
    h+                        S SK r SSKJr  SSKJr  SSKJrJr  SSKJ	r	  SS	K
JrJr  S
S/rS rS rS\S0\S0/0r\S0\S0\S0\S0\S0\S0\S0\S0\S0\S0\S0\S0\S0\S0\S0\S 0\S!0\S"0\S#04 H  r\/\\\   '   M     S$ H  r\\0/\\'   M     S% H=  r\\R)                  5       4 H$  rS& H  r\ S'\ 3r\\0\S(0\S)\-   0/\\'   M     M&     M?     S* H*  r\\R)                  5       4 H  r\\0\S+0/\\ S+3'   M     M,     S, H-  r\\R)                  5       4 H  r\\0\S-0\S+0/\\ S.3'   M     M/     S/ H  u  rr\\0\S)\-   0/\\ S)\ 3'   M     S0 H#  u  rrr\\S1-   0\\0\S)\-   0/\\ S1\ S)\ 3'   M%     / rS1rS)r\ H  r\1r\ V s/ s H  o \:w  d  M
  U PM     sn  H4  r\R?                  \ Vs/ s H  oRA                  \\5      PM     sn5        M6     S2 V s/ s H  o \:w  d  M
  U PM     sn  H4  r!\R?                  \ Vs/ s H  oRA                  \\!5      PM     sn5        M6     \R?                  \ Vs/ s H  n\" U5      PM     sn5        \RE                  \5        M     \ H  r\\0/\\'   M     / S3Qr#/ S4RI                  \\S59PS6RI                  \\S59PS7RI                  \\S59PS8RI                  \\S59PS9RI                  \\S59PS:RI                  \\S59PS;RI                  \\S59PS<RI                  \\S=9PS>RI                  \\S59PS?RI                  \\S59PS@RI                  \\S59PSARI                  \\S59PSBRI                  \\S59PSCRI                  \\S59PSDRI                  \\S59PSERI                  \\S59PSFRI                  \\S59PSGRI                  \\S59PSHRI                  \\S59PSIRI                  \\S59PSJRI                  \\S59PSKRI                  \\S59PSLRI                  \\S59PSMRI                  \\S59PSNRI                  \\S59PSORI                  \\S59PSPRI                  \\S59PSQRI                  \\S59PSRRI                  \\S59PSSRI                  \\S59PSTRI                  \\S59PSURI                  \\S59PSVRI                  \\S59PSWRI                  \\S=9PSXRI                  \\S59PSYRI                  \\S59PSZRI                  \\S59PS[RI                  \\S59PS\RI                  \\S59PS]RI                  \\S59PS^RI                  \\S59PS_RI                  \\S59PS`RI                  \\S59PSaRI                  \\S59PSbRI                  \\S59PScRI                  \\S59PSdRI                  \\S59PSeRI                  \\S59PSfRI                  \\S59PSgRI                  \\S=9Pr%\%\# Vs/ s H  nShRI                  U\\\Si9PM     sn-  r%/ SjQr&\%\& Vs/ s H  nSkRI                  U\\\Sl9PM     sn-  r%/ SmQr'\%\' Vs/ s H  nSnRI                  U\\\So9PM     sn-  r%\" \	\5      r(\ RR                  " SpSqRU                  Sr \% 5       5      -   5      RV                  r,gs  sn f s  snf s  sn f s  snf s  snf s  snf s  snf s  snf )s    N   )ORTH)
