-i`dZddlZddlZddlZddlmZddlmZddlZ ddl Z ddl Z ddl ZddlmZddlZddlZeZddlmZddlZddlmZddlmZejdejd d Z d Z!Gd d Z"d Z#dZ$e%dkr e$dSdS)z Arabic Tashkeel Class N)and_)count)unknown_tashkeel)tashkeel_constz../libz../FceZdZdZd,dZdZdZdZdZdZ d Z d Z d Z d Z d ZdZdZdZdZdZdZdZdZdZdZdZdZdZdZd-dZdZdZd Z d!Z!d.d"Z"d#Z#d$Z$d%Z% d/d'Z&d(Z'd)Z(d*Z) d0d+Z*d&S)1 TashkeelClassz# Arabic Tashkeel Class FcDtjtjtjt|_d|_d|_d|_d|_ d|_ d|_ d|_ d|_ d|_d|_t j||_|jt*j||_|jd|_t4j||_|jt<|j|jtAj!|j |_"tGj$|_%dS)N)leveliTF) cache_path)&logging basicConfigERROR getLogger__name__loggerlimitenabled_last_markenabled_stat_tashkeelenabled_verboseenabled_show_collocation_markselect_by_score_enabledenabled_syntaxic_analysisenabled_ajust_vocalizationenabled_semantic_analysisallow_syntax_last_markqalsadianalexAnalexanalyzerdisable_allow_cache_usesylajoneanasynSyntaxAnalyzeranasyntsyntax_train_enabledasmaianasemSemanticAnalyzer set_debugdebug set_limitcollCollocationClasscollorUnknownTashkeelunknown_vocalizer)self mycache_paths N/var/www/html/speakWrite/venv/lib/python3.11/site-packages/mishkal/tashkeel.py__init__zTashkeelClass.__init__1su'-0000'11  "& &*" %.3*(-$*.& +/'*.&'+# ---FF  --/// 55<5PP  ,,...%*!l333NN  &&&  +++*4+MNN "2!A!C!CcR||_|j|jdS)z; set the limit length of words to vocalize N)rr r,)r2rs r4r,zTashkeelClass.set_limits)   +++++r6cd|_dS)z1 Enable the stat tasheel option. TNrr2s r4enable_stat_tashkeelz"TashkeelClass.enable_stat_tashkeels&*"""r6cPd|j_|jdS)zQ Enable the syntaxique relation extraction from a vocalized text TN)r%r&r enable_fully_vocalized_inputr:s r4enable_syn_trainzTashkeelClass.enable_syn_trains*-1 ) 2244444r6cd|_dS)z2 disable the stat tasheel option. FNr9r:s r4disable_stat_tashkeelz#TashkeelClass.disable_stat_tashkeels&+"""r6cd|_dS)z enable logger TN)rr:s r4enable_verbosezTashkeelClass.enable_verboses $r6cj|j|jdS)z9 disable the morphological use of cache. N)r r!r%r:s r4 disable_cachezTashkeelClass.disable_caches2 --/// ,,.....r6cj|j|jdS)z8 enable the morphological use of cache. N)r enable_allow_cache_user%r:s r4 enable_cachezTashkeelClass.enable_caches2 ,,... ++-----r6c|jS)z return the the stat tasheel option value. @return: True if enabled, false else. @rtype: boolean. r9r:s r4get_enabled_stat_tashkeelz'TashkeelClass.get_enabled_stat_tashkeels ))r6cFd|_|jdS)z> Enable the show the collocation mark option. TN)rr/enable_show_delimiterr:s r4enable_show_collocation_markz*TashkeelClass.enable_show_collocation_marks&.2* ((*****r6cFd|_|jdS)z? disable the show the collocation mark option. FN)rr/disable_show_delimiterr:s