( @difffiles ) {
+  chomp($cwd=(`pwd`));
+  if  (defined($dir)) {
+    ( $diff =~ s/$dir\/?// ) ;
+    chdir $dir ; 
+  }
+  @ptmpfiles=();
+  ( $diffbase=$diff) =~ s/\.(tex)$// ;
+
+  # adapt magically changebar styles to [pdftex] display driver if pdf output was selected
+  if ( $pdf ) {
+    system("sed \"s/Package\\[dvips\\]/Package[pdftex]/\" \"$diff\" > \"$diff.tmp$$\" ; \\mv \"$diff.tmp$$\" \"$diff\"");
+  }
+
+  if ( $run ) {
+    print STDERR "PDF: $pdf Postscript: $postscript cwd $cwd\n" if $debug;
+
+#    if ( system("grep -q \'^[^%]*\\\\bibliography{\' \"$diff\"") == 0 ) { 
+    if ( greptex('^[^%]*\\\\bibliography{',$diff ) == 0 ) { 
+      system("$CFG{LATEX} -draftmode -interaction=batchmode \"$diff\"; $CFG{BIBTEX} \"$diffbase\";");
+      push @ptmpfiles, "$diffbase.bbl","$diffbase.bbl" ; 
+    }
+
+    # if special needs, as CHANGEBAR
+    if ( $extracomp ) {
+      # print "Extracomp\n";
+      system("$CFG{LATEX} -draftmode -interaction=batchmode \"$diff\";");
+    }
+
+    # final compilation
+    system("$CFG{LATEX} -draftmode -interaction=batchmode \"$diff\";"); # needed if cross-refs
+    system("$CFG{LATEX} \"$diff\";"); # final, with possible error messages
+
+    if ( $rundvi2 ) {
+      my $dvi="$diffbase.dvi";
+      #my $ps="$diffbase.ps";
+      my $ppoption="";
+      
+      if ( $onlychanges ) {
+	$ppoption="-pp ".join(",",findchangedpages("$diffbase.aux"));
+      }
+      #system("$CFG{DVI2} $ppoption -o $ps $dvi");
+      system("$CFG{DVI2} $ppoption $dvi");
+      push @ptmpfiles, "$diffbase.aux","$diffbase.log",$dvi ;
+      #print STDERR "Generated postscript file $ps\n";
+    } elsif ( $run ) {
+      if ( $onlychanges ) {
+	my @pages=compresspages(findchangedpages("$diffbase.aux"));
+        my $gs = `which gs`;
+        $gs =~ s/^\s+|\s+$//g;
+        my $qpdf = `which qpdf`;
+        $qpdf =~ s/^\s+|\s+$//g;
+        my $pdftk = `which pdftk`;
+        $pdftk =~ s/^\s+|\s+$//g;
+        my $command;
+        if (-x $gs && `$gs --version` >= 9.20) {
+          $command="gs -sDEVICE=pdfwrite -dNOPAUSE -dBATCH -dSAFER -sPageList=" . join(",", @pages) . " -sOutputFile=\"$diffbase-changedpage.pdf\" \"$diffbase.pdf\"";
+        } elsif (-x $pdftk) {
+          $command="pdftk \"$diffbase.pdf\" cat " . join(" ",@pages) . " output \"$diffbase-changedpage.pdf\"";
+        } elsif (-x $qpdf) {
+          $command="qpdf --linearize \"$diffbase.pdf\" --pages \"$diffbase.pdf\" " . join(",", @pages) . " -- \"$diffbase-changedpage.pdf\" ";
+        } else {
+          die ("could not find any of gs, pdftk or qpdf");
+        }
+        print "Executing ".$command;
+        system($command) == 0
+          or die("could not execute <".$command."> to strip pages. Return code: $?");
+	move("$diffbase-changedpage.pdf","$diffbase.pdf");
+      }
+      push @ptmpfiles, "$diffbase.aux","$diffbase.log";
+    }
+  }
+  unlink @ptmpfiles;
+  chdir $cwd;
+}
+
+# findchangedpages($auxfile)
+# returns a list of the page numbers on which changes were marked with the ZLABEL subtype of latexdiff
+sub findchangedpages {
+  my ($auxfile) =@_;
+  my $j;
+  my (%pages);		# note that I use a hash with page numbers as keys and arbitrary values - this way I can use Perl's built-in functions to get rid of duplicates
+  my %start;
+  open(AUX,$auxfile) or die ("Could open aux file $auxfile . System error: $!");
+  while (<AUX>) {
+    if (m/\\zref\@newlabel\{DIFchgb(\d*)\}\{.*\\abspage\{(\d*)\}.*\}/ ) { 
+      $start{$1}=$2; $pages{$2}=1;
+    }
+    if (m/\\zref\@newlabel\{DIFchge(\d*)\}\{.*\\abspage\{(\d*)\}.*\}/) { 
+      if (defined($start{$1})) {
+	for ($j=$start{$1}; $j<=$2; $j++) {
+	  $pages{$j}=1;
+	}
+      } else {
+	$pages{$2}=1;
+      }
+    }
+  }
+  close(AUX);
+  # sort pages numerically (they should be sorted already; this is just to make sure
+  return(sort {$a <=> $b} keys(%pages));
+}
+
+# inspired by a python version posted at https://stackoverflow.com/a/54714846/8136338
+# compresspages(@pages)
+# @pages: sorted array of unique integers (pages)
+# return a list of pages and page ranges. 3 or more consecutive numbers are merged into a "begin-end" string
+#
+# example: join(",", compresspages([1,3,4,5,7,8])) -> join(",", [1,"3-5",7,8]) -> "1,3-5,7,8"
+sub compresspages {
+  my (@pages) = @_;
+  my @res;
+  my $begin=$pages[0];
+  my $end =$pages[0];
+  my $page;
+  foreach $page ( @pages ) {
+    next if ($page == $begin);
+    if ($page == $end+1) { # handle continuous pages
+      $end = $page;
+      next;
+    } elsif ($begin == $end) { # push single continuous page
+      push @res, $begin;
+      $begin = $end = $page;
+    } elsif ($begin + 1 == $end) { # push two continuous pages
+      push @res, $begin, $end;
+      $begin = $end = $page;
+    } else { # push multiple continuous pages
+      push @res, "$begin-$end";
+      $begin = $end = $page;
+    }
+  }
+  # Finally just check in the same manner for the end element
+  if ($begin == $end) {
+    push @res, $begin;
+  } elsif ($begin + 1 == $end) {
+    push @res, $begin, $end;
+  } else {
+    push @res, "$begin-$end";
+  }
+  return @res;
+}
+
+# checkout_dir(rev,dirname, <file>)
+# checks out revision rev and stores it in dirname
+# If argument <file> is provided, check if this file imports a bibliography and generate it if necessary 
+# uses global variables: $vc, $rootdir, $CFG
+sub checkout_dir {
+  my ($rev,$dirname,$file)=@_;
+
+  my ($filebase,$filedir);
+  unless (-e $dirname) { mkpath([ $dirname ]) or die "Cannot mkdir $dirname ." ;}
+  if ( $vc eq "SVN" ) {
+    system("svn checkout -r $rev $rootdir $dirname")==0 or die "Something went wrong in executing:  svn checkout -r $rev $rootdir $dirname";
+  } elsif ( $vc eq "GIT" ) {
+    $rev="HEAD" if length($rev)==0;
+    system("git archive --format=tar $rev | ( cd $dirname ; tar -xf -)")==0 or die "Something went wrong in executing:  git archive --format=tar $rev | ( cd $dirname ; tar -xf -)";
+  } elsif ( $vc eq "HG" ) {
+    system("hg archive --type files -r $rev $dirname")==0 or die "Something went wrong in executing:  hg archive --type files -r $rev $dirname";
+  } else {
+    die "checkout_dir: only works with SVN, HG and GIT VCS system (selected: $vc)";
+  }
+  # Check if the main file needs a bbl generated
+  if ( defined($file) and greptex('^[^%]*\\\\bibliography\\{',$file) == 0)  {
+      ($filebase,$filedir)=fileparse($file,".tex");
+      if ( ! -e "$filedir$filebase.bbl" ){
+ 	  printf STDERR "Running $CFG{BIBTEX} to generate $filedir$filebase.bbl.\n";
+	  system("cd '$dirname'; $CFG{LATEX} -draftmode -interaction=batchmode '$file'; $CFG{BIBTEX} '$filedir$filebase'")==0
+	      or die  "Something went wrong in executing:  cd '$dirname'; $CFG{LATEX} -draftmode -interaction=batchmode '$file'; $CFG{BIBTEX} '$filedir$filebase'" ;
+      }
+#      else { # DEBUG
+#	  printf STDERR "DEBUG: Skipping bbl file generation as $filedir$filebase.bbl already exists.\n"
+#      } #DEBUG
+  }
+}
+
+# greptex($regex,$filename, <lines>) returns 1 if regex is not matched in filename
+# only the first <lines> non-comment lines are scanned; by default everythin is scanned
+# 0 if there is a match
+sub greptex {
+  my ($regex,$filename,$nlines)=@_;
+  my ($i)=0;
+  open (FH, $filename) or die("Couldn't open $filename: $!");
+  while (<FH>) {
+    next if /^\s*%/;    # skip comment lines
+    if ( m/$regex/ ) {
+      close(FH);
+      return(0);
+    }
+    # only scan <nlines> lines, if optional parameter proviede
+    $i++;    
+    last if defined($nlines) and $i>$nlines ;
+  }
+  close(FH);
+  return(1);
+}
+
+
+=head1 NAME
+
+latexdiff-vc - wrapper script that calls latexdiff for different versions of a file under version management (CVS, RCS or SVN)
+
+=head1 SYNOPSIS
+
+B<latexdiff-vc> [ F<latexdiff-options> ] [ F<latexdiff-vc-options> ]  B<-r> [F<rev1>] [B<-r> F<rev2>]  F<file1.tex> [ F<file2.tex> ...]
+
+ or
+
+B<latexdiff-vc> [ F<latexdiff-options> ]  [ F<latexdiff-vc-options> ][ B<--postscript> | B<--pdf> ]  F<old.tex> F<new.tex>
+
+=head1 DESCRIPTION
+
+I<latexdiff-vc> is a wrapper script that applies I<latexdiff> to a
+file, or multiple files under version control (git, subversion (SVN), mercurial (hg), CVS, RCS), and optionally runs the
+sequence of C<latex> and C<dvips> or C<pdflatex> commands necessary to
+produce pdf or postscript output of the difference tex file(s). It can
+also be applied to a pair of files to automatise the generation of difference
+file in postscript or pdf format.
+
+=head1 OPTIONS
+
+=over 4
+
+=item B<--rcs>, B<--svn>, B<--cvs>, B<--git> or B<--hg>
+
+Set the version control system used. 
+If no version system is specified, latexdiff-vc will venture a guess.
+
+latexdiff-cvs, latexdiff-rcs etc are variants of latexdiff-vc which default to 
+the respective versioning system. However, this default can still be overridden using the options above.
+
+Note that hg needs to support the C<--root> option (version >= 2.9)
+
+=item B<-r>, B<-r> F<rev> or B<--revision>, B<--revision=>F<rev>
+
+Choose revision (under RCS, CVS, SVN, GIT or HG). One or two B<-r> options can be
+specified, and they result in different behaviour:
+
+=over 4
+
+=item B<latexdiff-vc> -r F<file.tex> ...
+
+compares F<file.tex> with the most recently checked-in version checked.
+
+=item B<latexdiff-vc> -r F<rev1> F<file.tex> ...
+
+compares F<file.tex> with revision F<rev1>.
+
+=item B<latexdiff-vc> -r F<rev1> -r F<rev2> F<file.tex> ...
+
+compares revisions F<rev1> and F<rev2> of F<file.tex>.
+
+Multiple files can be specified for all of the above options. All files must have the
+extensions C<.tex>, C<.bbl>, or C<.flt>, though.
+
+=item B<latexdiff-vc>  F<old.tex> F<new.tex>
+
+compares two files.
+
+=back
+
+The name of the difference file is generated automatically and
+reported to stdout.
+
+=item B<-d> or B<--dir>  B<-d> F<path> or B<--dir=>F<path>
+
+Rather than appending the string C<diff> and optionally the version
+numbers given to the output-file, this will prepend a directory name C<diff> 
+to the
+original filename, creating the directory and sub-directories should they not exist already.  This is particularly useful in order to clone a
+complete directory hierarchy.  Optionally, a pathname F<path> can be specified, which is prepended instead of C<diff>.
+
+=item B<--flatten,--flatten=keep-intermediate>
+
+If combined with C<--git>, C<--svn> or C<--hg> option or the corresponding modes, check out the revisions to compare in a separate temporary directory, and then pass on option C<--flatten> to latexdiff. The directory in which C<latexdiff-vc> is invoked defines the subtree which will be checked out.
+Note that if additional files are needed which are not included in the flatten procedure (package files, included graphics), they need to be accessible in the current directory. If you use bibtex, it is recommended to include the C<.bbl> file in the version management. From latexdiff-vc >=1.3.4  there is an attempt to create the C<.bbl> file by running bibtex in the retrieved subdirectories; note that the bibtex is run in the temporary subdirectories so any .bib or .bst files either need to be under version management, or in the global search paths for bibtex (shell environment variables C<BIBINPUTS> and C<BSTINPUTS>). 
+
+The generic usage of this function is : C<latexdiff-vc --flatten -r rev1 [-r rev2] master.tex> where master.tex is the project file containing the highest level of includes etc.
+
+With C<--flatten=keep-intermediate>, the intermediate revision snapshots are kept in the current directory (Default is to store them in a temporary directory and delete them after generating the diff file.)
+
+
+=item B<--only-changes>
+
+Post-process the output such that only pages with changes on them are displayed. This requires the use of subtype ZLABEL 
+in latexdiff, which will be set automatically, but any manually set -s option will be overruled (also requires zref package to 
+be installed). This option also disables internal links (as implemented by hyperref package) and graphics markup.
+(note that this option must be combined with --ps or --pdf to make sense)
+
+=item B<--force>
+
+Overwrite existing diff files without asking for confirmation. Default 
+behaviour is to ask for confirmation before overwriting an existing difference
+file.
+
+=item B<--run>
+
+run latex command on diff file after generation of diff file.
+
+=item B<--dvi>
+
+run latex and dvixxx commands after generation of diff file.
+
+=item B<-c configfile>
+=item B<--config var1=val1,var2=val2,...> or B<-c var1=val1,..>
+
+Set configuration variables for latexdiff and latexdiff-vc.  The option can be repeated to set different
+variables (as an alternative to the comma-separated list).
+Available variables for latexdiff-vc:
+
+=over 8
+
+=item C<LATEXDIFF> latexdiff command (e.g. latexdiff-fast, latexdiff-so). This command should support the options C<-draftmode> and C<-interaction=batchmode>
+
+=item C<LATEX> latex command (e.g. pdflatex, lualatex)
+
+=item C<DVI2>  Command for conversion of dvi file (e.g. dvips, dvipdf)
+
+=item C<BIBTEX> Command replacing bibtex
+
+=back
+
+All other config variables are passed to latexdiff. Explicitly set configuration changes always override implicit
+changes by the following shortcut options B<--fast>, B<--so>, B<--ps> and B<--pdf>. 
+
+=item B<--fast> or B<--so>
+
+Use C<latexdiff-fast> or C<latexdiff-so>, respectively (instead of C<latexdiff>).
+
+=item B<--ps> or B<--postscript>
+
+Generate postscript output from difference file.  This will run the
+sequence C<latex; latex; dvips> on the difference file (do not use
+this option in the rare cases, where three C<latex> commands are
+required if you care about correct referencing).  If the difference
+file contains a C<\bibliography> tag, run the sequence C<latex;
+bibtex; latex; latex; dvips>.
+
+=item B<--pdf>
+
+Generate pdf output from difference file using C<pdflatex>. This will
+run the sequence C<pdflatex; pdflatex> on the difference file, or
+C<pdflatex; bibtex; pdflatex; pdflatex> for files requiring bibtex.
+Note that this is not just a shortcut for setting configuration variable but also triggers 
+some special behaviour.
+
+
+=item B<--show-config>
+
+Show values of configuration variables.
+
+=item B<--help> or B<-h>
+
+Show help text
+
+=item B<--version>
+
+Show version number
+
+=back
+
+All other options are passed on to C<latexdiff>.  Note that it is not always easy for latexdiff to correctly assign arguments to options when short form (single letter) options are used so it is recommended to use long-form options. 
+
+=head1 SEE ALSO
+
+L<latexdiff>
+
+=head1 PORTABILITY
+
+I<latexdiff-vc> uses external commands and is therefore dependent on the system architecture; it has been
+tested mainly on Unix-like systems. It also requires a version control
+system and latex to be installed on the system to make use of all features.  Modules from Perl 5.8
+or higher are required.
+
+=head1 BUG REPORTING
+
+Please submit bug reports using the issue tracker of the github repository page I<https://github.com/ftilmann/latexdiff.git>, 
+or send them to I<tilmann -- AT -- gfz-potsdam.de>.  Include the version number of I<latexdiff-vc>
+(option C<--version>).
+
+=head1 AUTHOR
+
+Version 1.3.4a
+Copyright (C) 2005-2017 Frederik Tilmann
+
+This program is free software; you can redistribute it and/or modify
+it under the terms of the GNU General Public License Version 3
+Contributors: S Utcke, H Bruyninckx; some ideas have been inspired by git-latexdiff bash script.
+C. Junghans: Mercurial Support.
+
+=cut
+
diff --git a/scripts/latexdiff.pl b/scripts/latexdiff.pl
new file mode 100644
index 0000000..5df02b7
--- /dev/null
+++ b/scripts/latexdiff.pl
@@ -0,0 +1,5470 @@
+#!/usr/bin/env perl
+##!/usr/bin/perl -w
+# latexdiff - differences two latex files on the word level
+#             and produces a latex file with the differences marked up.
+#
+#   Copyright (C) 2004-22  F J Tilmann (tilmann@gfz-potsdam.de)
+#
+# Repository/issue tracker:   https://github.com/ftilmann/latexdiff
+# CTAN page:          http://www.ctan.org/pkg/latexdiff
+#
+#    This program is free software: you can redistribute it and/or modify
+#    it under the terms of the GNU General Public License as published by
+#    the Free Software Foundation, either version 3 of the License, or
+#    (at your option) any later version.
+#
+#    This program is distributed in the hope that it will be useful,
+#    but WITHOUT ANY WARRANTY; without even the implied warranty of
+#    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+#    GNU General Public License for more details.
+#    You should have received a copy of the GNU General Public License
+#    along with this program.  If not, see <http://www.gnu.org/licenses/>.
+#
+# Detailed usage information at the end of the file
+#
+#
+# Note references to issue numbers are for the github repository of latexdiff: https://github.com/ftilmann/latexdiff
+#
+# Version 1.3.4:
+# New features:
+#  - Option to use lua-ul instead of ulem (for use with LuaLaTeX) (fix #17, #60, #188, #255, #270)
+# Enhancement
+#  - If amsmath is detected as one of the included (based on idea in PR #263 contributed by github user xlucn
+# Bug fix
+#  - sometimes the introduction of aux commands to show deleted lists or descriptions leaves in place empty auxiliary
+#    list environments, which would cause error messages (though skipping messages would result in a correct output pdf).
+#    These are now removed.
+#  - add more mboxsafecmd and safecmd commands for SIunitx to stay compatible with newer versionsof SIunitx (PR #283, fixing issue #282, contributed by github user joe6302413)
+#  - File added via --preamble option is no longer assumed to be ASCII, but read either with encoding as defined by --encoding option or using the encoding of the LOCALE (fixes issue #285 )
+#  - multicolumn argument is now treated as text
+#  - when tikz-dependency package is used, \& is no longer a safe command as it has special meaning inside dependency environment.  The fix is a little of a hack as really it should only be considered unsafe within dependency environment (fixes (mostly) issue #303 )
+#  - listings package had trouble with non-ASCII chars. The encoding is now set (thanks to github user anka-213 for finding this). Fixes #304
+#
+# Version 1.3.3:
+# New features:
+#  - Option --no-del to remove all deleted text (merge contributed by github user tdegeus PR #252, fixing issue #66
+#
+# Bug fixes:
+#  - Abbreviations involving punctuations within them. They need special treatment because otherwise in some
+#    circumstances the gnoring of white space differences in conjunction with merging according to MINWORDSBLOCK rule
+#    could turn 'i.e.' into 'i.\PAR e.' (see https://github.com/ftilmann/latexdiff/issues/269). A few abbreviations
+#    are now hard-coded and treated as atomic:
+#    English: i.e., e.g.   Deutsch: z.B.
+#    (fixes issue #269)
+#  - In WHOLE and COARSE math modes, now properly treat math environments with arguments such as \alignat. Fixes #251
+#  - For FINE math mode, multiple improvments to the processing work flow yield more robust outcomes. In particular, changes
+#    to the equation type, e.g. \begin{displaymath} -> \begin{equation}  without modifications  now usually no longer result
+#    in errors. (Partially) fixes issues #235  and #244.
+#  - When encountering deleted math array environments such as align or eqnarray, rather than replacing them with  a
+#    fixed replacement environment (e.g. align* or eqnarray*), an asterisk is now added to the original command, which
+#    in amsmath (and with eqnarray) will result in the same environment but without line numbers. Config variable MATHARRREPL
+#    is therefore (nearly) redundant, and a depracation warning is given when it is set. Reference to MATHARRREPL are have
+#    been removed from the manual (there is one exception, when it's still being used: sometimes latexdiff can figure out
+#    that there is a deleted array environment, but does not know which one. In this case, MATHARRREPL is still being used
+#    to encapsulate these parts of the source, and therefore it is still set internally.  But this is a quite rare situation).
+#    Fixes issue #216
+#  - Unlike 'array' environment, 'split' (amsmath) does not work in argument of \DIFadd or \DIFdl in UNDERLINE modes; therefore remove it from ARRENV configuration variable.
+#     Exclude \begin and \end in math environments in COARSE and WHOLE modes. Fixes #258. Fixes #109
+#  - --flatten now works for empty files. Fixes issue #242
+#  - improved processing of Chinese and Japanese texts in that splitting is done based on characters. Thanks to LuXu (Oliver Lew) in git for working this out. Fixes #229, fixes #145
+
+
+
+# Version 1.3.2
+# API adaptions:
+#  - latexdiff now completes with exit code 0 after --help or --version command (see issue #248)
+# New features / feature extensions
+#  - extend CUSTOMDIFCMD related postprocessing to deal properly with multiline commands, or a sequence of several commands in the same line (see github issue #204)
+#  - Support for additional macros from import package (\import, \inputfrom, \includefrom, \subimport,\subinputfrom, \subincludefrom). Provided by janniklasrose in PR #243 (fixes #239)
+#  - replace default driver dvips->pdftex
+# Bug fixes:
+#  - fix issue #206 affecting proper markup of text commands which are not also safe cmd's and have multiple arguments
+#  - fix issue #210 by adding \eqref (amsmath package) to the list of safe commands
+#  - fix bug reported in issue #168 mangled verbatim line environment
+#  - fix bug reported in issue #218 by replacing \hspace{0pt} after \mbox{..} auxiliary commands with \hskip0pt.
+#  - more ways to process \frac correctly with atomic arguments (committed by julianuu PR #246
+#  - fix a bug in biblatex mode, which prevented proper processing of modified \textcite (see: https://tex.stackexchange.com/questions/555157/latexdiff-and-biblatex-citation-commands)
+#  - -h string fix: add -driver option
+#
+# Version 1.3.1.1
+#  - remove spurious \n to fix error: Unknown regexp modifier "/n" at .../latexdiff line 1974, near "=~ " (see github issue #201)
+#
+# Version 1.3.1
+#    Bug fixes:
+#      - remove some uninitialised variable $2 warnings in string substitution in flatten function in case included file is not found
+#      - add minimal postprocessing to diff processing of preamble commands (replace \RIGHTBRACE by \} )
+#      - pre-processing: replace (contributed) routine take_comments_and_enter_from_frac() with take_comments_and_newline_from_frac(), which does the same thing
+#        (remove whitespace characters and comments between the argument of \frac commands) in an easier and more robust way. In addition, it
+#        will replace commands like \frac12 with \frac{1}{2} as pre-processing step.   Fixes issue #184
+#      - add "intertext" to list of unsafe math commands @UNSAFEMATHCMD . Fixes issue #179
+#      - provide citation command patterns for biblatex and protect them with mbox'es. Fixes issue #199
+#      - hardcode number of parameters for \href and \url commands to allow spaces between commands and arguments even if --allow-spaces option is not used (this
+#        is needed because some bibliography styles add such in-command-sequence spaces)  Fixes issues: #178 #198
+#      - bibitem is now kept even in deleted blocks such that deleted references show up properly (this implies that the actual numbers in numerical referencing schemes will change)
+#        (this is implemented by introducing a new class of commands KEEPCMD , which are kept as is in deleted environments (no effect in added environments). Currently
+#        \bibitem   is hardwired to be the only member of this class  (fixes issue #194, #174)
+#    Features:
+#      - add some special processing for revtex bibliography commands, so that the spaces between bibliography commands \bibfield and \bibinfo and their arguments are ignored.
+#         (fixes issue #194, should fix #174)
+#
+# Version 1.3.0 (7 October 2018):
+#    - treat options to \documentclass as potential package names (some packages allow implicit loading of or imply selected packages
+#    - improved pattern matching: now allows nested angular brackets, and is no longer confused by escaped curly braces
+#    - improved pattern matching in COARSE mode: occasionally, the closing bracket or some other elements would be matched in an 'unnatural' way due to another sequence being more minimal in the computational sense, sometimes even causing errors due to tokens moving in or out of the scope of math environments. This is now discouraged by adding internal \DIFANCHOR commands (which are removed again in post-processing) (fixes issues reported via email by li_ruomeng .
+#    - verbatim and lstlisting environments are marked-up with line-by-line in a similar style to non-verbatim text (requires the listing package to be installed)
+#       (see new configuration variable VERBATIMLINEENV) (several issues and pull requests by jprotze)
+#    - --flatten: now supports \verbatiminput and \lstlistinput
+#     - --flatten: if file is not found, do not fail, simply warn and leave command unexpanded (inspired by issue #112).  Don't warn if file name contains #[0-9] as it is then most likely an argument within a command definition rather than an actual file (applies to \input, \subfile, \include commands)
+#     - added to textcmds: \intertext
+#    - new config variable CUSTOMDIFCMD to allow defining special versions of  commands  in added or deleted blocks (Pull request by github user jprotze)
+#    - added option -no-links (mostly for use by latexdiff-vc in only-changes modes) (Pull request by github user jprotze)
+#    - new option --filter-script to run both input through a pre-processing script (PR jasonmccsmith  #167)
+#      new option --no-filter-stderr to hide stderr output from filter-script (potentially dangerous, as this might hide malfunctioning of filter scripts)
+#    - --flatten now can deal with imports made using the import package {PR jasonmccsmith #173)
+#   Bug fixes:
+#    - pattern matching of \verb and \lstinline commands had an error which meant they would trigger on commands beginning with \verb.
+#    - In description environments, mark up item descriptions by effectively reating the insides of item commannds as text commands (fixes #161)
+#
+#
+# Version 1.2.1 (22 June 2017)
+#    - add "DeclareOldFontCommand" to styles using \bf or \sf old style font commands (fixies issue #92 )
+#    - improved markup: process lstinline commands in listings package correctly
+#      for styles using colour, \verb and \lstinline arguments are marked up with colour (blue for added, red for deleted)
+#    - bug fix: protecting inline math expressions for mbox did not work as intended (see stack exchange question: http://tex.stackexchange.com/questions/359412/compiling-the-latexdiff-when-adding-a-subscript-before-a-pmatrix-environment-cau)
+#    - bug fix: when deleted \item commands are followed immediately by unsafe commands, they were not restored properly
+#      (thanks to J. Protze for pull request) (pull request #89)
+#    - treat lstlisting and comment as equivalent to verbatim environment
+#      make environments that are treated like verbatim environments configurable (config variable VERBATIMENV)
+#      treat lstinlne as equivalent to verb command
+#      partially addresses issue #38
+#    - refactoring: set default configuration variables in a hash, and those that correspond to lists
+#    - feature: option --add-to-config used to amend configuration variables, which are regex pattern lists
+#    - bug fix: deleted figures when endfloat package is activated
+#    - bug fix: alignat environment now always processed correctly (fix issues #65)
+#    - bug fix: avoid processing of commands as potential files in routine init_regex_arr (fix issue #70 )
+#    - minimal feature enhancement: treat '@' as allowed character in commands (strictly speaking requires prior \makeatletter statement, but always assuming it to be
+#       @       a letter if it is part of a command name will usually lead to the correct behaviour (see http://tex.stackexchange.com/questions/346651/latexdiff-and-let)
+#    - new feature/bug fix: --flatten option \endinput in included files now respected but only if \endinput stands right at the beginning of the line (issue #77)
+#    - bug fix: flatten would incorrectly attempt to process commented out \include commands (from discussion in issue #77 )
+#    - introduce an invisible space (\hspace{0pt} after \mbox{..} auxiliary commands (not in math mode), to allow line breaks between added and deleted citations (change should not cause adverse behaviour otherwise)
+#
+# Version 1.2.0:
+#    - highlight new and deleted figures
+#    - bug fix in title mark-up. Previously deleted commands in title (such as \title, \author or \date) were marked up erroneously
+#    - (minor) bug fixes in new 1.1.1 features: disabled label was commented out twice, additional spaces were introduced before list environment begin and end commands
+#    - depracation fix: left brace in RegEx now needs to be escaped
+#    - add type PDFCOMMENT based on issue #49 submitted by github user peci1 (Martin Pecka)
+#    - make utf8 the default encoding
+#
+# Version 1.1.1
+#    - patch mhchem: allow ce in equations
+#    - flatten now also expands \input etc. in the preamble (but not \usepackage!)
+#    - Better support for Japanese ( contributed by github user kshramt )
+#    - prevent duplicated verbatim hashes (patch contributed by github user therussianjig, issue #36)
+#    - disable deleted label commands (fixes issue #31)
+#    - introduce post-processing to reinstate most deleted environments and all needed item commands (fixes issue #1)
+#
+# Version 1.1.0
+#    - treat diacritics (\",\', etc) as safe commands
+#    - treat \_ and \& correctly as safe commands, even if used without spacing to the next word
+#    - Add a BOLD markup type that sets added text in bold face (Contribution by Victor Zabalza via pull request )
+#    - add append-mboxsafecmd list option to be able to specify special safe commands which need to be surrounded by mbox to avoid breaking (mostly this is needed with ulem package)
+#    - support for siunitx and cleveref packages: protect \SI command in siunitx package and \cref,\Cref{range}{*} in cleveref packages (thanks to Stefan Pinnow for testing)
+#    - experimental support for chemformula, mhchem packages: define \ch and \ce in packages as safe (but not \ch,\cee in equation array environments) - these unfortunately will not be marked up (thanks to Stefan Pinnow for testing)
+#    - bug fix: packages identified correctly even if \usepackage command options extend over several lines (previously \usepackage command needed to be fully contained in one line)
+#    - new subtype ONLYCHANGEDPAGE outputs only changed pages (might not work well for floating material)
+#    - new subtype ZLABEL operates similarly to LABEL but uses absolute page numbers (needs zref package)
+#    - undocumented option --debug/--nodebug to override default setting for debug mode (Default: 0 for release version, 1: for development version
+#
+# Version 1.0.4
+#    - introduce list UNSAFEMATHCMD, which holds list of commands which cannot be marked up with \DIFadd or \DIFdel commands  (only relevant for WHOLE and COARSE math markup modes)
+#    - new subtype LABEL which gives each change a label. This can later be used to only display pages where changes
+#      have been made (instructions for that are put as comments into the diff'ed file) inspired by answer on http://tex.stackexchange.com/questions/166049/invisible-markers-in-pdfs-using-pdflatex
+#    - Configuration variables take into accout some commands from additional packages:
+#      tikzpicture environment now treated as PICTUREENV, and \smallmatrix in ARRENV (amsmath)
+#    - --flatten: support for \subfile command (subfiles package)  (in response to http://tex.stackexchange.com/questions/167620/latexdiff-with-subfiles )
+#    - --flatten: \bibliography commands expand if corresponding bbl file present
+#    - angled bracket optional commands now parsed correctly (patch #3570) submitted by Dave Kleinschmidt (thanks)
+#    - \RequirePackage now treated as synonym of \usepackage with respect to setting packages
+#    - special rules for apacite package (redefine citation commands)
+#    - recognise /dev/null as 'file-like' arguments for --preamble and --config options
+#    - fix units package incompatibility with ulem for text maths statements $ ..$ (thanks to Stuart Prescott for reporting this)
+#    - amsmath environment cases treated correctly (Bug fix #19029) (thanks to Jalar)
+#    - {,} in comments no longer confuse latexdiff (Bug fix #19146)
+#    - \% in one-letter sub/Superscripts was not converted correctly
+#
+# Version 1.0.3
+#    - fix bug in add_safe_commands that made latexdiff hang on DeclareMathOperator
+#      command in preamble
+#    - \(..\) inline math expressions were not parsed correctly, if they contained a linebreak
+#    - applied patch contributed by tomflannaghan via Berlios: [ Patch #3431 ] Adds correct handling of \left< and \right>
+#    - \$ is treated correctly as a literal dollar sign (thanks to Reed Cartwright and Joshua Miller for reporting this bug
+#      and sketching out the solution)
+#    - \^ and \_ are correctly interpreted as accent and underlined space, respectively, not as superscript of subscript
+#      (thanks to Wail Yahyaoui for pointing out this bug)
+#
+# Version 1.0.1 - treat \big,\bigg etc. equivalently to \left and
+#              \right - include starred version in MATHENV - apply
+#            - flatten recursively and --flatten expansion is now
+#              aware of comments (thanks to Tim Connors for patch)
+#            - Change to post-processing for more reliability for
+#              deleted math environments
+#            - On linux systems, recognise  and remove DOS style newlines
+#            - Provide markup for some special preamble commands (\title,
+#              \author,\date,
+#            - configurable by setting context2cmd
+#            - for styles using ulem package, remove \emph and \text.. from list of
+#              safe commands in order to allow linebreaks within the
+#              highlighted sections.
+#            - for ulem style, now show citations by enclosing them in \mbox commands.
+#              This unfortunately implies linebreaks within citations no longer function,
+#              so this functionality can be turned off (Option --disable-citation-markup).
+#              With --enable-citation-markup, the mbox markup is forced for other styles)
+#            - new substyle COLOR.  This is particularly useful for marking up citations
+#              and some special post-processing is implemented to retain cite
+#              commands in deleted blocks.
+#            - four different levels of math-markup
+#            - Option --driver for choosing driver for modes employing changebar package
+#            - accept \\* as valid command (and other commands of form \.*). Also accept
+#              \<nl> (backslashed newline)
+#            - some typo fixes, include commands defined in preamble as safe commands
+#              (Sebastian Gouezel)
+#            - include compared filenames as comments as line 2 and 3 of
+#              the preamble (can be modified with option --label, and suppressed with
+#              --no-label), option --visible-label to show files in generated pdf or dvi
+#              at the beginning of main document
+#
+# Version 0.5  A number of minor improvements based on feedback
+#              Deleted blocks are now shown before added blocks
+#              Package specific processing
+#
+# Version 0.43 unreleased typo in list of styles at the end
+#              Add protect to all \cbstart, \cbend commands
+#              More robust substitution of deleted math commands
+#
+# Version 0.42 November 06  Bug fixes only
+#
+# Version 0.4   March 06 option for fast differencing using UNIX diff command, several minor bug fixes (\par bug, improved highlighting of textcmds)
+#
+# Version 0.3   August 05 improved parsing of displayed math, --allow-spaces
+#               option, several minor bug fixes
+#
+# Version 0.25  October 04 Fix bug with deleted equations, add math mode commands to safecmd, add | to allowed interpunctuation signs
+# Version 0.2   September 04 extension to utf-8 and variable encodings
+# Version 0.1   August 04    First public release
+
+use Algorithm::Diff qw(traverse_sequences);
+
+use Getopt::Long ;
+use strict ;
+use warnings;
+use utf8 ;
+
+use File::Spec ;
+
+
+my ($algodiffversion)=split(/ /,$Algorithm::Diff::VERSION);
+
+
+my ($versionstring)=<<EOF ;
+This is LATEXDIFF 1.3.4 (Algorithm::Diff $Algorithm::Diff::VERSION, Perl $^V)
+  (c) 2004-2024 F J Tilmann
+EOF
+
+# Hash with defaults for configuration variables. These marked undef have default values constructed from list defined in the DATA block
+# (under tag CONFIG)
+my %CONFIG=(
+   MINWORDSBLOCK => 3, # minimum number of tokens to form an independent block
+                        # shorter identical blocks will be merged to the previous word
+   SCALEDELGRAPHICS => 0.5, # factor with which deleted figures will be scaled down (i.e. 0.5 implies they are shown at half linear size)
+                             # this is only used for --dgraphics-markup=BOTH option
+   FLOATENV => undef ,   # Environments in which FL variants of defined commands are used
+   PICTUREENV => undef ,   # Environments in which all change markup is removed
+   MATHENV => undef ,           # Environments turning on display math mode (code also knows about \[ and \])
+   MATHREPL => 'displaymath',  # Environment introducing deleted maths blocks
+   MATHARRENV => undef ,           # Environments turning on eqnarray math mode
+   MATHARRREPL => 'eqnarray*',  # Environment introducing deleted maths blocks (note that now the starred varieties are being used, so this is only used to replace MATHMODE environments (where original environment is unknown)
+   ARRENV => undef , # Environments making arrays in math mode.  The underlining style does not cope well with those - as a result in-text math environments are surrounded by \mbox{ } if any of these commands is used in an inline math block
+   COUNTERCMD => undef,
+                                        # COUNTERCMD textcmds which are associated with a counter
+                                        # If any of these commands occur in a deleted block
+                                        # they will be followed by an \addtocounter{...}{-1}
+                                        # for the associated counter such that the overall numbers
+                                        # should be the same as in the new file
+   LISTENV => undef ,  # list making environments - they will generally be kept
+   VERBATIMENV => undef,      # Environments whose content should be treated as verbatim text and not be touched
+   VERBATIMLINEENV => undef,  # Environments whose content should be treated as verbatim text and processed in line diff mode
+   CUSTOMDIFCMD => undef,# Custom dif command. Is defined in the document as a \DELcommand and \ADDcommand version to be replaced by the diff
+   ITEMCMD => 'item'                    # command marking item in a list environment
+);
+# Configuration variables: these have to be visible from the subroutines
+my ($ARRENV,
+    $COUNTERCMD,
+    $FLOATENV,
+    $ITEMCMD,
+    $LISTENV,
+    $MATHARRENV,
+    $MATHARRREPL,
+    $MATHENV,
+    $MATHREPL,
+    $MINWORDSBLOCK,
+    $PICTUREENV,
+    $SCALEDELGRAPHICS,
+    $VERBATIMENV,
+    $VERBATIMLINEENV,
+    $CUSTOMDIFCMD
+    );
+
+# my $MINWORDSBLOCK=3; # minimum number of tokens to form an independent block
+#                      # shorter identical blocks will be merged to the previous word
+# my $SCALEDELGRAPHICS=0.5; # factor with which deleted figures will be scaled down (i.e. 0.5 implies they are shown at half linear size)
+#                       # this is only used for --graphics-markup=BOTH option
+# my $FLOATENV='(?:figure|table|plate)[\w\d*@]*' ;   # Environments in which FL variants of defined commands are used
+# my $PICTUREENV='(?:picture|tikzpicture|DIFnomarkup)[\w\d*@]*' ;   # Environments in which all change markup is removed
+# my $MATHENV='(?:equation[*]?|displaymath|DOLLARDOLLAR)[*]?' ;           # Environments turning on display math mode (code also knows about \[ and \])
+# my $MATHREPL='displaymath';  # Environment introducing deleted maths blocks
+# my $MATHARRENV='(?:eqnarray|align|alignat|gather|multline|flalign)[*]?' ;           # Environments turning on eqnarray math mode
+# my $MATHARRREPL='eqnarray*';  # Environment introducing deleted maths blocks
+# my $ARRENV='(?:aligned|gathered|multlined|array|[pbvBV]?matrix|smallmatrix|cases|split)'; # Environments making arrays in math mode.  The underlining style does not cope well with those - as a result in-text math environments are surrounded by \mbox{ } if any of these commands is used in an inline math block
+# my $COUNTERCMD='(?:footnote|part|chapter|section|subsection|subsubsection|paragraph|subparagraph)';  # textcmds which are associated with a counter
+#                                         # If any of these commands occur in a deleted block
+#                                         # they will be succeeded by an \addtocounter{...}{-1}
+#                                         # for the associated counter such that the overall numbers
+#                                         # should be the same as in the new file
+# my $LISTENV='(?:itemize|description|enumerate)'; # list making environments - they will generally be kept
+# my $ITEMCMD='item';   # command marking item in a list environment
+
+
+
+my $LABELCMD='(?:label)';                # matching commands are disabled within deleted blocks - mostly useful for maths mode, as otherwise it would be fine to just not add those to SAFECMDLIST
+my @UNSAFEMATHCMD=('qedhere','intertext','begin','end');           # Commands which are definitely unsafe for marking up in math mode (amsmath qedhere only tested to not work with UNDERLINE markup) (only affects WHOLE and COARSE math markup modes). Note that unlike text mode (or FINE math mode0 deleted unsafe commands are not deleted but simply taken outside \DIFdel
+my $MBOXINLINEMATH=0; # if set to 1 then surround marked-up inline maths expression with \mbox ( to get around compatibility
+                      # problems between some maths packages and ulem package
+
+
+# Markup strings
+# If at all possible, do not change these as parts of the program
+# depend on the actual name (particularly post-processing)
+# At the very least adapt subroutine postprocess to new tokens.
