-iY=dZddlmZmZmZmZddlZddlZddlZe dkrej dddl Z dZddlmZdddd d d d ZGd dZdZe dkr edSdS)z~ Arabic Word frequency Dictionary Class from Arramooz Al Waseet. Used in multiporpus morpholigical treatment )absolute_importprint_functionunicode_literalsdivisionN__main__z../zdata/wordfreq.sqlite)id vocalized unvocalized word_typefreq future_typecFeZdZdZedfdZdZdZdZdZ d d Z d d Z d S) WordFreqDictionaryuz Arabic dictionary Class Used to allow abstract acces to lexicon of arabic language, can get indexed and hashed entries from the basic lexicon add also, support to extract attributtes from entries Example: >>> mydict = WordFreqDictionary('wordfreq') >>> wordlist = [u"صلاة", u'كرة', u"قَطَرً", u"أَرْض"] >>> for word in wordlist: >>> print("word freq", mydict.get_freq(word)) >>> idlist = mydict.lookup(word) >>> for row in idlist: >>> row = dict(row) >>> print('frequency', row['freq']) >>> print(repr(row).decode("unicode-escape")) صلاة 0 word freq 0 كرة 0 word freq 0 قَطَرً 2 [(984, u'قَطَر', u'قطر', u'noun_prop', 154772, u'َ'), (13874, u'قَطَر', u'قطر', u'verb', 1859, u'َ')] word freq 154772 frequency 154772 {'vocalized': u'قَطَر', 'word_type': u'noun_prop', 'unvocalized': u'قطر', 'future_type': u'َ', 'freq': 154772, 'id': 984} frequency 1859 {'vocalized': u'قَطَر', 'word_type': u'verb', 'unvocalized': u'قطر', 'future_type': u'َ', 'freq': 1859, 'id': 13874} أَرْض 1 [(349, u'أَرْض', u'أرض', u'noun', 389839, u'َ')] word freq 389839 frequency 389839 {'vocalized': u'أَرْض', 'word_type': u'noun', 'unvocalized': u'أرض', 'future_type': u'َ', 'freq': 389839, 'id': 349} rci|_||_||_i|_|jD]}|j|}||j|<||_t tdr tj}nAtj tj t}tj |t}tj |rzd|zdz} t#j|dd|_t"j|j_|j|_nc#t.$rt1d|YnGwxYwt1dd |d tjgd d } || d S)zw initialisation of dictionary from a data dictionary, create indexes to speed up the access. frozenzfile:z?mode=roTF)uricheck_same_threadz(Fatal Error Can't find the database file zInexistant Filez current dir utf8rN) dictionary attrib_index key_attributeattrib_num_indexkeys table_namehasattrsysprefixospathdirnamerealpath__file__join FILE_DB_FREQexistssqliteconnect db_connectRow row_factorycursorIOErrorprintcurdirencodecreate_table_index) selfr rrkvalbase file_pathfile_uri index_fields ^/var/www/html/speakWrite/venv/lib/python3.11/site-packages/arramooz/wordfreqdictionaryclass.py__init__zWordFreqDictionary.__init__Ls(+ " "'')) + +A#A&C)*D !# & &$ 3 ! ! AZDDW__RW%5%5h%?%?@@DGLL|44 7>>) $ $ )y(3H 9$*N8Y^$`$`$`17 + $ 6 6 8 8  M M M@)LLLLL M 499/O Y ) ) )$   ,,,,,sEE54E5cnt|dr"|jr|jdSdSdS)zH Delete instance and close database connection r.N)r!r.close)r7s r>__del__zWordFreqDictionary.__del__}sN 4 % % ( (%%''''' ( ( ( (cd|jd|d} |j||jrdSdS#tj$rYdSwxYw)z create the database index if not exists @param index_field: the given to be indexed field @type index_field: text @return: void @rtype: void z&create index if not exists myindex on z ()TFN)r r1executer,OperationalError)r7r=sqls r>r6z%WordFreqDictionary.create_table_indexsw &  K   $ $ ${ t  &   55 s!5AAcd|jd|d} |j||jr|jSn#tj$rYdSwxYwdS)z Get dictionary entry by id from the dictionary @param idf :word identifier @type idf: integer @return: all attributes @rtype: dict select * FROM z WHERE id=''F)r r1rFfetchallr,rG)r7idfrHs r>get_entry_by_idz"WordFreqDictionary.get_entry_by_ids6:___cccJ . K   $ $ ${ .{++--- .