r4disable_show_collocation_markz+TashkeelClass.disable_show_collocation_marks&.3* ))+++++r6c|jS)z return the the show the collocation mark option value. @return: True if enabled, false else. @rtype: boolean. )rr:s r4get_show_collocation_markz'TashkeelClass.get_show_collocation_marks 11r6cd|_dS)z. Enable the last mark option. TNrr:s r4enable_last_markzTashkeelClass.enable_last_marks"&r6cd|_dS)z= disable the last mark vocalization option. FNrSr:s r4disable_last_markzTashkeelClass.disable_last_marks"'r6c|jS)z return the the last mark vocalization option value. @return: True if enabled, false else. @rtype: boolean. rSr:s r4get_enabled_last_markz#TashkeelClass.get_enabled_last_marks %%r6cd|_dS)z6 Enable the syntaxic analysis option. TNrr:s r4enable_syntaxic_analysisz&TashkeelClass.enable_syntaxic_analysis*.&&&r6cd|_dS)z7 disable the syntaxic analysis option. FNrZr:s r4disable_syntaxic_analysisz'TashkeelClass.disable_syntaxic_analysis*/&&&r6c|jS)z return the the syntaxic analysis option value. @return: True if enabled, false else. @rtype: boolean. rZr:s r4get_enabled_syntaxic_analysisz+TashkeelClass.get_enabled_syntaxic_analysis --r6cd|_dS)z6 Enable the Semantic analysis option. TNrr:s r4enable_semantic_analysisz&TashkeelClass.enable_semantic_analysisr\r6cd|_dS)z7 disable the Semantic analysis option. FNrdr:s r4disable_semantic_analysisz'TashkeelClass.disable_semantic_analysisr_r6c|jS)z return the the Semantic analysis option value. @return: True if enabled, false else. @rtype: boolean. rdr:s r4get_enabled_semantic_analysisz+TashkeelClass.get_enabled_semantic_analysisrbr6cd|_dS)z7 Enable the Ajust Vocalization option. TNrr:s r4enable_ajust_vocalizationz'TashkeelClass.enable_ajust_vocalization s+/'''r6cd|_dS)z8 disable the Ajust Vocalization option. FNrkr:s r4disable_ajust_vocalizationz(TashkeelClass.disable_ajust_vocalizations+0'''r6c|jS)z return the the Ajust Vocalization option value. @return: True if enabled, false else. @rtype: boolean. rkr:s r4get_enabled_ajust_vocalizationz,TashkeelClass.get_enabled_ajust_vocalizations ..r6c g}|j|}|D]7}|jdt t |8|r||j|\}}|D]}|jdt t ||D]v}|jdt | |jdt | w| r|j |}|D]{}|D]v}|jdt | |jdt | w|||fS)z Do the lexical, syntaxic and semantic analysis of the text. @param text: input text. @type text: unicode. @return: syntaxic and lexical tags. rtype: list of list of stemmedSynWord class. zSyn len before syntax: %sz Syn len: %szSyn sem next before: %szSyn next before: %szSyn sem next: %sz Syn next: %s) r check_textrr+reprlenrar%analyze get_sem_nextget_nextrir()r2textresultsyndl synodelistsynds r4 full_stemmerzTashkeelClass.full_stemmers))$// L LE K  9$s5zz:J:J K K K K  - - / / L!