update_exc   )ALPHAALPHA_LOWER)BASE_EXCEPTIONS   )ELISIONHYPHENSzaujourd'huizAujourd'huic                     [        U 5      S:X  a  U $ [        U 5      S:X  a  U R                  5       $ U S   R                  5       U SS  -   $ Nr   r
   )lenuppertexts    \/home/james-whalen/.local/lib/python3.13/site-packages/spacy/lang/fr/tokenizer_exceptions.pyupper_first_letterr      D    
4yA~
4yA~zz|7==?T!"X%%    c                     [        U 5      S:X  a  U $ [        U 5      S:X  a  U R                  5       $ U S   R                  5       U SS  -   $ r   )r   lowerr   s    r   lower_first_letterr      r   r   zJ.-C.zJ.z-C.zav.zjanv.u   févr.zavr.zjuill.zsept.zoct.znov.u   déc.zapr.zDr.zM.zMr.zMme.zMlle.u   n°u   d°zSt.zSte.)u   après-midiu   au-delàz	au-dessuszcelle-ciz	celles-cizcelui-cizcf.z
ci-dessousu
   elle-mêmez
en-dessouszetc.u
   jusque-làu	   lui-mêmezMM.zNo.u
   peut-êtrezpp.zquelques-unszrendez-vouszVol.)aestsembleindiquemoquepasse)elleilonz-t-z-t-)r   z-ce)zqu'zn'r   zest-ce))r   r!   )ESTIL))sr   r!   )Sr$   r%   ')r#   u   ‐)u   a[ée]roabatza[fg]roafterzaigues?u   am[ée]ricanoangloantiu	   apr[èe]sarabozarcs?archiu
   arrières?audioavantavionautobanczbas(?:ses?)?zbateaux?zbec?zbelles?beaubestzbio?bienblancu	   bo[îi]tez	bonn?e?s?boiszbou(?:c|rg)u   b[êe]tacachezcap(?:ello)?cassecastelchampchapelleu   ch[âa]teau(?:neuf)?chassezcha(?:ud|t)e?s?chauffechouchromoz	claire?s?zco(?:de|ca)?comptecontrecordonzcoupe?z	courte?s?couvrecrashcrisecrochecrosscyberu   côtedemizdi(?:sney)?dixu   d[ée]s?dyszex?u   émiratoentrer   ethnoexextrau   extrêmeu   [ée]cofauxfilfortz	franco?s?gallozgardes?gastrozgrande?gratteu	   gr[ée]cogrosu   g[ée]ozhaute?s?zhomm?es?horshyperindoinfrainterintraislamoitalojeanlabiolatinolivelotlouiszm[ai]cromalu   médiomesnilz	mi(?:ni)?monozmont?s?moyenmultiu
   m[ée]canou
   m[ée]dicou   m[ée]dou   m[ée]tazmots?neuronoixnonnordnotreu   n[ée]oouestoutreouvrer   percepharmacozph[oy]tozpieds?piquez	poissons?poncepontzpo[rs]tpousseprimou   pro(?:cès|to)?parez	petite?s?plessisporteu   préu   prêchiu   protègepseudou   pêleu   péripuyquasiquatreradiorecourtrythmoz(?:re)?doubles?u   r[ée]u	   r[ée]trorequinzsans?z
sa?inte?s?semiserresinosocioz
sociale?s?soixantesouszsu[bdrs]supertailletirethermotierstourneztoute?s?u   tra[iî]ne?transtrentetroistroussez
tr(?:i|ou)u   t[ée]l[ée]u   utérovasozvi[cd]eu	   vid[ée]ozvie(?:ux|i?lles?|i?l)z vill(?:e|eneuve|ers|ette|iers|y)vingtz	voitures?zwagons?ultra   àu   [ée]lectrou   [ée]quiFontainezLa ChapelleMariez	Le MesnilNeuvillePierreValVauxz^a[{hyphen}]sexualis[{al}]+$)hyphenalu"   ^arginine[{hyphen}]méthyl[{al}]+$z^binge[{hyphen}]watch[{al}]+$z^black[{hyphen}]out[{al}]*$z^bouche[{hyphen}]por[{al}]+$z^burn[{hyphen}]out[{al}]*$z^by[{hyphen}]pass[{al}]*$z^ch[{elision}]tiis[{al}]+$)elisionr   z^chape[{hyphen}]chut[{al}]+$z^down[{hyphen}]load[{al}]*$u   ^[ée]tats[{hyphen}]uni[{al}]*$z,^droits?