+my $ADDMARKOPEN='\DIFaddbegin ';   # Token to mark begin of appended text
+my $ADDMARKCLOSE='\DIFaddend ';   # Token to mark end of appended text
+my $ADDOPEN='\DIFadd{';  # To mark begin of added text passage
+my $ADDCLOSE='}';        # To mark end of added text passage
+my $ADDCOMMENT='DIF > ';   # To mark added comment line
+my $DELMARKOPEN='\DIFdelbegin ';   # Token to mark begin of deleted text
+my $DELMARKCLOSE='\DIFdelend ';   # Token to mark end of deleted text
+my $DELOPEN='\DIFdel{';  # To mark begin of deleted text passage
+my $DELCLOSE='}';        # To mark end of deleted text passage
+my $DELCMDOPEN='%DIFDELCMD < ';  # To mark begin of deleted commands (must begin with %, i.e., be a comment
+my $DELCMDCLOSE="%%%\n";    # To mark end of deleted commands (must end with a new line)
+my $AUXCMD='%DIFAUXCMD' ; #  follows auxiliary commands put in by latexdiff to make difference file legal
+                          # auxiliary commands must be on a line of their own
+                          # Note that for verbatim environment openings the %DIFAUXCMD cannot be placed in
+                          # the same line as this would mean they are shown
+                          # so the special form "%DIFAUXCMD NEXT" is used to indicate that the next line
+                          # is an auxiliary command
+                          # Similarly "%DIFAUXCMD LAST" would indicate the auxiliary command is in previous line (not currently used)
+my $DELCOMMENT='DIF < ';   # To mark deleted comment line
+my $VERBCOMMENT='DIFVRB ';  # to mark lines which are within a verbatim environment
+
+# main local variables:
+my @TEXTCMDLIST=();  # array containing patterns of commands with text arguments
+my @TEXTCMDEXCL=();  # array containing patterns of commands without text arguments (if a pattern
+                     # matches both TEXTCMDLIST and TEXTCMDEXCL it is excluded)
+my @CONTEXT1CMDLIST=();  # array containing patterns of commands with text arguments (subset of text commands),
+                         # but which cause confusion if used out of context (e.g. \caption).
+                         # In deleted passages, the command will be disabled but its argument is marked up
+                         # Otherwise they behave exactly like TEXTCMD's
+my @CONTEXT1CMDEXCL=();  # exclude list for above, but always empty
+my @CONTEXT2CMDLIST=();  # array containing patterns of commands with text arguments, but which fail or cause confusion
+                         # if used out of context (e.g. \title). They and their arguments will be disabled in deleted
+                         # passages
+my @CONTEXT2CMDEXCL=();  # exclude list for above, but always empty
+my @MATHTEXTCMDLIST=();  # treat like textcmd.  If a textcmd is in deleted or added block, just wrap the
+                         # whole content with \DIFadd or \DIFdel irrespective of content.  This functionality
+                         # is useful for pseudo commands \MATHBLOCK.. into which math environments are being
+                         # transformed
+my @MATHTEXTCMDEXCL=();  #
+
+# Note I need to declare this with "our" instead of "my" because later in the code I have to "local"ise these
+our @SAFECMDLIST=();  # array containing patterns of safe commands (which do not break when in the argument of DIFadd or DIFDEL)
+our @SAFECMDEXCL=();
+my @MBOXCMDLIST=();   # patterns for commands which are in principle safe but which need to be surrounded by an \mbox
+my @MBOXCMDEXCL=();           # all the patterns in MBOXCMDLIST will be appended to SAFECMDLIST
+
+my @KEEPCMDLIST=( qr/^bibitem$/ );   # patterns for commands which should not be deleted in nominally delete text passages
+my @KEEPCMDEXCL=();
+
+my ($i,$j,$l);
+my ($old,$new);
+my ($line,$key);
+my (@dumlist);
+my ($newpreamble,$oldpreamble);
+my (@newpreamble,@oldpreamble,@diffpreamble,@diffbody);
+my ($latexdiffpreamble);
+my ($oldbody, $newbody, $diffbo);
+my ($oldpost, $newpost);
+my ($diffall);
+# Option names
+my ($type,$subtype,$floattype,$config,$preamblefile,$encoding,$nolabel,$visiblelabel,
+    $filterscript,$ignorefilterstderr,
+    $showpreamble,$showsafe,$showtext,$showconfig,$showall,
+    $replacesafe,$appendsafe,$excludesafe,
+    $replacetext,$appendtext,$excludetext,
+    $replacecontext1,$appendcontext1,
+    $replacecontext2,$appendcontext2,
+    $help,$verbose,$driver,$version,$ignorewarnings,
+    $onlyadditions,
+    $enablecitmark,$disablecitmark,$allowspaces,$flatten,$nolinks,$debug,$earlylatexdiffpreamble);  ###$disablemathmark,
+my ($mboxsafe);
+# MNEMNONICS for mathmarkup
+my $mathmarkup;
+use constant {
+  OFF => 0,
+  WHOLE => 1,
+  COARSE => 2,
+  FINE => 3
+};
+# MNEMNONICS for graphicsmarkup
+my $graphicsmarkup;
+use constant {
+  NONE => 0,
+  NEWONLY => 1,
+  BOTH => 2
+};
+
+my ($mboxcmd);
+
+my (@configlist,@addtoconfiglist,@labels,
+    @appendsafelist,@excludesafelist,
+    @appendmboxsafelist,@excludemboxsafelist,
+    @appendtextlist,@excludetextlist,
+    @appendcontext1list,@appendcontext2list,
+    @packagelist);
+my ($assign,@config);
+# Hash where keys corresponds to the names of  all included packages (including the documentclass as another package
+# the optional arguments to the package are the values of the hash elements
+my ($pkg,%packages);
+
+# Defaults
+$mathmarkup=COARSE;
+$verbose=0;
+$onlyadditions=0;
+# output debug and intermediate files, set to 0 in final distribution
+$debug=0;
+# insert preamble directly after documentclass - experimental feature, set to 0 in final distribution
+# Note that this failed with mini example (or other files, where packages used in latexdiff preamble
+# are called again with incompatible options in preamble of resulting file)
+$earlylatexdiffpreamble=0;
+
+
+# define character properties
+sub IsNonAsciiPunct { return <<'END'    # Unicode punctuation but excluding ASCII punctuation
++utf8::IsPunct
+-utf8::IsASCII
+END
+}
+sub IsNonAsciiS { return <<'END'       # Unicode symbol but excluding ASCII
++utf8::IsS
+-utf8::IsASCII
+END
+}
+
+
+my %verbhash;
+
+Getopt::Long::Configure('bundling');
+GetOptions('type|t=s' => \$type,
+	   'subtype|s=s' => \$subtype,
+	   'floattype|f=s' => \$floattype,
+	   'config|c=s' => \@configlist,
+	   'add-to-config=s' => \@addtoconfiglist,
+	   'preamble|p=s' => \$preamblefile,
+	   'encoding|e=s' => \$encoding,
+	   'label|L=s' => \@labels,
+	   'no-label' => \$nolabel,
+	   'visible-label' => \$visiblelabel,
+	   'exclude-safecmd|A=s' => \@excludesafelist,
+	   'replace-safecmd=s' => \$replacesafe,
+	   'append-safecmd|a=s' => \@appendsafelist,
+	   'exclude-textcmd|X=s' => \@excludetextlist,
+	   'replace-textcmd=s' => \$replacetext,
+	   'append-textcmd|x=s' => \@appendtextlist,
+	   'replace-context1cmd=s' => \$replacecontext1,
+	   'append-context1cmd=s' => \@appendcontext1list,
+	   'replace-context2cmd=s' => \$replacecontext2,
+	   'append-context2cmd=s' => \@appendcontext2list,
+	   'exclude-mboxsafecmd=s' => \@excludemboxsafelist,
+	   'append-mboxsafecmd=s' => \@appendmboxsafelist,
+	   'show-preamble' => \$showpreamble,
+	   'show-safecmd' => \$showsafe,
+	   'show-textcmd' => \$showtext,
+	   'show-config' => \$showconfig,
+	   'show-all' => \$showall,
+           'packages=s' => \@packagelist,
+	   'allow-spaces' => \$allowspaces,
+           'math-markup=s' => \$mathmarkup,
+           'graphics-markup=s' => \$graphicsmarkup,
+           'enable-citation-markup|enforce-auto-mbox' => \$enablecitmark,
+           'disable-citation-markup|disable-auto-mbox' => \$disablecitmark,
+	   'verbose|V' => \$verbose,
+	   'ignore-warnings' => \$ignorewarnings,
+	   'driver=s'=> \$driver,
+	   'flatten' => \$flatten,
+	   'filter-script=s' => \$filterscript,
+	   'ignore-filter-stderr' => \$ignorefilterstderr,
+	   'no-links' => \$nolinks,
+	   'no-del' => \$onlyadditions,
+	   'version' => \$version,
+	   'help|h' => \$help,
+	   'debug!' => \$debug ) or die "Use latexdiff -h to get help.\n" ;
+
+if ( $help ) {
+  usage() ;
+}
+
+
+if ( $version ) {
+  print STDERR $versionstring ;
+  exit 0;
+}
+
+print STDERR $versionstring if $verbose;
+
+if (defined($showall)){
+  $showpreamble=$showsafe=$showtext=$showconfig=1;
+}
+# Default types
+$type='UNDERLINE' unless defined($type);
+$subtype='SAFE' unless defined($subtype);
+# set floattype to IDENTICAL for LABEL and ONLYCHANGEDPAGE subtype, unless it has been set explicitly on the command line
+$floattype=($subtype eq 'LABEL' || $subtype eq 'ONLYCHANGEDPAGE') ? 'IDENTICAL' : 'FLOATSAFE' unless defined($floattype);
+if ( $subtype eq 'LABEL' ) {
+  print STDERR "Note that LABEL subtype is deprecated. If possible, use ZLABEL instead (requires zref package)";
+}
+
+if (defined($mathmarkup)) {
+  $mathmarkup=~tr/a-z/A-Z/;
+  if ( $mathmarkup eq 'OFF' ){
+    $mathmarkup=OFF;
+  } elsif ( $mathmarkup eq 'WHOLE' ){
+    $mathmarkup=WHOLE;
+  } elsif ( $mathmarkup eq 'COARSE' ){
+    $mathmarkup=COARSE;
+  } elsif ( $mathmarkup eq 'FINE' ){
+    $mathmarkup=FINE;
+  } elsif ( $mathmarkup !~ m/^[0123]$/ ) {
+    die "latexdiff Illegal value: ($mathmarkup)  for option--math-markup. Possible values: OFF,WHOLE,COARSE,FINE,0-3\n";
+  }
+  # else use numerical value
+}
+
+# Give filterscript a default empty string
+$filterscript="" unless defined($filterscript);
+
+# setting extra preamble commands
+if (defined($preamblefile)) {
+  $latexdiffpreamble=join "\n",(extrapream($preamblefile),"");
+} else {
+  $latexdiffpreamble=join "\n",(extrapream($type,$subtype,$floattype),"");
+}
+
+if ( defined($driver) ) {
+  # for changebar only
+  $latexdiffpreamble=~s/\[pdftex\]/[$driver]/sg;
+}
+# setting up @SAFECMDLIST and @SAFECMDEXCL
+if (defined($replacesafe)) {
+  init_regex_arr_ext(\@SAFECMDLIST,$replacesafe);
+} else {
+  init_regex_arr_data(\@SAFECMDLIST, "SAFE COMMANDS");
+}
+foreach $appendsafe ( @appendsafelist ) {
+  init_regex_arr_ext(\@SAFECMDLIST, $appendsafe);
+}
+foreach $excludesafe ( @excludesafelist ) {
+  init_regex_arr_ext(\@SAFECMDEXCL, $excludesafe);
+}
+# setting up @MBOXCMDLIST and @MBOXCMDEXCL
+foreach $mboxsafe ( @appendmboxsafelist ) {
+  init_regex_arr_ext(\@MBOXCMDLIST, $mboxsafe);
+}
+foreach $mboxsafe ( @excludemboxsafelist ) {
+  init_regex_arr_ext(\@MBOXCMDEXCL, $mboxsafe);
+}
+
+
+
+# setting up @TEXTCMDLIST and @TEXTCMDEXCL
+if (defined($replacetext)) {
+  init_regex_arr_ext(\@TEXTCMDLIST,$replacetext);
+} else {
+  init_regex_arr_data(\@TEXTCMDLIST, "TEXT COMMANDS");
+}
+foreach $appendtext ( @appendtextlist ) {
+  init_regex_arr_ext(\@TEXTCMDLIST, $appendtext);
+}
+foreach $excludetext ( @excludetextlist ) {
+  init_regex_arr_ext(\@TEXTCMDEXCL, $excludetext);
+}
+
+
+# setting up @CONTEXT1CMDLIST ( @CONTEXT1CMDEXCL exist but is always empty )
+if (defined($replacecontext1)) {
+  init_regex_arr_ext(\@CONTEXT1CMDLIST,$replacecontext1);
+} else {
+  init_regex_arr_data(\@CONTEXT1CMDLIST, "CONTEXT1 COMMANDS");
+}
+foreach $appendcontext1 ( @appendcontext1list ) {
+  init_regex_arr_ext(\@CONTEXT1CMDLIST, $appendcontext1);
+}
+
+
+# setting up @CONTEXT2CMDLIST ( @CONTEXT2CMDEXCL exist but is always empty )
+if (defined($replacecontext2)) {
+  init_regex_arr_ext(\@CONTEXT2CMDLIST,$replacecontext2);
+} else {
+  init_regex_arr_data(\@CONTEXT2CMDLIST, "CONTEXT2 COMMANDS");
+}
+foreach $appendcontext2 ( @appendcontext2list ) {
+  init_regex_arr_ext(\@CONTEXT2CMDLIST, $appendcontext2);
+}
+
+# setting configuration variables
+@config=();
+foreach $config ( @configlist ) {
+  if (-f $config || lc $config eq '/dev/null' ) {
+    open(FILE,$config) or die ("Couldn't open configuration file $config: $!");
+    while (<FILE>) {
+      chomp;
+      next if /^\s*#/ || /^\s*%/ || /^\s*$/ ;
+      push (@config,$_);
+    }
+    close(FILE);
+  }
+  else {
+#    foreach ( split(",",$config) ) {
+#      push @config,$_;
+#    }
+     push @config,split(",",$config)
+  }
+}
+foreach $assign ( @config ) {
+  $assign=~ m/\s*(\w*)\s*=\s*(\S*)\s*$/ or die "Illegal assignment $assign in configuration list (must be variable=value)";
+  exists $CONFIG{$1} or die "Unknown configuration variable $1.";
+  $CONFIG{$1}=$2;
+}
+
+my @addtoconfig=();
+foreach $config ( @addtoconfiglist ) {
+  if (-f $config || lc $config eq '/dev/null' ) {
+    open(FILE,$config) or die ("Couldn't open addd-to-config file $config: $!");
+    while (<FILE>) {
+      chomp;
+      next if /^\s*#/ || /^\s*%/ || /^\s*$/ ;
+      push (@addtoconfig,$_);
+    }
+    close(FILE);
+  }
+  else {
+#    foreach ( split(",",$config) ) {
+#      push @addtoconfig,$_;
+#    }
+     push @addtoconfig,split(",",$config)
+  }
+}
+
+# initialise default lists from DATA
+# for those configuration variables, which have not been set explicitly, initiate from list in document
+foreach $key ( keys(%CONFIG) ) {
+  if (!defined $CONFIG{$key}) {
+    @dumlist=();
+    init_regex_arr_data(\@dumlist,"$key CONFIG");
+    $CONFIG{$key}=join(";",@dumlist)
+  }
+}
+
+
+foreach $assign ( @addtoconfig ) {
+  ###print STDERR "assign:|$assign|\n";
+  $assign=~ m/\s*(\w*)\s*=\s*(\S*)\s*$/ or die "Illegal assignment $assign in configuration list (must be variable=value)";
+  exists $CONFIG{$1} or die "Unknown configuration variable $1.";
+  $CONFIG{$1}.=";$2";
+}
+
+# Map from hash to variables (we do this to have more concise code later, change from comma-separated list)
+foreach  ( keys(%CONFIG) ) {
+  if ( $_ eq "MINWORDSBLOCK" ) { $MINWORDSBLOCK = $CONFIG{$_}; }
+  elsif ( $_ eq "FLOATENV" ) { $FLOATENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "ITEMCMD" ) { $ITEMCMD = $CONFIG{$_} ; }
+  elsif ( $_ eq "LISTENV" ) { $LISTENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "PICTUREENV" ) { $PICTUREENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "MATHENV" ) { $MATHENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "MATHREPL" ) { $MATHREPL = $CONFIG{$_} ; }
+  elsif ( $_ eq "MATHARRENV" ) { $MATHARRENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "MATHARRREPL" ) { $MATHARRREPL = $CONFIG{$_} ;
+				  print STDERR "WARNING: Setting MATHARRREPL is depracated. Generally deleted math array environments will be set to their starred varieties and the setting of MATHARREPL is ignored.\n\n" unless $MATHARRREPL =~ /eqnarray\*/ ;
+				}
+  elsif ( $_ eq "ARRENV" ) { $ARRENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "VERBATIMENV" ) { $VERBATIMENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "VERBATIMLINEENV" ) { $VERBATIMLINEENV = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "CUSTOMDIFCMD" ) { $CUSTOMDIFCMD = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "COUNTERCMD" ) { $COUNTERCMD = liststringtoregex($CONFIG{$_}) ; }
+  elsif ( $_ eq "SCALEDELGRAPHICS" ) { $SCALEDELGRAPHICS = $CONFIG{$_} ; }
+  else { die "Unknown configuration variable $_.";}
+}
+
+if ( $mathmarkup == COARSE || $mathmarkup == WHOLE ) {
+  push(@MATHTEXTCMDLIST,qr/^MATHBLOCK(?:$MATHENV|$MATHARRENV|SQUAREBRACKET)$/);
+}
+
+
+
+
+
+foreach $pkg ( @packagelist ) {
+  map { $packages{$_}="" } split(/,/,$pkg) ;
+}
+
+
+if ($showconfig || $showtext || $showsafe || $showpreamble) {
+
+  show_configuration();
+  exit 0;
+}
+
+
+if ( @ARGV != 2 ) {
+  print STDERR "2 and only 2 non-option arguments required.  Write latexdiff -h to get help\n";
+  exit(2);
+}
+
+# Are extra spaces between command arguments permissible?
+my $extraspace;
+if ($allowspaces) {
+  $extraspace='\s*';
+} else {
+  $extraspace='';
+}
+
+# append context lists to text lists (as text property is implied)
+push @TEXTCMDLIST, @CONTEXT1CMDLIST;
+push @TEXTCMDLIST, @CONTEXT2CMDLIST;
+
+push @TEXTCMDLIST, @MATHTEXTCMDLIST if $mathmarkup==COARSE;
+
+# internal additions to SAFECMDLIST
+push(@SAFECMDLIST, qr/^QLEFTBRACE$/, qr/^QRIGHTBRACE$/);
+
+
+# Patterns. These are used by some of the subroutines, too
+# I can only define them down here because value of extraspace depends on an option
+
+  my $pat0 = '(?:[^{}])*';
+  my $pat_n = $pat0;
+# if you get "undefined control sequence MATHBLOCKmath" error, increase the maximum value in this loop
+  for (my $i_pat = 0; $i_pat < 20; ++$i_pat){
+    $pat_n = '(?:[^{}]|\{'.$pat_n.'\}|\\\\\{|\\\\\})*';
+    # Actually within the text body, quoted braces are replaced in pre-processing. The only place where
+    # the last part of the pattern matters is when processing the arguments of context2cmds in the preamble
+    # and these contain a \{ or \} combination, probably rare.
+    # It should thus be fine to use the simpler version below.
+    ###  $pat_n = '(?:[^{}]|\{'.$pat_n.'\})*';
+  }
+
+  my $brat0 = '(?:[^\[\]]|\\\[|\\\])*';
+  my $brat_n = $brat0;
+  for (my $i_pat = 0; $i_pat < 4; ++$i_pat){
+    $brat_n = '(?:[^\[\]]|\['.$brat_n.'\]|\\\[|\\\])*';
+    ###  $brat_n = '(?:[^\[\]]|\['.$brat_n.'\])*';   # Version not taking into account escaped \[ and \]
+  }
+  my $abrat0 = '(?:[^<>])*';
+
+
+# variable definitions are in order that they are matched
+  my $and = '&';
+  my $quotemarks = '(?:\'\')|(?:\`\`)';
+  # some common abbreviations involving punctuations within them. They need special treatment because otherwise in some
+  # circumstances the gnoring of white space differences in conjunction with merging according to MINWORDSBLOCK rule
+  # could turn 'i.e.' into 'i.\PAR e.' (see https://github.com/ftilmann/latexdiff/issues/269)
+  # English: i.e., e.g.   Deutsch: z.B.
+  my $abbreviation='(?:i\. ?e\.|e\. ?g\.|z\. ?B\.)' ;
+  my $number='-?\d*\.\d*';
+
+  # word: sequence of letters or accents followed by letter
+  my $word_cj='\p{Han}|\p{InHiragana}|\p{InKatakana}';
+  my $word='(?:' . $word_cj . '|(?:(?:[-\w\d*]|\\\\[\"\'\`~^][A-Za-z\*])(?!(?:' . $word_cj . ')))+)';
+
+  # quoted underscore - this needs special treatment as perl treats _ as a letter (\w) but latex does not
+  # such that a\_b would otherwise be interpreted as a{\_}b by latex but a{\_b} by latexdiff
+  my $quotedunderscore='\\\\_';
+  # Handle tex \def macro: \def\MAKRONAME#1[#2]#3{DEFINITION}
+  my $defseq='\\\\def\\\\[\w\d@\*]+(?:#\d+|\[#\d+\])+(?:\{'. $pat_n . '\})?';
+  my $cmdleftright='\\\\(?:left|right|[Bb]igg?[lrm]?|middle)\s*(?:[<>()\[\]|\.]|\\\\(?:[|{}]|\w+))';
+
+  # for selected commands, the number of arguments is known, and we can therefore allow spaces between command and its argument
+  # Note that it is still expected that the arguments are blocks marked by parentheses rather than single characters, and that intervening comments will inhibit the association
+  my $predefinedcmdoptseq12='\\\\(?:href|bibfield|bibinfo)\s*(?:\['.$brat_n.'\])?\s*(?:\{'. $pat_n . '\}\s*){2}';  # Commands with one optional and two non-optional arguments
+  my $predefinedcmdoptseq01='\\\\(?:url|BibitemShut)\s*\s*(?:\{'. $pat_n . '\}\s*){1}';  # Commands with one non-optional argument
+  # \bibitem in revtex styles appears to be always followed by \BibItemOpen. We bind \BibItemOpen to the bibitem (if present) in order to prevent the comparison algorithm to interpret the \BibItemOpen as an identical part of the sequence; this interpretation can lead to added and removed entries to the reference list to become mixed.
+  my $predefinedbibitem='\\\\(?:bibitem)\s*(?:\['.$brat_n.'\])?\s*(?:\{'. $pat_n . '\})(?:%?\s*\\\\BibitemOpen)?';  # Commands with one optional and one non-optional arguments
+  my $predefinedcmdoptseq='(?:'.$predefinedcmdoptseq12.'|'.$predefinedcmdoptseq01.'|'.$predefinedbibitem.')';
+
+  # standard $cmdoptseq (default: no intrevening spaces, controlled by extraspcae) - a final open parentheses is merged to the commend if it exists to deal properly with multi-argument text command
+  my $coords= '[\-.,\s\d]*';
+  my $cmdoptseq='\\\\[\w\d@\*]+'.$extraspace.'(?:(?:<'.$abrat0.'>|\['.$brat_n.'\]|\{'. $pat_n . '\}|\(' . $coords .'\))'.$extraspace.')*\{?';
+
+  # inline math $....$ or \(..\)
+  ### the commented out version is simpler but for some reason cannot cope with newline (in spite of s option) - need to include \newline explicitly
+  ###  my $math='\$(?:[^$]|\\\$)*?\$|\\\\[(].*?\\\\[)]';
+  my $math='\$(?:[^$]|\\\$)*?\$|\\\\[(](?:.|\n)*?\\\\[)]';
+  ### test version (this seems to give the same results as version above)
+  ## the current maths command cannot cope with newline within the math expression
+  ### my $math='\$(?:[^$]|\\\$)*?\$|\\[(].*?\\[)]';
+  ###  my $math='\$(?:[^$]|\\\$)*\$';
+
+  my $backslashnl='\\\\\n';
+  my $oneletcmd='\\\\.\*?(?:\['.$brat_n.'\]|\{'. $pat_n . '\})*';
+  my $comment='%[^\n]*\n';
+  my $punct='[0.,\/\'\`:;\"\?\(\)\[\]!~\p{IsNonAsciiPunct}\p{IsNonAsciiS}]';
+  my $mathpunct='[+=<>\-\|]';
+
+  # Assembled pattern
+  my $pat=qr/(?:\A\s*)?(?:${abbreviation}|${and}|${quotemarks}|${number}|${word}|$quotedunderscore|${defseq}|$cmdleftright|${predefinedcmdoptseq}|${cmdoptseq}|${math}|${backslashnl}|${oneletcmd}|${comment}|${punct}|${mathpunct}|\{|\})\s*/ ;
+
+
+
+
+# now we are done setting up and can start working
+my ($oldfile, $newfile) = @ARGV;
+# check for existence of input files
+if ( ! -e $oldfile ) {
+  die "Input file $oldfile does not exist";
+}
+if ( ! -e $newfile ) {
+  die "Input file $newfile does not exist";
+}
+
+
+# set the labels to be included into the file
+# first find out which file name is longer for correct alignment
+my ($diff,$oldlabel_n_spaces,$newlabel_n_spaces);
+$oldlabel_n_spaces = 0;
+$newlabel_n_spaces = 0;
+$diff = length($newfile) - length($oldfile);
+if ($diff > 0) {
+  $oldlabel_n_spaces = $diff;
+}
+if ($diff < 0) {
+  $newlabel_n_spaces = abs($diff);
+}
+
+my ($oldtime,$newtime,$oldlabel,$newlabel);
+if (defined($labels[0])) {
+  $oldlabel=$labels[0] ;
+} else {
+  $oldtime=localtime((stat($oldfile))[9]);
+  $oldlabel="$oldfile   " . " "x($oldlabel_n_spaces) . $oldtime;
+}
+if (defined($labels[1])) {
+  $newlabel=$labels[1] ;
+} else {
+  $newtime=localtime((stat($newfile))[9]);
+  $newlabel="$newfile   " . " "x($newlabel_n_spaces) . $newtime;
+}
+
+$encoding=guess_encoding($newfile) unless defined($encoding);
+
+$encoding = "utf8" if $encoding =~ m/^utf8/i ;
+print STDERR "Encoding $encoding\n" if $verbose;
+if (lc($encoding) eq "utf8" ) {
+  binmode(STDOUT, ":utf8");
+  binmode(STDERR, ":utf8");
+}
+
+# filter($text)
+# Runs $text through the script provided in $filterscript argument, if set
+# If not set, just returns $text unchanged.
+# If flatten was set, defer filtering to flatten.  flatten will run the filter
+# on all incoming text prior to its own processing.
+# If flatten was not set, filter each of old and new once (see just below this def)
+sub filter {
+  my ($text)=@_;
+  my ($textout,$pid);
+  if ($filterscript ne "") {
+    print STDERR "Passing " . length($text) . " chars to filter script " . $filterscript . "\n" if $verbose;
+
+    if ($ignorefilterstderr) {
+        # If we need to capture and bury STDERR, use the Open3 version, and close CHLD_ERR below.
+        use IPC::Open3;
+        # We consume STDERR from the process, and hide it
+        $pid = open3(\*CHLD_IN, \*CHLD_OUT, \*CHLD_ERR, $filterscript) or die "open3() failed $!";
+    }
+    else {
+        # Capture STDOUT and use as our new $text.  Allow STDERR to go to console.
+        use IPC::Open2;
+        $pid = open2(\*CHLD_OUT, \*CHLD_IN, $filterscript) or die "open2() failed $!";
+    }
+    # Send in $text
+    print CHLD_IN $text."\n";  # Adding a newline just to make sure there is one.
+    close CHLD_IN;
+    # Wait for output and gather it up
+    while (<CHLD_OUT>) {
+      $textout = $textout.$_;
+    }
+    if ($ignorefilterstderr) {
+        close CHLD_ERR;       # Enable only if Open3 used above
+    }
+    # On the off chance a very long running and/or frequently called script is used.
+    waitpid( $pid, 0 );
+    $text = $textout;
+    print STDERR "Received " . length($text) . " chars after filtering\n" if $verbose;
+    print STDERR $text if $verbose;
+  }
+  return $text;
+}
+
+$old=read_file_with_encoding($oldfile,$encoding);
+$new=read_file_with_encoding($newfile,$encoding);
+
+if (not defined($flatten)) {
+    $old=filter($old);
+    $new=filter($new);
+}
+
+
+
+
+# reset time
+exetime(1);
+($oldpreamble,$oldbody,$oldpost)=splitdoc($old,'\\\\begin\{document\}','\\\\end\{document\}');
+
+($newpreamble,$newbody,$newpost)=splitdoc($new,'\\\\begin\{document\}','\\\\end\{document\}');
+
+
+if ($flatten) {
+  $oldbody=flatten($oldbody,$oldpreamble,File::Spec->rel2abs($oldfile),$encoding);
+  $newbody=flatten($newbody,$newpreamble,File::Spec->rel2abs($newfile),$encoding);
+  # flatten preamble
+  $oldpreamble=flatten($oldpreamble,$oldpreamble,File::Spec->rel2abs($oldfile),$encoding);
+  $newpreamble=flatten($newpreamble,$newpreamble,File::Spec->rel2abs($newfile),$encoding);
+
+}
+
+
+
+
+my @auxlines;
+
+# boolean variab
+my ($ulem)=0;
+
+if ( length $oldpreamble && length $newpreamble ) {
+  # pre-process preamble by looking for commands used in \maketitle (title, author, date etc commands)
+  # and marking up content with latexdiff markup
+  @auxlines=preprocess_preamble($oldpreamble,$newpreamble);
+
+  @oldpreamble = split /\n/, $oldpreamble;
+  @newpreamble = split /\n/, $newpreamble;
+
+  # If a command is defined in the preamble of the new file, and only uses safe commands, then it can be considered to be safe) (contribution S. Gouezel)
+  # Base this assessment on the new preamble
+  add_safe_commands($newpreamble);
+
+  # get a list of packages from preamble if not predefined
+  %packages=list_packages($newpreamble) unless %packages;
+  if ( %packages && $debug ) { my $key ; foreach $key (keys %packages) { print STDERR "DEBUG \\usepackage[",$packages{$key},"]{",$key,"}\n" ;} }
+}
+
+# have to return to all processing to properly add preamble additions based on packages found
+if (defined($graphicsmarkup)) {
+  $graphicsmarkup=~tr/a-z/A-Z/;
+  if ( $graphicsmarkup eq 'OFF' or $graphicsmarkup eq 'NONE' ) {
+    $graphicsmarkup=NONE;
+  } elsif ( $graphicsmarkup eq 'NEWONLY' or $graphicsmarkup eq 'NEW-ONLY' ) {
+    $graphicsmarkup=NEWONLY;
+  } elsif ( $graphicsmarkup eq 'BOTH' ) {
+    $graphicsmarkup=BOTH;
+  } elsif ( $graphicsmarkup !~ m/^[012]$/ ) {
+    die "latexdiff Illegal value: ($graphicsmarkup)  for option --highlight-graphics. Possible values: OFF,WHOLE,COARSE,FINE,0-2\n";
+  }
+  # else use numerical value
+} else {
+  # Default: no explicit setting in menu
+  if ( defined $packages{"graphicx"} or defined $packages{"graphics"} ) {
+    $graphicsmarkup=NEWONLY;
+  } else {
+    $graphicsmarkup=NONE;
+  }
+}
+
+if (defined $packages{"hyperref"} ) {
+  # deleted lines should not generate or appear in link names:
+  print STDERR "hyperref package detected.\n" if $verbose ;
+  $latexdiffpreamble =~ s/\{\\DIFadd\}/{\\DIFaddtex}/g;
+  $latexdiffpreamble =~ s/\{\\DIFdel\}/{\\DIFdeltex}/g;
+  $latexdiffpreamble .= join "\n",(extrapream("HYPERREF"),"");
+  if($nolinks){
+    $latexdiffpreamble .= "\n\\hypersetup{bookmarks=false}";
+  }
+  ###    $latexdiffpreamble .= '%DIF PREAMBLE EXTENSION ADDED BY LATEXDIFF FOR HYPERREF PACKAGE' . "\n";
+  ###    $latexdiffpreamble .= '\providecommand{\DIFadd}[1]{\texorpdfstring{\DIFaddtex{#1}}{#1}}' . "\n";
+  ###    $latexdiffpreamble .= '\providecommand{\DIFdel}[1]{\texorpdfstring{\DIFdeltex{#1}}{}}' . "\n";
+  ###    $latexdiffpreamble .= '%DIF END PREAMBLE EXTENSION ADDED BY LATEXDIFF FOR HYPERREF PACKAGE' . "\n";
+}
+
+# add commands for figure highlighting to preamble
+if ($graphicsmarkup != NONE ) {
+  my @matches;
+  # Check if \DIFaddbeginFL definition calls \DIFaddbegin - if so we will issue an error message that graphics highlighting is
+  # is not compatible with this.