%!5!5f!=!= FJ! S S !!-SZZ0@0@AAA!SSDK%%&?TEVEVEXEX@Y@YZZZK%%&;DO>O>Q>Q9R9RSSSK%%nT$--//5J5JKKKKLz!!r6rxc  ||}|r0||}|jd||g}d}d}g}g}g} t j} |D]} || \} } d}d}d}d}d}d}|jdttt| | D]l\}}}|dz dkr | |dz }nd}|dzt| kr | |dz}nd}|r|r| ||}| |||||mttt| | D]\}}}|dz dkr | |dz }nd}|dzt| kr | |dz}nd}|r|r| ||}|||||||}|r||}nd}|}||}|}|}|}|||}| |֐g}d}d}t%|D]\}}||}|s||}||}d ||||g}||d }|||d | ||jrA|jd ||t?| d t?| d | d dz|z}|dkr4||!||"} nd} tFj$j%&| d} ||'}!tPr|!dkr|}|)s,d||*||g}n+d||*||g}|||d| ||| |!d|}|+r|,|}|r|-|}|S|S)z Vocalize the text and give suggestion to improve tashkeel by user. @param text: input text. @type text: unicode. @return: vocalized text. rtype: dict of dict or text. zstat tashkeel %sNFzList of synodes **********rr:z0TaharZe Tags to display '%s' and tagmaker =='%s'[z]{}z
 ;)chosensemisuggestinflectlinkrule). pre_tashkeelrI stat_tashkeelrr+mysamtagcodertagCoderr}listiziprrt_TashkeelClass__reduce_cases_TashkeelClass__choose_tashkeelget_next_relationappendget_vocalizeds enumerate get_vocalizedget_wordget_semivocalizedjoinget_typeget_tags_to_displayget_tagssplitextendencoderinfostrrget_previous_relation get_orderr" syn_constDISPLAY_RELATIONgetget_ruleUNCERTAIN_TASHKEELrXdisplayrp_ajust_vocalized_resultajust_vocalized_suggestion)"r2 inputtext suggestionformat_displaytextsvocalized_textpreviousoutput_suggest_list _chosen_list suggests_list mytagmakerrxdetailled_syntaxr{ next_nodepre_nodeprevious_indexprevious_case_indexprevious_chosen_relation current_indexword_cases_listcurrent_synodecurrent_chosen_case_indexcurrent_chosen_relationcurrent_chosenr privous_orderivoc_word semivocalizedrtagslistrelationselection_rules" r4tashkeelzTashkeelClass.tashkeelCsF%%i00  ) ) + + =**955I K  0) < < <    ^,,.. M .M .D+/+<+  1$)))-/:HH#H 1$s+;'<'<<< *=? ;II $I!U&9U/?@STH###^h DHUWWUegqHrHrCsCs' .' .>  1$)))-/:HH $H 1$s+;'<'<<< *=1+< =II $I!U&9U/?@STH,0,B,B#^Xi)A--) 4.6.H.HIb.c.c++.3+,C("11J!K!.&?#)##N333)7799$$W----O' .P! !*>xLLH)!_5577N" )nr&9&9(--// e!%NDLLXf>vo{D99 r6crtt|dz D]}|dzt|krd||dzvr||dzdtjr||ddvrZ||ddkr d||d<||ddkr d||d<||ddkr d ||d<||dtjr,||dd d tjz||d<|S) u1 Ajust the resulted text after vocalization to correct some case like 'meeting of two queiscents = ألتقاء الساكنين' @param text: _suggest_list @type text: list of dict of unicode @return: _suggest_list. @rtype: list of dict of unicode rr)مَنْمِنْعَنْruمِنَruعَنِruمَنِNr)rangert startswithrALEFendswithrr)r2 _suggest_listrs r4rz(TashkeelClass.ajust_vocalized_suggestions]s=))!