-iDdZedkr8ddlZejdejdddlmZddlZ Gdd Z d Z edkr e dSdS) z0 A class to remove ambiguation in text analysis __main__Nz../z ../support)disambig_constceZdZdZdZdZedZedZedZ edZ edZ ed Z ed Z d S) Disambiguatorz8 A class to remove ambiguation in text analysis cLtj|_dS)N) naftawayhwordtag WordTaggertagger)selfs N/var/www/html/speakWrite/venv/lib/python3.11/site-packages/qalsadi/disambig.py__init__zDisambiguator.__init__"s '2244 c|r t|t|kr|Sg}tt||}t|D]$\}}|d}||r|dz dkr||dz d}nd}|||} | |kr| }n|dzt|kr||dzd} |j| r+||r| |}nD|j | r*| |r| |}| |&|S)a} Disambiguate some word according to tag guessing to reduce cases. return word list with dismbiguate. @param word_list: the given word lists. @type word_list: unicode list. @param tag_list: the given tag lists, produced by naftawayh @type tag_list: unicode list. @return: a new word list @rtype: unicode list rr)lenlistzip enumerate is_ambiguousget_disambiguated_by_prev_wordr is_verb_tagis_disambiguated_by_next_verbget_disambiguated_by_next_verb is_noun_tagis_disambiguated_by_next_nounget_disambiguated_by_next_nounappend) r word_listtag_list newwordlist wordtaglistir currentword previouswordtmpwordnexttags rdisambiguate_wordsz Disambiguator.disambiguate_words's) C NNc(mm;; Ks9h7788K' 44! 0! 0 7%aj $$[111uzz'21q5'9!'< ') "AA#\G+--&- Q[!1!111"-a!e"4Q"7  ;22# "@@MM +/*M*M +++KK"[44#"@@MM+/*M*M +++K"";//// rc|tjvS)ztest if the word is an ambiguous case @param word: input word. @type word: unicode. @return : if word is ambiguous @rtype: True/False. )dconstDISAMBIGUATATION_TABLEwords rrzDisambiguator.is_ambiguous`sv444rctj|idid|S)aget The disambiguated form of the word by the next word is noun. The disambiguated form can be fully or partially vocalized. @param word: input word. @type word: unicode. @return : if word is ambiguous @rtype: True/False. noun vocalizedr+r,getr-s rrz,Disambiguator.get_disambiguated_by_next_nounj;  ) - -dB 7 7 S__ Sd # # rctj|idi||S)aTget The disambiguated form of the word by the previous. The disambiguated form can be fully or partially vocalized. @param word: input word. @type word: unicode. @param previous: input previous word. @type previous: unicode. @return : if word is ambiguous @rtype: True/False. previousr2)r.r6s rrz,Disambiguator.get_disambiguated_by_prev_wordys=  ) - -dB 7 7 SR S4  rctj|idi||S)aDget The disambiguated form of the word by the next. The disambiguated form can be fully or partially vocalized. @param word: input word. @type word: unicode. @param next: input next word. @type next: unicode. @return : if word is ambiguous @rtype: True/False. nextr2)r.w_nexts rget_disambiguated_by_next_wordz,Disambiguator.get_disambiguated_by_next_words;  ) - -dB 7 7 S__ S   rctj|idid|S)aget The disambiguated form of the word by the next word is a verb. The disambiguated form can be fully or partially vocalized. @param word: input word. @type word: unicode. @return : if word is ambiguous @rtype: True/False. verbr1r2r-s rrz,Disambiguator.get_disambiguated_by_next_verbr4rcFdtj|ivS)ztest if the word can be disambiguated if the next word is a noun @param word: input word. @type word: unicode. @return : if word has an disambiguated. @rtype: True/False. r0r2r-s rrz+Disambiguator.is_disambiguated_by_next_noun"6::4DDDDrcFdtj|ivS)ztest if the word can be disambiguated if the next word is a verb @param word: input word. @type word: unicode. @return : if word has an disambiguated. @rtype: True/False. r<r2r-s rrz+Disambiguator.is_disambiguated_by_next_verbr>rN)__name__ __module__ __qualname____doc__rr) staticmethodrrrr:rrrrrrrs555 777r55\5   \    \   \    \  EE\EEE\EEErrc^d}|d}t}|j|}t d||||}t d|ddS)z Main test uv السلام أن العبادي كان أعلن فتح المنطقة أن السلام مفيد أن يركبوا  utf8N)splitrr word_taggingprintjoinr)encode)textwordlistdisambtaglistr"s rmainlyrSs DDzz#H __Fm((22G $))G  ++Hg>>K #((;   & &v . ./////r) rCr@syspathrrrr+naftawayh.wordtagr rrSrErrrWs zJJJHOOEHOOL!!!'&&&&&_E_E_E_E_E_E_E_ED 0 0 0  z FHHHHHr