-iJJdZedkr8ddlZejdejdddlmZddlZddlZddlZddl m Z ddl m Z ddl ZddlmZdd lmZdd lmZdd lmZdd lmZGd dZdZedkr edSdS)z Syntaxic Analysis __main__Nz../libz../)xor)arepr) syn_const)synnode)stemmedsynword)cache_pickledbceZdZdZd!dZdZdZdZdZdZ d Z d Z d"d Z dZ dZdZdZdZdZdZdZdZdZdZdZdZdZdZdZd!dZdZd!d Zd S)#SyntaxAnalyzerz$ Arabic Syntax analyzer Fctj|_d|_||_|rt j||_nd|_dS)NF)wordtag WordTagger wordtaggersyntax_train_enabled cache_pathcache)selfrs M/var/www/html/speakWrite/venv/lib/python3.11/site-packages/sylajone/anasyn.py__init__zSyntaxAnalyzer.__init__.sN!,..%*!%  Z00DJJDJ cx|jr2|j|jdS)N)rrupdate display_allrs r__del__zSyntaxAnalyzer.__del__?s<  $ % J      J " " $ $ $ rc|}|}|j|||dS)zN Add relation between previous and current with realtion code N) get_originalr add_relation)rpreviouscurrentrelation pre_original cur_originals radd_cache_relationz!SyntaxAnalyzer.add_cache_relationGsF ,,.. ++--   lHEEEEErcRg}g}|D]fdttD}|stjdg}|||t j||||\}}||fS)a Syntaxic analysis of stemming results. morphological Result is a list of stemmedword objects The syntaxic result have the same structure, but we add a field named 'syntax' to every word steming dictioionary @param detailed_stemming_dict: detailed stemming dict. @type detailed_stemming_dict:list of list of stemmedword objects @return: detailed syntaxic result with syntaxic tags. @rtype: list of list of stemmedsynword cFg|]}tj||S)r StemmedSynWord).0order stemming_lists r z*SyntaxAnalyzer.analyze..asA@@@!&4]55I5QQ@@@rr)rangelenr r)appendrSynNodestudy_syntax_by_synode)rdetailed_stemming_dictstemmedsynwordlistlist synnode_listtmplistr,s @ranalyzezSyntaxAnalyzer.analyzePs"$ 4 : :M@@@@%*3}+=+=%>%>@@@G M)8q9IJJL " ) )' 2 2 2    8 8 9 9 9 9/3/J/JKaco/p/p, %|33rc d}d}ttt||D]\}\}}|dzt|kr ||dz}nd}t|D]\\} } |s|d| d} $t||D]"\} } || | | | ||\} } #]|}|}||fS)zF Analyzing the texts words cases by one syntax node, NFr) enumeratelistzipr/bigram_analyze) rr4r5pre_nodeprevious_index current_indexstemmedsynwordlist current_node next_nodecurrent_case_indexstmwordprevious_case_indexr s rr2z%SyntaxAnalyzer.study_syntax_by_synodels2CLTRUVlmyRzRzM{M{C|C| * * =M=. q 3|#4#444(q9  /89K/L/L Q Q+"G%Q"11$AA!DGG9BCYZhCi9j9jQQ5+X,0,?,?'Sfhz}EGP-Q-Q)''Q+N)HH%|33rc d}|r|r tt|}n#t $rd}YnwxYw|dzt tddvs |dzdkr!|r tj }nK|dzt tddvs |dzdkr | r tj }|S)zT Return the weight if the previous and current have number relation rd c) is_noun is_tanwinintfloatget_word ValueErrorr:r. is_majroursconstJarMajrourRelation is_mansoubNasebMansoubRelation)rr r!weightnumbers ris_number_relationz!SyntaxAnalyzer.is_number_relations ??   9!2!2!4!4 9 U8#4#4#6#67788    }U1R[[ 1 111Vc\Q5F5F%%''7#6F3,$uR||"4"444 8I8I%%''9#8F s.A A*)A*c|o<|o(| o|S)zY Return the weight if the previous and current have verb object relation )rLrU has_encletic is_transitiverr r!s ris_verb_object_relationz&SyntaxAnalyzer.is_verb_object_relationsX!!Hg&8&8&:&:H((***H/7/E/E/G/G IrcN|rdS|r}|ri|rU|rA|r-|s||rdSdS|r|r|r|r|r|s||krY| | kr/| s||rdSdSdSdSdS)NFT) is_stopwordis_breakrL has_procletic has_jonctionhas_jareq_caseis_verb get_tense get_pronoun is_presentr]s ris_jonction_relationz#SyntaxAnalyzer.is_jonction_relations    ! ! 