+  # (A more elegant solution would be to suppress the redefinitions of the \DIFaddbeginFL etc commands, but for this narrow use case
+  #  I currently don't see this as an efficient use of time)
+  ### The foreach loop does not make sense here. I don't know why I put this in -  (F Tilmann)
+  ###foreach my $cmd ( "DIFaddbegin","DIFaddend","DIFdelbegin","DIFdelend" ) {
+  @matches=( $latexdiffpreamble =~ m/command\{\\DIFaddbeginFL}\{($pat_n)}/sg ) ;
+  # we look at the last one of the list to take into account possible redefinition but almost always matches should have exactly one element
+  if ( $matches[$#matches] =~ m/\\DIFaddbegin/ ) {
+    die "Cannot combine graphics markup with float styles defining \\DIFaddbeginFL in terms of \\DIFaddbegin. Use --graphics-markup=none option or choose a different float style.";
+    exit 10;
+  }
+  ###}
+  $latexdiffpreamble .= join "\n",("\\newcommand{\\DIFscaledelfig}{$SCALEDELGRAPHICS}",extrapream("HIGHLIGHTGRAPHICS"),"");
+
+  # only change required for highlighting both is to declare \includegraphics safe, as preamble already contains commands for deleted environment
+  if ( $graphicsmarkup == BOTH ) {
+    init_regex_arr_list(\@SAFECMDLIST,'includegraphics');
+  }
+}
+
+$ulem = ($latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{ulem\}/ || defined $packages{"ulem"});
+
+# If amsmath is defined and $ulem is used for markup, redefine the \sout command to also work (mostly) in math mode
+# See stack exchange https://tex.stackexchange.com/questions/20609/strikeout-in-math-mode/308647#308647 based on comment by Taylor Raine
+if ( defined($packages{'amsmath'}) and $ulem ) {
+    $latexdiffpreamble .= join "\n",(extrapream('AMSMATHULEM'),"");
+}
+
+# If listings is being used or can be found in the latexdiff search path, add to the preamble auxiliary code to enable line-by-line markup
+if ( defined($packages{"listings"}) or `kpsewhich listings.sty` ne "" ) {
+  my @listingpreamble=extrapream("LISTINGS");
+  if ($latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/ ) {
+    @listingpreamble=extrapream("COLORLISTINGS");
+  }
+  my @listingDIFcode=();
+  my $replaced;
+  # note that in case user supplies preamblefile the type might not reflect well the actual markup style
+  @listingDIFcode=extrapream("-nofail","DIFCODE_" . $type) unless defined($preamblefile);
+  if (!(@listingDIFcode)) {
+    # if listingDIFcode is empty try to guess a suitable one from the preamble
+    if ($latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/ and $ulem ) {
+      @listingDIFcode=extrapream("DIFCODE_UNDERLINE");
+    } elsif ( $latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/ ) {
+      # only colour used
+      @listingDIFcode=extrapream("DIFCODE_CFONT");
+    } else {
+      # fall-back solution
+      @listingDIFcode=extrapream("DIFCODE_BOLD");
+    }
+  }
+  # add configuration so that listings work with utf-8
+  push @listingpreamble, '\lstset{extendedchars=\true,inputencoding='.$encoding."}\n";
+
+
+  # now splice it in
+  $replaced=0;
+  ###print STDERR "DEBUG: listingDIFcode: ",join("\n",@listingDIFcode),"|||\n" if $debug;
+
+  @listingpreamble=grep {
+    # only replace if this has not been done already (use short-circuit property of and)
+    if (!$replaced and $_ =~ s/^.*%DIFCODE TEMPLATE.*$/join("\n",@listingDIFcode)/e ) {
+      ###print STDERR "DEBUG: Replaced text $_\n" if $debug;
+      $replaced=1;
+      1;
+    } else {
+      # return false for those lines matching %DIFCODE TEMPLATE (so that they are not included in output)
+      not m/%DIFCODE TEMPLATE/;
+    }
+  } @listingpreamble;
+  ###  print STDERR "DEBUG: listingpreamble @listingpreamble\n";
+  $latexdiffpreamble .= join "\n",(@listingpreamble,"");
+} else {
+  print STDERR "WARNING: listings package not detected. Disabling mark-up in verbatim environments \n" ;
+  # if listings does not exist disable line-by-line markup and treat all verbatim environments as opaque
+  $VERBATIMENV = liststringtoregex($CONFIG{VERBATIMENV}.";".$CONFIG{VERBATIMLINEENV});
+  $VERBATIMLINEENV = "";
+}
+
+# adding begin and end marker lines to preamble
+$latexdiffpreamble = "%DIF PREAMBLE EXTENSION ADDED BY LATEXDIFF\n" . $ latexdiffpreamble . "%DIF END PREAMBLE EXTENSION ADDED BY LATEXDIFF\n";
+
+# and return to preamble specific processing
+if ( length $oldpreamble && length $newpreamble ) {
+  print STDERR "Differencing preamble.\n" if $verbose;
+
+  # insert dummy first line such that line count begins with line 1 (rather than perl's line 0) - just so that line numbers inserted by linediff are correct
+  unshift @newpreamble,'';
+  unshift @oldpreamble,'';
+  @diffpreamble = linediff(\@oldpreamble, \@newpreamble);
+  # remove dummy line again
+  shift @diffpreamble;
+  # add filenames, modification time and latexdiff mark
+  defined($nolabel) or splice @diffpreamble,1,0,
+      "%DIF LATEXDIFF DIFFERENCE FILE",
+      ,"%DIF DEL $oldlabel",
+      "%DIF ADD $newlabel";
+  if ( @auxlines ) {
+    push @diffpreamble,"%DIF DELETED TITLE COMMANDS FOR MARKUP";
+    push @diffpreamble,join("\n",@auxlines);
+  }
+  if ( $earlylatexdiffpreamble) {
+    # insert latexdiff command directly after documentclass at beginning of preamble
+    # note that grep is only run for its side effect
+    ( grep { s/^([^%]*\\documentclass.*)$/$1$latexdiffpreamble/ } @diffpreamble )==1 or die "Could not find documentclass statement in preamble";
+  } else {
+    # insert latexdiff commands at the end of preamble (default behaviour)
+    push @diffpreamble,$latexdiffpreamble;
+  }
+  push @diffpreamble,'\begin{document}';
+  if (defined $packages{"hyperref"} && $nolinks) {
+    push @diffpreamble, '\begin{NoHyper}';
+  }
+}
+elsif ( !length $oldpreamble && !length $newpreamble ) {
+  @diffpreamble=();
+} else {
+  print STDERR "Either both texts must have preamble or neither text must have the preamble.\n";
+  exit(2);
+}
+
+# Special: treat all cite commands as safe except in UNDERLINE and FONTSTRIKE mode
+# (there is a conflict between citation and ulem package, see
+# package documentation)
+# Use post-processing
+# and $packages{"apacite"}!~/natbibpapa/
+
+
+if (defined $packages{"units"}  && $ulem ) {
+  # protect inlined maths environments by surrounding with an \mbox
+  # this is done to get around an incompatibility between the ulem and units package
+  # where spaces in the argument to underlined or crossed-out \unit commands cause an error message
+  print STDERR "units package detected at the same time as style using ulem.\n" if $verbose ;
+  $MBOXINLINEMATH=1;
+}
+
+if (defined $packages{"siunitx"} ) {
+  # protect SI command by surrounding them with an \mbox
+  # this is done to get around an incompatibility between the ulem and siunitx package
+  print STDERR "siunitx package detected.\n" if $verbose ;
+  my $mboxcmds='SI,ang,numlist,numrange,SIlist,SIrange,qty,qtylist,qtyproduct,qtyrange,complexqty';
+  init_regex_arr_list(\@SAFECMDLIST,'num,si,numproduct,unit,complexnum');
+  if ( $enablecitmark || ( $ulem  && ! $disablecitmark )) {
+    init_regex_arr_list(\@MBOXCMDLIST,$mboxcmds);
+  } else {
+    init_regex_arr_list(\@SAFECMDLIST,$mboxcmds);
+  }
+}
+
+if (defined $packages{"cleveref"} ) {
+  # protect selected command by surrounding them with an \mbox
+  # this is done to get around an incompatibility between ulem and cleveref package
+  print STDERR "cleveref package detected.\n" if $verbose ;
+  my $mboxcmds='[Cc]ref(?:range)?\*?,labelcref,(?:lc)?name[cC]refs?' ;
+  if ( $enablecitmark || ( $ulem  && ! $disablecitmark )) {
+    init_regex_arr_list(\@MBOXCMDLIST,$mboxcmds);
+  } else {
+    init_regex_arr_list(\@SAFECMDLIST,$mboxcmds);
+  }
+}
+
+if (defined $packages{"glossaries"} ) {
+  # protect selected command by surrounding them with an \mbox
+  # this is done to get around an incompatibility between ulem and glossaries package
+  print STDERR "glossaries package detected.\n" if $verbose ;
+  my $mboxcmds='[gG][lL][sS](?:|pl|disp|link|first|firstplural|desc|user[iv][iv]?[iv]?),[aA][cC][rR](?:long|longpl|full|fullpl),[aA][cC][lfp]?[lfp]?';
+  init_regex_arr_list(\@SAFECMDLIST,'[gG][lL][sS](?:(?:entry)?(?:text|plural|name|symbol)|displaynumberlist|entryfirst|entryfirstplural|entrydesc|entrydescplural|entrysymbolplural|entryuser[iv][iv]?[iv]?|entrynumberlist|entrydisplaynumberlist|entrylong|entrylongpl|entryshort|entryshortpl|entryfull|entryfullpl),[gG]lossentry(?:name|desc|symbol),[aA][cC][rR](?:short|shortpl),[aA]csp?');
+  if ( $enablecitmark || ( $ulem  && ! $disablecitmark )) {
+    init_regex_arr_list(\@MBOXCMDLIST,$mboxcmds);
+  } else {
+    init_regex_arr_list(\@SAFECMDLIST,$mboxcmds);
+  }
+}
+
+if (defined $packages{"chemformula"} or defined $packages{"chemmacros"} ) {
+  print STDERR "chemformula package detected.\n" if $verbose ;
+  init_regex_arr_list(\@SAFECMDLIST,'ch');
+  push(@UNSAFEMATHCMD,'ch');
+  # The next command would be needed to allow highlighting the interior of \ch commands in math environments
+  # but the redefinitions in chemformula are too deep to make this viable
+  # push(@MATHTEXTCMDLIST,'ch');
+}
+
+if (defined $packages{"mhchem"} ) {
+  print STDERR "mhchem package detected.\n" if $verbose ;
+  init_regex_arr_list(\@SAFECMDLIST,'ce');
+  push(@UNSAFEMATHCMD,'ce','cee');
+  # The next command would be needed to allow highlighting the interior of \cee commands in math environments
+  # but the redefinitions in chemformula are too deep to make this viable
+  # push(@MATHTEXTCMDLIST,'cee');
+}
+
+if ( defined $packages{"tikz-dependency"} ) {
+  init_regex_arr_ext(\@SAFECMDEXCL, 'AMPERSAND');
+}
+
+
+my ( $citpat);
+
+if ( defined $packages{"apacite"}  ) {
+  print STDERR "apacite package detected.\n" if $verbose ;
+  $citpat='(?:mask)?(?:full|short|no)?cite(?:A|author|year|meta)?(?:NP)?';
+} elsif (defined $packages{"biblatex"}) {
+  print STDERR "biblatex package detected.\n" if $verbose ;
+  $citpat='(?:[cC]ites?|(?:[pP]aren|foot|[Tt]ext|[sS]mart|super)cites?\*?|footnotecitetex)';
+  push(@TEXTCMDEXCL, qr/^textcite$/);
+} else {
+  # citation command pattern for all other citation schemes
+  $citpat='(?:cite\w*|nocite)';
+};
+
+if ( ! $ulem ) {
+  # modes not using ulem: citation is safe
+  push (@SAFECMDLIST, $citpat);
+} else {
+  ### Experimental: disable text and emph commands
+  push(@SAFECMDEXCL, qr/^emph$/, qr/^text..$/);
+  # replace \cite{..} by \mbox{\cite{..}} in added or deleted blocks in post-processing
+  push(@MBOXCMDLIST,$citpat) unless $disablecitmark;
+  if ( uc($subtype) eq "COLOR" or uc($subtype) eq "DVIPSCOL" ) {
+    # remove \cite command again from list of safe commands
+    pop @MBOXCMDLIST;
+    # deleted cite commands
+  }
+}
+push(@MBOXCMDLIST,$citpat) if $enablecitmark ;
+
+
+if (defined $packages{"amsmath"}  or defined $packages{"amsart"} or defined $packages{"amsbook"} ) {
+  print STDERR "amsmath package detected.\n" if $verbose ;
+  $MATHARRREPL='align*';
+}
+
+# add commands in MBOXCMDLIST to SAFECMDLIST
+foreach $mboxcmd ( @MBOXCMDLIST ) {
+  init_regex_arr_list(\@SAFECMDLIST, $mboxcmd);
+}
+
+# check if \label is in SAFECMDLIST, and if yes replace "label" in $LABELCMD by something that never matches (we hope!)
+if ( iscmd("label",\@SAFECMDLIST,\@SAFECMDEXCL) ) {
+  $LABELCMD=~ s/label/NEVERMATCHLABEL/;
+}
+
+
+
+print STDERR "Preprocessing body.  " if $verbose;
+preprocess($oldbody,$newbody);
+writedebugfile($oldbody,'old-preprocess');
+writedebugfile($newbody,'new-preprocess');
+
+# run difference algorithm
+@diffbody=bodydiff($oldbody, $newbody);
+$diffbo=join("",@diffbody);
+writedebugfile($diffbo,"bodydiff");
+
+print STDERR "(",exetime()," s)\n","Postprocessing body. \n" if $verbose;
+postprocess($diffbo);
+$diffall =join("\n",@diffpreamble) ;
+# add visible labels
+if (defined($visiblelabel)) {
+  # Give information right after \begin{document} (or at the beginning of the text for files without preamble
+  ### if \date command is used, add information to \date argument, otherwise give right after \begin{document}
+  ###  $diffall=~s/(\\date$extraspace(?:\[$brat0\])?$extraspace)\{($pat_n)\}/$1\{$2 \\ LATEXDIFF comparison \\ Old: $oldlabel \\ New: $newlabel \}/  or
+  $diffbo = "\\begin{verbatim}LATEXDIFF comparison\nOld: $oldlabel\nNew: $newlabel\\end{verbatim}\n$diffbo"   ;
+}
+
+$diffall .= "$diffbo" ;
+if (defined $packages{"hyperref"} && $nolinks) {
+  $diffall .= "\\end{NoHyper}\n";
+}
+$diffall .= "\\end{document}$newpost" if length $newpreamble ;
+if ( lc($encoding) ne "utf8" && lc($encoding) ne "ascii" ) {
+  print STDERR "Encoding output file to $encoding\n" if $verbose;
+  $diffall=Encode::encode($encoding,$diffall);
+  binmode STDOUT;
+}
+print $diffall;
+
+
+print STDERR "(",exetime()," s)\n","Done.\n" if $verbose;
+
+
+# liststringtoregex(liststring)
+# expands string with semi-colon separated list into a regular expression corresponding
+# matching any of the elements
+sub liststringtoregex {
+  my ($liststring)=@_;
+  my @elements=grep /\S/,split(";",$liststring);
+  if ( @elements) {
+    return('(?:(?:' . join(')|(?:',@elements) .'))');
+  } else {
+    return "";
+  }
+}
+
+# show_configuration
+# note that this is not encapsulated but uses variables from the main program
+# It is provided for convenience because in the future it is planned to allow output
+# to be modified based on what packages are read etc - this works only if the input files are actually read
+# whether or not additional files are provided
+sub show_configuration {
+  if ($showpreamble) {
+    print "\nPreamble commands:\n";
+    print $latexdiffpreamble ;
+  }
+
+  if ($showsafe) {
+    print "\nsafecmd: Commands safe within scope of $ADDOPEN $ADDCLOSE and $DELOPEN $DELCLOSE (unless excluded):\n";
+    print_regex_arr(@SAFECMDLIST);
+    print "\nsafecmd-exlude: Commands not safe within scope of $ADDOPEN $ADDCLOSE and $DELOPEN $DELCLOSE :\n";
+    print_regex_arr(@SAFECMDEXCL);
+    print "\nmboxsafecmd:  Commands safe only if they are surrounded by \\mbox command:\n";
+    print_regex_arr(@MBOXCMDLIST);
+    print "\nnmboxsafecmd: Commands not safe:\n";
+    print_regex_arr(@MBOXCMDEXCL);
+  }
+
+  if ($showtext) {
+    print "\nCommands with last argument textual (unless excluded) and safe in every context:\n";
+    print_regex_arr(@TEXTCMDLIST);
+    print "\nContext1 commands (last argument textual, command will be disabled in deleted passages, last argument will be shown as plain text):\n";
+    print_regex_arr(@CONTEXT1CMDLIST);
+    print "\nContext2 commands (last argument textual, command and its argument will be disabled in deleted passages):\n";
+    print_regex_arr(@CONTEXT2CMDLIST);
+    print "\nExclude list of Commands with last argument not textual (overrides patterns above):\n";
+    print_regex_arr(@TEXTCMDEXCL);
+  }
+
+
+  if ($showconfig) {
+    print "Configuration variables:\n";
+    print "ARRENV=$ARRENV\n";
+    print "COUNTERCMD=$COUNTERCMD\n";
+    print "FLOATENV=$FLOATENV\n";
+    print "ITEMCMD=$ITEMCMD\n";
+    print "LISTENV=$LISTENV\n";
+    print "MATHARRENV=$MATHARRENV\n";
+#    print "MATHARRREPL=$MATHARRREPL\n";     # this is not deprecated and thus no longer shown
+    print "MATHENV=$MATHENV\n";
+    print "MATHREPL=$MATHREPL\n";
+    print "MINWORDSBLOCK=$MINWORDSBLOCK\n";
+    print "PICTUREENV=$PICTUREENV\n";
+    print "SCALEDELGRAPHICS=$SCALEDELGRAPHICS\n";
+    print "VERBATIMENV=$VERBATIMENV\n";
+    print "VERBATIMLINEENV=$VERBATIMLINEENV\n";
+    print "CUSTOMDIFCMD=$CUSTOMDIFCMD\n";
+  }
+}
+
+
+
+## guess_encoding(filename)
+## reads the first 20 lines of filename and looks for call of inputenc package
+## if found, return the option of this package (encoding), otherwise return utf8
+sub guess_encoding {
+  my ($filename)=@_;
+  my ($i,$enc);
+  open (FH, $filename) or die("Couldn't open $filename: $!");
+  $i=0;
+  while (<FH>) {
+    next if /^\s*%/;    # skip comment lines
+    if (m/\\usepackage\[(\w*?)\]\{inputenc\}/) {
+      close(FH);
+      return($1);
+    }
+    last if (++$i > 20 ); # scan at most 20 non-comment lines
+  }
+  close(FH);
+  ### return("ascii");
+  return("utf8");
+}
+
+
+sub read_file_with_encoding {
+  my ($output);
+  my ($filename, $encoding) = @_;
+
+  if (lc($encoding) eq "utf8" ) {
+    open (FILE, "<:utf8",$filename) or die("Couldn't open $filename: $!");
+    local $/ ; # locally set record operator to undefined, ie. enable whole-file mode
+    $output=<FILE>;
+  } elsif ( lc($encoding) eq "ascii") {
+    open (FILE, $filename) or die("Couldn't open $filename: $!");
+    local $/ ; # locally set record operator to undefined, ie. enable whole-file mode
+    $output=<FILE>;
+  } else {
+    require Encode;
+    open (FILE, "<",$filename) or die("Couldn't open $filename: $!");
+    local $/ ; # locally set record operator to undefined, ie. enable whole-file mode
+    $output=<FILE>;
+    print STDERR "Converting $filename from $encoding to utf8\n" if $verbose;
+    $output=Encode::decode($encoding,$output);
+  }
+  close FILE;
+  if ($^O eq "linux" ) {
+    $output =~ s/\r\n/\n/g ;
+  }
+  return $output;
+}
+
+## %packages=list_packages(@preamble)
+## scans the arguments for \documentclass,\RequirePackage and \usepackage statements and constructs a hash
+## whose keys are the included packages, and whose values are the associated optional arguments
+#sub list_packages {
+#  my (@preamble)=@_;
+#  my %packages=();
+#  foreach $line ( @preamble ) {
+#    # get rid of comments
+#    $line=~s/(?<!\\)%.*$// ;
+#    if ( $line =~ m/\\(?:documentclass|usepackage|RequirePackage)(?:\[(.+?)\])?\{(.*?)\}/ ) {
+##      print STDERR "Found something: |$line|\n" if $debug;
+#      if (defined($1)) {
+#	$packages{$2}=$1;
+#      } else {
+#	$packages{$2}="";
+#      }
+#    }
+#  }
+#  return (%packages);
+#}
+
+
+# %packages=list_packages($preamble)
+# scans the arguments for \documentclass,\RequirePackage and \usepackage statements and constructs a hash
+# whose keys are the included packages, and whose values are the associated optional arguments
+# if argument of \usepackage or \RequirePackage is comma separated list, treat as different packages
+sub list_packages {
+  my ($preamble)=@_;
+  my %packages=();
+  my $pkg;
+
+  # remove comments
+  $preamble=~s/(?<!\\)%.*$//mg ;
+
+  while ( $preamble =~  m/\\(?:documentclass|usepackage|RequirePackage)(?:\[($brat_n)\])?\{(.*?)\}/gs ) {
+    if (defined($1)) {
+      foreach $pkg ( split /,/,$2 ) {
+	$packages{$pkg}=$1;
+      }
+    } else {
+      foreach $pkg ( split /,/,$2 ) {
+	$packages{$pkg}="";
+      }
+    }
+  }
+
+  # sometimes, class options are defined in such a way that they imply the loading and/or presence of a package
+  # so we also treat all class options as 'packages.
+  if ( $preamble =~  m/\\documentclass\s*\[($brat_n)\]\s*\{.*?\}/s ) {
+     foreach $pkg ( split /,/,$1 ) {
+       $pkg =~ s/\s//g ;   # remove space and newline characters
+       $packages{$pkg}="" unless exists($packages{$pkg});
+     }
+  }
+  return (%packages);
+}
+
+# Subroutine add_safe_commands modified from version provided by S. Gouezel
+# add_safe_commands($preamble)
+# scans the argument for \newcommand and \DeclareMathOperator,
+# and adds the created commands which are clearly safe to @SAFECMDLIST
+sub add_safe_commands {
+  my ($preamble)=@_;
+
+  # get rid of comments
+  $preamble=~s/(?<!\\)%.*$//mg ;
+
+  my $to_test = "";
+  # test for \DeclareMathOperator{\foo}{myoperator}
+  while ( $preamble =~ m/\DeclareMathOperator\s*\*?\{\\(\w*?)\}/osg) {
+    $to_test=$1;
+    if ($to_test ne "" and not iscmd($to_test,\@SAFECMDLIST,\@SAFECMDEXCL) and not iscmd($to_test, \@SAFECMDEXCL, [])) {
+      # one should add $to_test to the list of safe commands.
+      init_regex_arr_list(\@SAFECMDLIST, $to_test);
+      print STDERR "Adding $to_test to the list of safe commands\n" if $verbose;
+    }
+  }
+
+  while ( $preamble =~ m/\\(?:new|renew|provide)command\s*{\\(\w*)\}(?:|\[\d*\])\s*\{(${pat_n})\}/osg ) {
+    my $maybe_to_test  = $1;
+    my $should_be_safe = $2;
+    print STDERR "DEBUG Checking new command: maybe_to_test, should_be_safe: $1 $2\n" if $debug;
+    my $success = 0;
+    # skip custom diff commands
+    next if ($maybe_to_test =~ m/^(?:ADD|DEL)?${CUSTOMDIFCMD}$/);
+    ###print STDERR "DEBUG: really test it. \n";
+    # test if all latex commands inside it are safe
+    $success = 1;
+    if ($should_be_safe =~ m/\\\\/) {
+      $success = 0;
+    } else {
+      while ($should_be_safe =~ m/\\(\w+)/g) {
+	###	  print STDERR "DEBUG: Testing command $1 " if $debug;
+	$success = 0 unless iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL); ### or $1 eq "";
+	###        print STDERR " success=$success\n" if $debug;
+      }
+    }
+    ###      }
+    if ($success) {
+      $to_test = $maybe_to_test;
+      if (  not iscmd($to_test,\@SAFECMDLIST,\@SAFECMDEXCL) and not iscmd($to_test, \@SAFECMDEXCL, [])) {
+	#        # one should add $to_test to the list of safe commands.
+	init_regex_arr_list(\@SAFECMDLIST, $to_test);
+	print STDERR "Adding $to_test to the list of safe commands\n" if $verbose;
+      }
+    }
+  }
+}
+
+
+# helper function for flatten
+# remove \endinput at beginning of line and everything
+# following it, # if \endinput is not at the beginning of
+# the line, nothing will be removed. It is assumed that
+# this case is most common when \endinput is part of a
+# conditional clause.  The file will only be processed
+# correctly if the conditional is always false,
+# i.e. \endinput # not actually reached
+sub remove_endinput {
+  # s/// operates on default input
+  $_[0] =~ s/^\\endinput.*\Z//ms ;
+  return($_[0]);
+}
+
+# flatten($text,$preamble,$filename,$encoding)
+# expands \input and \include commands within text
+# expands \bibliography command with corresponding bbl file if available
+# expands \subfile command (from subfiles package - not part of standard text distribution)
+# expands \import etc commands (from import package - not part of standard text distribution)
+# preamble is scanned for includeonly commands
+# encoding is the encoding
+sub flatten {
+  my ($text,$preamble,$filename,$encoding)=@_;
+  my ($includeonly,$dirname,$fname,$newpage,$fullfile,$filecontent,$replacement,$begline,$inputcmd,$bblfile,$subfile,$command,$verbenv,$verboptions,$ignore,$fileonly);
+  my ($subpreamble,$subbody,$subpost);
+  my ($subdir,$subdirfull,$importfilepath);
+  require File::Basename ;
+  ###  require File::Spec ;    # now this is needed even if flatten option not given
+  $filename = File::Spec->rel2abs( $filename ) ;
+  ($ignore, $dirname, $fileonly) = File::Spec->splitpath($filename) ;
+  $bblfile = $filename;
+  $bblfile=~s/\.tex$//;
+  $bblfile.=".bbl";
+
+  if ( ($includeonly) = ($preamble =~ m/\\includeonly\{(.*?)\}/ ) ) {
+    $includeonly =~ s/,/|/g;
+  } else {
+    $includeonly = '.*?';
+  }
+
+  print STDERR "DEBUG: includeonly $includeonly\n" if $debug;
+
+
+  # Run through filter, to let filterscript have a pass if it was set
+  $text = filter($text);
+
+  # Recursively replace \\import, \\subimport, and related import commands
+  $text =~ s/(^(?:[^%\n]|\\%)*)(\\(sub)?(?:import|inputfrom|includefrom))\{(.*?)\}(?:[\s]*)\{(.*?)\}/{
+          #  (--------1-------)(--(=3=)-------------2-------------------)  (-4-)             (-5-)
+          # $1 is begline
+          # $2 is the import macro name
+          # $3 is (optional) prefix "sub"
+          # $4 is directory
+          # $5 is filename
+          $begline = (defined($1)? $1 : "");
+          $subdir = $4;
+          $fname = $5;
+          $fname .= ".tex" unless $fname =~ m|\.\w{3,4}$|;
+          print STDERR "DEBUG begline:", $begline, "\n" if $debug;
+          print STDERR "DEBUG", (defined($3)? "subimport_file:" : "import_file:"), $subdir, "\n" if $debug;
+          print STDERR "DEBUG file:", $fname, "\n" if $debug;
+
+          # subimport appends $subdir to the current $dirname.  import replaces it with an absolute path.
+          $subdirfull = (defined($3) ? File::Spec->catdir($dirname,$subdir) : $subdir);
+
+          $importfilepath = File::Spec->catfile($subdirfull, $fname);
+
+          print STDERR "importing importfilepath:", $importfilepath,"\n" if $verbose;
+          if ( -f $importfilepath ) {
+              # If file exists, replace input or include command with expanded input
+              #TODO: need remove_endinput & newpage similar to other replacements inside flatten
+              $replacement=flatten(read_file_with_encoding($importfilepath, $encoding), $preamble,$importfilepath,$encoding);
+          } else {
+              # if file does not exist, do not expand include or input command (do not warn if fname contains #[0-9] as it is then likely part of a command definition
+              # and is not meant to be expanded directly
+              print STDERR "WARNING: Could not find included file ",$importfilepath,". I will continue but not expand |$2|\n";
+              $replacement = $2;
+              $replacement .= "{$subdir}{$fname} % Processed";
+          }
+          "$begline$replacement";
+  }/exgm;
+
+  # recursively replace \\input and \\include files
+  $text =~ s/(^(?:[^%\n]|\\%)*)(\\input\{(.*?)\}|\\include\{(${includeonly}(?:\.tex)?)\})/{
+	    $begline=(defined($1)? $1 : "") ;
+	    $inputcmd=$2;
+	    $fname = $3 if defined($3) ;
+	    $fname = $4 if defined($4) ;
+            $newpage=(defined($4)? " \\newpage " : "") ;
+            #      # add tex extension unless there is a three or four letter extension already
+            $fname .= ".tex" unless $fname =~ m|\.\w{3,4}$|;
+            $fullfile = File::Spec->catfile($dirname,$fname);
+            print STDERR "DEBUG Beg of line match |$1|\n" if defined($1) && $debug ;
+            print STDERR "Include file $fname\n" if $verbose;
+            print STDERR "DEBUG looking for file ",$fullfile, "\n" if $debug;
+            # content of file becomes replacement value (use recursion), add \newpage if the command was include
+            if ( -f $fullfile ) {
+	      # If file exists, replace input or include command with expanded input
+	      $replacement=flatten(read_file_with_encoding($fullfile, $encoding), $preamble,$filename,$encoding);
+	      $replacement = remove_endinput($replacement);
+	      # \include always starts a new page; use explicit \newpage command to simulate this
+	    } else {
+	      # if file does not exist, do not expand include or input command (do not warn if fname contains #[0-9] as it is then likely part of a command definition
+              # and is not meant to be expanded directly
+	      print STDERR "WARNING: Could not find included file ",$fullfile,". I will continue but not expand |$inputcmd|\n" unless $fname =~ m(#[0-9]) ;
+	      $replacement = $inputcmd ;   # i.e. just the original command again -> make no change file does not exist
+	      $newpage="";
+	    }
+	    "$begline$newpage$replacement$newpage";
+          }/exgm;
+
+  # replace bibliography with bbl file if it exists
+  $text=~s/(^(?:[^%\n]|\\%)*)\\bibliography\{(.*?)\}/{
+           if ( -f $bblfile ){
+	     $replacement=read_file_with_encoding(File::Spec->catfile($bblfile), $encoding);
+	   } else {
+	     warn "Bibliography file $bblfile cannot be found. No flattening of \\bibliography done. Run bibtex on old and new files first";
+	     $replacement="\\bibliography{$2}";
+	   }
+	   $begline=(defined($1)? $1 : "") ;
+	   "$begline$replacement";
+  }/exgm;
+
+  # replace subfile with contents (subfile package)
+  $text=~s/(^(?:[^%\n]|\\%)*)\\subfile\{(.*?)\}/{
+           $begline=(defined($1)? $1 : "") ;
+     	   $fname = $2;
+           #      # add tex extension unless there is a three or four letter extension already
+           $fname .= ".tex" unless $fname =~ m|\.\w{3,4}|;
+           print STDERR "Include file as subfile $fname\n" if $verbose;
+           # content of file becomes replacement value (use recursion)
+           # now strip away everything outside and including \begin{document} and \end{document} pair#
+	   #             # note: no checking for comments is made
+           $fullfile=File::Spec->catfile($dirname,$fname);
+           if ( -f $fullfile) {
+	     # if file exists, expand \subfile command by contents of file
+	     $subfile=read_file_with_encoding($fullfile,$encoding) or die "Could not open included subfile ",$fullfile,": $!";
+	     ($subpreamble,$subbody,$subpost)=splitdoc($subfile,'\\\\begin\{document\}','\\\\end\{document\}');
+	     ###           $subfile=~s|^.*\\begin{document}||s;
+	     ###           $subfile=~s|\\end{document}.*$||s;
+	     $replacement=flatten($subbody, $preamble,$fullfile,$encoding);
+	     ### $replacement = remove_endinput($replacement);
+	   } else {
+	      # if file does not exist, do not expand subfile
+	      print STDERR "WARNING: Could not find subfile ",$fullfile,". I will continue but not expand |$2|\n" unless $fname =~ m(#[0-9]) ;
+	      $replacement = "\\subfile\{$2\}" ;   # i.e. just the original command again -> make no change file does not exist
+	    }
+
+	   "$begline$replacement";
+  }/exgm;
+
+  # replace \verbatiminput and \lstlistinginput
+  $text=~s/(^(?:[^%\n]|\\%)*)\\(verbatiminput\*?|lstinputlisting)$extraspace(\[$brat_n\])?$extraspace\{(.*?)\}/{
+     $begline=(defined($1)? $1 : "") ;
+     $command = $2 ;
+     $fname = $4 ;
+     $verboptions = defined($3)? $3 : "" ;
+     if ($command eq 'verbatiminput' ) {
+       $verbenv = "verbatim" ;
+     } elsif ($command eq 'verbatiminput*' ) {
+       $verbenv = "verbatim*" ;
+     } elsif ($command eq 'lstinputlisting' ) {
+       $verbenv = "lstlisting" ;
+     } else {
+       die "Internal errorL Unexpected verbatim input type $command.\n";
+     }
+     print STDERR "DEBUG Beg of line match |$begline|\n" if $debug ;
+     print STDERR "Include file $fname  verbatim\n" if $verbose;
+     print STDERR "DEBUG looking for file ",File::Spec->catfile($dirname,$fname), "\n" if $debug;
+     # content of file becomes replacement value (do not use recursion), add \newpage if the command was include
+     ###$replacement=read_file_with_encoding(File::Spec->catfile($dirname,$fname), $encoding) or die "Couldn't find file ",File::Spec->catfile($dirname,$fname),": $!";
+     $replacement=read_file_with_encoding(File::Spec->catfile($dirname,$fname), $encoding); # (cannot on apparent failure as this triggers for empty fie. Original:  or die "Couldn't find file ",File::Spec->catfile($dirname,$fname),": $!";
+     # Add a new line if it not already there (note that the matching operator needs to use different delimiters, as we are still inside an outer scope that takes precedence
+     $replacement .= "\n" unless  $replacement =~ m(\n$)  ;
+     "$begline\\begin{$verbenv}$verboptions\n$replacement\\end{$verbenv}\n";
+    }/exgm;
+
+  return($text);
+}
+
+
+# print_regex_arr(@arr)
+# prints regex array without x-ism expansion put in by pearl to stdout
+sub print_regex_arr {
+  my $dumstring;
+  $dumstring = join(" ",@_);     # PERL generates string (?-xism:^ref$) for quoted refex ^ref$
+  $dumstring =~ s/\(\?-xism:\^(.*?)\$\)/$1/g;   # remove string and ^,$ marks before output
+  print $dumstring,"\n";
+}
+
+
+# @lines=extrapream($type,...)
+# reads line from appendix or external file
+# (end of file after __END__ token)
+# if $type is a filename, it will read the file instead of reading from the appendix
+# otherwise it will screen appendix for line "%DIF $TYPE" and copy everything up to line
+# '%DIF END $TYPE' (where $TYPE is upcased version of $type)
+# extrapream('-nofail',$type) will---instead of failing---simply return nothing if
+# it does not find the matching line in appendix (do not use -nofail option with multiple types!)
+sub extrapream {
+  my @types=@_;
+  my ($type,$arg);
+  my $nofail=0;
+  ###my @retval=("%DIF PREAMBLE EXTENSION ADDED BY LATEXDIFF") ;
+  my @retval=();
+  my ($copy);
+
+  foreach $arg ( @types ) {
+    if ( $arg eq '-nofail' ) {
+      $nofail=1;
+      next;
+    }
+    $type=$arg;
+    $copy=0;
+    if ( -f $type || lc $type eq '/dev/null' ) {
+      print STDERR "Reading preamble file $type\n" if $verbose ;
+      open (FILE,$type) or die "Cannot open preamble file $type: $!";
+      if (defined($encoding)) {
+	binmode(FILE,":encoding($encoding)");
+      } else {
+	require Encode::Locale;
+	binmode(FILE,":encoding(locale)");
+      }
+      while (<FILE>) {
+	chomp ;
+	if ( $_ =~ m/%DIF PREAMBLE/ ) {
+	  push (@retval,"$_");
+	} else {
+	  push (@retval,"$_ %DIF PREAMBLE");
+	}
+      }
+    } else {    # not (-f $type)
+       $type=uc($type);   # upcase argument
+       print STDERR "Preamble Internal Type $type\n" if $verbose;
+       # save filehandle position (before first read this points to line after __END__)
+       # but seek DATA,0,0 resets it to the beginning of the file
+       # see https://stackoverflow.com/questions/4459601/how-can-i-use-data-twice
+       my $data_start = tell DATA;
+       while (<DATA>) {
+	 if ( m/^%DIF $type/ ) {
+	   $copy=1;
+	 } elsif ( m/^%DIF END $type/ ) {
+	   last;
+	 }
+	 chomp;
+	 push (@retval,"$_ %DIF PREAMBLE") if $copy;
+       }
+       if ( $copy == 0 ) {
+	 unless ($nofail) {
+	   print STDERR "\nPreamble style $type not implemented.\n";
+	   print STDERR "Write latexdiff -h to get help with available styles\n";
+	   exit(2);
+	 }
+       }
+       seek DATA,$data_start,0;    # rewind DATA handle to beginning of data record
+     }
+  }
+  ###push (@retval,"%DIF END PREAMBLE EXTENSION ADDED BY LATEXDIFF")  ;
+  return @retval;
+}
+
+
+# ($part1,$part2,$part3)=splitdoc($text,$word1,$word2)
+# splits $text into 3 parts at $word1 and $word2.
+# if neither $word1 nor $word2 exist, $part1 and $part3 are empty, $part2 is $text
+# If only $word1 or $word2 exist but not the other, output an error message.
+
+# NB this version avoids $` and $' for performance reason although it only makes a tiny difference
+# (in one test gain a tenth of a second for a 30s run)
+sub splitdoc {
+  my ($text,$word1,$word2)=@_;
+  my ($part1,$part2,$part3)=("","","");
+  my ($rest,$pos);
+
+  if ( $text =~ m/(^[^%]*)($word1)/mg ) {
+    $pos=pos $text;
+    $part1=substr($text,0,$pos-length($2));
+    $rest=substr($text,$pos);
+    if ( $rest =~ m/(^[^%]*)($word2)/mg ) {
+      $pos=pos $rest;
+      $part2=substr($rest,0,$pos-length($2));
+      $part3=substr($rest,$pos);
+    }
+    else {
+      die "$word1 and $word2 not in the correct order or not present as a pair." ;
+    }
+  } else {
+    $part2=$text;
+    die "$word2 present but not $word1." if ( $text =~ m/(^[^%]*)$word2/ms );
+  }
+  return ($part1,$part2,$part3);
+}
+
+
+
+
+
+# bodydiff($old,$new)
+sub bodydiff {
+  my ($oldwords, $newwords) = @_;
+  my @retwords;
+
+  print STDERR "(",exetime()," s)\n","Splitting into latex tokens \n" if $verbose;
+  print STDERR "Parsing $oldfile \n" if $verbose;
+  my @oldwords = splitlatex($oldwords);
+  print STDERR "Parsing $newfile \n" if $verbose;
+  my @newwords = splitlatex($newwords);
+
+  if ( $debug ) {
+    open(TOKENOLD,">","latexdiff.debug.tokenold");
+    print TOKENOLD join("***\n",@oldwords);
+    close(TOKENOLD);
+    open(TOKENNEW,">","latexdiff.debug.tokennew");
+    print TOKENNEW join("***\n",@newwords);
+    close(TOKENNEW);
+  }
+
+  print STDERR "(",exetime()," s)\n","Pass 1: Expanding text commands and merging isolated identities with changed blocks  " if $verbose;
+  pass1(\@oldwords, \@newwords);
+
+
+  print STDERR "(",exetime()," s)\n","Pass 2: inserting DIF tokens and mark up.  " if $verbose;
+  if ( $debug ) {
+    open(TOKENOLD,">","latexdiff.debug.tokenold2");
+    print TOKENOLD join("***\n",@oldwords);
+    close(TOKENOLD);
+    open(TOKENNEW,">","latexdiff.debug.tokennew2");
+    print TOKENNEW join("***\n",@newwords);
+    close(TOKENNEW);
+  }
+
+  @retwords=pass2(\@oldwords, \@newwords);
+
+  return(@retwords);
+}
+
+
+
+
+# @words=splitlatex($string)
+# split string according to latex rules
+# Each element of words is either
+# a word (including trailing spaces and punctuation)
+# a latex command
+# if there is white space in the beginning return that as first token
+sub splitlatex {
+  my ($inputstring) = @_ ;
+  my $string=$inputstring ;
+  # if input is empty, return empty list
+  length($string)>0 or return ();
+  $string=~s/^(\s*)//s;
+  my $leadin=$1;
+  length($string)>0 or return ($leadin);
+
+  my @retval=($string =~ m/$pat/osg);
+
+  if (length($string) != length(join("",@retval))) {
+    print STDERR "\nWARNING: Inconsistency in length of input string and parsed string:\n     This often indicates faulty or non-standard latex code.\n     In many cases you can ignore this and the following warning messages.\n Note that character numbers in the following are counted beginning after \\begin{document} and are only approximate." unless $ignorewarnings;
+    print STDERR "DEBUG Original length ",length($string),"  Parsed length ",length(join("",@retval)),"\n" if $debug;
+    print STDERR "DEBUG Input string:  |$string|\n" if (length($string)<500) && $debug;
+    print STDERR "DEBUG Token parsing: |",join("+",@retval),"|\n" if (length($string)<500) && $debug ;
+    @retval=();
+    # slow way only do this if other m//sg method fails
+    my $last = 0;
+    while ( $string =~ m/$pat/osg ) {
+      my $match=$&;
+      if ($last + length $& != pos $string  ) {
+	my $pos=pos($string);
+	my $offset=30<$last ? 30 : $last;
+	my $dum=substr($string,$last-$offset,$pos-$last+2*$offset);
+	my $dum1=$dum;
+	my $cnt=$#retval;
+	my $i;
+	$dum1 =~ s/\n/ /g;
+	unless ($ignorewarnings) {
+	  print STDERR "\n$dum1\n";
+	  print STDERR " " x 30,"^" x ($pos-$last)," " x 30,"\n";
+	  print STDERR "Missing characters near word " . (scalar @retval) . " character index: " . $last . "-" .  pos($string) . " Length: " . length($match) . " Match: |$match| (expected match marked above).\n";
+	}
+	  # put in missing characters `by hand'
+	push (@retval, substr($dum,$offset,$pos-$last-length($match)));
+#       Note: there seems to be a bug in substr with utf8 that made the following line output substr which were too long,
+#             using dum instead appears to work
+#	push (@retval, substr($string,$last, pos($string)-$last-length($match)));
+      }
+      push (@retval, $match);
+      $last=pos $string;
+    }
+
+  }
+
+  unshift(@retval,$leadin) if (length($leadin)>0);
+  return @retval;
+}
+
+
+# pass1( \@seq1,\@seq2)
+# Look for differences between seq1 and seq2.
+# Where an common-subsequence block is flanked by deleted or appended blocks,
+# and is shorter than $MINWORDSBLOCK words it is appended
+# to the last deleted or appended word.  If the block contains tokens other than words
+# or punctuation it is not merged.
