-i$dZddlmZddlZddlmZddlmZddlm Z m Z m Z m Z m Z mZmZmZmZmZmZmZmZmZmZmZmZmZmZmZdZdZdZdZ d Z!d Z"d Z#d Z$d Z%dZ&dZ'dZ(ddZ)e fdZ*dZ+dZ,dZ-dZ.dZ/dZ0dZ1dZ2dZ3dZ4dS)z' Basic routines to treat verbs ar_verb )print_functionN)FATHADAMMAKASRASHADDASUKUNHAMZAALEFWAWALEF_HAMZA_ABOVE ALEF_MADDA YEH_HAMZA ALEF_MAKSURAYEHTEHLAM_ALEFSIMPLE_LAM_ALEFLAM_ALEF_HAMZA_ABOVESIMPLE_LAM_ALEF_HAMZA_ABOVELAM_ALEF_MADDA_ABOVESIMPLE_LAM_ALEF_MADDA_ABOVEc8|d||z||dzdzS)a Replace a letter in string in position @param word: given string @type word: unicode @param rep: replecment letter @type rep: unicode char @param pos: replemcment position @type pos: int @return: modified string @rtype: unicode string rN)wordrepposs O/var/www/html/speakWrite/venv/lib/python3.11/site-packages/libqutrub/ar_verb.py replace_posr,s' #;s?4A< ''c"d}|dks|dkrd}n}|dkrttf}nh|dkrttf}nS|dkrttf}n>|dkrttf}n)|dkrttf}n|dkrttf}|S)a Get the the past and future marks by the bab sarf number - Bab: past future - 1 : FATHA DAMMA - 2 : FATHA KASRA - 3 : FATHA FATHA - 4 : KASRA FATHA - 5 : DAMMA DAMMA - 6 : KASRA KASRA @param number: Bab sarf number (1-6). @type number: integer(1-6) @return: a tuple of (past_mark, future_mark) @rtype: tuple Nrrrr)numberbabs rget_bab_sarf_harakatr*=s C zzVaZZ 1en 1en 1en 1en 1en 1en Jr c4d}|tkr|tkrd}n||tkr|tkrd}nc|tkr|tkrd}nJ|tkr|tkrd}n1|tkr|tkrd}n|tkr |tkrd}|S)a Get the bab sarf number by the past and future marks - Bab: past future - 1 : FATHA DAMMA - 2 : FATHA KASRA - 3 : FATHA FATHA - 4 : KASRA FATHA - 5 : DAMMA DAMMA - 6 : KASRA KASRA @param past_haraka: past haraka of the verb. @type past_haraka: unicode @param future_haraka: future haraka of the verb. @type future_haraka: unicode @return: Bab sarf number (1-6) @rtype: integer rrr#r$r%r&r"r') past_haraka future_harakar)s rget_bab_sarf_numberr.^s" Ce  6 6   -5"8"8   -5"8"8   -5"8"8   -5"8"8   -5"8"8 Jr cd}tdtdtdtdtjdtjdtjdtjd tj d tj d i }|D]}||vr|d ||zz }|d z }|S)z Write the harakat name in full in arabic @param harakat: list of harakat chars. @type harakat: unicode string @return: harakat in full @rtype: unicode فتحةضمةكسرةسكونuألفuواوuياءuىuوuئ-*) rrrrvconst ALEF_HARAKA WAW_HARAKA YEH_HARAKAALEF_YEH_HARAKAALEF_WAW_HARAKAALEF_YEH_ALTERNATIVE)harakatfull tab_harakathrks rwrite_harakat_in_fullrB~s D + ) + + y i i 5 5  K +   DS)) )DD DLDD Kr ctdz}|dvr tdz}nY|dvr(dtttg}n-|dkr'dtttg}|S)a Get the past harakat for the trileteral verb by bab sarf - Bab: past future - 1 : FATHA DAMMA - 2 : FATHA KASRA - 3 : FATHA FATHA - 4 : KASRA FATHA - 5 : DAMMA DAMMA - 6 : KASRA KASRA @param vtype: the bab sarf codification. @type vtype: unicode a string of number @return: harakat @rtype: unicode r$12346r05)rrjoinrvtypemarkss rget_past_harakat_by_babsarfrPsn !GE a *  %.