5      '//"3"3 9I9I9K9K ##%% *1*>*>*@*@ IPIZIZ x(( t        GOO$5$5 8;K;K;M;M ##%% *1*>*>*@*@ IPIZIZ  ""$$(:(:(<(<<<((**h.B.B.D.DDD))++ x00%#t$u 4u5rrNcd}|r|r|||||S|s|||||S|||rD||t j||t j||fS|r | r$||t j ||fS| rp| s| rH| s|r |r t j}|r<|dkr$||t j||fSn|r|||}n*| r|||}|s*| r|||}|r|||}|rJ|||||||jr||||||fS)a Syntaxic analysis of stemming results, two words. the positions are use to join related cases. @param previous : the first item in bigram. @type previous : stemmedSynWord @param current : the second item in bigram. @type current : stemmedSynWord @param currentSynode : the current synode in The phrase. @type currentSynode : synnode object @param previous : the first item position in the word case list. @type previous : stemmedSynWord @param current : the second item position in the word case list. @type current : stemmedSynWord @return: the updated previous and current stemmedSynWord. @rtype: (previous, current) ru قَالَ) is_initial treat_initial treat_finalrjadd_nextrSJonctionRelation add_previousrarMTanwinRelationrfr`rdis_indirect_transitive_stopwordis_indirect_transitiveVerbParticulRelation is_pounctrVerbObjectRelationrLtreat_previous_nountreat_previous_verbtreat_previous_stopword is_numberrYrr%)rr r!previous_positioncurrent_positionr=rBrWs rr<zSyntaxAnalyzer.bigram_analyzes4 '8..00 '%%h9JL\]] ] '##Hg7HJZ[[ [  & &x 9 9 '   /1H I I I  !2F4K L L Lg& &     & D!!## +!!"2F4IJJJ '**!! 9w':':'<'< 9@Q@Q 9__&&9'*Q*Q*S*S9YaYxYxYzYz9#8F  "" /((**m;;%%&68QRRR$g..!! E11(GDD!!## E11(GDD Ih2244 I55hHH !!## D007CC  C   . 7 7 7  !2F ; ; ;( C'''6BBB  rc|s(|s|r ||tj||fS)aP Treat Initial case, where the current case is the first in the list @param previous: the previous case @type previous: stemmedSynWord @param next: the current case @type current: stemmedSynWord @return: (previous, current) after treatment @rtype: tuple of stemmedSynWord ) is_marfou3is_pastr`rqrSPrimateRelationrr r!r|r}s rrmzSyntaxAnalyzer.treat_initial+sf      L7??#4#4 L8K8K8M8M L  !2F4J K K K'""rcr|r ||tj||fS)aI Treat final cases, where the current case is the end of line @param previous: the previous case @type previous: stemmedSynWord @param next: the current case @type current: stemmedSynWord @return: (previous, current) after treatment @rtype: tuple of stemmedSynWord )rMrorSrrrs rrnzSyntaxAnalyzer.treat_final>s>4      G   .0E F F F'""rcd}|s|sdS|r4|r |r tj}|||r!|r tj}n| ||r|r| r!| r tj }n| r!