&   55  usA AAcd|j|fz} |j||jr|jD] }||cSn%#tj$rt dYdSwxYwdS)a Get attribute value by id from the dictionary @param idf :word identifier @type idf: integer @param attribute :the attribute name @type attribute: unicode @return: The attribute value @rtype: mix. zselect * FROM %s WHERE id='%d'z,Fatal Error on query: wordfreq dict error 12F)r r1rFr,rGr3)r7rM attributerHrows r>get_attrib_by_idz#WordFreqDictionary.get_attrib_by_idsX2T_c4JJ + K   $ $ $ { +;++C N*** &    @ A A A55 usAA%$A%cg}tj|ddr |dd}tjtjtjztj|}|dkrd|jd|d}n%d|jd|d}|dkr|d z }n |d kr|d z } |j||jr|jD]}| ||S#tj $rtd gcYSwxYw) u look up for all word forms in the dictionary, according to word_type - 'verb': lookup for verb only. - 'noun': look up for nouns. - 'unknown': the word is not alayzed, then search for unvocalized word. - '': look for voaclize word without type Example: >>> mydict = WordFreqDictionary('wordfreq') >>> wordlist = [u"صلاة", u'كرة', u"قَطَرً", u"أَرْض"] >>> for word in wordlist: >>> print("word freq", mydict.get_freq(word)) >>> idlist = mydict.lookup(word) >>> for row in idlist: >>> row = dict(row) >>> print('frequency', row['freq']) >>> print(repr(row).decode("unicode-escape")) صلاة 0 word freq 0 كرة 0 word freq 0 قَطَرً 2 [(984, u'قَطَر', u'قطر', u'noun_prop', 154772, u'َ'), (13874, u'قَطَر', u'قطر', u'verb', 1859, u'َ')] word freq 154772 frequency 154772 {'vocalized': u'قَطَر', 'word_type': u'noun_prop', 'unvocalized': u'قطر', 'future_type': u'َ', 'freq': 154772, 'id': 984} frequency 1859 {'vocalized': u'قَطَر', 'word_type': u'verb', 'unvocalized': u'قطر', 'future_type': u'َ', 'freq': 1859, 'id': 13874} أَرْض 1 [(349, u'أَرْض', u'أرض', u'noun', 389839, u'َ')] word freq 389839 frequency 389839 {'vocalized': u'أَرْض', 'word_type': u'noun', 'unvocalized': u'أرض', 'future_type': u'َ', 'freq': 389839, 'id': 349} @param text:vocalized word. @type text: unicode. @param word_type: the word type can take 'verb', 'noun', 'unknwon', ''. @type word_type: unicode. @return: list of dictionary entries IDs. @rtype: list. NunknownrJz WHERE unvocalized='rKz WHERE vocalized='verbz AND word_type='verb' nounz AND word_type!='verb' z>Fatal Error can't execute query: file: wordfrequencydictionary) araby is_harakaresubFATHAALEFr r1rFappendr,rGr3)r7textridlistrHrQs r>lookupzWordFreqDictionary.lookupsNX ?49 % % 9Dvek%*,ej$??  ! ! ! OOOTTT#CC " OOOTTT#CF""//f$$00  K   $ $ $ { ';''CMM#&&&&M&    R S S SIII sC D?Dcg}|||}|r+t|ddd}|SdS)u return the word frequency from the in the dictionary Example: >>> mydict = WordFreqDictionary('wordfreq') >>> wordlist = [u"صلاة", u'كرة', u"قَطَرً", u"أَرْض"] >>> for word in wordlist: >>> print("word freq", mydict.get_freq(word)) >>> idlist = mydict.lookup(word) >>> for row in idlist: >>> row = dict(row) >>> print('frequency', row['freq']) >>> print(repr(row).decode("unicode-escape")) صلاة 0 word freq 0 كرة 0 word freq 0 قَطَرً 2 [(984, u'قَطَر', u'قطر', u'noun_prop', 154772, u'َ'), (13874, u'قَطَر', u'قطر', u'verb', 1859, u'َ')] word freq 154772 frequency 154772 {'vocalized': u'قَطَر', 'word_type': u'noun_prop', 'unvocalized': u'قطر', 'future_type': u'َ', 'freq': 154772, 'id': 984} frequency 1859 {'vocalized': u'قَطَر', 'word_type': u'verb', 'unvocalized': u'قطر', 'future_type': u'َ', 'freq': 1859, 'id': 13874} أَرْض 1 [(349, u'أَرْض', u'أرض', u'noun', 389839, u'َ')] word freq 389839 frequency 389839 {'vocalized': u'أَرْض', 'word_type': u'noun', 'unvocalized': u'أرض', 'future_type': u'َ', 'freq': 389839, 'id': 349} @param text:vocalized word. @type text: unicode. @param word_type: the word type can take 'verb', 'noun', 'unknwon', ''. @type word_type: unicode. @return: word freq. @rtype: integer. rr)rbdictget)r7r`rrars r>get_freqzWordFreqDictionary.get_freq5sPRT9--  q ??&&va00DK1rCN)rS) __name__ __module__ __qualname____doc__WORDFREQ_DICTIONARY_INDEXr?rBr6rNrRrbrfrCr>rr(s!!F1J[h*-*-*-*-b((("4666pHHHHT222222rCrcXtd}gd}|D]}td||||}|D]V}t |}td|dtt |dWdS)z main test wordfreq)uصلاةuكرةu قَطَرًu أَرْضz word freq frequencyrzunicode-escapeN)rr3rfrbrdreprdecode)mydictwordlistwordrarQs r>mainlyruhs + +FGGGH66 k6??400111t$$ 6 6Cs))C +s6{ + + + $s))""#344 5 5 5 5 666rC)rj __future__rrrrr[r"r$rgr%r_sqlite3r,r*pyarabic.arabyrYrkrrurlrCr>rys,    (//%&     @ 6 6 6   FHHHHHrC