+,, _ _AsS''''Xqs9K-K-Kac"8,77 CC.L!#H-1XXX(+H5DD9DM!,X66*1-h7;FF9DM!,X66*1-h7;FF9DM!,X6"1%h/88EE_1>q1A(1KCRC1PSXS^1^M!$X.r6cT|}|dkrd|zS|dkr|S|S)z format the vocalized word to be displayed on web interface. @param word: input vocalized word. @type word: unicode. @return: html code. rtype: unicode. rz0%srx)lower)r2wordrs r4rzTashkeelClass.display6s@(--// V # #FM M v % %KKr6c||}d}|D]2}|D]-}|}d||g}.3|S)z Vocalize the text. @param text: input text. @type text: unicode. @return: vocalized text. rtype: unicode. rr)r}rr)r2rxrrword_analyze_listitemvocs r4assistanttashkeelzTashkeelClass.assistanttashkeelFsy ,,T22!1 B B * B B((**!%NC+@!A!A Br6cttjD].}t jd|zdtj|z|}/|j|}tj |}|j dd |t|t|kr.|j dd |tj|}|j dd |d |S)z Vocalize the text by evident cases and by detecting numbers clauses @param text: input text. @type text: unicode. @return: statisticlly vocalized text. rtype: unicode. z\s%s\sz %s zpre_tashkeel numbers %srznb %s+znmd %s)rrCorrectedTashkeelkeysrrr tokenizepyarabicnumberpre_tashkeel_numberrr+rrtnamedpretashkeel_named)r2rxabrwordlistprevocalized_lists r4rzTashkeelClass.pre_tashkeelZs98==??@@ 9 9C6)C-  ,S 1*23799DD=))$//$O??II 3TYY?P5Q5QRRR  ! !S]] 2 2 K  gtyy1B'C'C D D D$N<<=NOO (DII.?$@$@AAAyy*+++r6c|j|}|j|}|j|\}}d|S)z Vocalize the text by statistical method according to the collocation dictionary @param text: input text. @type text: unicode. @return: statisticlly vocalized text. rtype: unicode. r)r/lookup4long_collocationsr rlookupr)r2rxrnewlisttaglists r4rzTashkeelClass.stat_tashkeelus^z22488=))$//  :,,X66yy!!!r6Ncd}d}d} |} | rd} nd} |r|} nd} |} g}t| dkrjdrNd|jdd} nd} | st| fd} | sY| r| rC| | | |\} } |j d t| | sm|j d t| | | | | ||\} } |j d t| | sA|| \} } |j d t| | d}|}| sd } tr;|j d| dfd| D|| |S)a Choose a tashkeel for the current word, according to the previous one. A new algorithm @param caselist: list of steming result of the word. @type caselist: list of stemmedSynword @param: the choosen previous word stemming. @type previous_chosen_case:stemmedSynwordhg @return: the choosen stemming of the current word. @rtype:stemmedSynword. NFrrfec8|SNget_freqxcaselists r4z1TashkeelClass.__choose_tashkeel..shqk6J6J6L6Lr6)keyz2:%sz2-1:%sz3: %sz4: %sdz 100: %d %s, cDg|]}|Sr.0rr s r4 z3TashkeelClass.__choose_tashkeel..s*=l=l=l^_hqk>W>W>Y>Y=l=l=lr6)get_chosen_indexesrt is_unknown set_vocalizedr1rrsorted is_initial_TashkeelClass__choose_initialrr+rs_TashkeelClass__choose_cases_TashkeelClass__choose_defaultrset_ruler)r2r rprevious_chosen_caserrrr chosen_indexrr pre_relationnext_chosen_indexesindxlisttmplists ` r4__choose_tashkeelzTashkeelClass.