[{hyphen}]de[{hyphen}]l'homm[{al}]+$z^fac[{hyphen}]simil[{al}]*$z^fleur[{hyphen}]bleuis[{al}]+$z^flic[{hyphen}]flaqu[{al}]+$z^fox[{hyphen}]trott[{al}]+$z^google[{hyphen}]is[{al}]+$z^hard[{hyphen}]discount[{al}]*$z^hip[{hyphen}]hop[{al}]*$z^jet[{hyphen}]set[{al}]*$z^knock[{hyphen}]out[{al}]*$u   ^lèche[{hyphen}]bott[{al}]+$z"^litho[{hyphen}]typographi[{al}]+$z^lock[{hyphen}]out[{al}]*$z ^lombri[{hyphen}]compost[{al}]+$z^mac[{hyphen}]adamis[{al}]+$z^marque[{hyphen}]pag[{al}]+$z^mouton[{hyphen}]noiris[{al}]+$z^new[{hyphen}]york[{al}]*$z^pair[{hyphen}]programm[{al}]+$z^people[{hyphen}]is[{al}]+$z^plan[{hyphen}]socialis[{al}]+$z!^premier[{hyphen}]ministr[{al}]*$z^prud[{elision}]hom[{al}]+$u   ^réarc[{hyphen}]bout[{al}]+$z^refox[{hyphen}]trott[{al}]+$z^remicro[{hyphen}]ond[{al}]+$z^repique[{hyphen}]niqu[{al}]+$u!   ^repetit[{hyphen}]déjeun[{al}]+$z^rick[{hyphen}]roll[{al}]*$z^rond[{hyphen}]ponn[{al}]+$z^shift[{hyphen}]cliqu[{al}]+$z^soudo[{hyphen}]bras[{al}]+$z^stabilo[{hyphen}]boss[{al}]+$z^strip[{hyphen}]teas[{al}]+$z^terra[{hyphen}]form[{al}]*$z^teuf[{hyphen}]teuf[{al}]*$z^yo[{hyphen}]yo[{al}]+$z^zig[{hyphen}]zag[{al}]*$z^z[{elision}]yeut[{al}]+$z2^{prefix}[{hyphen}][{al}][{hyphen}{al}{elision}]*$)prefixr   r   r   )u	   r?é?entrz	grande?s?rz3^{prefix}[{elision}][{al}][{hyphen}{al}{elision}]*$)r   r   r   r   )u   l[èe]s?laenzdes?zd[eu]surr   zaux?r   etu   prèssaintzA^[{a}]+[{hyphen}]{hyphen_combo}[{hyphen}](?:l[{elision}])?[{a}]+$)hyphen_combor   r   r   z(?iu)|c              #   D   #    U  H  nS R                  U5      v   M     g7f)z(?:{})N)format).0ms     r   	<genexpr>r     s     @<axq))<s    )-resymbolsr   utilr   char_classesr   r   tokenizer_exceptionsr	   punctuationr   r   FR_BASE_EXCEPTIONSr   r   _excexc_dataorthverbtitlepronountokenprer&   _infixes_excorig_elisionorig_hypheninfixvariants_infixelision_charupdatereplacehyphen_charextend_hyphen_prefixr   _regular_exp_elision_prefix_hyphen_combinationTOKENIZER_EXCEPTIONScompilejoinmatchTOKEN_MATCH)xwordphcs   0000r   <module>r      s
   	   - 2 ) $]3 && 	4,u./ 
5M	7O	8	6N	8	7O	6N	6N	7O	6N	5M	4L	5M	6N	7O	5M	5M	5M	6N'H* %:D$+0D, ,DJ-2D tzz|$+GfCy)E $<$cGm7LMDK , % Dtzz|$#TlT5M:vS\ % 
 Cciik""&e}tUm LvV_ # 
 4MD'"&cGm/D EDD67)	 4 AAtW	q3w	t	sW}%DA3avQwi	 ! A   EWN$+AGqL/@GABPQ.$\\,5.Q	
 B $0D<a3C<D@NO\\+{3O	
 E O-d3OP'   D,DJ JX5"))[)I5(//w;/O5 $**'k*J5 "((K(H	5
 #))[)I5 !''w;'G5  &&g+&F5 !''K'H5 #))[)I5 "((K(H5 &,,G,L5 399; : 5 "((K(H5  %++7{+K!5" #))[)I#5$ "((K(H%5& "((K(H'5( &,,G,L)5*  &&g+&F+5,  &&g+&F-5. "((K(H/50 $**'k*J152 )//w;/O354 !''w;'G556 '--W-M758 #))[)I95: #))[)I;5< &,,G,L=5> !''w;'G?5@ &,,G,LA5B "((K(HC5D &,,G,LE5F (..g+.NG5H "(([(II5J $**'k*JK5L $**'k*JM5N $**'k*JO5P %++7{+KQ5R (..g+.NS5T "((K(HU5V "((K(HW5X $**'k*JY5Z #))[)I[5\ %++7{+K]5^ #))[)I_5` #))[)Ia5b "((K(Hc5d $$G$De5f  &&g+&Fg5h  &&w;&Gi5n    9??	 @    2  	  :@@''k A  	    "	 " HNNE O  "	  "/48 jjchh@<@@@% {	 BQDOOT,s<   	Y$$Y$?Y))	Y.6Y.Y3Y8
0Y=ZZ