+# Deleted or appended block consisting of words and safe commands only are
+# also merged, to prevent break-up in pass2 (after previous isolated words have been removed)
+# If there are commands with textual arguments (e.g. \caption) both in corresponding
+# appended and deleted blocks split them such that the command and opening bracket
+# are one token, then the rest is split up following standard rules, and the closing
+# bracket is a separate token, ie. turn
+# "\caption{This is a textual argument}" into
+# ("\caption{","This ","is ","a ","textual ","argument","}")
+# No return value.  Destructively changes sequences
+sub pass1 {
+  my $seq1 = shift ;
+  my $seq2 = shift ;
+
+  my $len1 = scalar @$seq1;
+  my $len2 = scalar @$seq2;
+  my $wpat=qr/^(?:[a-zA-Z.,'`:;?()!]*)[\s~]*$/;   #'
+
+  my ($last1,$last2)=(-1,-1) ;
+  my $cnt=0;
+  my $block=[];
+  my $addblock=[];
+  my $delblock=[];
+  my $todo=[];
+  my $instruction=[];
+  my $i;
+  my (@delmid,@addmid,@dummy);
+
+  my ($addcmds,$delcmds,$matchindex);
+  my ($addtextblocks,$deltextblocks);
+  my ($addtokcnt,$deltokcnt,$mattokcnt)=(0,0,0);
+  my ($addblkcnt,$delblkcnt,$matblkcnt)=(0,0,0);
+
+  my $adddiscard = sub {
+                      if ($cnt > 0 ) {
+			$matblkcnt++;
+			# just after an unchanged block
+#			print STDERR "Unchanged block $cnt, $last1,$last2 \n";
+                        if ($cnt < $MINWORDSBLOCK
+			    && $cnt==scalar (
+				     grep { /^$wpat/ || ( /^\\((?:[`'^"~=.]|[\w\d@*]+))((?:\[$brat_n\]|\{$pat_n\})*)/o
+							   && iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL)
+							   && scalar(@dummy=split(" ",$2))<3 ) }
+					     @$block) )  {
+			  # merge identical blocks shorter than $MINWORDSBLOCK
+			  # and only containing ordinary words
+			  # with preceding different word
+			  # We cannot carry out this merging immediately as this
+			  # would change the index numbers of seq1 and seq2 and confuse
+			  # the algorithm, instead we store in @$todo where we have to merge
+                          push(@$todo, [ $last1,$last2,$cnt,@$block ]);
+			}
+			$block = [];
+			$cnt=0; $last1=-1; $last2=-1;
+		      }
+		    };
+  my $discard=sub { $deltokcnt++;
+                    &$adddiscard; #($_[0],$_[1]);
+		    push(@$delblock,[ $seq1->[$_[0]],$_[0] ]);
+		    $last1=$_[0] };
+
+  my $add =   sub { $addtokcnt++;
+                    &$adddiscard; #($_[0],$_[1]);
+		    push(@$addblock,[ $seq2->[$_[1]],$_[1] ]);
+		    $last2=$_[1] };
+
+  my $match = sub { $mattokcnt++;
+                    if ($cnt==0) {   # first word of matching sequence after changed sequence or at beginning of word sequence
+		      $deltextblocks = extracttextblocks($delblock);
+		      $delblkcnt++ if scalar @$delblock;
+		      $addtextblocks = extracttextblocks($addblock);
+		      $addblkcnt++ if scalar @$addblock;
+
+		      # make a list of all TEXTCMDLIST commands in deleted and added blocks
+		      $delcmds = extractcommands($delblock);
+      		      $addcmds = extractcommands($addblock);
+		      # now find those text commands, which are found in both deleted and added blocks, and expand them
+		      # keygen(third argument of _longestCommonSubsequence) implies to sort on command (0th elements of $addcmd elements)
+		      # the calling format for longestCommonSubsequence has changed between versions of
+		      # Algorithm::Diff so we need to check which one we are using
+		      if ( $algodiffversion  > 1.15 ) {
+			### Algorithm::Diff 1.19
+			$matchindex=Algorithm::Diff::_longestCommonSubsequence($delcmds,$addcmds, 0, sub { $_[0]->[0] } );
+		      } else {
+			### Algorithm::Diff 1.15
+			$matchindex=Algorithm::Diff::_longestCommonSubsequence($delcmds,$addcmds, sub { $_[0]->[0] } );
+		      }
+
+		      for ($i=0 ; $i<=$#$matchindex ; $i++) {
+			if (defined($matchindex->[$i])){
+			  $j=$matchindex->[$i];
+			  @delmid=splitlatex($delcmds->[$i][3]);
+			  @addmid=splitlatex($addcmds->[$j][3]);
+			  while (scalar(@$deltextblocks)  && $deltextblocks->[0][0]<$delcmds->[$i][1]) {
+			    my ($index,$block,$cnt)=@{ shift(@$deltextblocks) };
+			    push(@$todo, [$index,-1,$cnt,@$block]);
+			  }
+			  push(@$todo, [ $delcmds->[$i][1],-1,-1,$delcmds->[$i][2],@delmid,$delcmds->[$i][4]]);
+
+			  while (scalar(@$addtextblocks) && $addtextblocks->[0][0]<$addcmds->[$j][1]) {
+			    my ($index,$block,$cnt)=@{ shift(@$addtextblocks) };
+			    push(@$todo, [-1,$index,$cnt,@$block]);
+			  }
+			  push(@$todo, [ -1,$addcmds->[$j][1],-1,$addcmds->[$j][2],@addmid,$addcmds->[$j][4]]);
+			}
+		      }
+		      # mop up remaining textblocks
+		      while (scalar(@$deltextblocks)) {
+			my ($index,$block,$cnt)=@{ shift(@$deltextblocks) } ;
+			push(@$todo, [$index,-1,$cnt,@$block]);
+		      }
+		      while (scalar(@$addtextblocks)) {
+			my ($index,$block,$cnt)=@{ shift(@$addtextblocks) };
+			push(@$todo, [-1,$index,$cnt,@$block]);
+		      }
+
+		      $addblock=[];
+		      $delblock=[];
+		    }
+		    push(@$block,$seq2->[$_[1]]);
+		    $cnt++  };
+
+  my $keyfunc = sub { join("  ",split(" ",shift())) };
+
+  traverse_sequences($seq1,$seq2, { MATCH=>$match, DISCARD_A=>$discard, DISCARD_B=>$add }, $keyfunc );
+
+
+  # now carry out the merging/splitting.  Refer to elements relative from
+  # the end (with negative indices) as these offsets don't change before the instruction is executed
+  # cnt>0: merged small unchanged groups with previous changed blocks
+  # cnt==-1: split textual commands into components
+  foreach $instruction ( @$todo) {
+    ($last1,$last2,$cnt,@$block)=@$instruction ;
+    if ($cnt>=0) {
+      splice(@$seq1,$last1-$len1,1+$cnt,join("",$seq1->[$last1-$len1],@$block)) if $last1>=0;
+      splice(@$seq2,$last2-$len2,1+$cnt,join("",$seq2->[$last2-$len2],@$block)) if $last2>=0;
+    } else {
+      splice(@$seq1,$last1-$len1,1,@$block) if $last1>=0;
+      splice(@$seq2,$last2-$len2,1,@$block) if $last2>=0;
+    }
+  }
+
+  if ($verbose) {
+    print STDERR "\n";
+    print STDERR "  $mattokcnt matching  tokens in $matblkcnt blocks.\n";
+    print STDERR "  $deltokcnt discarded tokens in $delblkcnt blocks.\n";
+    print STDERR "  $addtokcnt appended  tokens in $addblkcnt blocks.\n";
+  }
+}
+
+
+# extracttextblocks(\@blockindex)
+# $blockindex has the following format
+# [ [ token1, index1 ], [token2, index2],.. ]
+# where index refers to the index in the original old or new word sequence
+# Returns: reference to an array of the form
+# [[ $index, $textblock, $cnt ], ..
+# where $index index of block to be merged
+#       $textblock contains all the words to be merged with the word at $index (but does not contain this word)
+#       $cnt   is length of block
+#
+# requires: iscmd
+#
+sub extracttextblocks {
+  my $block=shift;
+  my ($i,$token,$index);
+  my $textblock=[];
+  my $last=-1;
+  my $wpat=qr/^(?:[a-zA-Z.,'`:;?()!]*)[\s~]*$/;  #'
+  my $retval=[];
+
+  # we redefine locally $extraspace (shadowing the global definition) to capture command sequences with intervening spaces no matter what the global setting
+  # this is done so we can capture those commands with a predefined number of arguments without having to introduce them again explicitly here
+  my $extraspace='\s*';
+
+  for ($i=0;$i< scalar @$block;$i++) {
+    ($token,$index)=@{ $block->[$i] };
+    # store pure text blocks
+    if ($token =~ /$wpat/ ||  ( $token =~/^\\((?:[`'^"~=.]|[\w\d@\*]+))((?:${extraspace}\[$brat_n\]${extraspace}|${extraspace}\{$pat_n\})*)/
+				&& iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL)
+				&& !iscmd($1,\@TEXTCMDLIST,\@TEXTCMDEXCL))) {
+      # we have text or a command which can be treated as text
+      if ($last<0) {
+	# new pure-text block
+	$last=$index;
+      } else {
+	# add to pure-text block
+	push(@$textblock, $token);
+      }
+    } else {
+      # it is not text
+      if (scalar(@$textblock)) {
+	push(@$retval,[ $last, $textblock, scalar(@$textblock) ]);
+      }
+      $textblock=[];
+      $last=-1;
+    }
+  }
+  # finish processing a possibly unfinished block before returning
+  if (scalar(@$textblock)) {
+    push(@$retval,[ $last, $textblock, scalar(@$textblock) ]);
+  }
+  return($retval)
+}
+
+
+
+# extractcommands( \@blockindex )
+# $blockindex has the following format
+# [ [ token1, index1 ], [token2, index2],.. ]
+# where index refers to the index in the original old or new word sequence
+# Returns: reference to an array of the form
+# [ [ "\cmd1", index, "\cmd1[optarg]{arg1}{", "arg2" ,"} " ],..
+# where index is just taken from input array
+# command must have a textual argument as last argument
+#
+# requires: iscmd
+#
+sub extractcommands {
+  my $block=shift;
+  my ($i,$token,$index,$cmd,$open,$mid,$closing);
+  my $retval=[];
+
+  # we redefine locally $extraspace (shadowing the global definition) to capture command sequences with intervening spaces no matter what the global setting
+  # this is done so we can capture those commands with a predefined number of arguments without having to introduce them again explicitly here
+  my $extraspace='\s*';
+
+  for ($i=0;$i< scalar @$block;$i++) {
+    ($token,$index)=@{ $block->[$i] };
+    # check if token is an alphanumeric command sequence with at least one non-optional argument
+    # \cmd[...]{...}{last argument}
+    # Capturing in the following results in these associations
+    # $1: \cmd[...]{...}{
+    # $2: \cmd
+    # $3: last argument
+    # $4: }  + trailing spaces
+    if ( ( $token =~ m/^(\\([\w\d\*]+)(?:${extraspace}\[$brat_n\]|${extraspace}\{$pat_n\})*${extraspace}\{)($pat_n)(\}\s*)$/so )
+	 && iscmd($2,\@TEXTCMDLIST,\@TEXTCMDEXCL) ) {
+      print STDERR "DEBUG EXTRACTCOMMANDS Match |$1|$2|$3|$4|$index \n" if $debug;
+      #      push(@$retval,[ $2,$index,$1,$3,$4 ]);
+      ($cmd,$open,$mid,$closing) = ($2,$1,$3,$4) ;
+      $closing =~ s/\}/\\RIGHTBRACE/ ;
+      push(@$retval,[ $cmd,$index,$open,$mid,$closing ]);
+    }
+  }
+  return $retval;
+}
+
+# iscmd($cmd,\@regexarray,\@regexexcl) checks
+# return 1 if $cmd matches any of the patterns in the
+# array $@regexarray, and none of the patterns in \@regexexcl, otherwise return 0
+sub iscmd {
+  my ($cmd,$regexar,$regexexcl)=@_;
+  my ($ret)=0;
+  ### print STDERR "DEBUG: iscmd($cmd)=" if $debug;
+  foreach $pat ( @$regexar ) {
+    if ( $cmd =~ m/^${pat}$/ ) {
+      $ret=1 ;
+      last;
+    }
+  }
+  ### print STDERR "0\n" if ($debug && !$ret) ;
+  return 0 unless $ret;
+  foreach $pat ( @$regexexcl ) {
+    ### print STDERR "0\n" if ( $debug && $cmd =~ m/^${pat}$/) ;
+    return 0 if ( $cmd =~ m/^${pat}$/ );
+  }
+  ### print STDERR "1\n" if $debug;
+  return 1;
+}
+
+
+# pass2( \@seq1,\@seq2)
+# Look for differences between seq1 and seq2.
+# Mark begin and end of deleted and appended sequences with tags $DELOPEN and $DELCLOSE
+# and $ADDOPEN and $ADDCLOSE, respectively, however exclude { } & and all comands, unless
+# they match an element of the whitelist (SAFECMD)
+# For words in TEXTCMD but not in SAFECMD, enclose interior with $ADDOPEN and $ADDCLOSE brackets
+# Deleted comment lines are marked with %DIF <
+# Added comment lines are marked with %DIF >
+sub pass2 {
+  my $seq1 = shift ;
+  my $seq2 = shift ;
+
+  my ($addtokcnt,$deltokcnt,$mattokcnt)=(0,0,0);
+  my ($addblkcnt,$delblkcnt,$matblkcnt)=(0,0,0);
+
+  my $retval = [];
+  my $delhunk   = [];
+  my $addhunk   = [];
+
+  my $discard = sub { $deltokcnt++;
+                      push ( @$delhunk, $seq1->[$_[0]]) };
+
+  my $add = sub { $addtokcnt++;
+                  push ( @$addhunk, $seq2->[$_[1]]) };
+
+  my $match = sub { $mattokcnt++;
+		    if ( scalar @$delhunk ) {
+                      $delblkcnt++;
+		      # mark up changes, but comment out commands
+                      push @$retval,marktags($DELMARKOPEN,$DELMARKCLOSE,$DELOPEN,$DELCLOSE,$DELCMDOPEN,$DELCMDCLOSE,$DELCOMMENT,$delhunk);
+		      $delhunk = [];
+		    }
+                    if ( scalar @$addhunk ) {
+                      $addblkcnt++;
+                      # we mark up changes, but simply quote commands
+                      push @$retval,marktags($ADDMARKOPEN,$ADDMARKCLOSE,$ADDOPEN,$ADDCLOSE,"","",$ADDCOMMENT,$addhunk);
+		      $addhunk = [];
+		    }
+		    push(@$retval,$seq2->[$_[1]]) };
+
+  my $keyfunc = sub { join("  ",split(" ",shift())) };
+
+  traverse_sequences($seq1,$seq2, { MATCH=>$match, DISCARD_A=>$discard, DISCARD_B=>$add }, $keyfunc );
+  # clear up unprocessed hunks
+  push @$retval,marktags($DELMARKOPEN,$DELMARKCLOSE,$DELOPEN,$DELCLOSE,$DELCMDOPEN,$DELCMDCLOSE,$DELCOMMENT,$delhunk) if scalar @$delhunk;
+  push @$retval,marktags($ADDMARKOPEN,$ADDMARKCLOSE,$ADDOPEN,$ADDCLOSE,"","",$ADDCOMMENT,$addhunk) if scalar @$addhunk;
+
+
+  if ($verbose) {
+    print STDERR "\n";
+    print STDERR "  $mattokcnt matching  tokens. \n";
+    print STDERR "  $deltokcnt discarded tokens in $delblkcnt blocks.\n";
+    print STDERR "  $addtokcnt appended  tokens in $addblkcnt blocks.\n";
+  }
+  return(@$retval);
+}
+
+# marktags($openmark,$closemark,$open,$close,$opencmd,$closecmd,$comment,\@block)
+# returns ($openmark,$open,$block,$close,$closemark) if @block contains no commands (except white-listed ones),
+# braces, ampersands, or comments
+# mark comments with $comment
+# exclude all other exceptions from scope of open, close like this
+# ($openmark, $open,...,$close, $opencmd,command, command,$closecmd, $open, ..., $close, $closemark)
+# If $opencmd begins with "%" marktags assumes it is operating on a deleted block, otherwise on an added block
+sub marktags {
+  my ($openmark,$closemark,$open,$close,$opencmd,$closecmd,$comment,$block)=@_;
+  my $word;
+  my (@argtext);
+  my $retval=[];
+  my $noncomment=0;
+  my $cmd=-1;    # -1 at beginning 0: last token written is a ordinary word
+                 # 1: last token written is a command
+                # for keeping track whether we are just in a command sequence or in a word sequence
+  my $cmdcomment= ($opencmd =~ m/^%/);  # Flag to indicate whether opencmd is a comment (i.e. if we intend to simply comment out changed commands)
+  my ($command,$commandword,$closingbracket) ; # temporary variables needed below to remember sub-pattern matches
+
+
+
+# split this block to split sequences joined in pass1
+  ### print STDERR "DEBUG: marktags before splitlatex blocksplit ",join("|",@$block),"\n" if $debug;
+  @$block=splitlatex(join "",@$block);
+  ### print STDERR "DEBUG: marktags $openmark,$closemark,$open,$close,$opencmd,$closecmd,$comment\n" if $debug;
+  print STDERR "DEBUG: after splitlatex ",join("|",@$block),"\n" if $debug;
+
+  # we redefine locally $extraspace (shadowing the global definition) to capture command sequences with intervening spaces no matter what the global setting
+  # this is done so we can capture those commands with a predefined number of arguments without having to introduce them again explicitly here
+  my $extraspace_mt='\s*';
+
+
+  foreach (@$block) {
+    $word=$_;
+    if ( $word =~ s/^%/%$comment/ ) {
+      # a comment
+      if ($cmd==1) {
+	push (@$retval,$closecmd) ;
+	$cmd=-1;
+      }
+      push (@$retval,$word);
+      next;
+    }
+    if ( $word =~ m/^\s*$/ ) {
+      ### print STDERR "DEBUG MARKTAGS: whitespace detected |$word| cmdcom |$cmdcomment| |$opencmd|\n" if $debug;
+      # a sequence of white-space characters - this should only ever happen for the first element of block.
+      # in deleted block, omit, otherwise just copy it in
+      if ( ! $cmdcomment) {   # ignore in deleted blocks
+	push(@$retval,$word);
+      }
+      next;
+    }
+    if (! $noncomment) {
+      push (@$retval,$openmark);
+      $noncomment=1;
+    }
+    # negative lookahead pattern (?!) in second clause is put in to avoid matching \( .. \) patterns
+    # also note that second pattern will match \\
+    if (  $word =~ /^[&{}\[\]]/ || ( $word =~ /^\\(?!\()(\\|[`'^"~=.]|[\w*@]+)/ &&  !iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL)) ) {
+      ###    if ( $word =~ /^[&{}\[\]]/ || ( $word =~ /^\\([\w*@\\% ]+)/ && !iscmd($1,\@SAFECMDLIST,\@SAFECMDEXCL)) ) {
+      # word is a command or other significant token (not in SAFECMDLIST)
+	## same conditions as in subroutine extractcommand:
+	# check if token is an alphanumeric command sequence with at least one non-optional argument
+	# \cmd[...]{...}{last argument}
+	# Capturing in the following results in these associations
+	# $1: \cmd[...]{...}{
+	# $2: cmd
+	# $3: last argument
+	# $4: }  + trailing spaces
+	### pre-0.3    if ( ( $token =~ m/^(\\([\w\d\*]+)(?:\[$brat0\]|\{$pat_n\})*\{)($pat_n)(\}\s*)$/so )
+      if ( ( $word =~ m/^(\\([\w\d\*]+)(?:${extraspace_mt}\[$brat_n\]|${extraspace_mt}\{$pat_n\})*${extraspace_mt}\{)($pat_n)(\}\s*)$/so )
+	   && (iscmd($2,\@TEXTCMDLIST,\@TEXTCMDEXCL)|| iscmd($2,\@MATHTEXTCMDLIST,\@MATHTEXTCMDEXCL))
+           && ( !$cmdcomment || !iscmd($2,\@CONTEXT2CMDLIST, \@CONTEXT2CMDEXCL) )  ) {
+	# Condition 1: word is a command? - if yes, $1,$2,.. will be set as above
+        # Condition 2: word is a text command - we mark up the interior of the word. There is a separate check for MATHTEXTCMDLIST
+        #              because for $mathmarkup=WHOLE, the commands should not be split in pass1 (ie. math mode commands are not in
+        #              TEXTCMDLIST, but the interior of MATHTEXT commnds should be highlighted in both deleted and added blocks
+        # Condition 3: But if we are in a deleted block ($cmdcomment=1) and
+        #            $2 (the command) is in context2, just treat it as an ordinary command (i.e. comment it open with $opencmd)
+        # Because we do not want to disable this command
+	# here we do not use $opencmd and $closecmd($opencmd is empty)
+	print STDERR "DEBUG: Detected text |$word| but not safe command \$2: $2 \$3: $3\n." if $debug;
+	if ($cmd==1) {
+	  push (@$retval,$closecmd) ;
+	} elsif ($cmd==0) {
+	  push (@$retval,$close) ;
+	}
+        $command=$1; $commandword=$2; $closingbracket=$4;
+	@argtext=splitlatex($3);   # split textual argument into tokens
+	# and mark it up (but we do not need openmark and closemark)
+        # insert command with initial arguments, marked-up final argument, and closing bracket
+	if ( $cmdcomment && iscmd($commandword,\@CONTEXT1CMDLIST, \@CONTEXT1CMDEXCL) ) {
+	  # context1cmd in a deleted environment; delete command itself but keep last argument, marked up
+	  push (@$retval,$opencmd);
+	  $command =~ s/\n/\n${opencmd}/sg ; # repeat opencmd at the beginning of each line
+	  # argument, note that the additional comment character is included
+          # to suppress linebreak after opening parentheses, which is important
+          # for latexrevise
+          push (@$retval,$command,"%\n{$AUXCMD\n",marktags("","",$open,$close,$opencmd,$closecmd,$comment,\@argtext),$closingbracket);
+        } elsif ( iscmd($commandword,,\@MATHTEXTCMDLIST, \@MATHTEXTCMDEXCL) ) {
+	  # MATHBLOCK pseudo command: consider all commands safe, except & and \\, \begin and \end and a few package sprcific one (look at UNSAFEMATHCMD definition)
+	  # Keep these commands even in deleted blocks, hence set $opencmd and $closecmd (5th and 6th argument of marktags) to
+	  # ""
+	  local @SAFECMDLIST=(".*");
+	  local @SAFECMDEXCL=('\\','\\\\',@UNSAFEMATHCMD);
+	  push(@$retval,$command,marktags("","",$open,$close,"","",$comment,\@argtext)#@argtext
+                       ,$closingbracket);
+        } else {
+	  # normal textcmd or context1cmd in an added block
+	  push (@$retval,$command,marktags("","",$open,$close,$opencmd,$closecmd,$comment,\@argtext),$closingbracket);
+	}
+	push (@$retval,$AUXCMD,"\n") if $cmdcomment ;
+	$cmd=-1 ;
+      } elsif ( $cmdcomment &&
+	       ( $word =~ m/^(\\([\w\d\*]+)(?:${extraspace_mt}\[$brat_n\]|${extraspace_mt}\{$pat_n\})*${extraspace_mt}\{)($pat_n)(\}\s*)/so )
+	       && iscmd($2,\@KEEPCMDLIST, \@KEEPCMDEXCL) ) {
+	# 'keepcmd' in a deleted environment: keep  the command as is
+	push (@$retval,$close) if $cmd==0 ;
+	push (@$retval,$word);
+	$cmd=-1;  # pretend we are at the beginning of a sequence because we do not want to add an additional $closecmd or $close before the next token, no matter what it is
+      } else {
+	# ordinary command
+	push (@$retval,$opencmd) if $cmd==-1 ;
+	push (@$retval,$close,$opencmd) if $cmd==0 ;
+	$word =~ s/\n/\n${opencmd}/sg if $cmdcomment ;   # if opencmd is a comment, repeat this at the beginning of every line
+        ### print STDERR "MARKTAGS: Add command |$word|\n";
+	push (@$retval,$word);
+	$cmd=1;
+      }
+    } else {
+      ###print STDERR "DEBUG MARKTAGS is an ordinary word or SAFECMD command \n" if $debug;
+      # just an ordinary word or command in SAFECMD
+      push (@$retval,$open) if $cmd==-1 ;
+      push (@$retval,$closecmd,$open) if $cmd==1 ;
+      ###TODO:  check here if it is a command in MBOXCMD list, and surround it with \mbox{...}
+      ### $word =~ /^\\(?!\()(\\|[`'^"~=.]|[\w*@]+)/ &&  iscmd($1,\@MBOXCMDLIST,\@MBOXCMDEXCL))
+      ### but actually this check has been carried out already so can simply check if word begins with backslash
+      if ( $word =~ /^\\(?!\()(\\|[`'^"~=.]|[\w*@]+)(.*?)(\s*)$/s &&  iscmd($1,\@MBOXCMDLIST,\@MBOXCMDEXCL)) {
+	# $word is a safe command in MBOXCMDLIST
+	###print STDERR "DEBUG Mboxsafecmd detected:$word:\n" if $debug ;
+	push(@$retval,"\\mbox{$AUXCMD\n\\" . $1 . $2 . $3 ."}\\hskip0pt$AUXCMD\n" );
+      } else {
+	# $word is a normal word or a safe command (not in MBOXCMDLIST)
+	push (@$retval,$word);
+      }
+      $cmd=0;
+    }
+  }
+  push (@$retval,$close) if $cmd==0;
+  push (@$retval,$closecmd) if $cmd==1;
+
+  push (@$retval,$closemark) if ($noncomment);
+  return @$retval;
+}
+
+#used in preprocess
+sub take_comments_and_newline_from_frac() {
+  # some special magic for common usage of frac, which does not conform to the latexdiff requirements but can be made to fit
+  # note that this is a rare exception to the general rule that the new tex can be reconstructed from the diff file
+
+  # regex that matches space and comment characters
+  my $space = qr/\s|%[^\n]*?/;
+  # \frac {abc} -> \frac{abc}
+  # \frac1 -> \frac{1}
+  # \frac a -> \frac{a}
+  # \frac \lambda -> \frac{\lambda}
+  s/\\frac(?|${space}+\{($pat_n)\}|${space}*(\d)|${space}+(\w)|${space}*(\\[a-zA-Z]+))/\\frac\{$1\}/g;
+  # same as above for the second argument of frac
+  s/\\frac(\{$pat_n\})(?|${space}*\{($pat_n)\}|${space}*(\d)|${space}+(\w)|${space}*(\\[a-zA-Z]+))/\\frac$1\{$2\}/g;
+}
+
+# preprocess($string, ..)
+# carry out the following pre-processing steps for all arguments:
+# 1. Remove leading white-space
+#    Change \{ to \QLEFTBRACE and \} to \QRIGHTBRACE and \& to \AMPERSAND
+# #.   Change {,},\frac in comments to \CLEFTBRACE, \CRIGHTBRACE, \CFRAC
+# 2. mark all first empty line (in block of several) with \PAR tokens
+# 3. Convert all '\%' into '\PERCENTAGE ' and all '\$' into \DOLLAR to make parsing regular expressions easier
+# 4. Convert all \verb|some verbatim text| commands (where | can be an arbitrary character)
+#    into \verb{hash}  (also lstinline)
+# 5. Convert \begin{verbatim} some verbatim text \end{verbatim} into \verbatim{hash}  (not only verbatim, all patterns matching VERBATIMENV)
+# 6. Convert _n into \SUBSCRIPTNB{n} and _{nnn} into \SUBSCRIPT{nn}
+# 7. Convert ^n into \SUPERSCRIPTNB{n} and ^{nnn} into \SUPERSCRIPT{nn}
+# 8. a. Convert $$ $$ into \begin{DOLLARDOLLAR} \end{DOLLARDOLLAR}
+#    b. Convert \[ \] into \begin{SQUAREBRACKET} \end{SQUAREBRACKET}
+# 9. Convert all picture environmentent (\begin{PICTUREENV} .. \end{PICTUREENV} \PICTUREBLOCKenv
+#     For math-mode COARSE,WHOLE or NONE option -convert all \begin{MATH} .. \end{MATH}
+#    into \MATHBLOCKmath{...} commands, where MATH/math is any valid math environment
+
+# 10. Add final token STOP to the very end.  This is put in because the algorithm works better if the last token is identical.  This is removed again in postprocessing.
+#
+# NB: step 6 and 7 is likely to  convert some "_" inappropriately, e.g. in file
+#     names or labels but it does not matter because they are converted back in the postprocessing step
+# Returns: leading white space removed in step 1
+sub preprocess {
+  for (@_) {
+
+
+    # change in \verb and similar commands - note that I introduce an extra space here so that the
+    #       already hashed variants do not trigger again
+    # transform \lstinline{...}
+#    s/\\lstinline(\[$brat0\])?(\{(?:.*?)\})/"\\DIFlstinline". $1 ."{". tohash(\%verbhash,"$2") ."}"/esg;
+#    s/\\lstinline(\[$brat0\])?((\S).*?\2)/"\\DIFlstinline". $1 ."{". tohash(\%verbhash,"$2") ."}"/esg;
+    s/\\lstinline((?:\[$brat_n\])?)(\{(?:.*?)\})/"\\DIFlstinline". $1 ."{". tohash(\%verbhash,"$2") ."}"/esg;
+    s/\\lstinline((?:\[$brat_n\])?)(([^\s\w]).*?\3)/"\\DIFlstinline". $1 ."{". tohash(\%verbhash,"$2") ."}"/esg;
+    s/\\(verb\*?|lstinline)([^\s\w])(.*?)\2/"\\DIF${1}{". tohash(\%verbhash,"${2}${3}${2}") ."}"/esg;
+
+    #    Change \{ to \QLEFTBRACE, \} to \QRIGHTBRACE, and \& to \AMPERSAND
+    s/(?<!\\)\\\{/\\QLEFTBRACE /sg;
+    s/(?<!\\)\\\}/\\QRIGHTBRACE /sg;
+    s/(?<!\\)\\&/\\AMPERSAND /sg;
+# replace {,}, \frac in comments with \\CLEFTBRACE,\\CRIGHTBRACE, CFRAC to protect from special treatment
+    1 while s/((?<!\\)%.*)\{(.*)$/$1\\CLEFTBRACE $2/mg ;
+    1 while s/((?<!\\)%.*)\}(.*)$/$1\\CRIGHTBRACE $2/mg ;
+    1 while s/((?<!\\)%.*)\\frac(.*)$/$1\\CFRAC $2/mg ;
+
+    s/(?<!\\)\\%/\\PERCENTAGE /g ;  # (?<! is negative lookbehind assertion to prevent \\% from being converted
+    s/(?<!\\)\\\$/\\DOLLAR /g ;  # (?<! is negative lookbehind assertion to prevent \\$ from being converted
+    s/\\begin\{($VERBATIMENV)\}(.*?)\\end\{\1\}/"\\${1}{". tohash(\%verbhash,"${2}") . "}"/esg;
+    s/\\begin\{($VERBATIMLINEENV)\}(.*?)\\end\{\1\}/"\\begin{$1}". linecomment($2) . "\\end{$1}"/esg;
+
+    # mark all first empty line (in block of several) with \PAR tokens
+    s/\n(\s*?)\n((?:\s*\n)*)/\n$1\\PAR\n$2/g ;
+    # Convert _n or _\cmd into \SUBSCRIPTNB{n} or \SUBSCRIPTNB{\cmd} and _{nnn} into \SUBSCRIPT{nn}
+    1 while s/(?<!\\)_(\s*([^{\\\s]|\\\w+))/\\SUBSCRIPTNB{$1}/g ;
+    1 while s/(?<!\\)_(\s*{($pat_n)})/\\SUBSCRIPT$1/g ;
+    # Convert ^n into \SUPERSCRIPTNB{n} and ^{nnn} into \SUPERSCRIPT{nn}
+    1 while s/(?<!\\)\^(\s*([^{\\\s]|\\\w+))/\\SUPERSCRIPTNB{$1}/g ;
+    1 while s/(?<!\\)\^(\s*{($pat_n)})/\\SUPERSCRIPT$1/g ;
+    # Convert  \sqrt{n} into \SQRT{n}  and  \sqrt nn into SQRTNB{nn}
+    1 while s/(?<!\\)\\sqrt(\s*([^{\\\s]|\\\w+))/\\SQRTNB{$1}/g ;
+    1 while s/(?<!\\)\\sqrt(\s*{($pat_n)})/\\SQRT$1/g ;
+    # Convert $$ $$ into \begin{DOLLARDOLLAR} \end{DOLLARDOLLAR}
+    s/\$\$(.*?)\$\$/\\begin{DOLLARDOLLAR}$1\\end{DOLLARDOLLAR}/sg;
+    # Convert \[ \] into \begin{SQUAREBRACKET} \end{SQUAREBRACKET}
+    s/(?<!\\)\\\[/\\begin{SQUAREBRACKET}/sg;
+    s/\\\]/\\end{SQUAREBRACKET}/sg;
+    # Convert all picture environmentent (\begin{PICTUREENV} .. \end{PICTUREENV} \PICTUREBLOCKenv
+    s/\\begin\{($PICTUREENV)}(.*?)\\end\{\1}/\\PICTUREBLOCK$1\{$2\}/sg;
+    #    For math-mode COARSE,WHOLE or NONE option -convert all \begin{MATH} .. \end{MATH}
+    #    into \MATHBLOCKMATH{...} commands, where MATH is any valid math environment
+    #    Also convert all array environments into ARRAYBLOCK environments.
+    #    Where these environments have arguments they become optional arguments to the MATHBLOCK command enclosed in < > brackets
+    #    Example: \begin{alignat}{3} ... \end{alignat} will turn into \MATHBLOCKalignat[{3}]{ ... }
+
+    if ( $mathmarkup != FINE ) {
+      # DIFANCHORARRB and DIFANCHORARRE, DIFANCHORMATHB and DIFANCHORMATHE markers are inserted here to encourage the matching algorithm
+      # to always match up the closing brace. Otherwise sometimes one ends up with a situation where
+      # the closing brace is deleted and added at another point. The deleted closing brace is then
+      # prevented by a %DIFDELCMD, leading to material leaking in or out of the math environment.
+      # The anchors are removed in post-processing again. (note that they are simple text to cause least amount of complications
+      # Admittedly, this is something of a hack and will not always work. If it does not, then one needs to
+      # resort to WHOLE or FINE, or NONE math mode processing.
+      s/\\begin\{($ARRENV)}(.*?)\\end\{\1}/\\ARRAYBLOCK$1\{$2\\DIFANCHORARRB \}\\DIFANCHORARRE /sg;
+
+      take_comments_and_newline_from_frac();
+
+      # Convert Math environments with arguments
+      s/\\begin\{($MATHENV|$MATHARRENV|SQUAREBRACKET)\}((?:\[$brat_n\])|(?:\{$pat_n\}))+(.*?)\\end\{\1\}/\\MATHBLOCK$1\[$2\]\{$3\\DIFANCHORMATHB \}\\DIFANCHORMATHE /sg;
+      # Convert Math environments without arguments
+      s/\\begin\{($MATHENV|$MATHARRENV|SQUAREBRACKET)\}(.*?)\\end\{\1\}/\\MATHBLOCK$1\{$2\\DIFANCHORMATHB \}\\DIFANCHORMATHE /sg;
+    }
+
+    # add final token " STOP"
+    $_ .= " STOP"
+  }
+}
+
+
+# $expanded=linecomment($string)
+#preface all lines with verbatim marker (usually DIFVRB)
+sub linecomment {
+  my @verbatimlines=split("\n",$_[0]);
+  # the first line needs special treatment - we do want to retain optional arguments as is but wrap the remainder also with VERBCOMMENT
+  ### print STDERR "DEBUG: before verbatimlines[0] = ",$verbatimlines[0],"\n";
+  $verbatimlines[0]=~s/^((?:\s*\[$brat_n\])?\s*)([^\s\[].*)/ defined($2) ? ( "$1\%$VERBCOMMENT$2" ) : ( $1 )/e;
+  ### print STDERR "DEBUG: after  verbatimlines[0] = ",$verbatimlines[0],"\n";
+  return(join("\n%$VERBCOMMENT",@verbatimlines)."\n");
+}
+
+# $simple=reverselinecomment($env $string)
+# remove DIFVRB comments but leave changed lines marked
+sub reverselinecomment {
+  my ($environment, $verbatimtext)=@_;
+  ### print STDERR "DEBUG REVERSELINECOMMENT input: $environment,|$verbatimtext|\n" if $debug;
+  # remove markup added by latexdiff
+  # (this should occur only if the type of verbatim environment was changed)
+  # (note that this destroys some information in old file)
+  #  in theory I could save it by moving it out of the verbatim environment
+  #  but this requires more bookkeeping and is probably not necessary)
+  $verbatimtext =~ s/\\DIFaddbegin //g;
+  $verbatimtext =~ s/\\DIFaddend //g;
+  $verbatimtext =~ s/\\DIFdelbegin //g;
+  $verbatimtext =~ s/\\DIFdelend //g;
+  $verbatimtext =~ s/$DELCMDOPEN.*//g;
+
+  # remove DIFVRB mark
+  $verbatimtext=~ s/%$VERBCOMMENT//g;
+
+  # remove part of the markup in changed lines
+  # if any of these substitution was made, then there was at least
+  # one changed line, and we have to extend the style
+  if ( $verbatimtext=~ s/$VERBCOMMENT//g ) {
+    # in the next line we add ~alsolanguage~ modifier, but also deletes the rest of the line after the optional argument, as lstlisting commands gets sometimes
+    # very confused by what is there   (and othertimes seems to ignore this anyway)
+    unless ( $verbatimtext =~ s/^(\s*)\[($brat_n)\](.*)\n/$1\[$2,alsolanguage=DIFcode\]\n/ ) {
+      if ( $verbatimtext =~ m/^\s*\n/ ) {
+	$verbatimtext = "[alsolanguage=DIFcode]" . $verbatimtext;
+      } else {
+	$verbatimtext = "[alsolanguage=DIFcode]\n" . $verbatimtext;
+      }
+    }
+    # There is a bug in listings package (at least v1.5b) for empty comments where the actual comment command is not made invisible
+    # I therefore have to introduce an artificial '-' character at the end of empty added or deleted lines
+    $verbatimtext =~ s/($DELCOMMENT\s*)$/$1-/mg;
+    $verbatimtext = "\\DIFmodbegin\n\\begin{${environment}}${verbatimtext}\\end{${environment}}\n\\DIFmodend"
+  } else {
+    $verbatimtext = "\\begin{${environment}}${verbatimtext}\\end{${environment}}"
+  }
+  ### print STDERR "DEBUG REVERSELINECOMMENT output: |$verbatimtext|\n" if $debug;
+  return($verbatimtext);
+}
+
+
+#hashstring=tohash(\%hash,$string)
+# creates a hash value based on string and stores in %hash
+sub tohash {
+  my ($hash,$string)=@_;
+  my (@arr,$val);
+  my ($sum,$i)=(0,1);
+  my ($hstr);
+
+  @arr=unpack('c*',$string);
+
+  while (1) {
+    foreach $val (@arr) {
+      $sum += $i*$val;
+      $i++;
+    }
+    $hstr= "$sum";
+    last unless (defined($hash->{$hstr}) && $string ne $hash->{$hstr});
+    # else found a duplicate HASH need to repeat for a higher hash value
+  }
+  $hash->{$hstr}=$string;
+  ###  print STDERR "Hash:$hstr: Content:$string:\n";
+  return($hstr);
+}
+
+#string=fromhash(\%hash,$fromstring)
+# restores string value stored in hash
+#string=fromhash(\%hash,$fromstring,$prependstring)
+# additionally begins each line with prependstring
+sub fromhash {
+  my ($hash,$hstr)=($_[0],$_[1]);
+  my $retstr=$hash->{$hstr};
+  if ( $#_ >= 2) {
+    $retstr =~ s/^/$_[2]/mg;
+  }
+  return $retstr;
+}
+
+# stripdelcmpopen(string)
+# return string with $DELCMDOPEN removed
+sub stripdelcmdopen {
+  my ($str) = $_[0];
+  $str =~ s/${DELCMDOPEN}//mg;
+  return $str;
+}
+
+# writedebugfile(string, label)
+# if $debug set writes <string> to file latexdiff.debug.<label>
+# otherwise do nothing
+sub writedebugfile {
+  my ($string,$label)=@_;
+  if ( $debug ) {
+    open(RAWDIFF,">","latexdiff.debug." . $label);
+    print RAWDIFF $string;
+    close(RAWDIFF);
+  }
+}
+
+
+# postprocess($string, ..)
+# carry out the following post-processing steps for all arguments:
+# * Remove STOP token from the end
+# * Replace \RIGHTBRACE by }
+# * If option --no-del is set delete all deleted blocks
+# 1. Check all deleted blocks:
+#    a.where a deleted block contains a matching \begin and
+#      \end environment (these will be disabled by a %DIFDELCMD statements), for selected environments enable
+#      these commands again (such that for example displayed math in a deleted equation
+#      is properly within math mode.  For math mode environments replace numbered equation
+#      environments with their display only variety (so that equation numbers in new file and
+#      diff file are identical).  Where the correct type of math environment cannot be determined
+#      use a place holder MATHMODE
+#    b.where one of the commands matching $COUNTERCMD is used as a DIFAUXCMD, add a statement
+#      subtracting one from the respective counter to keep numbering consistent with new file
+#    Replace all MATHMODE environment commands by the correct environment to achieve matching
+#    pairs
+#    c. Convert MATHBLOCKmath commands to their uncounted numbers (e.g. convert equation -> displaymath
+#       (environments defined in $MATHENV will be replaced by $MATHREPL, and  environments in $MATHARRENV
+#       will be replaced by their equivalent with a * appended, e.g. align -> align*
+#    d. If in-line math mode contains array environment, enclose the whole environment in \mbox'es
+#    d. place \cite commands in mbox'es (for UNDERLINE style)
+#
+#   For added blocks:
+#    c. If in-line math mode contains array environment, enclose the whole environment in \mbox'es
+#    d. place \cite commands in mbox'es (for UNDERLINE style)
+#
+# 2.   If math-mode COARSE,WHOLE or NONE option set: Convert \MATHBLOCKmath{..} commands back to environments
+#
+#      Convert all PICTUREblock{..} commands back to the appropriate environments
+# 3. Convert DIFadd, DIFdel, DIFaddbegin , ... into FL varieties
+#    within floats (currently recognised float environments: plate,table,figure
+#    plus starred varieties).