__choose_tashkeels '  LLL  '"+">">"@"@  "& "4466 x==A  {%%''  ))$*@*G*GQR H\H\H^H^*_*_``` Nh,L,L,L,LMMMH 6X 6)<)<)>)> 6!228XH[]fggNHd K  fd8nn 5 5 5 7 K  hX 7 7 7!008XObdmpHIINHd K  gtH~~ 6 6 6 8!228XFFNHd K  gX 7 7 7 { ,'CT  o K  lD%**=l=l=l=lck=l=l=l2m2m n n n !!!r6c d}|s|r|jdt|fd|D}t t ||d\}}|jdt||jdt||sF|r1|jdt|D]j}|jd|||jd|t|jk|jdt|fd |D}|jd t|t ||d \}}|jd t|t tfd |}|jdt|t ||d\}}||fS)aQ Select the initial case to start a sentence @param caselist: list of steming result of the word. @type caselist: list of stemmedSynword @param indxlist: list of index. @type indxlist: list of integer @return: indexlist and rule. @rtype:( list of integer, integer/False) . Fz10 %scJg|]}|| Sr) has_sem_nextrrr r s r4rz2TashkeelClass.__choose_initial..s1\\\Qhqk.F.FGZ.[.[\q\\\r6z11 %sz12a %s next_chosenz12b [%d] : %s nextsz12e %s indxlistcJg|]}|| Srhas_nextr's r4rz2TashkeelClass.__choose_initial..s1XXXQhqk.B.BCV.W.WXqXXXr6z12c %s indxlistz12 %s indxlistc|pN|o| p|Sr) is_stopword is_marfou3 is_passiveis_pastrs r4r z0TashkeelClass.__choose_initial..sb!)@)@)B)B**1+((**K8A;3I3I3K3K/K**QK''))r6z13 %s) rirr+rs_get_indexlist_and_rulerrarwnextfilter)r2r!r r rrr"rs `` r4__choose_initialzTashkeelClass.__choose_initialsj 7::<< 7 K  gtH~~ 6 6 6\\\\\(\\\G4T']]HhXYZZNHd K  gtG}} 5 5 5 K  gtH~~ 6 6 6 @::<< @ K  2D9L4M4M N N N S S !!"78A;;O;O;Q;QRRR !!"74 @P;Q;QRRRR K  /h @ @ @XXXXX(XXXG K  /g ? ? ?5WhRTUUNHd K  .X ? ? ?v,,,,-57788 '4>>2220(HcRR$$r6cH d}|srjdt|t t fd|}t |||\}}jdt|t t fd|}jdt|t ||d\}}t t fd|}jd t|t ||d \}}t t fd |}jd t|t ||d \}}|s]t t fd|}t ||d\}}jd t||s`t t fd|}jdt|t ||d\}}|s^t t fd|}jdt|t ||d\}}|str`t t fd|}jdt|t ||d\}}|ssr_t t fd|}jdt|t ||d\}}|str`t t fd|}jdt|t ||d\}}|str`t t fd|}t ||d\}}jd t||srr^t t fd!|}jd"t|t ||d#\}}|ssr_t t fd$|}jd%t|t ||d&\}}|su|r|s|s]t t fd'|}jd(t|t ||d)\}}||fS)*aY Select the case in the middle, when it is possible @param caselist: list of steming result of the word. @type caselist: list of stemmedSynword @param indxlist: list of index. @type indxlist: list of integer @return: indexlist and rule. @rtype:( list of integer, integer/False) . rz14-a %sc|j|p|Srr( is_relatedr&rr r rr2s r4r z.TashkeelClass.__choose_cases..sHDK,B,B8XVW[,Y,Y-K]efg]h]u]uwJ^K^Kr6z14-b %sc|j|o|Srr9r;s r4r z.TashkeelClass.__choose_cases.. sHDK,B,B8XVW[,Y,Y-L^fgh^i^v^vxK_L_Lr6z14-c %sc|j|o|Sr)r(r:r+r;s r4r z.TashkeelClass.__choose_cases..(sHDK,B,B8XVW[,Y,Y-H^fgh^i^r^rtG_H_Hr6z15 %sc|j|o|Sr)r%r:r&r;s r4r z.TashkeelClass.