-iuhdZddlZejdejdejdddlZddlmZddlmZddl Z ddl m Z  dd l m Z n#YnxYwGd d e j j Zd Zed kree jdSdS)zJ stemmedWord represents the data resulted from the morpholocigal analysis Nz../z mishkal/libz../lib)xor) syn_const) stemmedwordceZdZdZdJdZdZdZdZdZd Z d Z d Z d Z d Z dZdZdZdKdZdZdLdZdLdZdKdZdZdZdZdZdZdKdZdZdZdLdZdKd Zd!Z d"Z!d#Z"d$Z#d%Z$d&Z%d'Z&d(Z'd)Z(d*Z)d+Z*d,Z+d-Z,d.Z-d/Z.d0Z/d1Z0d2Z1d3Z2d4Z3d5Z4d6Z5d7Z6d8Z7d9Z8d:Z9d;Z:d<Z;d=ZZ=d?Z>d@Z?dAZ@dBZAdCZBdDZCdEZDdFZEdGZFdHZGdIZHdS)MStemmedSynWordzR stemmedWord represents the data resulted from the morpholocigal analysis Nc|rH|j|_tj|j|_tj|j|_d|_d|_d|_ | r| |_ | rK| |_ | |_||_||_||_d|_d|_d|_d|_||_i|_i|_i|_i|_d|_d|_dS)NrF)__dict__arabystrip_tashkeel vocalized unvocalizedoriginal unvoriginaltag_verbal_factortag_nominal_factortag_kana_rafe3is_verb_is_kana_rafe3 is_stopword#_StemmedSynWord__get_nominal_factor"_StemmedSynWord__get_verbal_factor _is_addition tag_addition _is_break tag_breakforced_word_casesyntaxsemanticforced_wordtypeordernextprevioussem_next sem_previousscorerule)self result_dictr#s U/var/www/html/speakWrite/venv/lib/python3.11/site-packages/sylajone/stemmedsynword.py__init__zStemmedSynWord.__init__%sG  D'0DM % 4T^ D DD  % 4T] C CD $%$%$) <<>> :$($7$7$9$9D       A$($7$7$9$9D &*&?&?&A&AD #&*&>&>&@&@D "!..00..** %  $      cdS)z desctructor Nr*s r,__del__zStemmedSynWord.__del__Is r.c||jr|jS|jrtj|j|_ndS|jS)z Get the unvocalized form of the input word @return: the given unvocalized. @rtype: unicode string r )rrr rr1s r,get_unvocalizedzStemmedSynWord.get_unvocalizedOsH   # #~ #(#7#G#G  sr.c||jr|jS|jrtj|j|_ndS|jS)z Get the unvocalized original form of the input word @return: the given unvocalized original. @rtype: unicode string r )rrr rr1s r,get_unvoriginalzStemmedSynWord.get_unvoriginal^sH   $# #} $)$8$G$G  s# #r.c||_dS) Add a position number in the case word list. @param order: the order of the stemmed word in the word case list. @tyep order: integer Nr#)r*r#s r, set_orderzStemmedSynWord.set_orderms  r.c|jS)r8r9r1s r, get_orderzStemmedSynWord.get_orderu zr.c||_dS)z Set the word case score @param _score: the _score of the stemmed word in the word case list. @tyep _score: integer Nr()r*_scores r, set_scorezStemmedSynWord.set_score}s  r.c|jS)z get the case word score @param _score: the _score of the stemmed word in the word case list. @tyep _score: integer r?r1s r, get_scorezStemmedSynWord.get_scorer=r.c|jS)zy Get the syntax form of the input word @return: the given syntax. @rtype: unicode string r r1s r, get_syntaxzStemmedSynWord.get_syntaxs {r.c||_dS)z} Set the syntax word @param newsyntax: the new given syntax. @type newsyntax: unicode string NrE)r* newsyntaxs r, set_syntaxzStemmedSynWord.set_syntaxs ! r.cHd|j|g|_dS)zq Add a new tag to syntax field @param tag: the new added tag. @tyep tag: unicode :N)joinr )r*tags r, add_syntaxzStemmedSynWord.add_syntaxs" xxc 233 r.c|jS)z} Get the semantic form of the input word @return: the given semantic. @rtype: unicode string r!r1s r, get_semanticzStemmedSynWord.get_semantics }r.c||_dS)z Set the semantic word @param newsemantic: the new given semantic. @type newsemantic: unicode string NrP)r* newsemantics