| r tj }nk| r | r tj}n6|r"|||r tj}|s |r tj}|S)a; Treat noun cases, where the previous is a noun @param previous: the previous case @type previous: stemmedSynWord @param next: the current case @type current: stemmedSynWord @return: (previous, current) after treatment @rtype: tuple of stemmedSynWord r)rarL need_additionis_additionablerSAdditionRelationare_compatibleis_adjDescribedAdjectiveRelationare_nominal_compatiblerPrimatePredicateRelationrUPrimateMansoubPredicateRelationPrimatePredicateMansoubRelationrfcompatible_subject_verbRafe3Marfou3Relationis_confirmationConfirmationRelationrr r!rWs rrxz"SyntaxAnalyzer.treat_previous_noun]s      X%5%5%7%7 1 ??  $ 5%%'' 1G,C,C,E,E 10""8W55 H>>##?#>F,,Xw?? H>>## H ++--H'2D2D2F2FH!'!@"--//HG4F4F4H4HH!'!G"--//HG4F4F4H4HH!'!G__   5++Hg>> 54 1'1133 10F rcd}|s|sdS|rK|r|r|r|s/ | r tj }nitj }n\|r|s(|s | r tj}|s|rH|s4|s |r tj}|S)a; Treat verb cases, where the previous is a noun @param previous: the previous case @type previous: stemmedSynWord @param next: the current case @type current: stemmedSynWord @return: (previous, current) after treatment @rtype: tuple of stemmedSynWord r)rarfrLr is3rdperson is_single is_passive is_femininis3rdperson_femininis3rdperson_masculin is_kana_rafe3rSKanaRafe3Marfou3RelationVerbSubjectRelationVerbPassiveSubjectRelationrUis_mabnir[r\rwrs rryz"SyntaxAnalyzer.treat_previous_verbs      X%5%5%7%7 1 ??   ;!!## G''))Gh.@.@.B.BG$..00GD (5577D)/)H*0)C%//11Gh6R6R6T6TG#*#5#5#7#7G,>,@,@ 9/'')) 9g.@.@.B.B 98""$$ 9%%''9#8F &&(( @''))@g.@.@.B.B@#?FF))++@0B0B0D0D@#?FF))++@0B0B0D0D@#?F__   :++Hg>> 43!!## =$$&&=7+=+=+?+?= //11W<<"99;;@$*$?6!'!<++-- =73E3E3G3G =#8FF++--='2D2D2F2F= //11W<<&==??@$*$?6!'!<++-- 65..00 :9    ! ! -,F rc|rT|rR||vr*||vrdSdS)ar verify the syntaxic path from the previous to current stemmed word. If the current word is related with the previous word, return True. The previous word can contain a pointer to the next word. t he current can have a pointer to the previous if they ara realated @param previous: the previous stemmed word, choosen by the tashkeel process. @type previous:stemmedSynWord class @param current: the current stemmed word. @type current:stemmedSynWord class @return: return if the two words are related syntaxicly. @rtype: boolean TF) get_order get_previousget_nextr]s r is_relatedzSyntaxAnalyzer.is_relatedsg   h&8&8&:&:''#*#4#4#6#6(:K:K:M:M#M#M4UrcN|s|r|sdSd}||rd}ndS||rd}||sP|r?|r+|r|rd}ndS||rd}ndS| r| dvrd}ndS|S)u verify the gramatica relation between the two words. دراسة الترابط النخوي بين الكلمتين، اي توافقهما في الجمع والنوع، والحركة If the current word is related with the previous word, return True. The previous word can contain a pointer to the next word. the current can have a pointer to the previous if they ara realated @param previous: the previous stemmed word, choosen by the tashkeel process. @type previous:stemmedSynWord class @param current: the current stemmed word. @type current:stemmedSynWord class @return: return if the two words are related syntaxicly. @rtype: boolean FT)uالuفالuوالuوuف) rL is_addition eq_defined eq_gender eq_numberrr is_pluralrerb get_procleticrr r! compatibles rrzSyntaxAnalyzer.are_compatible1sf ""$$ (<(<(>(>  5    h ' ' JJ5   X & & J   X & & 7+<+<+>+> 7CUCUCWCW     ! ! &.