+# 4. Remove empty %DIFDELCMD < lines
+# 4. Convert \begin{SQUAREBRACKET} \end{SQUAREBRACKET} into \[ \]
+#    Convert \begin{DOLLARDOLLAR} \end{DOLLARDOLLAR} into $$ $$
+# 5. Convert  \SUPERSCRIPTNB{n} into ^n  and  \SUPERSCRIPT{nn} into ^{nnn}
+# 6. Convert  \SUBSCRIPTNB{n} into _n  and  \SUBCRIPT{nn} into _{nnn}
+# 7. Expand hashes of verb and verbatim environments
+# 8. Convert '\PERCENTAGE ' back into '\%' and '\DOLLAR ' into '\$'
+# 9.. remove all \PAR tokens
+# 10.  package specific processing:  endfloat: make sure \begin{figure} and \end{figure} are always
+#      on a line by themselves, similarly for table environment
+#  4, undo renaming of the \begin, \end,{,}  in comments
+#    Change \QLEFTBRACE, \QRIGHTBRACE,\AMPERSAND to \{,\},\&
+#
+# Note have to manually synchronize substitution commands below and
+# DIF.. command names in the header
+sub postprocess {
+  my ($begin,$len,$float,$delblock,$addblock); ### $cnt
+  # second level blocks
+  my ($begin2,$len2,$eqarrayblock,$eqblock,$mathblock); ### $cnt2
+
+  my (@textparts,@newtextparts,@liststack,$listtype,$listlast);
+
+  my (@itemargs, $itemarg);
+
+
+  for (@_) {
+    # change $'s in comments to something harmless
+    1 while s/(%.*)\$/$1DOLLARDIF/mg ;
+
+    # Remove final STOP token
+    s/ STOP$//;
+    # Replace \RIGHTBRACE in comments by \MBLOCKRIGHTBRACE
+    # the only way to get these is as %DIFDELCMD < \RIGHTBRACE construction
+    # This essentially marks closing right braces of MATHBLOCK environments, which did not get matched
+    # up. This case should be rare, so I just leave this in the diff file output. Not really elegant
+    # but can still be dealt with later if it results in problems.
+    s/%DIFDELCMD < \\RIGHTBRACE/%DIFDELCMD < \\MBLOCKRIGHTBRACE/g ;
+    # Replace \RIGHTBRACE by }
+    s/\\RIGHTBRACE/}/g;
+
+    # Optional: Remove deleted block entirely
+    if ($onlyadditions) {
+        s/\\DIFdelbegin.*?\\DIFdelend//sg ;
+    }
+
+    # Check all deleted blocks: where a deleted block contains a matching \begin and
+    #    \end environment (these will be disabled by a %DIFDELCMD statements), enable
+    #    these commands again (such that for example displayed math in a deleted equation
+    #    is properly within math mode).  For math mode environments replace numbered equation
+    #    environments with their display only variety (so that equation numbers in new file and
+    #    diff file are identical)
+    writedebugfile($_,'postprocess1');
+
+
+    while ( m/\\DIFdelbegin.*?\\DIFdelend/sg ) {
+      # special processing within each deleted block
+      ###    while ( m/\\DIFdelbegin.*?\\DIFdelend/sg ) {
+      ###      print STDERR "DEBUG Match delblock \n||||$&||||\n at ",pos,"\n";
+      my $arrenv;    # dummy variable only needed within this block
+      my @symbols;   # another temporary variable
+      $len=length($&);
+      $begin=pos($_) - $len;
+      $delblock=$&;
+      ###   A much simpler method for math replacement might follow this strategy (can recycle part of the commands below for following
+      ###   this strategy:
+      ###   1. a Insert aux commands \begin{MATHMODE} or \end{MATHMODE} for all deleted commands opening or closing displayed math mode
+      ###      b Insert aux commands \begin{MATHARRMODE} or \end{MATHARRMODE} for all deleted commands opening or closing math array mode
+      ###   2  Replace MATHMODE and MATHARRMODE by correct pairing if appropriate partner  math command is found in text
+      ###   3  a Replace remaining \begin{MATHMODE}...\end{MATHMODE} pairs with \begin{$MATHREPL}..\end{$MATHREPL}
+      ###      b Replace remaining \begin{MATHARRMODE}...\end{MATHARRMODE} pairs with \begin{$MATHREPL}..\end{$MATHREPL}
+      ###   4  Delete all aux command math mode pairs which have simply comments or empty lines between them
+      ###   As written this won't actually work!
+
+
+      ###   Most general case: allow all included environments
+      ###      $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{(\w*\*?)\}\s*?\n)(.*?)(\%DIFDELCMD < \s*\\end\{\2\})/$1\\begin{$2}$AUXCMD\n$3\n\\end{$2}$AUXCMD\n$4/sg;
+      ### (.*?[^\n]?)\n? construct is necessary to avoid empty lines in math mode, which result in
+      ### an error
+      # displayed math environments
+      ###0.5:     $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{((?:$MATHENV)|SQUAREBRACKET)\}\s*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$MATHREPL}$AUXCMD\n$1$3\n\\end{$MATHREPL}$AUXCMD\n$4/sg;
+
+      if ($mathmarkup == FINE ) {
+        # if block contains both beginning and and markers for an explicit math environment, restore opening and closing commands
+	$delblock=~ s/(\%DIFDELCMD < \s*\\begin\{((?:$MATHENV)|SQUAREBRACKET)\}.*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$MATHREPL}$AUXCMD\n$1$3\n\\end{$MATHREPL}$AUXCMD\n$4/sg;
+	# also transform the opposite pair \end{displaymath} .. \begin{displaymath} but we have to be careful not to interfere with the results of the transformation in the line directly above
+	### pre-0.42 obsolete version which did not work on eqnarray test      $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\end\{($MATHENV)\}\s*?\n)(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\begin\{\2\})/$1\\end{$MATHREPL}$AUXCMD\n$3\n\\begin{$MATHREPL}$AUXCMD\n$4/sg;
+	###0.5:      $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\end\{((?:$MATHENV)|SQUAREBRACKET)\}\s*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\begin\{\2\})/\\end{MATHMODE}$AUXCMD\n$1$3\n\\begin{MATHMODE}$AUXCMD\n$4/sg;
+	$delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\end\{((?:$MATHENV)|SQUAREBRACKET)\}.*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\begin\{\2\})/\\end\{MATHMODE\}$AUXCMD\n$1$3\n\\begin\{MATHMODE\}$AUXCMD\n$4/sg;
+
+        # now look for unpaired %DIFDELCMD < \begin{MATHENV}; if found add \begin{$MATHREPL} and insert \end{$MATHREPL}
+        # just before end of block; again we use look-behind assertion to avoid matching constructions which have already been converted
+        if ($delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\begin\{((?:$MATHENV)|SQUAREBRACKET)\}\s*?(?:$DELCMDCLOSE|\n))/$1\\begin{$MATHREPL}$AUXCMD\n/sg ) {
+	  ### print STDERR "BINGO: begin block: \nBefore: |" . substr($_,$begin,$len) . "|\n" if $debug ;
+	  $delblock =~ s/(\\DIFdelend$)/\\end{$MATHREPL}$AUXCMD\n$1/s ;
+	  ### print STDERR "After: |" . $delblock . "|\n\n" if $debug ;
+        }
+        # now look for unpaired %DIFDELCMD < \end{MATHENV}; if found add \end{MATHMODE} and insert \begin{MATHMODE}
+        # just before end of block; again we use look-behind assertion to avoid matching constructions which have already been converted
+        if ($delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\end\{((?:$MATHENV)|SQUAREBRACKET)\}\s*?(?:$DELCMDCLOSE|\n))/$1\\end\{MATHMODE\}$AUXCMD\n/sg ) {
+	  ### print STDERR "BINGO: end block:\nBefore: |" . substr($_,$begin,$len) . "|\n" if $debug;
+	  $delblock =~ s/(\\DIFdelend$)/\\begin\{MATHMODE\}$AUXCMD\n$1/s ;
+	  ### print STDERR "After: |" . $delblock . "|\n\n" if $debug ;
+	}
+
+	### pre-0.42      # same as above for special case \[.\] (latex abbreviation for displaymath)
+        ### pre-0.42      $delblock=~ s/(\%DIFDELCMD < \s*\\\[\s*?\n())(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\\])/$1\\\[$AUXCMD\n$3\n\\\]$AUXCMD\n$4/sg;
+        ### pre-0.42      $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\\]\s*?\n())(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\\[)/$1\\\]$AUXCMD\n$3\n\\\[$AUXCMD\n$4/sg;
+        # equation array environment
+        ###pre-0.3      $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($MATHARRENV)\}\s*?\n)(.*?)(\%DIFDELCMD < \s*\\end\{\2\})/$1\\begin{$MATHARRREPL}$AUXCMD\n$3\n\\end{$MATHARRREPL}$AUXCMD\n$4/sg;
+        ###0.5      $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($MATHARRENV)\}\s*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$MATHARRREPL}$AUXCMD\n$1$3\n\\end{$MATHARRREPL}$AUXCMD\n$4/sg;
+	###1.3.2         $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($MATHARRENV)\}.*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$MATHARRREPL}$AUXCMD\n$1$3\n\\end{$MATHARRREPL}$AUXCMD\n$4/sg;
+	# Example Input: %DIFDELCMD < \begin{alignat}{2} \n\exp(ix)&=&\cos(x)+i\sin(x)\n%DIFDECMD < \end{alignat}
+        #                < $1                <$2   >><$3>  < $4                      ><$5                       >
+	# Example Output:\begin{alignat*}{2}%DIFAUXCMD\n\%DIFDELCMD\begin{align}\n\exp(ix)&=&\cos(x)+i\sin(x)\n\end{align*}%DIFAUXCMD\n%DIFDELCMD > \end{alignat*}
+        $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($MATHARRENV)\}(.*?)(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(\%DIFDELCMD < \s*\\end\{\2\})/\\begin{$2*}$3$AUXCMD\n$1$4\n\\end{$2*}$AUXCMD\n$5/sg;
+        ###  pre-0.42 obsolete version which did not work on eqnarray test     $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\end\{($MATHARRENV)\}\s*?\n)(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \s*\\begin\{\2\})/$1\\end{$MATHARRREPL}$AUXCMD\n$3\n\\begin{$MATHARRREPL}$AUXCMD\n$4/sg;
+        $delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\end\{($MATHARRENV)\}\s*?(?:$DELCMDCLOSE|\n))(.*?[^\n]?)\n?(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\begin\{\2\})/\\end{MATHMODE}$AUXCMD\n$1$3\n\\begin{MATHMODE}$AUXCMD\n$4/sg;
+        ### print STDERR "STEP1: |" . $delblock . "|\n\n" if $debug ;
+
+        # now look for unpaired %DIFDELCMD < \begin{MATHARRENV}; if found add \begin{$MATHARRREPL} and insert \end{$MATHARRREPL}
+        # just before end of block; again we use look-behind assertion to avoid matching constructions which have already been converted
+        if ($delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\begin\{($MATHARRENV)\}(.*?)(?:$DELCMDCLOSE|\n))/$1\\begin{$2*}$3$AUXCMD\n/sg ) {
+	  $arrenv=$2;
+	  $delblock =~ s/(\\DIFdelend$)/\\end{$arrenv*}$AUXCMD\n$1/s ;
+        }
+        ### print STDERR "STEP2: |" . $delblock . "|\n\n" if $debug ;
+
+        # now look for unpaired %DIFDELCMD < \end{MATHENV}; if found add \end{MATHMODE} and insert \begin{MATHMODE}
+        # just before end of block; again we use look-behind assertion to avoid matching constructions which have already been converted
+        if ($delblock=~ s/(?<!${AUXCMD}\n)(\%DIFDELCMD < \h*\\end\{($MATHARRENV)\}\s*?(?:$DELCMDCLOSE|\n))/$1\\end{MATHMODE}$AUXCMD\n/sg ) {
+	  $delblock =~ s/(\\DIFdelend$)/\\begin{MATHMODE}$AUXCMD\n$1/s ;
+        }
+
+	# parse $delblock for deleted and reinstated eqnarray* environments - within those reinstate \\ and & commands
+	###      while ( $delblock =~ m/\\begin{$MATHARRREPL}$AUXCMD\n.*?\n\\end{$MATHARRREPL}$AUXCMD\n/sg ) {
+        ### while ( $delblock =~ m/\\begin\Q{$MATHARRREPL}$AUXCMD\E\n.*?\n\\end\Q{$MATHARRREPL}$AUXCMD\E\n/sg ) {
+	while ( $delblock =~ m/\\begin\{$MATHARRENV\*}[^\n]*?$AUXCMD\n.*?\n\\end\{$MATHARRENV\*\}$AUXCMD\n/sg ) {
+	  ###	      print STDERR "DEBUG Match eqarrayblock $& at ",pos,"\n";
+	  $len2=length($&);
+	  $begin2=pos($delblock) - $len2;
+	  $eqarrayblock=$&;
+	  # reinstate deleted & and \\ commands
+	  ### $eqarrayblock=~ s/(\%DIFDELCMD < \s*(\&|\\\\)\s*?(?:$DELCMDCLOSE|\n))/$1$2$AUXCMD\n/sg ;
+	  $eqarrayblock    =~ s/(\%DIFDELCMD < (.*?(?:\&|\\\\).*)(?:$DELCMDCLOSE|\n))/
+        	  # The pattern captures comments with at least one of & or \\
+                  @symbols= split(m@((?:&|\\\\)\s*)@,$2);   #  extract & and \\ and other material from sequence
+                  @symbols= grep ( m@&|\\\\\s*@, @symbols); #  select & and \\ (and subsequent spaces)
+                  "$1@symbols$AUXCMD\n"
+            /eg ;
+	  ### print STDERR "Modified block:|$eqarrayblock|\n" if  $debug;
+	  # splice in modified block
+	  substr($delblock,$begin2,$len2)=$eqarrayblock;
+	  pos($delblock) = $begin2 + length($eqarrayblock);
+	}
+
+      } elsif ( $mathmarkup == COARSE || $mathmarkup == WHOLE ) {
+	#       Convert MATHBLOCKmath commands to their uncounted numbers (e.g. convert equation -> displaymath
+	#       (environments defined in $MATHENV will be replaced by $MATHREPL, and  environments in $MATHARRENV
+	#       will be replaced by their starred variety
+	$delblock=~ s/\\MATHBLOCK($MATHENV)((?:\[$brat_n\])?)\{($pat_n)\}/\\MATHBLOCK$MATHREPL$2\{$3\}/sg;
+	$delblock=~ s/\\MATHBLOCK($MATHARRENV)((?:\[$brat_n\])?)\{($pat_n)\}/\\MATHBLOCK$1\*$2\{$3\}/sg;
+      }
+
+      # Replacing math array environments with their starred varieties in deleted blocks as implemented now causes double asterisks
+      # in situation where there had been starred already, e.g. \begin{alignat*} is turned into \begin{alignat**}
+      # The following command seeks to undo this double-starring. It is a little bit of a hack because it relies on the fact that **
+      # double-starred math array environment does not occur naturally
+      $delblock =~ s/($MATHARRENV)(?<=\*)\*/$1/sg;
+
+      print STDERR "DELBLOCK after maths processing: |" . $delblock . "|\n\n" if $debug ;
+
+
+      # Reinstate completely deleted list environments. note that items within the
+      # environment will still be commented out.  They will be restored later
+      $delblock=~ s/(\%DIFDELCMD < \s*\\begin\{($LISTENV)\}\s*?(?:\n|$DELCMDCLOSE))(.*?)(\%DIFDELCMD < \s*\\end\{\2\})/{
+															###   # block within the search; replacement environment
+															###   "$1\\begin{$2}$AUXCMD\n". restore_item_commands($3). "\n\\end{$2}$AUXCMD\n$4";
+															"$1\\begin{$2}$AUXCMD\n$3\n\\end{$2}$AUXCMD\n$4";
+														       }/esg;
+      ###      $delblock=~ s/\\begin\{$MATHENV}$AUXCMD/\\begin{$MATHREPL}$AUXCMD/g;
+      ###      $delblock=~ s/\\end\{$MATHENV}$AUXCMD/\\end{$MATHREPL}$AUXCMD/g;
+      ###      $delblock=~ s/\\begin\{$MATHARRENV}$AUXCMD/\\begin{$MATHARRREPL}$AUXCMD/g;
+      ###      $delblock=~ s/\\end\{$MATHARRENV}$AUXCMD/\\end{$MATHARRREPL}$AUXCMD/g;
+
+      #    b.where one of the commands matching $COUNTERCMD is used as a DIFAUXCMD, add a statement
+      #      subtracting one from the respective counter to keep numbering consistent with new file
+      $delblock=~ s/\\($COUNTERCMD)((?:${extraspace}\[$brat_n\]${extraspace}|${extraspace}\{$pat_n\})*\s*${AUXCMD}\n)/\\$1$2\\addtocounter{$1}{-1}${AUXCMD}\n/sg ;
+
+      #    bb. disable active labels within deleted blocks (i.e. those not commented out) (as these are not safe commands, this should normally only
+      #        happen within deleted maths blocks
+      ###      $delblock=~ s/(?<!$DELCMDOPEN)(\\$LABELCMD(?:${extraspace})\{(?:[^{}])*\}[\t ]*)\n?/${DELCMDOPEN}$1${DELCMDCLOSE}/smg ;
+      ###      previous line caused trouble as by issue #90 I might need to modify this
+      $delblock=~ s/^([^%]*)(\\$LABELCMD(?:${extraspace})\{(?:[^{}])*\}[\t ]*)\n?/$1${DELCMDOPEN}$2${DELCMDCLOSE}/smg ;
+      ###      print STDERR "<<<$delblock>>>\n" if $debug;
+
+
+      #     c. If in-line math mode contains array environment, enclose the whole environment in \mbox'es
+      while ( $delblock =~ m/($math)(\s*)/sg ) {
+##		      print STDERR "DEBUG Delblock Match math $& at ",pos,"\n";
+	$len2=length($&);
+	$begin2=pos($delblock) - $len2;
+	$mathblock="%\n\\mbox{$AUXCMD\n$1\n}$AUXCMD\n";
+        next unless ( $mathblock =~ /ARRAYBLOCK/ or $mathblock =~ m/\{$ARRENV\}/ );
+	substr($delblock,$begin2,$len2)=$mathblock;
+	pos($delblock) = $begin2 + length($mathblock);
+      }
+      ###      if ($CITE2CMD) {
+      ######   ${extraspace}(?:\[$brat0\]${extraspace}){0,2}\{$pat_n\}))  .*?%%%\n
+      ###	$delblock=~s/($DELCMDOPEN\s*\\($CITE2CMD)(.*)$DELCMDCLOSE)/
+      ###	  # Replacement code
+      ###	  {my ($aux,$all);
+      ###	   $aux=$all=$1;
+      ###	   $aux=~s#\n?($DELCMDOPEN|$DELCMDCLOSE)##g;
+      ###	   $all."$aux$AUXCMD\n";}/sge;
+      ###      }
+      ###      # or protect \cite commands with \mbox
+      ###      if ($CITECMD) {
+      ######	$delblock=~s/(\\($CITECMD)${extraspace}(?:\[$brat0\]${extraspace}){0,2}\{$pat_n\})(\s*)/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/msg ;
+      ###	$delblock=~s/(\\($CITECMD)${extraspace}(?:<$abrat0>${extraspace})?(?:\[$brat0\]${extraspace}){0,2}\{$pat_n\})(\s*)/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/msg ;
+      ###      }
+      # if MBOXINLINEMATH is set, protect inlined math environments with an extra mbox
+
+
+      if ( $MBOXINLINEMATH ) {
+	# note additional \newline after command is omitted from output if right at the end of deleted block (otherwise a spurious empty line is generated)
+	$delblock=~s/($math)(?:[\s\n]*)?/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/sg;
+      }
+      ###if ( defined($packages{"listings"} and $latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/))   {
+      ###  #     change included verbatim environments
+      ###  $delblock =~ s/\\DIFverb\{/\\DIFDIFdelverb\{/g;
+      ###  $delblock =~ s/\\DIFlstinline/\\DIFDIFdellstinline/g;
+      ###}
+      # Mark deleted verbatim commands
+      $delblock =~ s/(${DELCMDOPEN}\\DIF((?:verb\*?|lstinline(?:\[$brat_n\])?)\{([-\d]*?)\}\s*).*)$/%\n\\DIFDIFdel$2${AUXCMD}\n$1/gm;
+      if ( $CUSTOMDIFCMD ) {
+        ###$delblock =~ s/(${DELCMDOPEN}.*)\\($CUSTOMDIFCMD)((?:\[${brat_n}\])*?(?:\s*\{${pat_n}\})*)/"$1${DELCMDCLOSE}\\DEL$2". stripdelcmdopen($3) ." ${DELCMDOPEN}"/egms;
+        ###$delblock =~ s/(${DELCMDOPEN}.*)\\($CUSTOMDIFCMD)/$1${DELCMDCLOSE}\\DEL$2/gm;
+        # ($1 ? "${DELCMDOPEN}$1${DELCMDCLOSE}":"") : only add the DELCMDOPEN / DELCMDCLOSE pair if there are actually any commands in between, otherwise this is redundant
+	$delblock    =~ s/${DELCMDOPEN}(.*?)\\($CUSTOMDIFCMD)((?:\[${brat_n}\])*?(?:\s*\{${pat_n}\})*)/($1 ? "${DELCMDOPEN}$1${DELCMDCLOSE}":"") ."\\DEL$2". stripdelcmdopen($3)/egs;
+        # if there is a sequence of several commands in the same row only the first will be converted due to the need to be connected to the DELCMDOPEN. To mop these up, just add the DEL to the front of any remaining cmd's in the deleted block
+        $delblock =~ s/\\($CUSTOMDIFCMD)/\\DEL$1/g;     # this will also convert comments but I guess it does not matter much
+      }
+
+      #     splice in modified delblock
+      substr($_,$begin,$len)=$delblock;
+      pos = $begin + length($delblock);
+    }
+
+    ### print STDERR "<<<$_>>>\n" if $debug;
+
+
+    # make the array modification in added blocks
+    while ( m/\\DIFaddbegin.*?\\DIFaddend/sg ) {
+      $len=length($&);
+      $begin=pos($_) - $len;
+      $addblock=$&;
+      while ( $addblock =~ m/($math)(\s*)/sg ) {
+	$len2=length($&);
+	$begin2=pos($addblock) - $len2;
+	$mathblock="%\n\\mbox{$AUXCMD\n$1\n}$AUXCMD\n";
+        next unless ( $mathblock =~ /ARRAYBLOCK/ or $mathblock =~ m/\{$ARRENV\}/) ;
+	substr($addblock,$begin2,$len2)=$mathblock;
+	pos($addblock) = $begin2 + length($mathblock);
+      }
+      # if MBOXINLINEMATH is set, protect inlined math environments with an extra mbox
+      if ( $MBOXINLINEMATH ) {
+	##$addblock=~s/($math)/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/sg;
+	$addblock=~s/($math)(?:[\s\n]*)?/\\mbox{$AUXCMD\n$1\n}$AUXCMD\n/sg;
+      }
+      ###if ( defined($packages{"listings"} and $latexdiffpreamble =~ /\\RequirePackage(?:\[$brat0\])?\{color\}/))   {
+	# mark added verbatim commands
+      $addblock =~ s/\\DIFverb/\\DIFDIFaddverb/g;
+      $addblock =~ s/\\DIFlstinline/\\DIFDIFaddlstinline/g;
+      if( $CUSTOMDIFCMD ) {
+        $addblock =~ s/\\($CUSTOMDIFCMD)/\\ADD$1/g;     # this will also convert comments but I guess it does not matter much
+      }
+      # markup the optional arguments of \item
+      $addblock =~ s/(\\$ITEMCMD$extraspace(?:<$abrat0>)?$extraspace)\[($brat_n)\]/
+	@itemargs=splitlatex(substr($2,0,length($2)));
+        $itemarg="[".join("",marktags("","",$ADDOPEN,$ADDCLOSE,"","",$ADDCOMMENT,\@itemargs))."]";
+      "$1$itemarg"/sge;   # old substitution: $1\[$ADDOPEN$2$ADDCLOSE\]
+      ###}
+#     splice in modified addblock
+      substr($_,$begin,$len)=$addblock;
+      pos = $begin + length($addblock);
+    }
+
+    # Go through whole text, and by counting list environment commands, find out when we are within a list environment.
+    # Within those restore deleted \item commands
+    @textparts=split /(?<!$DELCMDOPEN)(\\(?:begin|end)\{$LISTENV\})/ ;
+    @liststack=();
+    @newtextparts=map {
+      ### print STDERR ":::::::: $_\n";
+      if ( ($listtype) = m/^\\begin\{($LISTENV)\}$/ ) {
+	print STDERR "DEBUG: postprocess \\begin{$listtype}\n" if $debug;
+	push @liststack,$listtype;
+      } elsif ( ($listtype) = m/^\\end\{($LISTENV)\}$/ ) {
+ 	print STDERR "DEBUG: postprocess \\end{$listtype}\n" if $debug;
+	if (scalar  @liststack > 0) {
+ 	  $listlast=pop(@liststack);
+ 	  ($listtype eq $listlast) or warn "Invalid nesting of list environments: $listlast environment closed by \\end{$listtype}.";
+ 	} else {
+ 	  warn "WARNING: Invalid nesting of list environments: \\end{$listtype} encountered without matching \\begin{$listtype}.\n";
+ 	}
+      } else {
+	print STDERR "DEBUG: postprocess \@liststack=(",join(",",@liststack),")\n" if $debug;
+	if (scalar  @liststack > 0 ) {
+	  # we are within a list environment and should replace all item commands
+	  $_=restore_item_commands($_);
+	}
+	# else: we are outside a list environment and do not need to do anything
+      }
+      $_ } @textparts;     # end of map command
+    # replace the main text with the modified version
+    $_= join("",@newtextparts);
+
+
+
+
+    # Replace MATHMODE environments from step 1a above by the correct Math environment and remove unncessary pairings
+
+    if ( $mathmarkup == FINE ) {
+      # look for AUXCMD math-mode pairs which have only comments (or empty lines between them), and remove the added commands
+      # \begin{..} ... \end{..} pairs
+      s/\\begin\{((?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET|MATHMODE)\}$AUXCMD\n((?:\s*%.[^\n]*\n)*)\\end\{\1\}$AUXCMD\n/$2/sg;
+      # \end{..} .... \begin{..} pairs
+      s/\\end\{((?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET|MATHMODE)\}$AUXCMD\n((?:\s*%.[^\n]*\n)*)\\begin\{\1\}$AUXCMD\n/$2/sg;
+
+      writedebugfile($_,'postprocess15');
+      # The next line is complicated.  The negative look-ahead insertion makes sure that no \end{$MATHENV} (or other mathematical
+      # environments) are between the \begin{$MATHENV} and \end{MATHMODE} commands. This is necessary as the minimal matching
+      # is not globally minimal but only 'locally' (matching is beginning from the left side of the string)
+      # [NB: Do not be tempted to prettify the expression with /x modified. It seems this is applied after strings are expanded so will ignore spaces in strings
+      1 while s/(?<!$DELCMDOPEN)\\begin\{((?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET)}((?:${DELCMDOPEN}[^\n]*|.(?!(?:\\end\{(?:(?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET)}|\\begin\{MATHMODE})))*?)\\end\{MATHMODE}/\\begin{$1}$2\\end{$1}/s;
+      writedebugfile($_,'postprocess16');
+      1 while s/\\begin\{MATHMODE}((?:.(?!\\end\{MATHMODE}))*?)(?<!$DELCMDOPEN)\\end\{((?:$MATHENV)|(?:$MATHARRENV)|SQUAREBRACKET)}/\\begin{$2}$1\\end{$2}/s;
+      # convert remaining \begin{MATHMODE} \end{MATHMODE} (and not containing & or \\ )into MATHREPL environments
+      s/\\begin\{MATHMODE\}((?:(.(?!(?<!\\)\&|\\\\))*)?)\\end\{MATHMODE\}/\\begin{$MATHREPL}$1\\end{$MATHREPL}/sg;
+      # others into MATHARRREPL
+      s/\\begin\{MATHMODE\}(.*?)\\end\{MATHMODE\}/\\begin{$MATHARRREPL}$1\\end{$MATHARRREPL}/sg;
+      # Final cleanup of all equation environments: Sometimes I end up with %DIFDELCMD < \PAR lines, which I need to make sure do not get expanded into
+      # actual empty lines within equation environments (later part)
+      while (  m/\\begin\{((?:$MATHENV|$MATHARRENV|SQUAREBRACKET)\*?)}.*?(?<!\%DIFDELCMD < )\\end\{\1}/sg ) {
+	$len=length($&);
+	$begin=pos($_) - $len;
+	$eqblock=$&;
+	$eqblock =~ s/(%DIFDELCMD < )([^\n]*?)\\PAR\n/$1$2\n$1\n/sg;
+	#     splice in modified delblock
+	print STDERR "DEBUG EQBLOCK: |$eqblock|\n\n" if $debug;
+	substr($_,$begin,$len)=$eqblock;
+	pos = $begin + length($eqblock);
+      }
+    } else {
+      #   math modes OFF,WHOLE,COARSE: Convert \MATHBLOCKmath{..} commands back to environments
+      # with optionl argument to MATHBLOCK, e.g \MATHBLOCKalignat[{3}]{ ...}
+      s/\\MATHBLOCK($MATHENV|$MATHARRENV\*?|SQUAREBRACKET)\[($brat0)\]\{($pat_n)\}/\\begin{$1}$2$3\\end{$1}/sg;
+      # without optional argument e.g \MATHBLOCKalign{ ...}
+      s/\\MATHBLOCK($MATHENV|$MATHARRENV\*?|SQUAREBRACKET)\{($pat_n)\}/\\begin{$1}$2\\end{$1}/sg;
+      # convert ARRAYBLOCK.. commands back to environments
+      s/\\ARRAYBLOCK($ARRENV)\{($pat_n)\}/\\begin{$1}$2\\end{$1}/sg;
+      # get rid of the DIFANCHOR markers, first the delete comments, then everywhere
+      s/%DIFDELCMD < \\DIFANCHOR(?:MATH|ARR)[BE] (?:\n%DIFDELCMD < )?%%%\n//g ;
+      s/\\DIFANCHOR(?:MATH|ARR)[BE] //g;
+    }
+    writedebugfile($_,'postprocess2');
+
+    #  Convert all PICTUREblock{..} commands back to the appropriate environments
+    s/\\PICTUREBLOCK($PICTUREENV)\{($pat_n)\}/\\begin{$1}$2\\end{$1}/sg;
+#0.5:    # Remove all mark up within picture environments
+#     while ( m/\\begin\{($PICTUREENV)\}.*?\\end\{\1\}/sg ) {
+#       $cnt=0;
+#       $len=length($&);
+#       $begin=pos($_) - $len;
+#       $float=$&;
+#       $float =~ s/\\DIFaddbegin //g;
+#       $float =~ s/\\DIFaddend //g;
+#       $float =~ s/\\DIFadd\{($pat_n)\}/$1/g;
+#       $float =~ s/\\DIFdelbegin //g;
+#       $float =~ s/\\DIFdelend //g;
+#       $float =~ s/\\DIFdel\{($pat_n)\}//g;
+#       $float =~ s/$DELCMDOPEN.*//g;
+#       substr($_,$begin,$len)=$float;
+#       pos = $begin + length($float);
+#     }
+    # Convert DIFadd, DIFdel, DIFFaddbegin , ... into  varieties
+    #    within floats (currently recognised float environments: plate,table,figure
+    #    plus starred varieties).
+    while ( m/\\begin\{($FLOATENV)\}.*?\\end\{\1\}/sg ) {
+      $len=length($&);
+      $begin=pos($_) - $len;
+      $float=$&;
+      $float =~ s/\\DIFaddbegin /\\DIFaddbeginFL /g;
+      $float =~ s/\\DIFaddend /\\DIFaddendFL /g;
+      $float =~ s/\\DIFadd\{/\\DIFaddFL{/g;
+      $float =~ s/\\DIFdelbegin /\\DIFdelbeginFL /g;
+      $float =~ s/\\DIFdelend /\\DIFdelendFL /g;
+      $float =~ s/\\DIFdel\{/\\DIFdelFL{/g;
+      substr($_,$begin,$len)=$float;
+      pos = $begin + length($float);
+    }
+    ### former location of undo renaming of \begin and \end in comments
+
+    # remove empty DIFCMD < lines
+    s/^\Q${DELCMDOPEN}\E\n//msg;
+
+    # Expand hashes of verb and verbatim environments
+    s/${DELCMDOPEN}\\($VERBATIMENV)\{([-\d]*?)\}/"${DELCMDOPEN}\\begin{${1}}".fromhash(\%verbhash,$2,$DELCMDOPEN)."${DELCMDOPEN}\\end{${1}}"/esg;
+    # revert changes to verbatim environments for line diffs (and add code to mark up changes) (note negative look behind assertions to not leak out of DIFDELCMD comments)
+    # Example:
+    # < \begin{verbatim}
+    # < %DIF < DIFVRB old verbatim line
+    # < %DIF > DIFVRB new verbatim line
+    # < \end{verbatim}
+    # ---
+    # > \DIFmodbegin
+    # > \begin{verbatim}[alsolanguage=DIFcode]
+    # > %DIF < old verbatim line
+    # > %DIF > new verbatim line
+    # > \end{verbatim}
+    # > \DIFmodend
+    s/(?<!$DELCMDOPEN)\\begin\{($VERBATIMLINEENV)\}(.*?)(?<!$DELCMDOPEN)\\end\{\1\}/"". reverselinecomment($1, $2) .""/esg;
+#    # we do the same for deleted environments but additionally reinstate the framing commands
+#   s/$DELCMDOPEN\\begin\{($VERBATIMLINEENV)\}$extraspace(?:\[$brat0\])?$DELCMDCLOSE(.*?)$DELCMDOPEN\\end\{\1\}$DELCMDCLOSE/"\\begin{$1}". reverselinecomment($2) . "\\end{$1}"/esg;
+##    s/$DELCMDOPEN\\begin\{($VERBATIMLINEENV)\}($extraspace(?:\[$brat0\])?\s*)(?:\n|$DELCMDOPEN)*$DELCMDCLOSE((?:\%$DELCOMMENT$VERBCOMMENT.*?\n)*)($DELCMDOPEN\\end\{\1\}(?:\n|\s|$DELCMDOPEN)*$DELCMDCLOSE)/"SUBSTITUTION: \\begin{$1}$2 INTERIOR: |$3| END: |$4|"/esg;
+     s/ # Deleted \begin command of verbatim environment (Captures $1: whole deleted command, $2: environment, $3: optional arguments with white space
+          (\Q$DELCMDOPEN\E\\begin\{($VERBATIMLINEENV)\}(\Q$extraspace\E(?:\[$brat_n\])?\s*)(?:\n|\Q$DELCMDOPEN\E)*\Q$DELCMDCLOSE\E)
+        # Interior of deleted verbatim environment should consist entirely of delete DIFVRB comments, i.e. match only lines beginning with % DIF < DIFVRB
+        #   Captures: $4: all lines combined
+          ((?:\%\Q$DELCOMMENT$VERBCOMMENT\E[^\n]*?\n)*)
+        # Deleted \end command of verbatim environment. Note that the type is forced to match the opening. Captures: $5: Whole deleted environment  (previous way this line was written: (\Q$DELCMDOPEN\E\\end\{\2\}(?:\n|\s|\Q$DELCMDOPEN\E)*\Q$DELCMDCLOSE\E)
+          (\Q$DELCMDOPEN\E\\end\{\2\})
+      / # Substitution part
+            $1                   # Leave expression as is
+            . "$AUXCMD NEXT\n"   # Mark the following line as an auxiliary command
+            . ""    # reinstate the original environment without options
+            . reverselinecomment($2, "$3$4")   # modify the body to change the markup; reverselinecomment parses for options
+            . " $AUXCMD\n"  # close the auxiliary environment
+            . $5               # and again leave the original deleted closing environment as is
+      /esgx;  # Modifiers of substitution command
+    writedebugfile($_,'postprocess2');
+    # where changes have occurred in verbatim environment, change verbatim to DIFverbatim to allow mark-up
+    # (I use the presence of optional paramater to verbatim environment as the marker - normal verbatim
+    # environment does not take optional arguments)
+    s/(?<!$DELCMDOPEN)\\begin\{(verbatim[*]?)\}(\[$brat_n\].*?)\\end\{\1\}/\\begin{DIF$1}$2\\end{DIF$1}/sg;
+
+    s/\\($VERBATIMENV)\{([-\d]*?)\}/"\\begin{${1}}".fromhash(\%verbhash,$2)."\\end{${1}}"/esg;
+
+    # remove all \PAR tokens (taking care to properly keep commented out PAR's
+    # from introducing uncommented newlines - next line)
+    s/(%DIF < )([^\n]*?)\\PAR\n/$1$2\n$1\n/sg;
+    # convert PAR commands which are on a line by themselves
+    s/\n(\s*?)\\PAR\n/\n\n/sg;
+    # convert remaining PAR commands (which are preceded by non-white space characters, usually "}" ($ADDCLOSE)
+    s/\\PAR\n/\n\n/sg;
+
+    #  package specific processing:
+    if ( defined($packages{"endfloat"})) {
+      #endfloat: make sure \begin{figure} and \end{figure} are always
+      #      on a line by themselves, similarly for table environment
+      print STDERR "endfloat package detected.\n" if $verbose ;
+      # eliminate whitespace before and after
+      s/^(\s*)(\\(?:end|begin)\{(?:figure|table)\})(\s*?)$/$2/mg;
+      # split lines with remaining characters before float environment conmmand
+      s/^([^%]+)(\\(?:begin|end)\{(?:figure|table)\})/$1\n$2/mg;
+      # split lines with remaining characters after float environment conmmand
+      s/^((?:[^%]+)\\(?:begin|end)\{(?:figure|table)\}(?:\[[a-zA-Z]+\])?)(.+)((?:%.*)?)$/$1\n$2$3/mg;
+    }
+
+    # Remove empty auxiliary LISTENV  (sometmes these are generated if list environment occurs within the argument of a deleted comment)
+    #   (slightly hacky but I could not see an easy way to see if in argument of another command when this extra markup is added)
+    s/\\begin\{($LISTENV)\}$AUXCMD\n((?:\s*\%[^\n]*\n)*\n?)\\end\{\1\}$AUXCMD\n/$2\n/msg ;
+
+    # Convert '\PERCENTAGE ' back into '\%' (the final question mark catches a special situation where due to a latter pre-processing step the ' ' becomes separated
+    s/\\PERCENTAGE ?/\\%/g;
+    # Convert '\DOLLAR ' back into '\$'
+    s/\\DOLLAR /\\\$/g;
+
+    # undo renaming of the \begin and \end,{,}  and dollars in comments
+
+    # although we only renamed $ in comments to DOLLARDIFF, we might have lost the % in unchanged verbatim blocks, so rename all
+    s/DOLLARDIF/\$/g;
+#   Convert \begin{SQUAREBRACKET} \end{SQUAREBRACKET} into \[ \]
+    s/\\end\{SQUAREBRACKET\}/\\\]/sg;
+    s/\\begin\{SQUAREBRACKET\}/\\\[/sg;
+# 4. Convert \begin{DOLLARDOLLAR} \end{DOLLARDOLLAR} into $$ $$
+    s/\\begin\{DOLLARDOLLAR\}(.*?)\\end\{DOLLARDOLLAR\}/\$\$$1\$\$/sg;
+# 5. Convert  \SUPERSCRIPTNB{n} into ^n  and  \SUPERSCRIPT{nn} into ^{nnn}
+    1 while s/\\SUPERSCRIPT(\s*\{($pat_n)\})/^$1/g ;
+    1 while s/\\SUPERSCRIPTNB\{(\s*$pat0)\}/^$1/g ;
+    # Convert  \SUBSCRIPNB{n} into _n  and  \SUBCRIPT{nn} into _{nnn}
+    1 while s/\\SUBSCRIPT(\s*\{($pat_n)\})/_$1/g ;
+    1 while s/\\SUBSCRIPTNB\{(\s*$pat0)\}/_$1/g ;
+    # Convert  \SQRT{n} into \sqrt{n}  and  \SQRTNB{nn} into \sqrt nn
+    1 while s/\\SQRT(\s*\{($pat_n)\})/\\sqrt$1/g ;
+    1 while s/\\SQRTNB\{(\s*$pat0)\}/\\sqrt$1/g ;
+
+    1 while s/(%.*)\\CFRAC (.*)$/$1\\frac$2/mg ;
+    1 while s/(%.*)\\CRIGHTBRACE (.*)$/$1\}$2/mg ;
+    1 while s/(%.*)\\CLEFTBRACE (.*)$/$1\{$2/mg ;
+
+
+#    Change \QLEFTBRACE, \QRIGHTBRACE to \{,\}
+    s/\\QLEFTBRACE /\\\{/sg;
+    s/\\QRIGHTBRACE /\\\}/sg;
+    s/\\AMPERSAND /\\&/sg;
+    # Highlight added inline verbatim commands if possible
+    if ( $latexdiffpreamble =~ /\\RequirePackage(?:\[$brat_n\])?\{color\}/ )   {
+      # wrap added verb commands with color commands
+      s/\\DIFDIFadd((?:verb\*?|lstinline(?:\[$brat_n\])?)\{[-\d]*?\}[\s\n]*)/\{\\color{blue}$AUXCMD\n\\DIF$1%\n\}$AUXCMD\n/sg;
+      s/\\DIFDIFdel((?:verb\*?|lstinline(?:\[$brat_n\])?)\{[-\d]*?\}[\s\n]*$AUXCMD)/\{\\color{red}${AUXCMD}\n\\DIF$1\n\}${AUXCMD}/sg;
+    } else {
+      # currently if colour markup is not used just remove the added mark
+      s/\\DIFDIFadd(verb\*?|lstinline)/\\DIF$1/sg;
+      s/\\DIFDIFdel((?:verb\*?|lstinline(?:\[$brat_n\])?)\{[-\d]*?\}[\s\n]*$AUXCMD\n)//sg;
+    }
+    # expand \verb and friends inline arguments
+    s/\\DIF((?:DIFadd|DIFdel)?(?:verb\*?|lstinline(?:\[$brat_n\])?))\{([-\d]*?)\}/"\\${1}". fromhash(\%verbhash,$2)/esg;
+    # add basicstyle color{blue} to added lstinline commands
+    # finally add the comment to the ones not having an optional argument before
+    ###s/\\DIFaddlstinline(?!\[)/\\lstinline\n[basicstyle=\\color{blue}]$AUXCMD\n/g;
+
+  }
+  return;
+}
+
+# $out = restore_item_commands($listenviron)
+# short helper function for post-process, which restores deleted \item commands in its argument (as DIFAUXCMDs)
+sub restore_item_commands {
+  my ($string)=@_ ;
+  my ($itemarg,@itemargs);
+  $string =~ s/(\%DIFDELCMD < \s*(\\$ITEMCMD$extraspace)((?:<$abrat0>)?$extraspace)((?:\[$brat_n\])?)\s*((?:${cmdoptseq}\s*?)*)(?:\n|$DELCMDCLOSE))/
+     # if \item has an []argument, then mark up the argument as deleted)
+     if (length($4)>0) {
+       # use substr to exclude square brackets at end points
+       @itemargs=splitlatex(substr($4,1,length($4)-2));
+       $itemarg="[".join("",marktags("","",$DELOPEN,$DELCLOSE,$DELCMDOPEN,$DELCMDCLOSE,$DELCOMMENT,\@itemargs))."]";
+     } else {
+       $itemarg="";
+     }
+     "$1$2$3$itemarg$AUXCMD\n";  ###.((length($5)>0) ? "%DIFDELCMD $5 $DELCMDCLOSE\n" : "")
+     /sge;
+  return($string);
+}
+
+
+# @auxlines=preprocess_preamble($oldpreamble,$newpreamble);
+  # pre-process preamble by looking for commands used in \maketitle (title, author, date etc commands)
+  # the list of commands is defined in CONTEXT2CMD
+  # if found then use a bodydiff to mark up content, and replace the corresponding commands
+  # in both preambles by marked up version to 'fool' the linediff (such that only body is marked up).