// #%.// Lr cttztz}|dvrttztz}n9|dvrttztz}n|dkrttztz}|S)a Get the future harakat for the trileteral verb by bab sarf - Bab: past future - 1 : FATHA DAMMA - 2 : FATHA KASRA - 3 : FATHA FATHA - 4 : KASRA FATHA - 5 : DAMMA DAMMA - 6 : KASRA KASRA @param vtype: the bab sarf codification. @type vtype: unicode a string of number @return: harakat @rtype: unicode rDrHrK)rrrrMs rget_future_harakat_by_babsarfrRsd %K E e E! *  e E! #e E! Lr cb|dkrtS|dvrtS|dvrtS|dvrtSdS)u4 Get the future_type value from different codifications. used also in comand line in différent context the future_type is codified as: values or values used as Conjugation mode ( Bab Tasrif باب التصريف) - Bab: past future - 1 : FATHA DAMMA - 2 : FATHA KASRA - 3 : FATHA FATHA - 4 : KASRA FATHA - 5 : DAMMA DAMMA - 6 : KASRA KASRA @param vtype: the bab sarf codification. @type vtype: unicode a string of number @return: haraka @rtype: unicode char rE)rFrJ)rGrI)rErKr0)rrr)rNs rget_future_haraka_by_babsarfrTsG( || *   *   *   rr ctj|r|S|dkrtS|dkrtS|dkrtS|dkrt SdS)u( Convert an arabic named harakat to a real haraka values - Fahta:(فتحة) - DAMMA:(ضمة) - KASRA:(كسرة) @param haraka_name: the arabic name of haraka. @type haraka_name: unicode @return: the arabic name of haraka . @rtype: unicode char r1r2r3r4F)arabyis_shortharakarrrr) haraka_names rget_haraka_by_namerYsb K((k!!  ! !  # #  # # ur c6t|}|r|StS)u Get the future_type value by haraka arabic name. values - FATHA:(فتحة) - DAMMA:(ضمة) - KASRA:(كسرة) @param haraka_name: the arabic name of haraka. @type haraka_name: unicode @return: haraka @rtype: unicode char )rYr)rXharakas rget_future_type_by_namer\s# , ,F   r cd|z}|dvrtS|dvrtS|dvrtStS)u Get the future_type value from different codifications. used also in comand line in différent context the future_type is codified as: values - Fahta:(fatha, فتحة, ف, f) - DAMMA:(damma, ضمة, ض, d) - KASRA:(kasra, كسرة, ك, k) or values used as Conjugation mode ( Bab Tasrif باب التصريف) - Bab: past future - 1 : FATHA DAMMA - 2 : FATHA KASRA - 3 : FATHA FATHA - 4 : KASRA FATHA - 5 : DAMMA DAMMA - 6 : KASRA KASRA @param future_type: the future_type codification. @type future_type: unicode @return: extract the future