&9&9&;&; JJ5 ??8 $ $ JJ5$$&& '*?*?*A*AF.+.+.JJ5 rc\|sdSd}|r|s|rd}ndS|r|sP|r|s(|r|rd}ndS|sd}ndS|sd}ndS||rd}ndS||rd}ndS|r|rd}ndS|sd}ndS|sd}ndS||rd}ndS| |rd}ndS||rd}ndS| s(| r|rd}ndS|S)u, verify the gramatica relation between the two words, for nominal relational المبتدأ والخبر دراسة الترابط النخوي بين الكلمتين، اي توافقهما في الجمع والنوع، والحركة If the current word is related with the previous word, return True. The previous word can contain a pointer to the next word. the current can have a pointer to the previous if they ara realated @param previous: the previous stemmed word, choosen by the tashkeel process. @type previous:stemmedSynWord class @param current: the current stemmed word. @type current:stemmedSynWord class @return: return if the two words are related syntaxicly. @rtype: boolean FT) rL is_definedrUrrbrMrrrf eq_personrrirs rrz%SyntaxAnalyzer.are_nominal_compatibles !! 5 ??  v &&(( X-@-@-B-B ! u$$&& 7+=+=+?+? $$&& +2+=+=+?+? ##%%  +2*<*<*>*>   " u((** ! u%%'' ! u  ** ! u  ** !  __  : ""$$ ! u((** ! u%%'' ! u   ** ! u   ** !   ** !    W%7%7%9%9     ! ! ! urcd}|s|rdS|r|rr|r^|rJt j|g}| |vrd}| rJt j|g}| |vrd}n| s(| s|rd}ndS||rd}ndS||rd}ndS||rd}ndS|s(|r|rd}ndS|S)u verify the gramatical relation between the two words, for subject and verb دراسة الترابط بين الفاعل والفعل، حين يسبق الفاعل الفعل If the current word is related with the previous word, return True. The previous word can contain a pointer to the next word. the current can have a pointer to the previous if they ara realated @param previous: the previous stemmed word, choosen by the tashkeel process. @type previous:stemmedSynWord class @param current: the current stemmed word. @type current:stemmedSynWord class @return: return if the two words are related syntaxicly. @rtype: boolean FT)rrUr` is_pronounrfrS TABLE_PRONOUNgetrhris_added get_encleticrrMr[rrrrrir)rr r!rexpected_pronounss rrz&SyntaxAnalyzer.compatible_subject_verbs"        7#5#5#7#7 5    ! !; ##%% *'//*;*; *&&((*(.(<(@(@ATATAVAVXZ([([%,,..2CCC%)   "" &$*$8$<$ > >xSYSrGrGr5 f5 5 56:::uV<<<utrcg}|D]-}d|vr||.t|dkr|S|S)z exclode imcompatible cases @param word_result: The steming dict of the previous word. @type word_result: list of dict @return: the filtred word result dictionary with related tags. @rtype: list of dict #r) get_syntaxr0r/)r word_resultnew_word_result stemming_dicts r exclode_caseszSyntaxAnalyzer.exclode_casessc( 6 6Mm..0000&&}555    ! !" " rcg}|D]}g}g}|r|d|r|d|r|d|r|d|s|}n"|D]}|D]}|||z |r|}dS)z Detect Possible syntaxical nodes sequences @param synnode_list: list of synNode. @type synnode_list: list of synNode. VNS.T)hasVerbr0hasNoun hasStopword has_pounct)rr5 grammar_listsndr6taglistgrmtags rdetect_accepted_gammarz%SyntaxAnalyzer.detect_accepted_gammars   ' 'CGG{{}} $s###{{}} $s###   $s###~~ $s### 0& '00C&00s3w////0 '& trcbd|_|js tj|j|_dSdS)zD