r, set_semanticzStemmedSynWord.set_semantics % r.rc||j|<dS)z Add next word position number, if the word is realted @param next: the next of the stemmed word in the word case list. @tyep next: integer N)r$r*nextwweights r,add_nextzStemmedSynWord.add_nexts" %r.cNt|jSz get the next positions. @param next: the next of the stemmed word in the word case list. @tyep next: integer )listr$keysr1s r,get_nextzStemmedSynWord.get_nexts DINN$$%%%r.c|s |jikStt|j|S)a get True if current word has next relations. If The next is given, it returns if the next has relation with current. @param next: a stemmedsynword as next of the current word case. if it's None, the fucntion return if there are relation. @type next: stemmedsynword. @return: if the word has next relations @rtype: boolean )r$boolsetr] intersectionr*rWs r,has_nextzStemmedSynWord.has_nextsK C9? "DINN,,--::5AABB Br.ct|s |jikS|j|S)a get True if current word has previous relations. If The previous is given, it returns if the previous has relation with current. @param previous: a stemmedsynword as previous of the current word case. if it's None, the fucntion return if there are relation. @type previous: stemmedsynword. @return: if the word has previous relations @rtype: boolean )r%has_keyr<)r*r%s r, has_previouszStemmedSynWord.has_previouss; ?=B& &=((););)=)=>> >r.c||j|<dS)z Add the previous position of the related word. @param previous: the previous of the stemmed word in the word case list. @tyep previous: integer N)r%r*r%rXs r, add_previouszStemmedSynWord.add_previouss #) hr.c|jS)z) get the selection rule. r)r1s r,get_rulezStemmedSynWord.get_rules yr.c||_dS)z( set the selection rule Nrl)r*r)s r,set_rulezStemmedSynWord.set_rules r.cNt|jS)z9 get all privous word list of relations. )r\r%r]r1s r, get_previouszStemmedSynWord.get_previouss DM&&(()))r.c8|j|dS)z get the previous relation between the current case and the previous given by order @param previous_order: the previous of the stemmed word in the word case list. @tyep previous_order: integer r)r%get)r*previous_orders r,get_previous_relationz$StemmedSynWord.get_previous_relations}  333r.c8|j|dS)z get the next relation between the current case and the next given by order @param next_order: the next of the stemmed word in the word case list. @tyep next_order: integer r)r$rs)r* next_orders r,get_next_relationz StemmedSynWord.get_next_relationsy}}Z+++r.c||j|<dS)z Add next word position number, if the word is semanticly related @param next: the next of the stemmed word in the word case list. @tyep next: integer N)r&rVs r, add_sem_nextzStemmedSynWord.add_sem_next s & er.cNt|jSr[)r\r&r]r1s r, get_sem_nextzStemmedSynWord.get_sem_nexts DM&&(()))r.c|jikS)z get True if current word has semantic previout relations @return: if the word has previous relations @rtype: boolean r'r1s r,has_sem_previouszStemmedSynWord.has_sem_previouss  B&&r.c|s |jikStt|j|S)z get True if current word has semantic next relations @return: if the word has next relations @rtype: boolean )r&r`rar]rbrcs r, has_sem_nextzStemmedSynWord.has_sem_next%sK  '=B& &C **,,--::5AABBBr.c||j|<dS)z Add the previous position of the semantic related word. @param previous: the previous of the stemmed word in the word case list. @tyep previous: integer Nr~ris r,add_sem_previouszStemmedSynWord.add_sem_previous0s(.