type mark @rtype: unicode char r0)fathar1uفfrGrI)dammar2uضdrErK)kasrar3كkrFrJ)lowerrrr) future_types rget_future_type_entreergsR,k'')))KFFF DDD FFF ,r c |dvS)ug Get the transitive value from different codifications. in différent context the transitivity is codified as: - "t", "transitive", - u"متعدي", u"م", u"مشترك", u"ك" - True @param transitive: the transitive codification. @type transitive: unicode @return: True if is transitive @rtype: boolean )u متعديuمu مشتركrct transitiveTr)rjs rget_transitive_entreerk8s & &&r c|trtj|}t |dkr(|tt tzSt |dkrN|tj vrtj |dS|tt tzS|tt tzS|S)z Convert Alef madda into two letters. @param word: given word. @type word: unicode. @return: converted word. @rtype: unicode. r#r$r) startswithr rV strip_harakatlenreplacer r r7ALEF_MADDA_VERB_TABLE)rword_nms rnormalize_alef_maddarsJs z""%d++ w<<1  ??:uTz:: : \\Q  &6663Grrryrs r geminatingrsHK A \\F||Fa!! f** EEacFllwqs|v'='= ueUE222! 2  e##! (=(=GAJ&%gwqz1Q3??wqz) Q!*%%'!A#,%*?*?GAJ&%gwqz1Q3??u$ Q Q3!88 AaC 0E 9 9 6 11HqzU**&1#2#&6wqz&A!0'1181'> '2#2#&6u&<-'..5acl'; qSAXX+ac"2#V%6!##### 6 11H6#81#EEKKQZ5(( 6 11H6#81#EEKK 6 11H6#81#EEK Q UUqsV||!  (B(B u1%1F!F!F  6) )H 601= =K FAA  "H 71: %K FA f**@ k ""r ct|t|kr^t|dt|dd|Dt|dSd}d}t ||\}}t|t|krdSt|dkrg|t rMt|z}t|dkr&|dttj fvr t|z}n t|z}t|t|krdS|}|}t||\}}t|t|krtd t||dtd t|t|tt||ddSt||}t|}||krX||tjvr%|||tj||zz }n|||||zz }|dz }||kXtjD]\}}|||}|S) aS join the harakat and the letters to the give word in the standard script, it return one strings ( the word with harakat and the harakat). @param word_nm: given unvocalized word. @type word_nm: unicode. @param harakat: given harakat. @type harakat: unicode. @return: vocalized word. @rtype: unicode. utf8r5c6g|]}tj|Sr)rVname).0xs r zstandard2..s!=]=]=]PQejmm=]=]=]r r6r0rr#rz len word: z len harakat: )roprintencoderLrrmrr rr7r9r homogenizereprtahmeez2WRITTEN_HARAKASTANDARD_REPLACEMENTrp) rrr>rry word_beforeharakat_before len_word_nmpatrs r standard2rs 7||s7||## gnnV$$S\\499=]=]U\=]=]=]3^3^`cdk`l`lmmmt %gw77 w<<3w<< ' '3 w<<1  !3!3E!:!: 7lG7||q  WQZF-.&/&/-- w<<3w<< ' '3  %gw77 w<<3w<< ' ' ,G gnnV.D.D F F F /3w<<g ? ? ? $~&& (:(:6(B(B C C C37G,,'ll +ooqzV222 6#8#DDD 71:-- FA+oo1'' c||S#&& Kr c,t|t|krdSt|vr|Sd}|D]3}|tjks|tjkr tj}||z }4|}d}t t|D]}||tkr||tkr |||z }1|dkr*||}tj ||}n&||dz }||}|dzt|krV|tj ks |tkr3||dztkr||dztkr ||dz}|tj krt}|tj krt}|tjvr|tj|vrtj||}|ttjtjtjfvrk|tkr&||dz t"jkr t"j}n||dz t"jkr!