__choose_cases..0sHDL,C,CHhWXk,Z,Z-M_ghi_j_w_wyL`M`Mr6z16 %sr(c8|Sr)r.rs r4r z.TashkeelClass.__choose_cases..8sXa[-D-D-F-Fr6c|o;j|o|Sr)r.r%r:r+r;s r4r z.TashkeelClass.__choose_cases..BsYXa[-D-D-F-F.TDLLcLcdlnvwxnyLzLzMS)1!)=)=>Q)R)Rr6z17 %s)cn|o|Srr.r+rr r s r4r z.TashkeelClass.__choose_cases..Js0Xa[-D-D-F-F-t8TU;K_K_`sKtKtr6z18 %s c|j|o|Sr)r%r:r+r;s r4r z.TashkeelClass.__choose_cases..\sHT\-D-DXxXY{-[-[.J`hij`k`t`tvIaJaJr6z19 %s cFj|Sr)r%r:rr rr2s r4r z.TashkeelClass.__choose_cases..bsT\-D-DXxXY{-[-[r6z20 %s cj|o-j|Srr%r:are_compatible_relationsrrr rrr2s r4r z.TashkeelClass.__choose_cases..jwT\-D-DXxXY{-[-[.`d`laFaFG_aia{a{|}a~a~aar6z21 %s cj|o-j|SrrOrQs r4r z.TashkeelClass.__choose_cases..{rRr6z22 %sc:|Srr*rGs r4r z.TashkeelClass.__choose_cases..sHQK,@,@AT,U,Ur6z23 %sr,cj|o4|o| Sr)r%r:is_noun is_mansoubrLs r4r z.TashkeelClass.__choose_cases..swT\-D-DXxXY{-[-[.W`hij`k`s`s`u`u.WGHIJUUWW{Wr6z24 %scn|p| Sr) is_tanwinrWrs r4r z.TashkeelClass.__choose_cases..s0 0E0E0G0G0dxXY{ObObOdOdKdr6z25 %s) rirr+rsrr5r3rais_break) r2r!r rr rrrr"s ` ``` ` r4__choose_caseszTashkeelClass.__choose_cases s U::<< U K  ih 8 8 86#K#K#K#K#K#K#KMUVVWWG5WhRVWWNHd K  ih 8 8 86#L#L#L#L#L#L#LNVWWXXG K  ih 8 8 84WhRSTTNHd6#H#H#H#H#H#H#HJRSSTTG K  gtH~~ 6 6 64WhRSTTNHd6#M#M#M#M#M#M#MOWXXYYG K  gtH~~ 6 6 64WhRSTTNHd 76"F"F"F"FQQRRG5WhRSTTNHd K  gtH~~ 6 6 6  V6#T#T#T#T#T#T#TV^``aaG K  gtH~~ 6 6 64WhRTUUNHd V6"t"t"t"t"twAABBG K  gtH~~ 6 6 65WhRTUUNHd V::<< V6#K#K#K#K#K#K#KMUVVWWG K  gtH~~ 6 6 64WhRTUUNHd V;;== V6"\"\"\"\"\"\^fgghhG K  gtH~~ 6 6 64WhRTUUNHd V;;== V6#A#A#A#A#A#A#ACKLLMMG K  gtH~~ 6 6 64WhRTUUNHd 7;;== 76#A#A#A#A#A#A#ACKLLMMG5WhRTUUNHd K  gtH~~ 6 6 6 V::<< V6"U"U"U"U"UW_``aaG K  gtH~~ 6 6 64WhRTUUNHd V::<< V6#X#X#X#X#X#XZbccddG K  gtH~~ 6 6 64WhRTUUNHd Z Zi0022 Z9 Zv&d&d&d&dgoppqq !!'4>>:::"9(HVX!Y!Y$~r6cHdtfd|Dttfd|}t||d\}}|s5ttfd|}t||d\}}|s5ttfd|}t||d\}}|s5ttfd |}t||d \}}|s5ttfd |}t||d \}}|s5ttfd |}t||d\}}|s5ttfd|}t||d\}}||fS)aW Select default cases when all others methods fail @param caselist: list of steming result of the word. @type caselist: list of stemmedSynword @param indxlist: list of index. @type indxlist: list of integer @return: indexlist and rule. @rtype:( list of integer, integer/False) . rcDg|]}|Srrrs r4rz2TashkeelClass.__choose_default..s)@@@!x{++--@@@r6c@|kSrr)rr maxfreqs r4r z0TashkeelClass.__choose_default..s (<(<(>(>'(Ir6cn|o| Sr)rWis_dualrs r4r z0TashkeelClass.__choose_default..s0hqk.A.A.C.C.bXVW[M`M`MbMbIbr6 cl|o|Sr)rWrXrs r4r z0TashkeelClass.