(###r.cNt|jS)z Add a position number in the case word list. @param previous: the previous of the stemmed word in the word case list. @tyep previous: integer )r\r'r]r1s r,get_sem_previouszStemmedSynWord.get_sem_previous:s!D%**,,---r.cd|_dS)z Add a new tag to syntax field as foced case @param tag: the new added tag. @tyep tag: unicode TNrr1s r, forced_casezStemmedSynWord.forced_caseCs !%r.cd|_dS)z Add a new tag to syntax field as foced word type (noun, verb) @param tag: the new added tag. @tyep tag: unicode TNr"r1s r,force_wordtypezStemmedSynWord.force_wordtypeLs $r.c|jSz~ verify if the word is a forced word type (noun, verb) @return: True/False @tyep tag: Boolean rr1s r,is_forced_casezStemmedSynWord.is_forced_caseTs $$r.c|jSrrr1s r,is_forced_wordtypez!StemmedSynWord.is_forced_wordtype\s ##r.c|sd|vrdS|tjvrdSdS)j Return True if the word is a Rafe3. @return: Rafe3. @rtype: True/False uكان و أخواتهاTF) has_encleticget_tagsr6rKanaSisters_LISTr1s r,rzStemmedSynWord._is_kana_rafe3gsT!!## #t}}664  ! ! # #y'A A A4ur.cd|vsd|vr|sdSdS)u Return True if the word is a nominal addition اسم إضافة مثل نحو ومعاذ @rtype: True/False uاسم إضافةu إضافةTF)rrr1s r,rzStemmedSynWord._is_additionvsE !DMMOO 3 3} 7W7Waeararatat7W4ur.c|pRd|vp;|p'|p| S)uw Return True if the word can be added like مديرو and not مديرون @rtype: True/False u لايضاف) is_additionr is_defined is_tanwinrr1s r, need_additionzStemmedSynWord.need_additions !!O/T]]__*L+NPTP_P_PaPa+Nfjftftfvfv+Nz~{L{L{N{N&O Or.c|r*|r|rdSdS)ui Return True if the word can be an addition مضاف إليه @rtype: True/False TF) is_majrouris_adjrr1s r,is_additionablezStemmedSynWord.is_additionablesC ??   ;;== T__%6%6 tur.cVd|vr|syd|_d|vr|xjdz c_d|vr|xjdz c_d|vr|xjdz c_|jS) u Return int code of verbal factor. the inflected cases are coded in binary like not defined : 0 00000 factor : 1 00001 Rafe3 : 2 00010 Naseb : 4 00100 Jar : 8 01000 Jazem : 16 10000 active : 32 100000 this codification allow to have two verb factor for the same case, like feminin plural which ahve the same mark for Nasb and jar هذا الترميز يرمز حالتين في وقت واحد النصب والجر @return: verbal factor state numeric code @rtype: uفعلrرافعناصبuجازم)get_object_typerr get_actionr1s r,__get_verbal_factorz"StemmedSynWord.__get_verbal_factors2 ,,.. . .t7H7H7J7J .%&D "doo////&&!+&&doo////&&!+&&doo////&&",&&%%r.cd|vrd|_d|vr5|s|xjdz c_n|xjdz c_d|vr5|s|xjdz c_n|xjdz c_d|vr&d|vr|xjd z c_d |vr$|s|xjd z c_d |vrd |_|jS) ui Return int code of nominal factor. the inflected cases are coded in binary like not defined : 0 00000 factor : 1 00001 Rafe3 : 2 00010 Naseb : 4 00100 Jar : 8 01000 this codification allow to have two noun factor for the same case, like feminin plural which ahve the same mark for Nasb and jar هذا الترميز يرمز حالتين في وقت واحد النصب والجر @return: nominal factor state numeric code @rtype: uاسمrrrrrاسم إشارةuجرuجارضمير)rrrrrr1s r,__get_nominal_factorz#StemmedSynWord.__get_nominal_factorse0 ,,.. . .&'D #doo////((**1++q0+++++q0++doo////))++1++q0+++++q0++#t}}66t}}..