|t*tjfvr t"j}n||}n|tj krt}|tj krt}|tjvrb|tj|vrN||dz t"jfvr|tt.fvr t"j}n!tj||}n||}||z }|S)u Transform hamza on the standard script. in entry the word without harakat and the harakat seperately return the word with non uniform hamza. إعلال و إبدال الهمزة. @param word_nm: given unvocalized word. @type word_nm: unicode. @param harakat: given harakat. @type harakat: unicode. @return: (letters, harakat) after treatment. @rtype: tuple of unicode. r0rr)ror r7r;r<r8ranger INITIAL_TAHMEEZ_TABLEgetrrrrMIDDLE_TAHMEEZ_TABLEr:r9rVrrr rFINAL_TAHMEEZ_TABLEr) rrr>ha2rArryactualswapbefores rrr$s 7||s7||##s w    Cf,,,V+++( 3JCCs7||$$3 3 AqzU""wqz5E'E'E "66$QZF!7;;FFKKDD$QqS\F$QZFsS\\))!V%:::foo&qs|v55'!A#,%:O:O)01!V%:::%*F!V%:::%*F"f&AAA!v#Fv#ND%5&2CVEWY_Yj*kkk#)U??wqs|uy7P7P+0?DD&-QqS\UY%>%>6SXZ`ZkRlClCl+0;D#*1:DD"V%:::%*F!V%:::%*F"f&@@@6#=f#EEE&qs|~==&UTYNBZBZ&+ktt'-'A&'I&'Q#*1:D  Kr c@|tdkr|St|}t|}||kr|Sd}t|D]}|dz|kr||dztkr||tjkrI|dz|kr$||dzt kr|tjz }b|dkr |tz }s|tz }~||tj vr|tj ||z }|||z }|||z }|S)aq Treat the rencontre of sukun. in entry the word without harakat and the harakat seperately, and the probably haraka return the new sequence of harakat @param word_nm: given unvocalized word. @type word_nm: unicode. @param harakat: given harakat. @type harakat: unicode. @return: (letters, harakat). @rtype: tuple of unicode. rr0rr#) rrrorr7r8rrrCONVERSION_TABLE)rrr>rz len_harakatrrys r treat_sukun2rvs4||EQ7||Hg,,K; x * *As[  WQqS\U%:%:1:!333 sX~~'!A#,&*@*@#v'99 a#u, #u, QZ6#:::6#:71:#FFKK71:-KKwqz) r c8t|t|kr/tdt|t|||fStjdtt jttd|s||fS|d}|d}d}t|}||dz kru||}||}|dz dkr||dz }||dz } nd}d} |dz|kr||dz} ||dz} nd} d} |dz|kr||dztkrd} nd } |tks |tkr|tkr| tkr|dz }n|ttt j fvr&| tkr| s|d d t j z}nw|tvr| tkr | r|dz }nZ|tkr<| r:| ttfvr|d d tz}n)|tz }|tz }n|tt jfvr#| tkr| s|tz }|tz }n|tt jfvr| tkr| r||z }|tz }n|tkr1| tkr&| tkr| s|d d t jz}nl|t j kr!| tkr|tz }|tz }n;|t jkr$| tkr|d d t jz}n||z }|tz }n|t jks |tkr|t jkr| tkr|dz }n|t jkrA|ttt jfvr&| tkr| s|d d t jz}ng|tkr6|ttfvr&| tkr| s|d d t jz}n&|tt jfvr!| tkr|tz }|tz }n|tkr&| tkr| s|d d t j z}n|tkr| r|d d tz}n|t jkrB|tkr7| r5| tkr|d d tz}nb|tz }|tz }nL|tkr)|t jkr|d d t jz}n|tkr|tkr| tkr| r|dz }n|t jkr*|t j kr| s|d d t j z}n|tkr,|tkr!| tfvr| r|tz }|tz }n~|tkr,|tkr!| tfvr| r|tz }|tz }nG|tkr"|tkr| r|tz }|tz }n||z }|tz }n ||z }||z }|dz }||dz ku||} ||}|dz dkr||dz }||dz } nd}d} | tks | tkr|ttfvr$| tkr|d d t j z}nm|t j vr$| tkr|d d t j z}n;|tkr |dz }|dz }n$|tkr<|tttfvr&| tkr|t jz }|t"z }n|tkr<|tttfvr&| tkr|t jz }|tz }n|t jvr$| tkr|d d t jz}nd|t j kr!