__choose_default..s-hqk.A.A.C.C.aRSI_I_IaIar6!cn|o| Sr)is_verbr0rs r4r z0TashkeelClass.__choose_default..s0hqk.A.A.C.C.dHUVKLbLbLdLdHdr6#c|o3|p|Sr)rjr/r1rs r4r z0TashkeelClass.__choose_default..sEhqk.A.A.C.C.{RSI_I_IaIaIzemnoepexexezezr6$cl|o|Sr)rj is3rdpersonrs r4r z0TashkeelClass.__choose_default..-hqk.A.A.C.C.aQR H_H_HaHar6%cl|o|Sr)rj is1stpersonrs r4r z0TashkeelClass.__choose_default..rpr6&)maxrr5r3)r2r!r r"rrbs ` @r4__choose_defaultzTashkeelClass.__choose_defaults@@@@x@@@AAvIIIII8TTUU0(HbQQ$ V6"c"c"c"cemnnooG5WhRTUUNHd V6"b"b"b"bdlmmnnG5WhRTUUNHd V6"e"e"e"egoppqqG5WhRTUUNHd V6"|"|"|"|GHHIIG4WhRTUUNHd V6"b"b"b"bdlmmnnG4WhRTUUNHd V6"b"b"b"bdlmmnnG4WhRTUUNHd~r6c d}d}|r| nd |}g} d}|r|r|r| r| }|rtt fd|} | r| }t |dkrd}|rCt d|dfd|Dd ||dS) a reduce possible cases for the current word, according to the previous one and nexts. A new algorithm @param: list of steming result of the word. @type caselist: list of stemmedSynword @param: the choosen previous word stemming. @type previous_chosen_case:stemmedSynwordhg @return: the choosen stemming of the current word. @rtype:stemmedSynword. rFNcn|o|SrrFrGs r4r z.TashkeelClass.__reduce_cases..s2Xa[-D-D-F-F.HUVKL`L`atLuLur6rrBr cDg|]}|Srrrs r4rz0TashkeelClass.__reduce_cases..#s*/^/^/^PQ 0I0I0K0K/^/^/^r6utf8) rr]rr5rtprintrrset_chosen_indexes) r2r rrrrrr+r!r"r s ` @r4__reduce_caseszTashkeelClass.__reduce_casessM  '"+">">"@"@  "& "4466 #Y #9+=+=+?+? #  #"  p6######%%&&G 1"x==A%%q oeAdEJJ/^/^/^/^U]/^/^/^$_$_$f$fgm$n$nooo ))(33333r6)F)Frx)rx)NNNF)NNN)+r __module__ __qualname____doc__r5r,r;r>r@rBrDrGrIrLrOrQrTrVrXr[r^rarergrirlrnrpr}rrrrrrrrrrrrrrrr6r4r r -sJDJDJDJD\,,,*** 555+++ $$$ ///...***+++,,,222&&& ''' &&&... /// ...... /// .../// 000 ///#"#"#"Je"e"e"e"NKKKLLL88 (,,,6"""4>B@EJ"J"J"J"X---`SSSj@@@L>B!%C4C4C4C4C4C4r6r c d}|r|}t|dkr|nd}|rDtd||dfd|Dd||fS)z Just a macro to avoid repetetion if tmplist is not empty, change indexlist, if the indexlist has one element, the rule is applyed Frrzchoose tashkeelr cDg|]}|Srrrs r4rz+_get_indexlist_and_rule..8s*?o?o?o`a @Y@Y@[@[?o?o?or6rz)rtr{rr)r" indexlistr rr+ applied_rules ` r4r3r3*s E y>>Q..44AL A lEJJ?o?o?o?oen?o?o?o4p4p4w4wx~44 A A A l ""r6ctdt}d}||}t|ddS)z main test testu%يعبد الله تطلع الشمسrzN)r{r rr) vocalizerrxrs r4mainlyr<sR &MMMI 4D   T " "C #**V  r6__main__)&rsysrr operatorrpyarabic.arabyrqalsadi.analexrsylajone.anasynr"sylajone.syn_const asmai.anasemr'maskouk.collocations collocationsr-pyarabic.numberrpyarabic.namedzipr itertoolsrmysam.tagcoderrrrrpathrrr+r r3rrrr6r4rs ###### z4z4z4z4z4z4z4z4z###$    z FHHHHHr6