++q0++ )) ) )$2C2C2E2E )  # #q ( # # $--// ) )&+D #&&r.c.d|vS)u Return True if the word is jonction عطف منفصل. @return: is jonction. @rtype: True/False uمتبع)rr1s r, is_jonctionzStemmedSynWord.is_jonctionst0000r.c.d|vS)u Return True if the word is substitute مبدل منه. @return: is substituted. @rtype: True/False rrr1s r,is_substitutedzStemmedSynWord.is_substituteds% 77r.c.d|vS)u Return True if the word is confirmation توكيد. @return: is confirmation. @rtype: True/False u توكيدrr1s r,is_confirmationzStemmedSynWord.is_confirmations00r.ctj|rdS|rd|vrdSdS)zk Return True if the word has break. @return: is break. @rtype: True/False TbreakF)r StemmedWordis_break is_pounctrr1s r,rzStemmedSynWord._is_break sT  " + +D 1 1 4^^   'T]]__"<"<45r.c|jS)zj Return True if the word has break. @return: is break. @rtype: True/False )rr1s r,rzStemmedSynWord.is_break!s ~r.c|jS)z Return True if the word mark the begin of next sentence. @return: direct initial. @rtype: True/False ) tag_initialr1s r, is_initialzStemmedSynWord.is_initial3s r.cd|vp+d|vpd|vS)z Return True if the word has the 3rd person. @return: has the 3rd persontense. @rtype: True/False ruاسم موصولrrr1s r, is_pronounzStemmedSynWord.is_pronoun;s_ dmmoo-E1E1XE]quyvCvCvEvE^E Er.c.d|vS)z7 return True if the word accept Tanwin uينونrr1s r,canhave_tanwinzStemmedSynWord.canhave_tanwinDsdmmoo--r.c6t|jdz dzS)zy Return True if the word is a Rafe3 of verb @return: Rafe3 of verb. @rtype: True/False rr`rr1s r, is_verb_rafe3zStemmedSynWord.is_verb_rafe3J D*Q.2333r.c6t|jdz dzS)zr Return True if the word is a Naseb of verb. @return: Naseb. @rtype: True/False rrrr1s r, is_verb_nasebzStemmedSynWord.is_verb_nasebRrr.c6t|jdz dzS)zj Return True if the word is a Jazem. @return: Jazem. @rtype: True/False rrrr1s r,is_jazemzStemmedSynWord.is_jazemZs D*R/!3444r.c2t|jdkS)zy Return True if the word is a a jobless of verb. @return: Njobless. @rtype: True/False rrr1s r,is_verb_jobless_factorz%StemmedSynWord.is_verb_jobless_factoras D*a/000r.c*t|jS)z Return True if the word is a nominal factor. @return: is a nominal factor. @rtype: True/False r`rr1s r,is_nominal_factorz StemmedSynWord.is_nominal_factorisD+,,,r.c6t|jdz dzS)rrrr1s r,is_rafe3zStemmedSynWord.is_rafe3r D+a/!3444r.c6t|jdz dzS)zz Return True if the word is a Naseb of noun. @return: Naseb of noun. @rtype: True/False rrrr1s r,is_nasebzStemmedSynWord.is_nasebzrr.c`t|jdz dzo| S)zf Return True if the word is a Jar. @return: Jar. @rtype: True/False rr)r`rrr1s r,is_jarzStemmedSynWord.is_jars3 D+a/!344PT=N=N=P=P9PPr.cB|tjvS)z Return True if the word is a condition factor. @return: condition factor. @rtype: True/False ) get_originalrCONDITION_FACTORSr1s r,is_condition_factorz"StemmedSynWord.is_condition_factors   ""i&AAAr.c|jS)r)rr1s r, is_kana_rafe3zStemmedSynWord.is_kana_rafe3s ""r.c|jS)u Return True if the word is a Addition اسم إضافة مثل نحو ومعاذ. @return: is a addition. @rtype: True/False )rr1s r,rzStemmedSynWord.is_additions   r.c*t|jS)z Return True if the word is a verbal factor. @return: is a verbal factor. @rtype: True/False rr1s r,is_verbal_factorzStemmedSynWord.is_verbal_factors D*+++r.ct|| }t|| }|o|S)z Return