+  # A special case are e.g. author commands being added (or removed)
+  # 1. If commands are added, then the entire content is marked up as new, but also the lines are marked as new in the linediff
+  # 2. If commands are removed, then the linediff will mark the line as deleted.  The program returns
+  #    with $auxlines a text to be appended at the end of the preamble, which shows the respective fields as deleted
+sub preprocess_preamble {
+  my ($oldpreambleref,$newpreambleref)=(\$_[0],\$_[1]) ;
+  my @auxlines=();
+  # Remember to use $$oldpreambleref to refer to oldpreamble
+  my ($titlecmd,$titlecmdpat);
+  my (@oldtitlecommands,@newtitlecommands );
+  my  %oldhash  = ();
+  my  %newhash  = ();
+  my ($line,$cmd,$optarg,$arg,$optargnew,$optargold,$optargdiff,$argold,$argnew,$argdiff,$auxline);
+
+  my $warnmsgdetail = <<EOF ;
+     This should not occur for standard styles, but can occur for some specific styles, document classes,
+     e.g. journal house styles.
+     Workaround: Use --replace-context2cmd option to specifically set those commands, which are not repeated.
+EOF
+
+  # resuse context2cmdlist to define these commands to  look out for in preamble
+  $titlecmd = "(?:".join("|",@CONTEXT2CMDLIST).")";
+  # as context2cmdlist is stored as regex, e.g. ((?-xism:^title$), we need to remove ^- fo
+  # resue in a more complex regex
+  $titlecmd =~ s/[\$\^]//g;
+  # make sure to not match on comment lines:
+  $titlecmdpat=qr/^(?:[^%\n]|\\%)*(\\($titlecmd)$extraspace(?:\[($brat_n)\])?(?:\{($pat_n)\}))/ms;
+  ###print STDERR "DEBUG:",$titlecmdpat,"\n";
+  @oldtitlecommands= ( $$oldpreambleref =~ m/$titlecmdpat/g );
+  @newtitlecommands= ( $$newpreambleref =~ m/$titlecmdpat/g );
+
+
+  while ( @oldtitlecommands ) {
+    $line=shift @oldtitlecommands;
+    $cmd=shift @oldtitlecommands;
+    $optarg=shift @oldtitlecommands;
+    $arg=shift @oldtitlecommands;
+
+    if ( defined($oldhash{$cmd})) {
+      warn "WARNING: $cmd is used twice in preamble of old file. Reverting to pure line diff mode for preamble.\n";
+      print STDERR $warnmsgdetail;
+      return;
+    }
+    $oldhash{$cmd}=[ $line, $optarg, $arg ];
+  }
+  while ( @newtitlecommands ) {
+    $line=shift @newtitlecommands;
+    $cmd=shift @newtitlecommands;
+    $optarg=shift @newtitlecommands;
+    $arg=shift @newtitlecommands;
+
+    if ( defined($newhash{$cmd})) {
+      warn "$cmd is used twice in preamble of new file. Reverting to pure line diff mode for preamble.\n";
+      print STDERR $warnmsgdetail;
+      return;
+    }
+    $newhash{$cmd}=[ $line, $optarg, $arg ];
+  }
+  foreach $cmd ( keys %newhash ) {
+    if ( defined($newhash{$cmd}->[1])) {
+       $optargnew=$newhash{$cmd}->[1];
+    } else {
+      $optargnew="";
+    }
+    if ( defined($oldhash{$cmd}->[1])) {
+       $optargold=$oldhash{$cmd}->[1];
+    } else {
+      $optargold="";
+    }
+
+    if ( defined($oldhash{$cmd}->[2]) ) {
+      $argold=$oldhash{$cmd}->[2];
+    } else {
+      $argold="";
+    }
+    $argnew=$newhash{$cmd}->[2];
+    $argdiff="{" . join("",bodydiff($argold,$argnew)) ."}";
+    # Replace \RIGHTBRACE by }
+    $argdiff =~ s/\\RIGHTBRACE/}/g;
+
+    if ( length $optargnew ) {
+      $optargdiff="[".join("",bodydiff($optargold,$optargnew))."]" ;
+      $optargdiff =~ s/\\DIFaddbegin /\\DIFaddbeginFL /g;
+      $optargdiff =~ s/\\DIFaddend /\\DIFaddendFL /g;
+      $optargdiff =~ s/\\DIFadd\{/\\DIFaddFL{/g;
+      $optargdiff =~ s/\\DIFdelbegin /\\DIFdelbeginFL /g;
+      $optargdiff =~ s/\\DIFdelend /\\DIFdelendFL /g;
+      $optargdiff =~ s/\\DIFdel\{/\\DIFdelFL{/g;
+    } else {
+      $optargdiff="";
+    }
+    ### print STDERR "DEBUG s/\\Q$newhash{$cmd}->[0]\\E/\\$cmd$optargdiff$argdiff/s\n";
+    # Note: \Q and \E force literal interpretation of what it between them but allow
+    #      variable interpolation, such that e.g. \title matches just that and not TAB-itle
+    $$newpreambleref=~s/\Q$newhash{$cmd}->[0]\E/\\$cmd$optargdiff$argdiff/s;
+    # replace this in old preamble if necessary
+    if ( defined($oldhash{$cmd}->[0])) {
+      $$oldpreambleref=~s/\Q$oldhash{$cmd}->[0]\E/\\$cmd$optargdiff$argdiff/s ;
+    }
+    ### print STDERR "DEBUG NEW PRE ".$$newpreambleref."\n";
+  }
+
+  foreach $cmd ( keys %oldhash ) {
+    # if this has already been dealt with above can just skip
+    next if defined($newhash{$cmd}) ;
+    $argold=$oldhash{$cmd}->[2];
+    $argdiff="{" . join("",bodydiff($argold,"")) ."}";
+    if ( defined($oldhash{$cmd}->[1])) {
+      $optargold=$oldhash{$cmd}->[1];
+      $optargdiff="[".join("",bodydiff($optargold,""))."]" ;
+      $optargdiff =~ s/\\DIFdelbegin /\\DIFdelbeginFL /g;
+      $optargdiff =~ s/\\DIFdelend /\\DIFdelendFL /g;
+      $optargdiff =~ s/\\DIFdel\{/\\DIFdelFL{/g;
+    } else {
+      $optargdiff="";
+    }
+    $auxline = "\\$cmd$optargdiff$argdiff";
+    $auxline =~s/$/$AUXCMD/sg;
+    push @auxlines,$auxline;
+  }
+  # add auxcmd comment to highlight added lines
+  return(@auxlines);
+}
+
+
+
+# @diffs=linediff(\@seq1, \@seq2)
+# mark up lines like this
+#%DIF mm-mmdnn
+#%< old deleted line(s)
+#%DIF -------
+#%DIF mmann-nn
+#new appended line %<
+#%DIF -------
+# Future extension: mark change explicitly
+# Assumes: traverse_sequence traverses deletions before insertions in changed sequences
+#          all line numbers relative to line 0 (first line of real file)
+sub linediff {
+  my $seq1 = shift ;
+  my $seq2 = shift ;
+
+  my $block = [];
+  my $retseq = [];
+  my @begin=('','',''); # dummy initialisation
+  my $instring ;
+
+  my $discard = sub { @begin=('d',$_[0],$_[1]) unless scalar @$block ;
+                      push(@$block, "%DIF < " . $seq1->[$_[0]]) };
+  my $add = sub { if (! scalar  @$block) {
+		    @begin=('a',$_[0],$_[1]) ;}
+		  elsif ( $begin[0] eq 'd' ) {
+                    $begin[0]='c'; $begin[2]=$_[1];
+		    push(@$block, "%DIF -------") }
+                  push(@$block,  $seq2->[$_[1]] . " %DIF > " ) };
+  my $match = sub { if ( scalar @$block ) {
+                      if ( $begin[0] eq 'd' && $begin[1]!=$_[0]-1) {
+			$instring = sprintf "%%DIF %d-%dd%d",$begin[1],$_[0]-1,$begin[2]; }
+		      elsif ( $begin[0] eq 'a' && $begin[2]!=$_[1]-1) {
+			$instring = sprintf "%%DIF %da%d-%d",$begin[1],$begin[2],$_[1]-1; }
+		      elsif ( $begin[0] eq 'c' ) {
+			$instring = sprintf "%%DIF %sc%s",
+			                     ($begin[1]==$_[0]-1) ? "$begin[1]" : $begin[1]."-".($_[0]-1)  ,
+			                     ($begin[2]==$_[1]-1) ? "$begin[2]" : $begin[2]."-".($_[1]-1)  ; }
+		      else {
+			$instring = sprintf "%%DIF %d%s%d",$begin[1],$begin[0],$begin[2]; }
+		      push @$retseq, $instring,@$block, "%DIF -------" ;
+		      $block = [];
+		    }
+		    push @$retseq, $seq2->[$_[1]]
+		  };
+  # key function: remove multiple spaces (such that insertion or deletion of redundant white space is not reported)
+  my $keyfunc = sub { join("  ",split(" ",shift())) };
+
+  traverse_sequences($seq1,$seq2, { MATCH=>$match, DISCARD_A=>$discard, DISCARD_B=>$add }, $keyfunc );
+  push @$retseq, @$block if scalar @$block;
+
+  return wantarray ? @$retseq : $retseq ;
+}
+
+
+
+# init_regex_arr_data(\@array,"TOKEN INIT")
+# scans DATA file handel for line "%% TOKEN INIT" line
+# then appends each line not beginning with % into array (as a quoted regex)
+# This is used for command lists and configuration variables, but the processing is slightly
+# different:
+# For lists, the regular expression is extended to include beginning (^) and end ($) markers, to require full-string matching
+# For configuration variables (and all others), simply an unadorned list is copied
+sub init_regex_arr_data {
+  my ($arr,$token)=@_;
+  my $copy=0;
+  my ($mode);
+  if ($token =~ m/COMMANDS/ ) {
+    $mode=0;  # Reading command list
+  } else {
+    $mode=1;  # Reading configuration variables
+  }
+
+  while (<DATA>) {
+    if ( m/^%%BEGIN $token\s*$/ ) {
+      $copy=1;
+      next;
+    } elsif ( m/^%%END $token\s*$/ )  {
+      last; }
+    chomp;
+    if ( $mode==0 ) {
+#      print STDERR "DEBUG init_regex_arr_data regex >$_<\n" if ($debug && $copy);
+      push (@$arr,qr/^$_$/) if ( $copy && !/^%/ ) ;
+    } elsif ($mode==1) {
+      push (@$arr,"$_") if ( $copy && !/^%/ ) ;
+    }
+  }
+  seek DATA,0,0;    # rewind DATA handle to file begin
+}
+
+
+# init_regex_arr_ext(\@array,$arg)
+# appends array with regular expressions.
+# if arg is a file name, then read in list of regular expressions from that file
+# (one expression per line)
+# Otherwise treat arg as a comma separated list of regular expressions
+sub init_regex_arr_ext {
+  my ($arr,$arg)=@_;
+  if ( -f $arg ) {
+    init_regex_arr_file($arr,$arg);
+  } else {
+    init_regex_arr_list($arr,$arg);
+  }
+}
+
+# init_regex_arr_file(\@array,$fname)
+# appends array with regular expressions.
+# Read in list of regular expressions from $fname
+# (one expression per line)
+sub init_regex_arr_file {
+  my ($arr,$fname)=@_;
+  open(FILE,"$fname") or die ("Couldn't open $fname: $!");
+  while (<FILE>) {
+    chomp;
+    next if /^\s*#/ || /^\s*%/ || /^\s*$/ ;
+    push (@$arr,qr/^$_$/);
+  }
+  close(FILE);
+}
+
+# init_regex_arr_list(\@array,$arg)
+# appends array with regular expressions.
+# read from comma separated list of regular expressions ($arg)
+sub init_regex_arr_list {
+  my ($arr,$arg)=@_;
+  my $regex;
+  ###    print STDERR "DEBUG init_regex_arr_list arg >$arg<\n" if $debug;
+  foreach $regex (split(qr/(?<!\\),/,$arg)) {
+    $regex =~ s/\\,/,/g;
+    print STDERR "DEBUG init_regex_arr_list regex >$regex<\n" if $debug;
+    push (@$arr,qr/^$regex$/);
+  }
+}
+
+
+#exetime() returns time since last execution of this command
+#exetime(1) resets this time
+my $lasttime=-1;   # global variable for persistence
+sub exetime {
+  my $reset=0;
+  my $retval;
+  if ((scalar @_) >=1) {
+    $reset=shift;
+  }
+  if ($reset) {
+    $lasttime=times();
+  }
+  else {
+    $retval=times()-$lasttime;
+    $lasttime=$lasttime+$retval;
+    return($retval);
+  }
+}
+
+
+sub usage {
+  print STDERR <<"EOF";
+Usage: $0 [options] old.tex new.tex > diff.tex
+
+Compares two latex files and writes tex code to stdout, which has the same format as new.tex but
+has all changes relative to old.tex marked up or commented. Note that old.tex and new.tex need to
+be real files (not pipes or similar) as they are opened twice.
+
+--type=markupstyle
+-t markupstyle         Add code to preamble for selected markup style
+                       Available styles: UNDERLINE CTRADITIONAL TRADITIONAL CFONT FONTSTRIKE INVISIBLE
+                                         CHANGEBAR CCHANGEBAR CULINECHBAR CFONTCHBAR BOLD PDFCOMMENT
+                                         LUAUNDERLINE
+                       [ Default: UNDERLINE ]
+
+--subtype=markstyle
+-s markstyle           Add code to preamble for selected style for bracketing
+                       commands (e.g. to mark changes in  margin)
+                       Available styles: SAFE MARGIN DVIPSCOL COLOR ZLABEL ONLYCHANGEDPAGE (LABEL)*
+                       [ Default: SAFE ]
+                       * LABEL subtype is deprecated
+
+--floattype=markstyle
+-f markstyle           Add code to preamble for selected style which
+                       replace standard marking and markup commands within floats
+                       (e.g., marginal remarks cause an error within floats
+                       so marginal marking can be disabled thus)
+                       Available styles: FLOATSAFE IDENTICAL
+                       [ Default: FLOATSAFE ]
+
+--encoding=enc
+-e enc                 Specify encoding of old.tex and new.tex. Typical encodings are
+                       ascii, utf8, latin1, latin9.  A list of available encodings can be
+                       obtained by executing
+                       perl -MEncode -e 'print join ("\\n",Encode->encodings( ":all" )) ;'
+                       [Default encoding is utf8 unless the first few lines of the preamble contain
+                       an invocation "\\usepackage[..]{inputenc} in which case the
+                       encoding chosen by this command is asssumed. Note that ASCII (standard
+                       latex) is a subset of utf8]
+
+--preamble=file
+-p file                Insert file at end of preamble instead of auto-generating
+                       preamble.  The preamble must define the following commands
+                       \\DIFaddbegin,\\DIFaddend,\\DIFadd{..},
+                       \\DIFdelbegin,\\DIFdelend,\\DIFdel{..},
+                       and varieties for use within floats
+                       \\DIFaddbeginFL,\\DIFaddendFL,\\DIFaddFL{..},
+                       \\DIFdelbeginFL,\\DIFdelendFL,\\DIFdelFL{..}
+                       (If this option is set -t, -s, and -f options
+                       are ignored.)
+
+--exclude-safecmd=exclude-file
+--exclude-safecmd="cmd1,cmd2,..."
+-A exclude-file
+--replace-safecmd=replace-file
+--append-safecmd=append-file
+--append-safecmd="cmd1,cmd2,..."
+-a append-file         Exclude from, replace or append to the list of regex
+                       matching commands which are safe to use within the
+                       scope of a \\DIFadd or \\DIFdel command.  The file must contain
+                       one Perl-RegEx per line (Comment lines beginning with # or % are
+                       ignored). A literal comma within the comma-separated list must be
+                       escaped thus "\\,",   Note that the RegEx needs to match the whole of
+                       the token, i.e., /^regex\$/ is implied and that the initial
+                       "\\" of the command is not included. The --exclude-safecmd
+                       and --append-safecmd options can be combined with the --replace-safecmd
+                       option and can be used repeatedly to add cumulatively to the lists.
+
+--exclude-textcmd=exclude-file
+--exclude-textcmd="cmd1,cmd2,..."
+-X exclude-file
+--replace-textcmd=replace-file
+--append-textcmd=append-file
+--append-textcmd="cmd1,cmd2,..."
+-x append-file         Exclude from, replace or append to the list of regex
+                       matching commands whose last argument is text.  See
+                       entry for --exclude-safecmd directly above for further details.
+
+--replace-context1cmd=replace-file
+--append-context1cmd=append-file
+--append-context1cmd="cmd1,cmd2,..."
+                       Replace or append to the list of regex matching commands
+                       whose last argument is text but which require a particular
+                       context to work, e.g. \\caption will only work within a figure
+                       or table.  These commands behave like text commands, except when
+                       they occur in a deleted section, when they are disabled, but their
+                       argument is shown as deleted text.
+
+--replace-context2cmd=replace-file
+--append-context2cmd=append-file
+--append-context2cmd="cmd1,cmd2,..."
+                       As corresponding commands for context1.  The only difference is that
+                       context2 commands are completely disabled in deleted sections, including
+                       their arguments.
+                       context2 commands are also the only commands in the preamble, whose argument will
+                       be processed in word-by-word mode (which only works, if they occur no more than
+		       once in the preamble).
+
+--exclude-mboxsafecmd=exclude-file
+--exclude-mboxsafecmd="cmd1,cmd2,..."
+--append-mboxsafecmd=append-file
+--append-mboxsafecmd="cmd1,cmd2,..."
+                       Define safe commands, which additionally need to be protected by encapsulating
+                       in an \\mbox{..}. This is sometimes needed to get around incompatibilities
+                       between external packages and the ulem package, which is  used for highlighting
+                       in the default style UNDERLINE as well as CULINECHBAR CFONTSTRIKE
+
+
+
+--config var1=val1,var2=val2,...
+-c var1=val1,..        Set configuration variables.
+-c configfile           Available variables:
+                          ARRENV (RegEx)
+                          COUNTERCMD (RegEx)
+                          FLOATENV (RegEx)
+                          ITEMCMD (RegEx)
+                          LISTENV (RegEx)
+                          MATHARRENV (RegEx)
+                          MATHENV (RegEx)
+                          MATHREPL (String)
+                          MINWORDSBLOCK (Integer)
+                          PICTUREENV (RegEx)
+                          SCALEDELGRAPHICS (Float)
+                          VERBATIMENV (RegEx)
+                          VERBATIMLINEENV (RegEx)
+                          CUSTOMDIFCMD (RegEx)
+                       This option can be repeated.
+
+--add-to-config  varenv1=pattern1,varenv2=pattern2
+                       For configuration variables containing a regular expression (essentially those ending
+                       in ENV, and COUNTERCMD) this provides an alternative way to modify the configuration
+                       variables. Instead of setting the complete pattern, with this option it is possible to add an
+                       alternative pattern. varenv must be one of the variables listed above that take a regular
+                       expression as argument, and pattern is any regular expression (which might need to be
+                       protected from the shell by quotation). Several patterns can be added at once by using semi-colons
+                       to separate them, e.g. --add-to-config "LISTENV=myitemize;myenumerate,COUNTERCMD=endnote"
+
+--packages=pkg1,pkg2,..
+                       Tell latexdiff that .tex file is processed with the packages in list
+                       loaded.  This is normally not necessary if the .tex file includes the
+                       preamble, as the preamble is automatically scanned for \\usepackage commands.
+                       Use of the --packages option disables automatic scanning, so if for any
+                       reason package specific parsing needs to be switched off, use --packages=none.
+                       The following packages trigger special behaviour:
+                       endfloat hyperref amsmath apacite siunitx cleveref glossaries mhchem chemformula/chemmacros
+                       biblatex
+                       [ Default: scan the preamble for \\usepackage commands to determine
+                         loaded packages.]
+
+--show-preamble        Print generated or included preamble commands to stdout.
+
+--show-safecmd         Print list of regex matching and excluding safe commands.
+
+--show-textcmd         Print list of regex matching and excluding commands with text argument.
+
+--show-config          Show values of configuration variables
+
+--show-all             Show all of the above
+
+   NB For all --show commands, no old.tex or new.tex file needs to be given, and no
+      differencing takes place.
+
+Other configuration options:
+
+--allow-spaces         Allow spaces between bracketed or braced arguments to commands
+                       [Default requires arguments to directly follow each other without
+                                intervening spaces]
+
+--math-markup=level    Determine granularity of markup in displayed math environments:
+                      Possible values for level are (both numerical and text labels are acceptable):
+                      off or 0: suppress markup for math environments.  Deleted equations will not
+                               appear in diff file. This mode can be used if all the other modes
+                               cause invalid latex code.
+                      whole or 1: Differencing on the level of whole equations. Even trivial changes
+                               to equations cause the whole equation to be marked changed.  This
+                               mode can be used if processing in coarse or fine mode results in
+                               invalid latex code.
+                      coarse or 2: Detect changes within equations marked up with a coarse
+                               granularity; changes in equation type (e.g.displaymath to equation)
+                               appear as a change to the complete equation. This mode is recommended
+                               for situations where the content and order of some equations are still
+                               being changed. [Default]
+                      fine or 3: Detect small change in equations and mark up and fine granularity.
+                               This mode is most suitable, if only minor changes to equations are
+                               expected, e.g. correction of typos.
+
+--graphics-markup=level   Change highlight style for graphics embedded with \\includegraphics commands
+                      Possible values for level:
+                      none,off or 0: no highlighting for figures
+                      new-only or 1: surround newly added or changed figures with a blue frame [Default]
+                      both or 2:     highlight new figures with a blue frame and show deleted figures
+                                at reduced scale, and crossed out with a red diagonal cross. Use configuration
+                                variable SCALEDELGRAPHICS to set size of deleted figures.
+                      Note that changes to the optional parameters will make the figure appear as changed
+                      to latexdiff, and this figure will thus be highlighted.
+                      In some circumstances "Misplaced \\noalign" errors can occur if there are certain types
+                      of changes in tables. In this case please use option --graphics-markup=none as a
+                      work-around.
+
+--no-del               Suppress deleted text from the diff. It is similar in effect to the BOLD style,
+                       but the deleted text ist not just invisible in the output, it is also not included in the
+                       diff text file. This can be more robust than just making it invisible.
+
+--disable-citation-markup
+--disable-auto-mbox    Suppress citation markup and markup of other vulnerable commands in styles
+                       using ulem (UNDERLINE,FONTSTRIKE, CULINECHBAR)
+                       (the two options are identical and are simply aliases)
+
+--enable-citation-markup
+--enforce-auto-mbox    Protect citation commands and other vulnerable commands in changed sections
+                       with \\mbox command, i.e. use default behaviour for ulem package for other packages
+                       (the two options are identical and are simply aliases)
+
+
+
+Miscellaneous options
+
+--label=label
+-L label               Sets the labels used to describe the old and new files.  The first use
+                       of this option sets the label describing the old file and the second
+                       use of the option sets the label for the new file.
+                       [Default: use the filename and modification dates for the label]
+
+--no-label             Suppress inclusion of old and new file names as comment in output file
+
+--visible-label         Include old and new filenames (or labels set with --label option) as
+                       visible output
+
+--flatten              Replace \\input and \\include commands within body by the content
+                       of the files in their argument.  If \\includeonly is present in the
+                       preamble, only those files are expanded into the document. However,
+                       no recursion is done, i.e. \\input and \\include commands within
+                       included sections are not expanded.  The included files are assumed to
+                       be located in the same directories as the old and new master files,
+                       respectively, making it possible to organise files into old and new directories.
+                       --flatten is applied recursively, so inputted files can contain further
+                       \\input statements.  Also handles files included by the import package
+                       (\\import and \\subimport), and \\subfile command.
+
+--filter-script=filterscript    Run files through this filterscript (full path preferred) before processing.
+                       The filterscript must take STDIN input and output to STDOUT.
+                       When coupled with --flatten, each file will be run through the filter as it is brought in.
+
+--ignore-filter-stderr When running with --filter-script, STDERR from the script may cause readability issues.
+                       Turn this flag on to ignore STDERR from the filter script.
+
+--driver=type          Choose driver for changebar package (only relevant for styles using
+                       changebar: CCHANGEBAR CFONTCHBAR CULINECHBAR CHANGEBAR). Possible
+                       drivers are listed in changebar manual, e.g. pdftex,dvips,dvitops
+                       [Default: pdftex]
+
+--help
+-h                     Show this help text.
+
+--ignore-warnings      Suppress warnings about inconsistencies in length between input
+                       and parsed strings and missing characters.
+
+--verbose
+-V                     Output various status information to stderr during processing.
+                       Default is to work silently.
+
+--version              Show version number.
+
+Internal options:
+These options are mostly for automated use by latexdiff-vc. They can be used directly, but
+the API should be considered less stable than for the other options.
+
+--no-links             Suppress generation of hyperreferences, used for minimal diffs
+                       (option --only-changes of latexdiff-vc).
+EOF
+  exit 0;
+}
+
+=head1 NAME
+
+latexdiff - determine and markup differences between two latex files
+
+=head1 SYNOPSIS
+
+B<latexdiff> [ B<OPTIONS> ] F<old.tex> F<new.tex> > F<diff.tex>
+
+=head1 DESCRIPTION
+
+Briefly, I<latexdiff> is a utility program to aid in the management of
+revisions of latex documents. It compares two valid latex files, here
+called C<old.tex> and C<new.tex>, finds significant differences
+between them (i.e., ignoring the number of white spaces and position
+of line breaks), and adds special commands to highlight the
+differences.  Where visual highlighting is not possible, e.g. for changes
+in the formatting, the differences are
+nevertheless marked up in the source. Note that old.tex and new.tex need to
+be real files (not pipes or similar) as they are opened twice (unless C<--encoding> option is used)
+
+The program treats the preamble differently from the main document.
+Differences between the preambles are found using line-based
+differencing (similarly to the Unix diff command, but ignoring white
+spaces).  A comment, "S<C<%DIF E<gt>>>" is appended to each added line, i.e. a
+line present in C<new.tex> but not in C<old.tex>.  Discarded lines
+ are deactivated by prepending "S<C<%DIF E<lt>>>". Changed blocks are preceded  by
+comment lines giving information about line numbers in the original files.  Where there are insignificant
+differences, the resulting file C<diff.tex> will be similar to
+C<new.tex>.  At the end of the preamble, the definitions for I<latexdiff> markup commands are inserted.
+In differencing the main body of the text, I<latexdiff> attempts to
+satisfy the following guidelines (in order of priority):
+
+=over 3
+
+=item 1
+
+If both C<old.tex> and C<new.tex> are valid LaTeX, then the resulting
+C<diff.tex> should also be valid LateX. (NB If a few plain TeX commands
+are used within C<old.tex> or C<new.tex> then C<diff.tex> is not
+guaranteed to work but usually will).
+
+=item 2
+
+Significant differences are determined on the level of
+individual words. All significant differences, including differences
+between comments should be clearly marked in the resulting source code
+C<diff.tex>.
+
+=item 3
+
+If a changed passage contains text or text-producing commands, then
+running C<diff.tex> through LateX should produce output where added
+and discarded passages are highlighted.
+
+=item 4
+
+Where there are insignificant differences, e.g. in the positioning of
+line breaks, C<diff.tex> should follow the formatting of C<new.tex>
+
+=back
+
+For differencing the same algorithm as I<diff> is used but words
+instead of lines are compared.  An attempt is made to recognize
+blocks which are completely changed such that they can be marked up as a unit.
+Comments are differenced line by line
+but the number of spaces within comments is ignored. Commands including
+all their arguments are generally compared as one unit, i.e., no mark-up
+is inserted into the arguments of commands.  However, for a selected
+number of commands (for example, C<\caption> and all sectioning
+commands) the last argument is known to be text. This text is
+split into words and differenced just as ordinary text (use options to
+show and change the list of text commands, see below). As the
+algorithm has no detailed knowledge of LaTeX, it assumes all pairs of
+curly braces immediately following a command (i.e. a sequence of
+letters beginning with a backslash) are arguments for that command.
+As a restriction to condition 1 above it is thus necessary to surround
+all arguments with curly braces, and to not insert
+extraneous spaces.  For example, write
+
+  \section{\textem{This is an emphasized section title}}
+
+and not
+
+  \section {\textem{This is an emphasized section title}}
+
+or
+
+  \section\textem{This is an emphasized section title}
+
+even though all varieties are the same to LaTeX (but see
+B<--allow-spaces> option which allows the second variety).
+
+For environments whose content does not conform to standard LaTeX or
+where graphical markup does not make sense all markup commands can be
+removed by setting the PICTUREENV configuration variable, set by
+default to C<picture> and C<DIFnomarkup> environments; see B<--config>
+option).  The latter environment (C<DIFnomarkup>) can be used to
+protect parts of the latex file where the markup results in illegal
+markup. You have to surround the offending passage in both the old and
+new file by C<\begin{DIFnomarkup}> and C<\end{DIFnomarkup}>. You must
+define the environment in the preambles of both old and new
+documents. I prefer to define it as a null-environment,
+
+C<\newenvironment{DIFnomarkup}{}{}>
+
+but the choice is yours.  Any markup within the environment will be
+removed, and generally everything within the environment will just be
+taken from the new file.
+
+It is also possible to difference files which do not have a preamble.
+ In this case, the file is processed in the main document
+mode, but the definitions of the markup commands are not inserted.
+
+All markup commands inserted by I<latexdiff> begin with "C<\DIF>".  Added
+blocks containing words, commands or comments which are in C<new.tex>
+but not in C<old.tex> are marked by C<\DIFaddbegin> and C<\DIFaddend>.
+Discarded blocks are marked by C<\DIFdelbegin> and C<\DIFdelend>.
+Within added blocks all text is highlighted with C<\DIFadd> like this:
+C<\DIFadd{Added text block}>
+Selected `safe' commands can be contained in these text blocks as well
+(use options to show and change the list of safe commands, see below).
+All other commands as well as braces "{" and "}" are never put within
+the scope of C<\DIFadd>.  Added comments are marked by prepending
+"S<C<%DIF E<gt> >>".
+
+Within deleted blocks text is highlighted with C<\DIFdel>.  Deleted
+comments are marked by prepending "S<C<%DIF E<lt> >>".  Non-safe command
+and curly braces within deleted blocks are commented out with
+"S<C<%DIFDELCMD E<lt> >>".
+
+
+
+=head1 OPTIONS
+
+=head2 Preamble
+
+The following options determine the visual markup style by adding the appropriate
+command definitions to the preamble. See the end of this section for a description of
+available styles.
+
+=over 4
+
+=item B<--type=markupstyle> or
+B<-t markupstyle>
+
+Add code to preamble for selected markup style. This option defines
+C<\DIFadd> and C<\DIFdel> commands.
+Available styles:
+
+C<UNDERLINE CTRADITIONAL TRADITIONAL CFONT FONTSTRIKE INVISIBLE
+CHANGEBAR CCHANGEBAR CULINECHBAR CFONTCHBAR BOLD PDFCOMMENT
+LUAUNDERLINE>
+
+[ Default: C<UNDERLINE> ]
+
+=item B<--subtype=markstyle> or
+B<-s markstyle>
+
+Add code to preamble for selected style for bracketing
+commands (e.g. to mark changes in  margin). This option defines
+C<\DIFaddbegin>, C<\DIFaddend>, C<\DIFdelbegin> and C<\DIFdelend> commands.
+Available styles: C<SAFE MARGIN COLOR DVIPSCOL  ZLABEL ONLYCHANGEDPAGE (LABEL)*>
+
+[ Default: C<SAFE> ]
+* Subtype C<LABEL> is deprecated
+
+=item B<--floattype=markstyle> or
+B<-f markstyle>
+
+Add code to preamble for selected style which
+replace standard marking and markup commands within floats
+(e.g., marginal remarks cause an error within floats
+so marginal marking can be disabled thus). This option defines all
+C<\DIF...FL> commands.
+Available styles: C<FLOATSAFE TRADITIONALSAFE IDENTICAL>
+
+[ Default: C<FLOATSAFE> ]
+
+=item B<--encoding=enc> or
+B<-e enc>
+
+Specify encoding of old.tex and new.tex. Typical encodings are
+C<ascii>, C<utf8>, C<latin1>, C<latin9>.  A list of available encodings can be
+obtained by executing
+
+C<perl -MEncode -e 'print join ("\n",Encode->encodings( ":all" )) ;' >
+
+If this option is used, then old.tex, new.tex are only opened once.
+[Default encoding is utf8 unless the first few lines of the preamble contain
+an invocation C<\usepackage[..]{inputenc}> in which case the
+encoding chosen by this command is asssumed. Note that ASCII (standard
+latex) is a subset of utf8]
+
+=item B<--preamble=file> or
+B<-p file>
+
+Insert file at end of preamble instead of generating
+preamble.  The preamble must define the following commands
+C<\DIFaddbegin, \DIFaddend, \DIFadd{..},
+\DIFdelbegin,\DIFdelend,\DIFdel{..},>
+and varieties for use within floats
+C<\DIFaddbeginFL, \DIFaddendFL, \DIFaddFL{..},
+\DIFdelbeginFL, \DIFdelendFL, \DIFdelFL{..}>
+(If this option is set B<-t>, B<-s>, and B<-f> options
+are ignored.)
+
+=item B<--packages=pkg1,pkg2,..>
+
+Tell latexdiff that .tex file is processed with the packages in list
+loaded.  This is normally not necessary if the .tex file includes the
+preamble, as the preamble is automatically scanned for C<\usepackage> commands.
+Use of the B<--packages> option disables automatic scanning, so if for any
+reason package specific parsing needs to be switched off, use B<--packages=none>.
+The following packages trigger special behaviour:
+
+=over 8
+
+=item C<endfloat>
+
+Ensure that C<\begin{figure}> and C<\end{figure}> always appear by themselves on a line.
+
+=item C<hyperref>
+
+Change name of C<\DIFadd> and C<\DIFdel> commands to C<\DIFaddtex> and C<\DIFdeltex> and
+define new C<\DIFadd> and C<\DIFdel> commands, which provide a wrapper for these commands,
+using them for the text but not for the link defining command (where any markup would cause
+errors).
+
+=item C<apacite>, C<biblatex>
+
+Redefine the commands recognised as citation commands.
+
+=item C<siunitx>
+
+Treat C<\SI> as equivalent to citation commands (i.e. protect with C<\mbox> if markup style uses ulem package.
+
+=item C<cleveref>
+
+Treat C<\cref,\Cref>, etc as equivalent to citation commands (i.e. protect with C<\mbox> if markup style uses ulem package.
+
+=item C<glossaries>
+
+Define most of the glossaries commands as safe, protecting them with \mbox'es where needed
+
+=item C<mhchem>
+
+Treat C<\ce> as a safe command, i.e. it will be highlighted (note that C<\cee> will not be highlighted in equations as this leads to processing errors)
+
+=item C<chemformula> or C<chemmacros>
+
+Treat C<\ch> as a safe command outside equations, i.e. it will be highlighted (note that C<\ch> will not be highlighted in equations as this leads to processing errors)
+
+
+=back
+
+[ Default: scan the preamble for C<\usepackage> commands to determine
+  loaded packages. ]
+
+
+
+=item B<--show-preamble>
+
+Print generated or included preamble commands to stdout.
+
+=back
+
+=head2 Configuration
+
+=over 4
+
+=item B<--exclude-safecmd=exclude-file> or
+B<-A exclude-file> or  B<--exclude-safecmd="cmd1,cmd2,...">
+
+=item B<--replace-safecmd=replace-file>
+
+=item B<--append-safecmd=append-file> or
+B<-a append-file> or B<--append-safecmd="cmd1,cmd2,...">
+
+Exclude from, replace or append to the list of regular expressions (RegEx)
+matching commands which are safe to use within the
+scope of a C<\DIFadd> or C<\DIFdel> command.  The file must contain
+one Perl-RegEx per line (Comment lines beginning with # or % are
+ignored).  Note that the RegEx needs to match the whole of
+the token, i.e., /^regex$/ is implied and that the initial
+"\" of the command is not included.
+The B<--exclude-safecmd> and B<--append-safecmd> options can be combined with the -B<--replace-safecmd>
+option and can be used repeatedly to add cumulatively to the lists.