Allow the anasyntax to use Cache to reduce calcul. TN)rrCacherrs renable_allow_cache_usez%SyntaxAnalyzer.enable_allow_cache_uses7%)!z 6T_55DJJJ 6 6rcd|_dS)zH Not allow the anasyntax to use Cache to reduce calcul. FN)rrs rdisable_allow_cache_usez&SyntaxAnalyzer.disable_allow_cache_uses%*!!!rc\tjtj|}d|D}|S)a/ Decode objects result from analysis. helps to display result. @param stemmed_synwordlistlist: list of list of StemmedSynWord. @type word_result: list of list of StemmedSynWord @return: the list of list of dict to display. @rtype: list of list of dict cg|] }|j Sr()__dict__)r*xs rr-z)SyntaxAnalyzer.decode..s444AQZ444r) functoolsreduceoperatorconcatrstemmed_synwordlistlist flat_lists rdecodezSyntaxAnalyzer.decodes2$X_6MNN 44444 rc d}|D]X}|dz }|D]I}|dz }|j}t|D]}|d|d||dz }|dz }J|dz }Y|d z }|S) z display objects result from analysis @param stemmed_synwordlistlist: list of list of StemmedSynWord. @type word_result: list of list of StemmedSynWord [z [z {z u'z' = u'z',z }z ]z ])rsortedkeys)rrtextrlistitemrDkeys rdisplayzSyntaxAnalyzer.displays ,  E H D # # "-!',,..11JJCDsssGCLLLIIDD " H DD  rc tg}g}g}d}|D]g}||d|D]-}|r|dn:.|r"|ddkr|dn|dg} |D]}|r|r|j} |D]} | } | | vrv| || || | | dg|| |}itt|||} g}|s"| D]\}}}|r|||} | S)a display objects result from analysis @param stemmed_synwordlistlist: list of list of StemmedSynWord. @type word_result: list of list of StemmedSynWord @param all: display all words with tags @type all: boolean (default False) NrBIOi') r0rPhas_next has_previousr r get_vocalizedrrr:r;)rrall wordsemlistwordlisttag_listprevious_rlistrr forms_listsyn_prevwordstemr+resulttmp_listwordrsyns r display_synzSyntaxAnalyzer.display_syns , # #E OOE!H--// 0 0 0 ) )==??OOC(((E )  3 3OOC((((OOC(((J ; ;$$&&;>;#}H$2;; ( 2 2 4 4 H,,&--t/A/A/C/CXE[E[E]E]!..00(2G2G2I2I&ll5%88/:;;;   z * * *"NNc(Hk::;; "( ) )c3)OOC(((F rcX||}tj|dS)z print objects result from analysis @param stemmed_synwordlistlist: list of list of StemmedSynWord. @type word_result: list of list of StemmedSynWord N)rpprintrs rrzSyntaxAnalyzer.pprintKs,KK 788  i     rctjd}||}||\}}|r||fS|S)zg Text Analysis syntacticly @param text: input text @type text: unicode here)qalsadianalexAnalex check_textr7)rrsynodesrr synodelists r analyze_textzSyntaxAnalyzer.analyze_textVsZ >((00&&t,,#||F33  :% %Mr)F)rrNN) __name__ __module__ __qualname____doc__rrr%r7r2rYr^rjr<rmrnrxryrzrrrrrrrrrrrrrr!r(rrr r *s    "   FFF4448$4$4$4T&III>EF?CY!Y!Y!Y!x###&###>;;;z---^SSSj&TTTnLLLbSSSl999v&<666****   (1111f ! ! !      rr cd}ddl}g}|j}t}||}||\}}||} tt|dS#t$r&t| dYdSwxYw)z main test uإلى البيتrNutf8) qalsadi.analexrrr rr7rprintstrUnicodeEncodeErrorencode)rrranalyzeranasyntr  text2displays rmainlyr0ds D F~$$&&HG   & &F!//&11FJ??6**L+ c,     +++ l!!&))******+s3B,CC)r%r"syspathr0rrrrpyarabic.arabyarabynaftawayh.wordtagrr(rpyarabic.arabreprrrrrSrr r rr r0r(rrr7sm zJJJHOOHHOOE ############ #"""""%%%%%%xxxxxxxxt!+++& z FHHHHHr