True if the current word is equal in definition with otherword. @return: equal or not. @rtype: True/False )rrr)r* otherwordeq_defndeq_tnwns r, eq_definedzStemmedSynWord.eq_defineds_ 4??,,y/C/C/E/EFFF4>>++i.A.A.C.CDDD#G#r.c|r|pO|o|p'|o|S)u Return True if the current word is equal in case إعراب with otherword. @return: equal or not. @rtype: True/False )r is_mansoub is_marfou3r*rs r,eq_casezStemmedSynWord.eq_casesm //##> (<(<(>(>=!!r.c|r|p'|o|S)u Return True if the current word is equal in gender النوع ذكر أنثى with otherword. @return: equal or not. @rtype: True/False ) is_feminin is_masculinrs r, eq_genderzStemmedSynWord.eq_gendersL //##> (<(<(>(>?!!=i&;&;&=&= r.c|r|pO|o|p'|o|S)u Return True if the current word is equal in number العدد مفرد مثنى جمع with otherword. @return: equal or not. @rtype: True/False ) is_pluralis_dual is_singlers r, eq_numberzStemmedSynWord.eq_numbersk ..""> O$"2"2"4"4 OT__=N=N O!!## O,0OO,=,= O~&&u{33 O!%!4U^!C((55 O!%!4U^!C(():5;)FGG O!%!4U^!C((UZ)?@@ O!%!4U^!CEJ!N O O O O O O O O O O O Or.cd}|j|r6t|j||dz|j|<||j|z }|j|r9t|j||dz|j|<||j|dzz }|t |jdzt |jdzzz }|t |jdzt |jdzzz}|ttj |j dzdz}t|j ||_ dS)aN Recalculate score according to previous node. @param previous_case_position: the pervious position in previos table @type previous_case_position: integer @param previous_score: the word case previous score @type previous_score: integer. @return: nothing @rtype: void rrN) r'rfmaxr%lenr&r$roundmathlogfreqr()r*previous_case_positionprevious_scorer(s r,recalculate_scorez StemmedSynWord.recalculate_scoresX   $ $%; < < @8; M0 1>!3C9E9ED 4 5 d'(>? ?E = !7 8 8 >58 M0 1>13D6F6FDM0 1 dm$:1$<= =E #d'((*C ,>,>q,@@@ C &&q(TY)99: %1--q111 $*e,, r.c8d|_t|jdz|_|xjt|jdzz c_|xjt|jdzz c_|xjt|jdzz c_|xjt tj|j dzdz c_|xj| dzz c_| dt|jzdS) zR Recalculate score. @return: nothing @rtype: void r drr2zScr:N) r(rr$r%r'r&r r r r rrNstrr1s r,calculate_scorezStemmedSynWord.calculate_scores  $)nnR'  s4=))",,  s4,--c11  s4=))#--  uTXdik22A666  t''))",,  s4:./////r.cT|j|j|j|jd}|j}||d<|S)z/ get dictionary of attributes )SPSNPNr )r'r&r%r$r )r*r ret_dicts r,get_dictzStemmedSynWord.get_dict1s=(]MI=#r.cd|jdd|jdd}|jD]}|d|d|j|dz }|dS) z/ represent the class as string 'r#z':rz, z 'z': utf8)r r]encode)r*textks r,__repr__zStemmedSynWord.__repr__@s#'-"8"8"8 k"""$##%% A AA DAAAt}Q/?/?/?@ @DD{{6"""r.)Nr )r)N)I__name__ __module__ __qualname____doc__r-r2r4r6r:r<rArCrFrIrNrQrTrYr^rdrgrjrmrorqrurxrzr|rrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr#r0r.r,rr!so""""H       $ $ $!!!444%%%""""&&&CCCC????"))))  *** 444,,,&&&&***''' C C C C.......%%%$$$%%%$$$   OOO   %&%&%&P1'1'1'f1118881110   $   EEE... 444444555111---555555QQQBBB###!!!,,,$$$>>>      LLLOOO&%-%-%-T000(#####r.rc4tdi}ddddddddd ddd d ddd }tj|}t|}t|t|t|d S)z main test testu الحياةuالْحَيَاةُuالr uحياةuُuتعريف::مرفوع*ru Noun:مصدرuحَيَاةٌ)wordr procleticprefixstemsuffixencletictagsr roottemplatetyperr r!N)printrrrrr)rdictstmwrds r,mainlyr7Ks &MMM E%0 .$)'E*  #E * *E E " "F &//   &     &MMMMMr.__main__)r'syspathappendr operatorrpyarabic.arabyr qalsadir rrrrr7r$NOMINAL_FACTOR_LISTr0r.r,r@s>   $$$$$$ $$$$$$$ Dg #g #g #g #g #g)5g #g #g #T@ z FHHH !!!!s 0A77A;