+ B<--exclude-safecmd>
+and B<--append-safecmd> can also take a comma separated list as input. If a
+comma for one of the regex is required, escape it thus "\,". In most cases it
+will be necessary to protect the comma-separated list from the shell by putting
+it in quotation marks.
+
+=item B<--exclude-textcmd=exclude-file> or
+B<-X exclude-file> or B<--exclude-textcmd="cmd1,cmd2,...">
+
+=item B<--replace-textcmd=replace-file>
+
+=item B<--append-textcmd=append-file> or
+B<-x append-file> or B<--append-textcmd="cmd1,cmd2,...">
+
+Exclude from, replace or append to the list of regular expressions
+matching commands whose last argument is text.  See
+entry for B<--exclude-safecmd> directly above for further details.
+
+
+=item B<--replace-context1cmd=replace-file>
+
+=item B<--append-context1cmd=append-file> or
+
+=item B<--append-context1cmd="cmd1,cmd2,...">
+
+Replace or append to the list of regex matching commands
+whose last argument is text but which require a particular
+context to work, e.g. C<\caption> will only work within a figure
+or table.  These commands behave like text commands, except when
+they occur in a deleted section, when they are disabled, but their
+argument is shown as deleted text.
+
+=item B<--replace-context2cmd=replace-file>
+
+=item B<--append-context2cmd=append-file> or
+
+=item B<--append-context2cmd="cmd1,cmd2,...">
+
+As corresponding commands for context1.  The only difference is that
+context2 commands are completely disabled in deleted sections, including
+their arguments.
+
+context2 commands are also the only commands in the preamble, whose argument will be processed in
+word-by-word mode (which only works, if they occur no more than once in the preamble). The algorithm currently cannot cope with repeated context2 commands in the preamble, as they occur e.g. for the C<\author> argument in some journal styles (not in the standard styles, though
+If such a repetition is detected, the whole preamble will be processed in line-by-line mode. In such a case, use C<--replace-context2cmd> option to just select the commands, which should be processed and are not used repeatedly in the preamble.
+
+
+
+=item B<--exclude-mboxsafecmd=exclude-file> or B<--exclude-mboxsafecmd="cmd1,cmd2,...">
+
+=item B<--append-mboxsafecmd=append-file> or B<--append-mboxsafecmd="cmd1,cmd2,...">
+
+Define safe commands, which additionally need to be protected by encapsulating
+in an C<\mbox{..}>. This is sometimes needed to get around incompatibilities
+between external packages and the ulem package, which is  used for highlighting
+in the default style UNDERLINE as well as CULINECHBAR CFONTSTRIKE
+
+
+
+
+
+=item B<--config var1=val1,var2=val2,...> or B<-c var1=val1,..>
+
+=item B<-c configfile>
+
+Set configuration variables.  The option can be repeated to set different
+variables (as an alternative to the comma-separated list).
+Available variables (see below for further explanations):
+
+C<ARRENV> (RegEx)
+
+C<COUNTERCMD> (RegEx)
+
+C<CUSTODIFCMD> (RegEx)
+
+C<FLOATENV> (RegEx)
+
+C<ITEMCMD> (RegEx)
+
+C<LISTENV>  (RegEx)
+
+C<MATHARRENV> (RegEx)
+
+C<MATHENV> (RegEx)
+
+C<MATHREPL> (String)
+
+C<MINWORDSBLOCK> (Integer)
+
+C<PICTUREENV> (RegEx)
+
+C<SCALEDELGRAPHICS> (Float)
+
+
+=item B<--add-to-config varenv1=pattern1,varenv2=pattern2,...>
+
+For configuration variables, which are a regular expression (essentially those ending
+in ENV, COUNTERCMD and CUSTOMDIFCMD, see list above) this option provides an alternative way to modify the configuration
+variables. Instead of setting the complete pattern, with this option it is possible to add an
+alternative pattern. C<varenv> must be one of the variables listed above that take a regular
+expression as argument, and pattern is any regular expression (which might need to be
+protected from the shell by quotation). Several patterns can be added at once by using semi-colons
+to separate them, e.g. C<--add-to-config "LISTENV=myitemize;myenumerate,COUNTERCMD=endnote">
+
+=item B<--show-safecmd>
+
+Print list of RegEx matching and excluding safe commands.
+
+=item B<--show-textcmd>
+
+Print list of RegEx matching and excluding commands with text argument.
+
+=item B<--show-config>
+
+Show values of configuration variables.
+
+=item B<--show-all>
+
+Combine all --show commands.
+
+NB For all --show commands, no C<old.tex> or C<new.tex> file needs to be specified, and no
+differencing takes place.
+
+=back
+
+=head2 Other configuration options:
+
+=over 4
+
+=item B<--allow-spaces>
+
+Allow spaces between bracketed or braced arguments to commands.  Note
+that this option might have undesirable side effects (unrelated scope
+might get lumpeded with preceding commands) so should only be used if the
+default produces erroneous results.  (Default requires arguments to
+directly follow each other without intervening spaces).
+
+=item B<--math-markup=level>
+
+Determine granularity of markup in displayed math environments:
+Possible values for level are (both numerical and text labels are acceptable):
+
+C<off> or C<0>: suppress markup for math environments.  Deleted equations will not
+appear in diff file. This mode can be used if all the other modes
+cause invalid latex code.
+
+C<whole> or C<1>: Differencing on the level of whole equations. Even trivial changes
+to equations cause the whole equation to be marked changed.  This
+mode can be used if processing in coarse or fine mode results in
+invalid latex code.
+
+C<coarse> or C<2>: Detect changes within equations marked up with a coarse
+granularity; changes in equation type (e.g.displaymath to equation)
+appear as a change to the complete equation. This mode is recommended
+for situations where the content and order of some equations are still
+being changed. [Default]
+
+C<fine> or C<3>: Detect small change in equations and mark up at fine granularity.
+This mode is most suitable, if only minor changes to equations are
+expected, e.g. correction of typos.
+
+=item B<--graphics-markup=level>
+
+ Change highlight style for graphics embedded with C<\includegraphics> commands.
+
+Possible values for level:
+
+C<none>, C<off> or C<0>: no highlighting for figures
+
+C<new-only> or C<1>: surround newly added or changed figures with a blue frame [Default if graphicx package loaded]
+
+C<both> or C<2>:     highlight new figures with a blue frame and show deleted figures at reduced
+scale, and crossed out with a red diagonal cross. Use configuration
+variable SCALEDELGRAPHICS to set size of deleted figures.
+
+Note that changes to the optional parameters will make the figure appear as changed
+to latexdiff, and this figure will thus be highlighted.
+
+In some circumstances "Misplaced \noalign" error can occur if there are certain types
+of changes in tables. In this case please use C<--graphics-markup=none> as a
+work-around.
+
+
+=item B<--no-del>
+
+Suppress deleted text from the diff. It is similar in effect to the BOLD style,
+but the deleted text ist not just invisible in the output, it is also not included in the diff text file.
+This can be more robust than just making it invisible.
+
+=item B<--disable-citation-markup> or B<--disable-auto-mbox>
+
+Suppress citation markup and markup of other vulnerable commands in styles
+using ulem (UNDERLINE,FONTSTRIKE, CULINECHBAR)
+(the two options are identical and are simply aliases)
+
+=item B<--enable-citation-markup> or B<--enforce-auto-mbox>
+
+Protect citation commands and other vulnerable commands in changed sections
+with C<\mbox> command, i.e. use default behaviour for ulem package for other packages
+(the two options are identical and are simply aliases)
+
+=back
+
+=head2 Miscellaneous
+
+=over 4
+
+=item B<--verbose> or B<-V>
+
+Output various status information to stderr during processing.
+Default is to work silently.
+
+=item B<--driver=type>
+
+Choose driver for changebar package (only relevant for styles using
+   changebar: CCHANGEBAR CFONTCHBAR CULINECHBAR CHANGEBAR). Possible
+drivers are listed in changebar manual, e.g. pdftex,dvips,dvitops
+  [Default: pdftex]
+
+=item B<--ignore-warnings>
+
+Suppress warnings about inconsistencies in length between input and
+parsed strings and missing characters.  These warning messages are
+often related to non-standard latex or latex constructions with a
+syntax unknown to C<latexdiff> but the resulting difference argument
+is often fully functional anyway, particularly if the non-standard
+latex only occurs in parts of the text which have not changed.
+
+=item B<--label=label> or
+B<-L label>
+
+Sets the labels used to describe the old and new files.  The first use
+of this option sets the label describing the old file and the second
+use of the option sets the label for the new file, i.e. set both
+labels like this C<-L labelold -L labelnew>.
+[Default: use the filename and modification dates for the label]
+
+=item B<--no-label>
+
+Suppress inclusion of old and new file names as comment in output file
+
+=item B<--visible-label>
+
+Include old and new filenames (or labels set with C<--label> option) as
+visible output.
+
+=item B<--flatten>
+
+Replace C<\input> and C<\include> commands within body by the content
+of the files in their argument.  If C<\includeonly> is present in the
+preamble, only those files are expanded into the document. However,
+no recursion is done, i.e. C<\input> and C<\include> commands within
+included sections are not expanded.  The included files are assumed to
+ be located in the same directories as the old and new master files,
+respectively, making it possible to organise files into old and new directories.
+--flatten is applied recursively, so inputted files can contain further
+C<\input> statements.  Also handles files included by the import package
+(C<\import> and C<\subimport>), and C<\subfile> command.
+
+Use of this option might result in prohibitive processing times for
+larger documents, and the resulting difference document
+no longer reflects the structure of the input documents.
+
+=item B<--filter-script=filterscript>
+
+Run files through this filterscript (full path preferred) before processing.
+The filterscript must take STDIN input and output to STDOUT.
+When coupled with --flatten, each file will be run through the filter as it is brought in.
+
+=item B<--ignore-filter-stderr>
+
+When running with --filter-script, STDERR from the script may cause readability issues.
+Turn this flag on to ignore STDERR from the filter script.
+
+
+
+=item B<--help> or
+B<-h>
+
+Show help text
+
+=item B<--version>
+
+Show version number
+
+=back
+
+
+=head2 Internal options
+
+These options are mostly for automated use by latexdiff-vc. They can be used directly, but the API should be considered less stable than for the other options.
+
+=over 4
+
+=item B<--no-links>
+
+Suppress generation of hyperreferences, used for minimal diffs (option --only-changes of latexdiff-vc)
+
+=back
+
+
+=head2 Predefined styles
+
+=head2 Major types
+
+The major type determine the markup of plain text and some selected latex commands outside floats by defining the markup commands C<\DIFadd{...}> and C<\DIFdel{...}> .
+
+=over 10
+
+=item C<UNDERLINE>
+
+Added text is wavy-underlined and blue, discarded text is struck out and red
+(Requires color and ulem packages).  Overstriking does not work in displayed math equations such that deleted parts of equation are underlined, not struck out (this is a shortcoming inherent to the ulem package).
+
+=item C<LUAUNDERLINE>
+
+Added text is underlined and blue, discarded text is struck out and red
+(Requires lua-ul package + LuaLaTeX).
+
+=item C<CTRADITIONAL>
+
+Added text is blue and set in sans-serif, and a red footnote is created for each discarded
+piece of text. (Requires color package)
+
+=item C<TRADITIONAL>
+
+Like C<CTRADITIONAL> but without the use of color.
+
+=item C<CFONT>
+
+Added text is blue and set in sans-serif, and discarded text is red and very small size.
+
+=item C<FONTSTRIKE>
+
+Added tex is set in sans-serif, discarded text small and struck out
+
+=item C<CCHANGEBAR>
+
+Added text is blue, and discarded text is red.  Additionally, the changed text is marked with a bar in the margin (Requires color and changebar packages).
+
+=item C<CFONTCHBAR>
+
+Like C<CFONT> but with additional changebars (Requires color and changebar packages).
+
+=item C<CULINECHBAR>
+
+Like C<UNDERLINE> but with additional changebars (Requires color, ulem and changebar packages).
+
+=item C<CHANGEBAR>
+
+No mark up of text, but mark margins with changebars (Requires changebar package).
+
+=item C<INVISIBLE>
+
+No visible markup (but generic markup commands will still be inserted.
+
+=item C<BOLD>
+
+Added text is set in bold face, discarded is not shown. (also see --no-del option for another possibility to hide deleted text)
+
+=item C<PDFCOMMENT>
+
+The pdfcomment package is used to underline new text, and mark deletions with a PDF comment. Note that this markup might appear differently or not at all based on the pdf viewer used. The viewer with best support for pdf markup is probably acroread. This style is only recommended if the number of differences is small.
+
+=back
+
+=head2 Subtypes
+
+The subtype defines the commands that are inserted at the begin and end of added or discarded blocks, irrespectively of whether these blocks contain text or commands (Defined commands: C<\DIFaddbegin, \DIFaddend, \DIFdelbegin, \DIFdelend>)
+
+=over 10
+
+=item C<SAFE>
+
+No additional markup (Recommended choice)
+
+=item C<MARGIN>
+
+Mark beginning and end of changed blocks with symbols in the margin nearby (using
+the standard C<\marginpar> command - note that this sometimes moves somewhat
+from the intended position.
+
+=item C<COLOR>
+
+An alternative way of marking added passages in blue, and deleted ones in red.
+(It is recommeneded to use instead the main types to effect colored markup,
+although in some cases coloring with dvipscol can be more complete, for example
+with citation commands).
+
+=item C<DVIPSCOL>
+
+An alternative way of marking added passages in blue, and deleted ones in red. Note
+that C<DVIPSCOL> only works with the dvips converter, e.g. not pdflatex.
+(it is recommeneded to use instead the main types to effect colored markup,
+although in some cases coloring with dvipscol can be more complete).
+
+
+=item C<ZLABEL>
+
+can be used to highlight only changed pages, but requires post-processing. It is recommend to not call this option manually but use C<latexdiff-vc> with C<--only-changes> option. Alternatively, use the script given within preamble of diff files made using this style.
+
+=item C<ONLYCHANGEDPAGE>
+
+also highlights changed pages, without the need for post-processing, but might not work reliably if
+there is floating material (figures, tables).
+
+=item C<LABEL>
+
+is similar to C<ZLABEL>, but does not need the zref package and works less reliably (deprecated).
+
+=back
+
+=head2 Float Types
+
+Some of the markup used in the main text might cause problems when used within
+floats (e.g. figures or tables).  For this reason alternative versions of all
+markup commands are used within floats. The float type defines these alternative commands.
+
+=over 10
+
+=item C<FLOATSAFE>
+
+Use identical markup for text as in the main body, but set all commands marking the begin and end of changed blocks to null-commands.  You have to choose this float type if your subtype is C<MARGIN> as C<\marginpar> does not work properly within floats.
+
+=item C<TRADITIONALSAFE>
+
+Mark additions the same way as in the main text.  Deleted environments are marked by angular brackets \[ and \] and the deleted text is set in scriptscript size. This float type should always be used with the C<TRADITIONAL> and  C<CTRADITIONAL> markup types as the \footnote command does not work properly in floating environments.
+
+=item C<IDENTICAL>
+
+Make no difference between the main text and floats.
+
+=back
+
+
+=head2 Configuration Variables
+
+=over 10
+
+=item C<ARRENV>
+
+If a match to C<ARRENV> is found within an inline math environment within a deleted or added block, then the inlined math
+is surrounded by C<\mbox{>...C<}>.  This is necessary as underlining does not work within inlined array environments.
+
+[ Default: C<ARRENV>=S<C<(?:array|[pbvBV]matrix)> >
+
+=item C<COUNTERCMD>
+
+If a command in a deleted block which is also in the textcmd list matches C<COUNTERCMD> then an
+additional command C<\addtocounter{>F<cntcmd>C<}{-1}>, where F<cntcmd> is the matching command, is appended in the diff file such that the numbering in the diff file remains synchronized with the
+numbering in the new file.
+
+[ Default: C<COUNTERCMD>=C<(?:footnote|part|section|subsection> ...
+
+C<|subsubsection|paragraph|subparagraph)>  ]
+
+=item C<CUSTOMDIFCMD>
+
+This option is for advanced users and allows definition of special versions of commands, which do not work as safe commands.
+
+Commands in C<CUSTOMDIFCMD> that occur in added or deleted blocks will be given an ADD or DEL prefix.
+The prefixed versions of the command must be defined in the preamble, either by putting them
+in the preamble of at least the new file, or by creating a custom preamble file (Option --preamble).
+For example the command C<\blindtext> (from package blindtext) does not interact well with underlining, so that
+for the standard markup type, it is not satisfactory to define it as a safe command. Instead, a customised versions
+without underlining can be defined in the preamble:
+
+C<\newcommand{\DELblindtext}{{\color{red}\blindtext}}>
+
+C<\newcommand{\ADDblindtext}{{\color{blue}\blindtext}}>
+
+and then latexdiff should be invoked with the option C<-c CUSTOMDIFCMD=blindtext>.
+
+[ Default: none ]
+
+=item C<FLOATENV>
+
+Environments whose name matches the regular expression in C<FLOATENV> are
+considered floats.  Within these environments, the I<latexdiff> markup commands
+are replaced by their FL variaties.
+
+[ Default: S<C<(?:figure|table|plate)[\w\d*@]*> >]
+
+=item C<ITEMCMD>
+
+Commands representing new item line with list environments.
+
+[ Default: \C<item> ]
+
+=item C<LISTENV>
+
+Environments whose name matches the regular expression in C<LISTENV> are list environments.
+
+[ Default: S<C<(?:itemize|enumerate|description)> >]
+
+=item C<MATHENV>,C<MATHREPL>
+
+If both \begin and \end for a math environment (environment name matching C<MATHENV> or \[ and \])
+are within the same deleted block, they are replaced by a \begin and \end commands for C<MATHREPL>
+rather than being commented out.
+
+[ Default: C<MATHENV>=S<C<(?:displaymath|equation)> >, C<MATHREPL>=S<C<displaymath> >]
+
+=item C<MINWORDSBLOCK>
+
+Minimum number of tokens required to form an independent block. This value is
+used in the algorithm to detect changes of complete blocks by merging identical text parts of less than C<MINWORDSBLOCK> to the preceding added and discarded parts.
+
+[ Default: 3 ]
+
+=item C<PICTUREENV>
+
+Within environments whose name matches the regular expression in C<PICTUREENV>
+all latexdiff markup is removed (in pathologic cases this might lead to
+inconsistent markup but this situation should be rare).
+
+[ Default: S<C<(?:picture|DIFnomarkup)[\w\d*@]*> >]
+
+=item C<SCALEDELGRAPHICS>
+
+If C<--graphics-markup=both> is chosen, C<SCALEDELGRAPHICS> is the factor, by which deleted figures will be scaled (i.e. 0.5 implies they are shown at half linear size).
+
+[ Default: 0.5 ]
+
+=item C<VERBATIMENV>
+
+RegEx describing environments like verbatim, whose contents should be taken verbatim. The content of these environments will not be processed in any way:
+deleted content is commented out, new content is not marked up
+
+[ Default:  S<C<comment> > ]
+
+=item C<VERBATIMLINEENV>
+
+RegEx describing environments like verbatim, whose contents should be taken verbatim. The content of environments described by VERBATIMLINEENV are compared in
+line mode, and changes are marked up using the listings package. The markup style is set based on the chosen mains markup type (Option -t), or on an analysis
+of the preamble.
+Note that "listings.sty" must be installed. If this file is not found the fallback solution is to
+treat VERBATIMLINEENV environments treated exactly the same way as VERBATIMENV environments.
+
+[ Default:  S<C<(?:verbatim[*]?|lstlisting> > ]
+
+=back
+
+=head1 COMMON PROBLEMS AND FAQ
+
+=over 10
+
+=item Changed citations result in overfull boxes
+
+There is an incompatibility between the C<ulem> package, which C<latexdiff> uses for underlining and striking out in the UNDERLINE style,
+the default style, and the way citations are generated. In order to be able to mark up citations properly, they are enclosed with an C<\mbox>
+command. As mboxes cannot be broken across lines, this procedure frequently results in overfull boxes, possibly obscuring the content as it extends beyond the right margin.  The same occurs for some other packages (e.g., siunitx). If this is a problem, you have several possibilities.
+
+1. Use C<CFONT> type markup (option C<-t CFONT>): If this markup is chosen, then changed citations are no longer marked up
+with the wavy line (additions) or struck out (deletions), but are still highlighted in the appropriate color, and deleted text is shown with a different font. Other styles not using the C<ulem> package will also work.
+
+2. Choose option C<--disable-citation-markup> which turns off the marking up of citations: deleted citations are no longer shown, and
+added citations are shown without markup. (This was the default behaviour of latexdiff at versions 0.6 and older).
+For custom packages you can define the commands which need to be protected by C<\mbox> with C<--append-mboxsafecmd> and C<--excludemboxsafecmd> options
+(submit your lists of command as feature request at github page to set the default behaviour of future versions, see section 6)
+
+3. If you are using luatex >= 1.12.0 you can use option LUAUNDERLINE that uses lua-ul instead of ulem for underlining, which does not have this problem (experimental feature).
+
+=item Changes in complicated mathematical equations result in latex processing errors.
+
+Try option C<--math-markup=whole>.   If even that fails, you can turn off mark up for equations with C<--math-markup=off>.
+
+=item Deleted parts in equations are not struck out but underlined.
+
+This is a limitation of the ulem package that implements the strike-out. If you use the amsmath package, then the strike-out command is redefined in such a way that deleted passages are wrapped with C<\text> command; adding C<\usepackage{amsmath}> to your preamble will trigger this behaviour. (If amsmath is not included directly, but imported by another package, latexdiff will not be able to detect its availability; in this case you can give latexdiff a hint with option C<--packages=amsmath>.
+
+
+=item How can I just show the pages where changes had been made?
+
+Use options C<--s ZLABEL>  (some postprocessing required) or C<-s ONLYCHANGEDPAGE>. C<latexdiff-vc --ps|--pdf> with C<--only-changes> option takes care of
+the post-processing for you (requires zref package to be installed).
+
+=item The character encoding breaks when running latexdiff from powershell.
+
+This problem is not limited to C<latexdiff> and has to do with the default settings of C<powershell> in Windows. It is recommended to use C<cmd> instead.
+
+=back
+
+=head1 BUGS
+
+=over 10
+
+=item Option allow-spaces not implemented entirely consistently. It breaks
+the rules that number and type of white space does not matter, as
+different numbers of inter-argument spaces are treated as significant.
+
+=back
+
+Please submit bug reports using the issue tracker of the github repository page I<https://github.com/ftilmann/latexdiff.git>,
+or send them to I<tilmann -- AT -- gfz-potsdam.de>.  Include the version number of I<latexdiff>
+(from comments at the top of the source or use B<--version>).  If you come across latex
+files that are error-free and conform to the specifications set out
+above, and whose differencing still does not result in error-free
+latex, please send me those files, ideally edited to only contain the
+offending passage as long as that still reproduces the problem. If your
+file relies on non-standard class files, you must include those.  I will not
+look at examples where I have trouble to latex the original files.
+
+=head1 SEE ALSO
+
+L<latexrevise>, L<latexdiff-vc>
+
+=head1 PORTABILITY
+
+I<latexdiff> does not make use of external commands and thus should run
+on any platform  supporting Perl 5.6 or higher.  If files with encodings
+other than ASCII or UTF-8 are processed, Perl 5.8 or higher is required.
+
+The standard version of I<latexdiff> requires installation of the Perl package
+C<Algorithm::Diff> (available from I<www.cpan.org> -
+I<http://search.cpan.org/~nedkonz/Algorithm-Diff-1.15>) but a stand-alone
+version, I<latexdiff-so>, which has this package inlined, is available, too.
+I<latexdiff-fast> requires the I<diff> command to be present.
+
+=head1 AUTHOR
+
+Version 1.3.4
+Copyright (C) 2004-2024 Frederik Tilmann
+
+This program is free software; you can redistribute it and/or modify
+it under the terms of the GNU General Public License Version 3
+
+Contributors of fixes and additions: V. Kuhlmann, J. Paisley, N. Becker, T. Doerges, K. Huebner,
+T. Connors, Sebastian Gouezel and many others.
+Thanks to the many people who sent in bug reports, feature suggestions, and other feedback.
+
+=cut
+
+__END__
+%%BEGIN SAFE COMMANDS
+% Regex matching commands which can safely be in the
+% argument of a \DIFadd or \DIFdel command (leave out the \)
+arabic
+dashbox
+emph
+fbox
+framebox
+hspace\*?
+math.*
+makebox
+mbox
+pageref
+ref
+eqref
+symbol
+raisebox
+rule
+text.*
+shortstack
+usebox
+dag
+ddag
+copyright
+pounds
+S
+P
+oe
+OE
+ae
+AE
+aa
+AA
+o
+O
+l
+L
+frac
+ss
+sqrt
+ldots
+cdots
+vdots
+ddots
+alpha
+beta
+gamma
+delta
+epsilon
+varepsilon
+zeta
+eta
+theta
+vartheta
+iota
+kappa
+lambda
+mu
+nu
+xi
+pi
+varpi
+rho
+varrho
+sigma
+varsigma
+tau
+upsilon
+phi
+varphi
+chi
+psi
+omega
+Gamma
+Delta
+Theta
+Lambda
+Xi
+Pi
+Sigma
+Upsilon
+Phi
+Psi
+Omega
+ps
+mp
+times
+div
+ast
+star
+circ
+bullet
+cdot
+cap
+cup
+uplus
+sqcap
+vee
+wedge
+setminus
+wr
+diamond
+(?:big)?triangle.*
+lhd
+rhd
+unlhd
+unrhd
+oplus
+ominus
+otimes
+oslash
+odot
+bigcirc
+d?dagger
+amalg
+leq
+prec
+preceq
+ll
+(?:sq)?su[bp]set(?:eq)?
+in
+vdash
+geq
+succ(?:eq)?
+gg
+ni
+dashv
+equiv
+sim(?:eq)?
+asymp
+approx
+cong
+neq
+doteq
+propto
+models
+perp
+mid
+parallel
+bowtie
+Join
+smile
+frown
+.*arrow
+(?:long)?mapsto
+.*harpoon.*
+leadsto
+aleph
+hbar
+imath
+jmath
+ell
+wp
+Re
+Im
+mho
+prime
+emptyset
+nabla
+surd
+top
+bot
+angle
+forall
+exists
+neg
+flat
+natural
+sharp
+backslash
+partial
+infty
+Box
+Diamond
+triangle
+clubsuit
+diamondsuit
+heartsuit
+spadesuit
+sum
+prod
+coprod
+int
+oint
+big(?:sq)?c[au]p
+bigvee
+bigwedge
+bigodot
+bigotimes
+bigoplus
+biguplus
+(?:arc)?(?:cos|sin|tan|cot)h?
+csc
+arg
+deg
+det
+dim
+exp
+gcd
+hom
+inf
+ker
+lg
+lim
+liminf
+limsup
+ln
+log
+max
+min
+Pr
+sec
+sup
+quad
+qquad
+bibfield
+bibinfo
+[Hclbkdruvt]
+[`'^"~=.]
+_
+AMPERSAND
+(SUPER|SUB)SCRIPTNB
+(SUPER|SUB)SCRIPT
+SQRT
+SQRTNB
+PERCENTAGE
+DOLLAR
+%%END SAFE COMMANDS
+
+%%BEGIN TEXT COMMANDS
+% Regex matching commands with a text argument (leave out the \)
+addcontents.*
+cc
+closing
+chapter
+dashbox
+emph
+encl
+fbox
+framebox
+footnote
+footnotetext
+framebox
+href
+intertext
+part
+(sub){0,2}section\*?
+(sub)?paragraph\*?
+makebox
+mbox
+opening
+parbox
+raisebox
+savebox
+sbox
+shortintertext
+shortstack
+sidenote
+signature
+text.*
+value
+underline
+sqrt
+(SUPER|SUB)SCRIPT
+%%END TEXT COMMANDS
+
+%%BEGIN CONTEXT1 COMMANDS
+% Regex matching commands with a text argument (leave out the \), which will fail out of context. These commands behave like text commands, except when they occur in a deleted section, where they are disabled, but their argument is shown as deleted text.
+caption
+subcaption
+multicolumn
+%%END CONTEXT1 COMMANDS
+
+%%BEGIN CONTEXT2 COMMANDS
+% Regex matching commands with a text argument (leave out the \), which will fail out of context.  As corresponding commands for context1.  The only difference is that context2 commands are completely disabled in deleted sections, including their arguments.
+title
+author
+date
+institute
+%%END CONTEXT2 COMMANDS
+
+%% CONFIGURATION variabe defaults
+%%BEGIN LISTENV CONFIG
+itemize
+description
+enumerate
+%%END LISTENV CONFIG
+
+%%BEGIN FLOATENV CONFIG
+figure[\w\d*@]*
+table[\w\d*@]*
+plate[\w\d*@]*
+%%END FLOATENV CONFIG
+
+%%BEGIN PICTUREENV CONFIG
+picture[\w\d*@]*
+tikzpicture[\w\d*@]*
+DIFnomarkup
+%%END PICTUREENV CONFIG
+
+%%BEGIN MATHENV CONFIG
+equation[*]?
+displaymath
+math
+DOLLARDOLLAR
+%%END MATHENV CONFIG
+
+%%BEGIN MATHARRENV CONFIG
+eqnarray[*]?
+align[*]?
+alignat[*]?
+gather[*]?
+multline[*]?
+flalign[*]?
+%%END MATHARRENV CONFIG
+
+%%BEGIN ARRENV CONFIG
+aligned
+gathered
+multlined
+array
+[pbvBV]?matrix
+smallmatrix
+cases
+%%END ARRENV CONFIG
+# split
+
+
+%%BEGIN COUNTERCMD CONFIG
+footnote
+part
+chapter
+section
+subsection
+subsubsection
+paragraph
+subparagraph
+%%END COUNTERCMD CONFIG
+
+%%BEGIN VERBATIMENV CONFIG
+comment
+%%END VERBATIMENV CONFIG
+
+%%BEGIN VERBATIMLINEENV CONFIG
+lstlisting
+verbatim[*]?
+%%END VERBATIMLINEENV CONFIG
+
+%%BEGIN CUSTOMDIFCMD CONFIG
+%%END CUSTOMDIFCMD CONFIG
+
+%%% TYPES (Commands for highlighting changed blocks)
+
+%DIF UNDERLINE PREAMBLE
+\RequirePackage[normalem]{ulem}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{{\protect\color{blue}\uwave{#1}}}
+\providecommand{\DIFdel}[1]{{\protect\color{red}\sout{#1}}}
+%DIF END UNDERLINE PREAMBLE
+
+%DIF LUAUNDERLINE PREAMBLE
+\RequirePackage{lua-ul}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{{\protect\color{blue}\underLine{#1}}}
+\providecommand{\DIFdel}[1]{{\protect\color{red}\strikeThrough{#1}}}
+%DIF END LUAUNDERLINE PREAMBLE
+
+%DIF CTRADITIONAL PREAMBLE
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\RequirePackage[stable]{footmisc}
+\DeclareOldFontCommand{\sf}{\normalfont\sffamily}{\mathsf}
+\providecommand{\DIFadd}[1]{{\protect\color{blue} \sf #1}}
+\providecommand{\DIFdel}[1]{{\protect\color{red} [..\footnote{removed: #1} ]}}
+%DIF END CTRADITIONAL PREAMBLE
+
+%DIF TRADITIONAL PREAMBLE
+\RequirePackage[stable]{footmisc}
+\DeclareOldFontCommand{\sf}{\normalfont\sffamily}{\mathsf}
+\providecommand{\DIFadd}[1]{{\sf #1}}
+\providecommand{\DIFdel}[1]{{[..\footnote{removed: #1} ]}}
+%DIF END TRADITIONAL PREAMBLE
+
+%DIF CFONT PREAMBLE
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\DeclareOldFontCommand{\sf}{\normalfont\sffamily}{\mathsf}
+\providecommand{\DIFadd}[1]{{\protect\color{blue} \sf #1}}
+\providecommand{\DIFdel}[1]{{\protect\color{red} \scriptsize #1}}
+%DIF END CFONT PREAMBLE
+
+%DIF FONTSTRIKE PREAMBLE
+\RequirePackage[normalem]{ulem}
+\DeclareOldFontCommand{\sf}{\normalfont\sffamily}{\mathsf}
+\providecommand{\DIFadd}[1]{{\sf #1}}
+\providecommand{\DIFdel}[1]{{\footnotesize \sout{#1}}}
+%DIF END FONTSTRIKE PREAMBLE
+
+%DIF CCHANGEBAR PREAMBLE
+\RequirePackage[pdftex]{changebar}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{\protect\cbstart{\protect\color{blue}#1}\protect\cbend}
+\providecommand{\DIFdel}[1]{\protect\cbdelete{\protect\color{red}#1}\protect\cbdelete}
+%DIF END CCHANGEBAR PREAMBLE
+
+%DIF CFONTCHBAR PREAMBLE
+\RequirePackage[pdftex]{changebar}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{\protect\cbstart{\protect\color{blue}\sf #1}\protect\cbend}
+\providecommand{\DIFdel}[1]{\protect\cbdelete{\protect\color{red}\scriptsize #1}\protect\cbdelete}
+%DIF END CFONTCHBAR PREAMBLE
+
+%DIF CULINECHBAR PREAMBLE
+\RequirePackage[normalem]{ulem}
+\RequirePackage[pdftex]{changebar}
+\RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1}
+\providecommand{\DIFadd}[1]{\protect\cbstart{\protect\color{blue}\uwave{#1}}\protect\cbend}
+\providecommand{\DIFdel}[1]{\protect\cbdelete{\protect\color{red}\sout{#1}}\protect\cbdelete}
+%DIF END CULINECHBAR PREAMBLE
+
+%DIF CHANGEBAR PREAMBLE
+\RequirePackage[pdftex]{changebar}
+\providecommand{\DIFadd}[1]{\protect\cbstart{#1}\protect\cbend}
+\providecommand{\DIFdel}[1]{\protect\cbdelete}
+%DIF END CHANGEBAR PREAMBLE
+
+%DIF INVISIBLE PREAMBLE
+\providecommand{\DIFadd}[1]{#1}
+\providecommand{\DIFdel}[1]{}
+%DIF END INVISIBLE PREAMBLE
+
+%DIF BOLD PREAMBLE
+\DeclareOldFontCommand{\bf}{\normalfont\bfseries}{\mathbf}
+\providecommand{\DIFadd}[1]{{\bf #1}}
+\providecommand{\DIFdel}[1]{}
+%DIF END BOLD PREAMBLE
+
+%DIF PDFCOMMENT PREAMBLE
+\RequirePackage{pdfcomment} %DIF PREAMBLE
+\providecommand{\DIFadd}[1]{\pdfmarkupcomment[author=ADD:,markup=Underline]{#1}{}}
+\providecommand{\DIFdel}[1]{\pdfcomment[icon=Insert,author=DEL:,hspace=12pt]{#1}}
+%DIF END PDFCOMMENT PREAMBLE
+
+%% SUBTYPES (Markers for beginning and end of changed blocks)
+
+%DIF SAFE PREAMBLE
+\providecommand{\DIFaddbegin}{}
+\providecommand{\DIFaddend}{}
+\providecommand{\DIFdelbegin}{}
+\providecommand{\DIFdelend}{}
+\providecommand{\DIFmodbegin}{}
+\providecommand{\DIFmodend}{}
+%DIF END SAFE PREAMBLE
+
+%DIF MARGIN PREAMBLE
+\providecommand{\DIFaddbegin}{\protect\marginpar{a[}}
+\providecommand{\DIFaddend}{\protect\marginpar{]}}
+\providecommand{\DIFdelbegin}{\protect\marginpar{d[}}
+\providecommand{\DIFdelend}{\protect\marginpar{]}}
+\providecommand{\DIFmodbegin}{\protect\marginpar{m[}}
+\providecommand{\DIFmodend}{\protect\marginpar{]}}
+%DIF END MARGIN PREAMBLE
+
+%DIF DVIPSCOL PREAMBLE
+%Note: only works with dvips converter
+\RequirePackage{color}
+\RequirePackage{dvipscol}
+\providecommand{\DIFaddbegin}{\protect\nogroupcolor{blue}}
+\providecommand{\DIFaddend}{\protect\nogroupcolor{black}}
+\providecommand{\DIFdelbegin}{\protect\nogroupcolor{red}}
+\providecommand{\DIFdelend}{\protect\nogroupcolor{black}}
+\providecommand{\DIFmodbegin}{}
+\providecommand{\DIFmodend}{}
+%DIF END DVIPSCOL PREAMBLE
+
+%DIF COLOR PREAMBLE
+\RequirePackage{color}
+\providecommand{\DIFaddbegin}{\protect\color{blue}}
+\providecommand{\DIFaddend}{\protect\color{black}}
+\providecommand{\DIFdelbegin}{\protect\color{red}}
+\providecommand{\DIFdelend}{\protect\color{black}}
+\providecommand{\DIFmodbegin}{}
+\providecommand{\DIFmodend}{}
+%DIF END COLOR PREAMBLE
+
+%DIF LABEL PREAMBLE
+% To show only pages with changes (pdf) (external program pdftk needs to be installed)
+% (only works for simple documents with non-repeated page numbers, otherwise use ZLABEL)
+% pdflatex diff.tex
+% pdflatex diff.tex
+%pdftk diff.pdf cat \
+%`perl -lne '\
+% if (m/\\newlabel{DIFchg[b](\d*)}{{.*}{(.*)}}/) { $start{$1}=$2; print $2}\
+% if (m/\\newlabel{DIFchg[e](\d*)}{{.*}{(.*)}}/) { \
+%      if (defined($start{$1})) { \
+%         for ($j=$start{$1}; $j<=$2; $j++) {print "$j";}\
+%      } else { \
+%         print "$2"\
+%      }\
+% }' diff.aux \
+% | uniq \
+% | tr  \\n ' '` \
+% output diff-changedpages.pdf
+% To show only pages with changes (dvips/dvipdf)
+% dvips -pp `\
+% [ put here the perl script from above]
+% | uniq | tr -s \\n ','`
+\typeout{Check comments in preamble of output for instructions how to show only pages where changes have been made}
+\newcount\DIFcounterb
+\global\DIFcounterb 0\relax
+\newcount\DIFcountere
+\global\DIFcountere 0\relax
+\providecommand{\DIFaddbegin}{\global\advance\DIFcounterb 1\relax\label{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFaddend}{\global\advance\DIFcountere 1\relax\label{DIFchge\the\DIFcountere}}
+\providecommand{\DIFdelbegin}{\global\advance\DIFcounterb 1\relax\label{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFdelend}{\global\advance\DIFcountere 1\relax\label{DIFchge\the\DIFcountere}}
+\providecommand{\DIFmodbegin}{\global\advance\DIFcounterb 1\relax\label{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFmodend}{\global\advance\DIFcountere 1\relax\label{DIFchge\the\DIFcountere}}
+%DIF END LABEL PREAMBLE
+
+%DIF ZLABEL PREAMBLE
+% To show only pages with changes (pdf) (external program pdftk needs to be installed)
+% (uses zref for reference to absolute page numbers)
+% pdflatex diff.tex
+% pdflatex diff.tex
+%pdftk diff.pdf cat \
+%`perl -lne 'if (m/\\zref\@newlabel{DIFchgb(\d*)}{.*\\abspage{(\d*)}}/ ) { $start{$1}=$2; print $2 } \
+%  if (m/\\zref\@newlabel{DIFchge(\d*)}{.*\\abspage{(\d*)}}/) { \
+%      if (defined($start{$1})) { \
+%         for ($j=$start{$1}; $j<=$2; $j++) {print "$j";}\
+%      } else { \
+%         print "$2"\
+%      }\
+% }' diff.aux \
+% | uniq \
+% | tr  \\n ' '` \
+% output diff-changedpages.pdf
+% To show only pages with changes (dvips/dvipdf)
+% latex diff.tex
+% latex diff.tex
+% dvips -pp `perl -lne 'if (m/\\newlabel{DIFchg[be]\d*}{{.*}{(.*)}}/) { print $1 }' diff.aux | uniq | tr -s \\n ','` diff.dvi
+\typeout{Check comments in preamble of output for instructions how to show only pages where changes have been made}
+\usepackage[user,abspage]{zref}
+\newcount\DIFcounterb
+\global\DIFcounterb 0\relax
+\newcount\DIFcountere
+\global\DIFcountere 0\relax
+\providecommand{\DIFaddbegin}{\global\advance\DIFcounterb 1\relax\zlabel{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFaddend}{\global\advance\DIFcountere 1\relax\zlabel{DIFchge\the\DIFcountere}}
+\providecommand{\DIFdelbegin}{\global\advance\DIFcounterb 1\relax\zlabel{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFdelend}{\global\advance\DIFcountere 1\relax\zlabel{DIFchge\the\DIFcountere}}
+\providecommand{\DIFmodbegin}{\global\advance\DIFcounterb 1\relax\zlabel{DIFchgb\the\DIFcounterb}}
+\providecommand{\DIFmodend}{\global\advance\DIFcountere 1\relax\zlabel{DIFchge\the\DIFcountere}}
+%DIF END ZLABEL PREAMBLE
+
+%DIF ONLYCHANGEDPAGE PREAMBLE
+\RequirePackage{atbegshi}
+\RequirePackage{etoolbox}
+\RequirePackage{zref}
+% redefine label command to write immediately to aux file - page references will be lost
+\makeatletter \let\oldlabel\label% Store \label
+\renewcommand{\label}[1]{% Update \label to write to the .aux immediately
+\zref@wrapper@immediate{\oldlabel{#1}}}
+\makeatother
+\newbool{DIFkeeppage}
+\newbool{DIFchange}
+\boolfalse{DIFkeeppage}
+\boolfalse{DIFchange}
+\AtBeginShipout{%
+  \ifbool{DIFkeeppage}
+        {\global\boolfalse{DIFkeeppage}}  % True DIFkeeppage
+         {\ifbool{DIFchange}{\global\boolfalse{DIFkeeppage}}{\global\boolfalse{DIFkeeppage}\AtBeginShipoutDiscard}} % False DIFkeeppage
+}
+\providecommand{\DIFaddbegin}{\global\booltrue{DIFkeeppage}\global\booltrue{DIFchange}}
+\providecommand{\DIFaddend}{\global\booltrue{DIFkeeppage}\global\boolfalse{DIFchange}}
+\providecommand{\DIFdelbegin}{\global\booltrue{DIFkeeppage}\global\booltrue{DIFchange}}
+\providecommand{\DIFdelend}{\global\booltrue{DIFkeeppage}\global\boolfalse{DIFchange}}
+\providecommand{\DIFmodbegin}{\global\booltrue{DIFkeeppage}\global\booltrue{DIFchange}}
+\providecommand{\DIFmodend}{\global\booltrue{DIFkeeppage}\global\boolfalse{DIFchange}}
+%DIF END ONLYCHANGEDPAGE PREAMBLE
+
+%% FLOAT TYPES
+
+%DIF FLOATSAFE PREAMBLE
+\providecommand{\DIFaddFL}[1]{\DIFadd{#1}}
+\providecommand{\DIFdelFL}[1]{\DIFdel{#1}}
+\providecommand{\DIFaddbeginFL}{}
+\providecommand{\DIFaddendFL}{}
+\providecommand{\DIFdelbeginFL}{}
+\providecommand{\DIFdelendFL}{}
+%DIF END FLOATSAFE PREAMBLE
+
+%DIF IDENTICAL PREAMBLE
+\providecommand{\DIFaddFL}[1]{\DIFadd{#1}}
+\providecommand{\DIFdelFL}[1]{\DIFdel{#1}}
+\providecommand{\DIFaddbeginFL}{\DIFaddbegin}
+\providecommand{\DIFaddendFL}{\DIFaddend}
+\providecommand{\DIFdelbeginFL}{\DIFdelbegin}
+\providecommand{\DIFdelendFL}{\DIFdelend}
+%DIF END IDENTICAL PREAMBLE
+
+%DIF TRADITIONALSAFE PREAMBLE
+% procidecommand color to make this work for TRADITIONAL and CTRADITIONAL
+\providecommand{\color}[1]{}
+\providecommand{\DIFaddFL}[1]{\DIFadd{#1}}
+\providecommand{\DIFdel}[1]{{\protect\color{red}[..{\scriptsize {removed: #1}} ]}}
+\providecommand{\DIFaddbeginFL}{}
+\providecommand{\DIFaddendFL}{}
+\providecommand{\DIFdelbeginFL}{}
+\providecommand{\DIFdelendFL}{}
+%DIF END TRADITIONALSAFE PREAMBLE
+
+% see:
+%  http://tex.stackexchange.com/questions/47351/can-i-redefine-a-command-to-contain-itself
+
+%DIF HIGHLIGHTGRAPHICS PREAMBLE
+\RequirePackage{settobox}
+\RequirePackage{letltxmacro}
+\newsavebox{\DIFdelgraphicsbox}
+\newlength{\DIFdelgraphicswidth}
+\newlength{\DIFdelgraphicsheight}
+% store original definition of \includegraphics
+\LetLtxMacro{\DIFOincludegraphics}{\includegraphics}
+\newcommand{\DIFaddincludegraphics}[2][]{{\color{blue}\fbox{\DIFOincludegraphics[#1]{#2}}}}
+\newcommand{\DIFdelincludegraphics}[2][]{%
+\sbox{\DIFdelgraphicsbox}{\DIFOincludegraphics[#1]{#2}}%
+\settoboxwidth{\DIFdelgraphicswidth}{\DIFdelgraphicsbox}
+\settoboxtotalheight{\DIFdelgraphicsheight}{\DIFdelgraphicsbox}
+\scalebox{\DIFscaledelfig}{%
+\parbox[b]{\DIFdelgraphicswidth}{\usebox{\DIFdelgraphicsbox}\\[-\baselineskip] \rule{\DIFdelgraphicswidth}{0em}}\llap{\resizebox{\DIFdelgraphicswidth}{\DIFdelgraphicsheight}{%
+\setlength{\unitlength}{\DIFdelgraphicswidth}%
+\begin{picture}(1,1)%
+\thicklines\linethickness{2pt}
+{\color[rgb]{1,0,0}\put(0,0){\framebox(1,1){}}}%
+{\color[rgb]{1,0,0}\put(0,0){\line( 1,1){1}}}%
+{\color[rgb]{1,0,0}\put(0,1){\line(1,-1){1}}}%
+\end{picture}%
+}\hspace*{3pt}}}
+}
+\LetLtxMacro{\DIFOaddbegin}{\DIFaddbegin}
+\LetLtxMacro{\DIFOaddend}{\DIFaddend}
+\LetLtxMacro{\DIFOdelbegin}{\DIFdelbegin}
+\LetLtxMacro{\DIFOdelend}{\DIFdelend}
+\DeclareRobustCommand{\DIFaddbegin}{\DIFOaddbegin \let\includegraphics\DIFaddincludegraphics}
+\DeclareRobustCommand{\DIFaddend}{\DIFOaddend \let\includegraphics\DIFOincludegraphics}
+\DeclareRobustCommand{\DIFdelbegin}{\DIFOdelbegin \let\includegraphics\DIFdelincludegraphics}
+\DeclareRobustCommand{\DIFdelend}{\DIFOaddend \let\includegraphics\DIFOincludegraphics}
+\LetLtxMacro{\DIFOaddbeginFL}{\DIFaddbeginFL}
+\LetLtxMacro{\DIFOaddendFL}{\DIFaddendFL}
+\LetLtxMacro{\DIFOdelbeginFL}{\DIFdelbeginFL}
+\LetLtxMacro{\DIFOdelendFL}{\DIFdelendFL}
+\DeclareRobustCommand{\DIFaddbeginFL}{\DIFOaddbeginFL \let\includegraphics\DIFaddincludegraphics}
+\DeclareRobustCommand{\DIFaddendFL}{\DIFOaddendFL \let\includegraphics\DIFOincludegraphics}
+\DeclareRobustCommand{\DIFdelbeginFL}{\DIFOdelbeginFL \let\includegraphics\DIFdelincludegraphics}
+\DeclareRobustCommand{\DIFdelendFL}{\DIFOaddendFL \let\includegraphics\DIFOincludegraphics}
+%DIF END HIGHLIGHTGRAPHICS PREAMBLE
+
+%% SPECIAL PACKAGE PREAMBLE COMMANDS
+
+%% Redefine strike out command to wrap in text box if amsmath is used and markup style with ulem is used
+%DIF AMSMATHULEM PREAMBLE
+\makeatletter
+\let\sout@orig\sout
+\renewcommand{\sout}[1]{\ifmmode\text{\sout@orig{\ensuremath{#1}}}\else\sout@orig{#1}\fi}
+\makeatother
+%DIF END AMSMATHULEM PREAMBLE
+
+
+% Standard \DIFadd and \DIFdel are redefined as \DIFaddtex and \DIFdeltex
+% when hyperref package is included.