| tkr|tz }|tz }n3||z }|tz }n"| t jkr|ttt jfvr$| tkr|d d t jz}n|t jvr$| tkr|d d t j z}n|t j kr|d d t j z}|dz }nr|tkr | tkrnZ|tkr |dz }|dz }nC|tkr+| tkr |t jz }|t jz }n ||z }|tz }n| tkr|ttfvr%| tkr|t jz }|tz }n|tfvr%| tkr|t jz }|tz }n|t j fvr-| ttfvr|d d t j z}|dz }nR|tfvr| ttfvr |dz }|dz }n-|||z }|||z }n|||z }|||z }||fS) u لإreat the jonction of WAW, YEH. معالجة التحولات التي تطرا على الياء أو الوا في وسط الكلمة أو في اخرها @param word_nm: given unvocalized word. @type word_nm: unicode. @param harakat: given harakat. @type harakat: unicode. @return: (letters, harakat)after treatment. @rtype: tuple of unicode. zHomogenize:inequal length[]rrr0r#TFNr~)rorresearchrr7rrr rrrr:rr9rrr r8)rrr>rrryr actual_letter actual_harakaprevious_letterprevious_haraka next_letter next_harakashadda_in_next last_letter last_harakas rrrs  7||s7||## )3w<<WFFF!!YYllF4G4GS###F'!!aj 1: 'll +a-#AJM#AJMsaxx")!A#,")!A#,"$"$s[  %acl %acl   s[  WQqS\V%;%;!%!&--#1E1E!L00[E5I5I2%KK$uf6G'HHH E)).)"-crc"263D"DKK$..Ou4L4L$5M2%KK$--N-&5%.88&1#2#&6u&< #u,  C$v/@'AAA!U**>*5(KOHH#v/@'AAA E))n)=0KOHH%..Ou4L4L C'''"-crc"263D"DKK#(9995((5(KOHH$):::SX@X@X"-crc"263D"DKK =0KOHH F$777 c ! ! F$777u$$2%KK"f&999%0A!BBB!U**N*"-crc"263D"DKK"c))}u~/M/M%..."-crc"263D"DKK%0A(BBB#u,,5(KOHH#// E))>)"-crc"263D"DKK$--.-"-crc"25"8KK#v':::%''^'&%//&1#2#&6u&< #u,  S($s** AR0R0R"-crc"2F4E"EKK#s** 0F0FO_dLdLdiwLd"$KK#v':::&"333#4"-crc"263D"DKK$++=E3I3IchbjOjOjPk5(KOHH#s** 0F0FO`e_gLgLgMh5(KOHH#s** 0F0F1G5(KOHH  =0KOHH}, M) FAA+a-Jaj aj Q3!88%aclO%aclOO O O < ' ';#+=+=u~--Ou4L4L)#2#.v/@@ &"344  % %)#2#.v/@@ &&r! B!S(( E5%0 0 0o6N6Nv44 D !S(( E5%0 0 0o6N6Nv44 L(&"344 u $ $)#2#.v/@@ &"333 u $ $u$ C{* C F/ / /ueV->??? u $ $)#2#.v/@@ !344  % %)#2#.v/@@ &"333)#2#.v/@@ B&&Ou,D,D&&r! B&&?e+C+Cv44 F//{* CS  u~--Ou4L4Lv44 L(%))50H0Hv|+ C6#4"666_QVX]P^=^=^)#2#.v/@@ B((5%./P/Pr! Bwqz) GAJ& 71: %K  "H+&&r ctj|}|ttt z}t |dkrdSdS)z Test if the verb is triliteral, used in selectionof verbs from the triliteral verb dictionnary @param verb: given verb. @type verb: unicode. @return: True if the verb is triliteral. @rtype: Boolean. r$TF)rVrnrpr r r ro)verbverb_nms ris_triliteral_verbrsF!$''Gooj%*55G 7||qt%r )rt)5__doc__ __future__rrlibqutrub.verb_const verb_constr7pyarabic.arabyrVrrrrrr r r r r rrrrrrrrrrrr*r.rBrPrRrTrYr\rgrkrsr{rrrrrrrrrrrrr rrs,  %%%%%%OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO ( ( ("B@:02@4&> & & &$6....b:?@   666rc#c#c#R   FX#X#X#v===@PPPd,,,`V'V'V'r     r