+%DIF HYPERREF PREAMBLE
+\providecommand{\DIFadd}[1]{\texorpdfstring{\DIFaddtex{#1}}{#1}}
+\providecommand{\DIFdel}[1]{\texorpdfstring{\DIFdeltex{#1}}{}}
+%DIF END HYPERREF PREAMBLE
+
+%DIF LISTINGS PREAMBLE
+\RequirePackage{listings}
+\lstdefinelanguage{DIFcode}{
+  % note that the definitions in the following two lines are overwritten dependent on the markup type selected %DIFCODE TEMPLATE
+  morecomment=[il]{\%DIF\ <\ },          %DIFCODE TEMPLATE
+  moredelim=[il][\bfseries]{\%DIF\ >\ }  %DIFCODE TEMPLATE
+}
+\lstdefinestyle{DIFverbatimstyle}{
+	language=DIFcode,
+	basicstyle=\ttfamily,
+	columns=fullflexible,
+	keepspaces=true
+}
+\lstnewenvironment{DIFverbatim}{\lstset{style=DIFverbatimstyle}}{}
+\lstnewenvironment{DIFverbatim*}{\lstset{style=DIFverbatimstyle,showspaces=true}}{}
+%DIF END LISTINGS PREAMBLE
+
+%DIF COLORLISTINGS PREAMBLE
+\RequirePackage{listings}
+\RequirePackage{color}
+\lstdefinelanguage{DIFcode}{
+  % note that the definitions in the following two lines are overwritten dependent on the markup type selected %DIFCODE TEMPLATE
+  morecomment=[il]{\%DIF\ <\ },          %DIFCODE TEMPLATE
+  moredelim=[il][\bfseries]{\%DIF\ >\ }  %DIFCODE TEMPLATE
+}
+\lstdefinestyle{DIFverbatimstyle}{
+	language=DIFcode,
+	basicstyle=\ttfamily,
+	columns=fullflexible,
+	keepspaces=true
+}
+\lstnewenvironment{DIFverbatim}{\lstset{style=DIFverbatimstyle}}{}
+\lstnewenvironment{DIFverbatim*}{\lstset{style=DIFverbatimstyle,showspaces=true}}{}
+%DIF END COLORLISTINGS PREAMBLE
+
+%DIF DIFCODE_UNDERLINE
+  moredelim=[il][\color{red}\sout]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\uwave]{\%DIF\ >\ }
+%DIF END DIFCODE_UNDERLINE
+
+%DIF DIFCODE_CTRADITIONAL
+  moredelim=[il][\color{red}\scriptsize]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_CTRADITIONAL
+
+%DIF DIFCODE_TRADITIONAL
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il][\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_TRADITIONAL
+
+%DIF DIFCODE_CFONT
+  moredelim=[il][\color{red}\scriptsize]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_CFONT
+
+%DIF DIFCODE_FONTSTRIKE
+  moredelim=[il][\scriptsize \sout]{\%DIF\ <\ },
+  moredelim=[il][\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_FONTSTRIKE
+
+%DIF DIFCODE_INVISIBLE
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il]{\%DIF\ >\ }
+%DIF END DIFCODE_INVISIBLE
+
+%DIF DIFCODE_CHANGEBAR
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il]{\%DIF\ >\ }
+%DIF END DIFCODE_CHANGEBAR
+
+%DIF DIFCODE_CCHANGEBAR
+  moredelim=[il][\color{red}]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}]{\%DIF\ >\ }
+%DIF END DIFCODE_CCHANGEBAR
+
+%DIF DIFCODE_CULINECHBAR
+  moredelim=[il][\color{red}\sout]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\uwave]{\%DIF\ >\ }
+%DIF END DIFCODE_CULINECHBAR
+
+%DIF DIFCODE_CFONTCHBAR
+  moredelim=[il][\color{red}\scriptsize]{\%DIF\ <\ },
+  moredelim=[il][\color{blue}\sffamily]{\%DIF\ >\ }
+%DIF END DIFCODE_CFONTCHBAR
+
+%DIF DIFCODE_BOLD
+  % unfortunately \bfseries cannot be combined with ttfamily without extra packages
+  % also morecomment=[il] is broken as of v1.5b of listings at least
+  % workaround: plot in white with tiny font
+  % morecomment=[il]{\%DIF\ <\ },
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il][\sffamily\bfseries]{\%DIF\ >\ }
+%DIF END DIFCODE_BOLD
+
+%DIF DIFCODE_PDFCOMMENT
+
+  moredelim=[il][\color{white}\tiny]{\%DIF\ <\ },
+  moredelim=[il][\sffamily\bfseries]{\%DIF\ >\ }
+%DIF END DIFCODE_PDFCOMMENT
+
diff --git a/scripts/latexrevise.pl b/scripts/latexrevise.pl
new file mode 100644
index 0000000..5332ed1
--- /dev/null
+++ b/scripts/latexrevise.pl
@@ -0,0 +1,584 @@
+#!/usr/bin/env perl 
+#   latexrevise - takes output file of latexdiff and removes either discarded
+#                 or appended passages, then deletes all other latexdiff markup
+#
+#   Copyright (C) 2004  F J Tilmann (tilmann@gfz-potsdam.de)
+#
+# Repository:         https://github.com/ftilmann/latexdiff
+# CTAN page:          http://www.ctan.org/pkg/latexdiff
+#
+#    This program is free software: you can redistribute it and/or modify
+#    it under the terms of the GNU General Public License as published by
+#    the Free Software Foundation, either version 3 of the License, or
+#    (at your option) any later version.
+#
+#    This program is distributed in the hope that it will be useful,
+#    but WITHOUT ANY WARRANTY; without even the implied warranty of
+#    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+#    GNU General Public License for more details.
+#    You should have received a copy of the GNU General Public License
+#    along with this program.  If not, see <http://www.gnu.org/licenses/>.
+#
+# Detailed usage information at the end of the file
+# Note: version number now keeping up with latexdiff
+# Version 1.3.2 
+#   - Functionality -c option: if comment to be deleted is the last in the line replace new line by space character. This is done to avoid errors or unintended line breaks from empty lines after deletion (which are implicit \par)
+#   - Introduce option -r / --replace which removes markup of the form \replaced{new text}{old text} in the changes package
+# Version 1.3.1  (Bug fix) Remove some deprecation warnings due to uncommented left parentheses 
+# Version 1.0.2  Option --version
+# Version 1.0.1   no changes to latexrevise
+# Version 0.3   Updated for compatibility with latexdiff 0.3 output (DIFAUXCMD removal)
+# Version 0.1   First public release
+
+use Getopt::Long ;
+use strict;
+use warnings;
+
+my $versionstring=<<EOF ;
+This is LATEXREVISE 1.3.2
+  (c) 2005-2021 F J Tilmann
+EOF
+
+# Markup strings (make sure that this are set to the same value as in 
+# latexdiff)
+# If at all possible, do not change these as parts of the program
+# depend on the actual name.
+# At the very least adapt subroutine postprocess to new tokens.
+
+# note: have to use three backslashes to mean one literal one (reason
+# unknown
+my $ADDMARKOPEN='\\\DIFaddbegin(?:FL)? ';   # Token to mark begin of appended text
+my $ADDMARKCLOSE='\\\DIFaddend(?:FL)?(?: |\z)';   # Token to mark end of appended text (|\z only deals with the case of \DIFaddend right before \end{document}
+my $ADDOPEN='\\\DIFadd(?:FL)?{';  # To mark begin of added text passage
+my $ADDCLOSE='}';        # To mark end of added text passage
+my $ADDCOMMENT='DIF > ';   # To mark added comment line
+my $DELMARKOPEN='\\\DIFdelbegin(?:FL)? ';   # Token to mark begin of deleted text
+my $DELMARKCLOSE='\\\DIFdelend(?:FL)?(?: |\z)';   # Token to mark end of deleted text
+my $DELOPEN='\\\DIFdel(?:FL)?{';  # To mark begin of deleted text passage
+my $DELCLOSE='}';        # To mark end of deleted text passage
+my $ALTEXT='FL';         # string which might be appended to above commands
+my $DELCMDOPEN='%DIFDELCMD < ';  # To mark begin of deleted commands (must begin with %, i.e., be a comment
+my $DELCMDCLOSE="%%%\n";    # To mark end of deleted commands (must end with a new line)
+my $AUXCMD='%DIFAUXCMD' ; #  follows auxiliary commands put in by latexdiff to make difference file legal
+my $DELCOMMENT='DIF < ';   # To mark deleted comment line
+
+my $PREAMBLEXTBEG='^%DIF PREAMBLE EXTENSION ADDED BY LATEXDIFF$';
+my $PREAMBLEXTEND='^%DIF END PREAMBLE EXTENSION ADDED BY LATEXDIFF$';
+
+
+#my $pat0 = '(?:[^{}]|\\\{|\\\})*';
+#my $pat1 = '(?:[^{}]|\\\{|\\\}|\{'.$pat0.'\})*';
+#my $pat2 = '(?:[^{}]|\\\{|\\\}|\{'.$pat1.'\})*';
+#my $pat3 = '(?:[^{}]|\\\{|\\\}|\{'.$pat2.'\})*';
+#my $pat4 = '(?:[^{}]|\\\{|\\\}|\{'.$pat3.'\})*';
+
+my $pat0 = '(?:[^{}])*';
+my $pat_n = $pat0;
+# if you get "undefined control sequence MATHBLOCKmath" error, increase the maximum value in this loop
+for (my $i_pat = 0; $i_pat < 20; ++$i_pat){
+  $pat_n = '(?:[^{}]|\{'.$pat_n.'\}|\\\\\{|\\\\\})*';
+  # Actually within the text body, quoted braces are replaced in pre-processing. The only place where 
+  # the last part of the pattern matters is when processing the arguments of context2cmds in the preamble
+  # and these contain a \{ or \} combination, probably rare.
+  # It should thus be fine to use the simpler version below.
+  ###  $pat_n = '(?:[^{}]|\{'.$pat_n.'\})*';
+}
+
+
+my $brat0 = '(?:[^\[\]]|\\\[|\\\])*'; 
+
+
+
+
+my ($input,$preamble,$body,$post);
+my (@matches);
+my ($cnt,$prematch,$postmatch);
+my ($help,$version);
+my ($verbose,$quiet,$accept,$decline,$simplify)=(0,0,0,0,0);
+my ($comment,$comenv,$markup,$markenv,$replace);
+
+# A word unlikely ever to be used in a real latex file
+my $someword='gobbledegooksygook';
+
+Getopt::Long::Configure('bundling');
+GetOptions('accept|a' => \$accept,
+	   'decline|d'=> \$decline,
+	   'simplify|s' => \$simplify,
+	   'comment|c=s' => \$comment,
+	   'comment-environment|e=s' => \$comenv,
+	   'markup|m=s' => \$markup,
+	   'markup-environment|n=s' => \$markenv,
+	   'replace|r=s' => \$replace,
+	   'no-warnings|q' => \$verbose,
+           'version' => \$version,
+	   'verbose|V' => \$verbose,
+	   'help|h|H' => \$help);
+
+if ( $help ) {
+  usage() ;
+}
+
+if ( $version ) {
+  print STDERR $versionstring ; 
+  exit 0;
+}
+
+
+if ( ($accept &&  $decline) || ($accept && $simplify) || ($decline && $simplify) ) {
+  die '-a,-d and -s options are mutually exclusive. Type latexrevise -h to get more help.';
+}
+
+
+
+print STDERR "ACCEPT mode\n" if $verbose && $accept;
+print STDERR "DECLINE mode\n" if $verbose && $decline;
+print STDERR "SIMPLIFY mode.  WARNING: The output will not normally be valid latex,\n" if $verbose && $simplify;
+
+# Slurp old and new files
+{ 
+  local $/ ; # locally set record operator to undefined, ie. enable whole-file mode
+  $input=<>;
+}
+
+# split into parts
+($preamble,$body,$post)=splitdoc($input,'\begin{document}','\end{document}');
+
+if (length $preamble && ( $accept || $decline ) ) {
+  #
+  # WORK ON PREAMBLE
+  #
+  # (compare subroutine linediff in latexdiff to make sure correct strings are used)
+  
+  # remove extra commands added to preamble by latexdiff 
+  $preamble =~ s/${PREAMBLEXTBEG}.*?${PREAMBLEXTEND}\n{0,1}//smg ;
+
+  if ( $accept ) {
+    # delete mark up in appended lines
+    $preamble =~ s/^(.*) %DIF > $/$1/mg ;
+  } elsif ( $decline ) {
+    # delete appended lines
+    #  $preamble =~ s/^(.*) %DIF > $//mg ;
+    $preamble =~ s/^(.*) %DIF > \n//mg ;
+    # delete markup in deleted lines
+    $preamble =~ s/^%DIF < //mg ;
+  }
+  # remove any remaining DIF markups
+  #$preamble =~ s/%DIF.*$//mg ;
+  $preamble =~ s/%DIF.*?\n//sg ;
+}
+#print $preamble ;
+
+#
+# WORK ON BODY
+#
+if ($accept) {
+  # remove ADDMARKOPEN, ADDMARKCLOSE tokens
+  @matches= $body =~ m/${ADDMARKOPEN}(.*?)${ADDMARKCLOSE}/sg;
+  checkpure(@matches);
+  $body =~ s/${ADDMARKOPEN}(.*?)${ADDMARKCLOSE}/$1/sg;
+  # remove text flanked by DELMARKOPEN, DELMARKCLOSE tokens
+  @matches= $body =~ m/${DELMARKOPEN}(.*?)${DELMARKCLOSE}/sg;
+  checkpure(@matches);
+  $body =~ s/${DELMARKOPEN}(.*?)${DELMARKCLOSE}//sg;
+  # remove markup of added comments
+  $body =~ s/%${ADDCOMMENT}(.*?)$/%$1/mg ;
+  # remove deleted comments (full line)
+  $body =~ s/^%${DELCOMMENT}.*?\n//mg ;
+  # remove deleted comments (part of line)
+  $body =~ s/%${DELCOMMENT}.*?$//mg ;
+} 
+elsif ( $decline) {
+  # remove DELMARKOPEN, DELMARKCLOSE tokens
+  @matches= $body =~ m/${DELMARKOPEN}(.*?)${DELMARKCLOSE}/sg;
+  checkpure(@matches);
+  $body =~ s/${DELMARKOPEN}(.*?)${DELMARKCLOSE}/$1/sg;
+  # remove text flanked by ADDMARKOPEN, ADDMARKCLOSE tokens
+  # as latexdiff algorithm keeps the formatting and white spaces
+  # of the new text, sometimes whitespace might be inserted or
+  # removed inappropriately.  We try to guess whether this has
+  # happened
+
+  # Mop up tokens. This must be done already now as otherwise 
+  # detection of white-space problems does not work
+  $cnt = $body =~ s/${DELOPEN}($pat_n)${DELCLOSE}/$1/sg;
+  # remove markup of deleted commands 
+  $cnt +=   $body =~ s/${DELCMDOPEN}(.*?)${DELCMDCLOSE}/$1/sg ;
+  $cnt +=   $body =~ s/${DELCMDOPEN}//g ;
+  # remove aux commands
+  $cnt +=   $body =~ s/^.*${AUXCMD}$/${someword}/mg; $body =~ s/${someword}\n//g;
+
+  while ( $body =~ m/${ADDMARKOPEN}(.*?)${ADDMARKCLOSE}/s ) {
+    $prematch=$`;
+    $postmatch=$';
+    checkpure($1);
+    if ( $prematch =~ /\w$/s && $postmatch =~ /^\w/ ) {
+      # apparently no white-space between word=>Insert white space
+      $body =~ s/${ADDMARKOPEN}(.*?)${ADDMARKCLOSE}/ /s ;
+    }
+    elsif ( $prematch =~ /\s$/s && $postmatch =~ /^[.,;:]/ ) {
+      # space immediately before one of ".,:;" => remove this space
+      $body =~ s/\s${ADDMARKOPEN}(.*?)${ADDMARKCLOSE}//s ;
+    }
+    else {
+      # do not insert or remove any extras
+      $body =~ s/${ADDMARKOPEN}(.*?)${ADDMARKCLOSE}//s;
+    }
+  }
+# Alternative without special cases treatment
+#  @matches= $body =~ m/${ADDMARKOPEN}(.*?)${ADDMARKCLOSE}/sg;
+#  checkpure(@matches);
+#  $body =~ s/${ADDMARKOPEN}(.*?)${ADDMARKCLOSE}//sg;
+  # remove markup of deleted comments
+  $body =~ s/%${DELCOMMENT}(.*?)$/%$1/mg ;
+  # remove added comments (full line)
+  $body =~ s/^%${ADDCOMMENT}.*?\n//mg ;
+  # remove added comments (part of line)
+  $body =~ s/%${ADDCOMMENT}.*?$//mg ;
+}
+
+# remove any remaining tokens
+if ( $accept || $decline || $simplify ) {
+  # first substitution command deals with special case of added paragraph
+  $cnt = $body =~ s/${ADDOPEN}($pat_n)\n${ADDCLOSE}\n/$1\n/sg;
+  $cnt += $body =~ s/${ADDOPEN}($pat_n)${ADDCLOSE}/$1/sg;
+  $cnt==0 || warn 'Remaining $ADDOPEN tokens in DECLINE mode\n' unless ( $quiet || $accept || $simplify );
+}
+if ($accept || $simplify ) {
+  # Note: in decline mode these commands have already been removed above
+  $cnt = $body =~ s/${DELOPEN}($pat_n)${DELCLOSE}/$1/sg;
+  #### remove markup of deleted commands 
+  $cnt +=   $body =~ s/${DELCMDOPEN}(.*?)${DELCMDCLOSE}/$1/sg ;
+  $cnt +=   $body =~ s/${DELCMDOPEN}//g ;
+  # remove aux commands
+  # $cnt +=   
+  $body =~ s/^.*${AUXCMD}$/${someword}/mg; $body =~ s/${someword}\n//g;
+
+  #### remove deleted comments 
+  ###$cnt += $body =~ s/${DIFDELCMD}.*?$//mg ;
+  $cnt==0 || warn 'Remaining $DELOPEN or $DIFDELCMD tokens in ACCEPT mode\n' unless ( $quiet || $simplify );
+}
+
+# Remove comment commands
+if (defined($comment)) {
+  print STDERR "Removing \\$comment\{..\} sequences (incl. argument)..." if $verbose;
+  # protect $comments in comments by making them look different
+  $body =~ s/(%.*)${comment}(.*)$/$1${someword}$2/mg ;
+  # carry out the substitution
+  $cnt = 0 + $body =~ s/\\${comment}(?:\[${brat0}\])?\{${pat_n}\}(?: *\n)?//sg ;
+  print STDERR "$cnt matches found and removed.\n" if $verbose;
+  # and undo the protection substitution
+  $body =~ s/(%.*)${someword}(.*)$/$1${comment}$2/mg ;
+}
+if (defined($comenv)) {
+  print STDERR "Removing $comenv environments ..." if $verbose;
+  $body =~ s/(%.*)${comenv}/$1${someword}/mg ;
+##  $cnt = 0 + $body =~ s/\\begin(?:\[${brat0}\])?\{\$comenv\}.*?\\end\{\$comenv\}//sg ;
+    $cnt = 0 + $body =~ s/\\begin(?:\[${brat0}\])?\{${comenv}\}.*?\\end\{${comenv}\}\s*?\n//sg ;
+  print STDERR "$cnt matches found and removed.\n" if $verbose;
+  $body =~ s/(%.*)${someword}/$1${comenv}/mg ;
+}
+
+if (defined($markup)) {
+  print STDERR "Removing \\$markup\{..\} commands (leaving argument)..." if $verbose;
+  # protect $markups in comments by making them look different
+  $body =~ s/(%.*)${markup}(.*)$/$1${someword}$2/mg ;
+  # carry out the substitution
+  $cnt = 0 + $body =~ s/\\${markup}(?:\[${brat0}\])?\{(${pat_n})\}/$1/sg ;
+  print STDERR "$cnt matches found and removed.\n" if $verbose;
+  # and undo the protection substitution
+  $body =~ s/(%.*)${someword}(.*)$/$1${markup}$2/mg ;
+}
+if (defined($markenv)) {
+  print STDERR "Removing $markenv environments ..." if $verbose;
+  $body =~ s/(%.*)${markenv}/$1${someword}/mg ;
+  $cnt = 0 + $body =~ s/\\begin(?:\[${brat0}\])?\{${markenv}\}\n?//sg;
+  $cnt += 0 + $body =~ s/\\end\{${markenv}\}\n?//sg;
+  print STDERR $cnt/2, " matches found and removed.\n" if $verbose;
+  $body =~ s/(%.*)${someword}/$1${markenv}/mg ;
+}
+
+if (defined($replace)) {
+  print STDERR "Removing \\$replace\{..\}\{..\} commands (leaving 1st and discarding 2nd argument))..." if $verbose;
+  # protect $markups in comments by making them look different
+  $body =~ s/(%.*)${replace}(.*)$/$1${someword}$2/mg ;
+  # carry out the substitution
+  $cnt = 0 + $body =~ s/\\${replace}(?:\[${brat0}\])?\{(${pat_n})\}\s?\{(${pat_n})\}/$1/sg ;
+  print STDERR "$cnt matches found and removed.\n" if $verbose;
+  # and undo the protection substitution
+  $body =~ s/(%.*)${someword}(.*)$/$1${replace}$2/mg ;
+}
+
+if ( length $preamble ) {
+  print "$preamble\\begin{document}${body}\\end{document}$post";
+} else {
+  print $body;
+}
+
+# checkpure(@matches)
+# checks whether any of the strings in matches contains
+# $ADDMARKOPEN, $ADDMARKCLOSE,$DELMARKOPEN, or $DELMARKCLOSE
+# If so, die reporting nesting problems, otherwise return to caller
+sub checkpure {
+  while (defined($_=shift)) {
+    if (  /$ADDMARKOPEN/ || /$ADDMARKCLOSE/
+	  || /$DELMARKOPEN/ || /$DELMARKCLOSE/ ) {
+      die <<EOF ;
+There is a problem with nesting of  $ADDMARKOPEN, $ADDMARKCLOSE,
+$DELMARKOPEN, and $DELMARKCLOSE tokens that prevents me from 
+processing the file.  Exiting ...
+EOF
+     }
+  }
+}
+
+# ($part1,$part2,$part3)=splitdoc($text,$word1,$word2)
+# splits $text into 3 parts at $word1 and $word2.
+# if neither $word1 nor $word2 exist, $part1 and $part3 are empty, $part2 is $text
+# If only $word1 or $word2 exist but not the other, output an error message.
+#
+# same subroutine as in latexdiff
+sub splitdoc {
+  my ($text,$word1,$word2)=@_;
+  my $l1 = length $word1 ;
+  my $l2 = length $word2 ;
+
+  my $i = index($text,$word1);
+  my $j = index($text,$word2);
+
+  my ($part1,$part2,$part3)=("","","");
+
+  if ( $i<0 && $j<0) {
+    # no $word1 or $word2
+    print STDERR "Old Document not a complete latex file. Assuming it is a tex file with no preamble.\n";
+    $part2 = $text;
+  } elsif ( $i>=0 && $j>$i ) {
+    $part1 = substr($text,0,$i) ; 
+    $part2 = substr($text,$i+$l1,$j-$i-$l1);
+    $part3 = substr($text,$j+$l2) unless $j+$l2 >= length $text;
+  } else {
+    die "$word1 or $word2 not in the correct order or not present as a pair."
+  }
+  return ($part1,$part2,$part3);
+}
+
+
+
+sub usage {
+  print STDERR <<"EOF"; 
+Usage: $0 [OPTIONS] [diff.tex] > revised.tex
+
+Read a file diff.tex (output of latexdiff), and remove its markup. 
+If no filename is given read from standard input. The command can be used
+in ACCEPT, DECLINE, or SIMPLIFY mode, and be used to remove user-defined
+latex commands from the input (see options -c, -e, -m, -n below). 
+In ACCEPT mode, all appended text fragments  (or preamble lines)
+are kept, and all discarded text fragments (or preamble lines) are
+deleted.  
+In DECLINE mode, all discarded text fragments are kept, and all appended 
+text fragments are deleted.  
+If you wish to keep some changes, edit the diff.tex file in
+advance, and manually remove those tokens  which would otherwise be
+deleted.  Note that latexrevise only pays attention to the \\DIFaddbegin,
+\\DIFaddend, \\DIFdelbegin, and \\DIFdelend tokens and corresponding FL
+varieties.  All \\DIFadd and \\DIFdel commands (but not their content) are 
+simply deleted.   The commands added by latexdiff to the preamble are also
+removed.
+In SIMPLIFY mode all latexdiff markup is removed from the body of the text (after
+\\begin{document}) except for \\DIFaddbegin, \\DIFaddend, \\DIFdelbegin, \\DIFdelend
+tokens and the corresponding FL varieties of those commands.  The result
+will not in general be valid latex-code but might be easier to read and edit in 
+preparation for a subsequent run in ACCEPT or DECLINE mode.  
+In SIMPLIFY mode the preamble is left unmodified.
+
+-a
+--accept          Run in ACCEPT mode (delete all blocks marked by \\DIFdelbegin
+                  and \\DIFdelend).
+
+-d
+--decline         Run in DECLINE mode (delete all blocks marked by \\DIFaddbegin
+                  and \\DIFaddend).
+
+-s
+--simplify        Run in SIMPLIFY mode (Keep all \\DIFaddbegin, \\DIFaddend, 
+                  \\DIFdelbegin, \\DIFdelend tokens, but remove all other latexdiff
+                  markup from body.  
+
+Note that the three mode options are mutually exclusive. If no mode option is given,
+latexrevise simply removes user annotations and markup according to the following four
+options; these functions can be very useful outside the latexdiff context, too.
+
+
+-c cmd
+--comment=cmd     Remove \\cmd{...}.  cmd is supposed to mark some explicit 
+                  annotations which should be removed from the file before 
+                  release.
+
+-e envir
+--comment-environment=envir 
+                  Remove explicit annotation environments from the text, i.e. remove
+                  \\begin{envir}
+                  ...
+                  \\end{envir}
+                  blocks.
+
+-m cmd
+--markup=cmd      Remove the markup command cmd but leave its argument, i.e.
+                  turn \\cmd{abc} into abc.  
+
+-n envir
+--markup-environment=envir  
+                  Similarly, remove \\begin{envir} and \\end{envir} commands,
+                  but leave content of the environment in the text.
+
+
+-r cmd
+--replace=cmd    For constructions \\cmd{..}{..}, remove the command, leave the 
+                 content of first argument, and delete second argument. 
+                  
+-q
+--no-warnings     Do not warn users about \\DIDadd{..} or \\DIFdel statements
+                  which should not be there anymore
+
+-V
+--verbose         Verbose output
+
+EOF
+  exit 0;
+}
+
+=head1 NAME
+
+latexrevise - selectively remove markup and text from latexdiff output
+
+=head1 SYNOPSIS
+
+B<latexrevise> [ B<OPTIONS> ] [ F<diff.tex> ] > F<revised.tex>
+
+=head1 DESCRIPTION
+
+I<latexrevise> reads a file C<diff.tex> (output of I<latexdiff>), and remove the markup commands. 
+If no filename is given the input is read from standard input. The command can be used
+in I<ACCEPT>, I<DECLINE>, or I<SIMPLIFY> mode, or can be used to remove user-defined
+latex commands from the input (see B<-c>, B<-e>, B<-m>, and B<-n> below). 
+In I<ACCEPT> mode, all appended text fragments  (or preamble lines)
+are kept, and all discarded text fragments (or preamble lines) are
+deleted.  
+In I<DECLINE> mode, all discarded text fragments are kept, and all appended 
+text fragments are deleted.  
+If you wish to keep some changes, edit the diff.tex file in
+advance, and manually remove those tokens  which would otherwise be
+deleted.  Note that I<latexrevise> only pays attention to the C<\DIFaddbegin>,
+C<\DIFaddend>, C<\DIFdelbegin>, and C<\DIFdelend> tokens and corresponding FL
+varieties.  All C<\DIFadd> and C<\DIFdel> commands (but not their contents) are 
+simply deleted.   The commands added by latexdiff to the preamble are also
+removed.
+In I<SIMPLIFY> mode, C<\DIFaddbegin, \DIFaddend, \DIFdelbegin, \DIFdelend>
+tokens and their corresponding C<FL> varieties are kept but all other markup (e.g. C<DIFadd> and <\DIFdel>) is removed.  The result
+will not in general be valid latex-code but it will be easier to read and edit in 
+preparation for a subsequent run in I<ACCEPT> or I<DECLINE> mode.  
+In I<SIMPLIFY> mode the preamble is left unmodified.
+
+=head1 OPTIONS
+
+=over 4
+
+=item B<-a> or B<--accept>
+
+Run in I<ACCEPT> mode (delete all blocks marked by C<\DIFdelbegin> and C<\DIFdelend>).
+
+=item B<-d> or B<--decline>
+
+Run in I<DECLINE> mode (delete all blocks marked by C<\DIFaddbegin>
+and C<\DIFaddend>).
+
+=item B<-s> or B<--simplify>
+
+Run in I<SIMPLIFY> mode (Keep all C<\DIFaddbegin>, C<\DIFaddend>, 
+C<\DIFdelbegin>, C<\DIFdelend> tokens, but remove all other latexdiff
+markup from body).  
+
+=back
+
+Note that the three mode options are mutually exclusive.  If no mode option is given,
+I<latexrevise> simply removes user annotations and markup according to the following four
+options. These functions can be very useful outside the latexdiff context, too
+
+=over 4
+
+=item B<-c cmd> or B<--comment=cmd>
+
+Remove C<\cmd{...}> sequences.  C<cmd> is supposed to mark some explicit 
+annotations which should be removed from the file before 
+release.
+
+=item B<-e envir> or B<--comment-environment=envir> 
+
+Remove explicit annotation environments from the text, i.e. remove
+
+            \begin{envir}
+            ...
+            \end{envir}
+
+blocks.
+
+=item B<-m cmd> or B<--markup=cmd>
+
+Remove the markup command C<\cmd> but leave its argument, i.e.
+turn C<\cmd{abc}> into C<abc>.  
+
+=item B<-n envir> or B<--markup-environment=envir>
+
+Similarly, remove C<\begin{envir}> and C<\end{envir}> commands but 
+leave content of the environment in the text.
+
+=item B<-r cmd> or B<--replace=cmd>
+
+For constructions C<\\cmd{..}{..}>, remove the command, leave the 
+content of first argument, and delete second argument, i.e. turn 
+C<\cmd{abc}{def}> into C<abc>.
+
+                  
+
+=item B<-V> or B<--verbose>
+
+Verbose output
+
+=item B<-q> or B<--no-warnings>
+
+Do not warn users about C<\DIDadd{..}> or C<\DIFdel{..}> statements
+which should have been removed already.
+
+=back
+
+=head1 BUGS
+
+The current version is a beta version which has not yet been
+extensively tested. It has not been actively maintained so might not process output of newer versions of latexdiff entirely correctly.
+Please submit bug reports using the issue tracker of the github repository page I<https://github.com/ftilmann/latexdiff.git>, 
+or send them to I<tilmann -- AT -- gfz-potsdam.de>.  Include the serial number of I<latexrevise>
+(Option --version).  If you come across latexdiff
+output which is not processed correctly by I<latexrevise> please include the
+problem file as well as the old and new files on which it is based,
+ideally edited to only contain the offending passage as long as that still
+reproduces the problem.
+
+Note that I<latexrevise> gets confused by commented C<\begin{document}> or 
+C<\end{document}> statements
+
+=head1 SEE ALSO
+
+L<latexdiff>
+
+=head1 PORTABILITY
+
+I<latexrevise> does not make use of external commands and thus should run
+on any platform  supporting PERL v5 or higher. 
+
+=head1 AUTHOR
+
+Copyright (C) 2004 Frederik Tilmann
+
+This program is free software; you can redistribute it and/or modify
+it under the terms of the GNU General Public License Version 3
+
+=cut
-- 
GitLab