'[fxddlZddlmZddlmZddlmZGddeZdZe dkr edSdS) N) defaultdict)reduce) CorpusReadercxeZdZdZejdZedZd fd Z d dZ d dZ d d Z d Z xZS) LinThesaurusCorpusReaderzEWrapper for the LISP-formatted thesauruses distributed by Dekang Lin.z \("?([^"]+)"? \(desc [0-9.]+\).+c*ttS)z6Factory for creating defaultdict of defaultdict(dict)s)rdictJ/var/www/piapp/venv/lib/python3.11/site-packages/nltk/corpus/reader/lin.py__defaultdict_factoryz.LinThesaurusCorpusReader.__defaultdict_factorys4   r czt|dttj|_||_|ddD]\}}}t|5}d}|D]}| }|r#tj d|} d};|dkrd}D| d} t| dkr;| \} } t| |j|| | d < d d d n #1swxYwYd S) a Initialize the thesaurus. :param root: root directory containing thesaurus LISP files :type root: C{string} :param badscore: the score to give to words which do not appear in each other's sets of synonyms :type badscore: C{float} z sim[A-Z]\.lspT)include_encodinginclude_fileidz\1Fz)) "N)super__init__rr._LinThesaurusCorpusReader__defaultdict_factory _thesaurus _badscoreabspathsopenstrip_key_resubsplitlenfloat)selfrootbadscorepathencodingfileidlin_filefirstlinekey split_linengramscore __class__s r rz!LinThesaurusCorpusReader.__init__s /000%&>&TUU!&*mm!$'4' '    "D(Fd x$D::<BB5$OO % $&*ZZ%5%5 z??a//+5LE5MR %NNDOF3C8S9I9IJ                 s5B-D//D3 6D3 Nckr|rdSdjDS|r5j|vrj|njSfdjDS)a Returns the similarity score for two ngrams. :param ngram1: first ngram to compare :type ngram1: C{string} :param ngram2: second ngram to compare :type ngram2: C{string} :param fileid: thesaurus fileid to search in. If None, search all fileids. :type fileid: C{string} :return: If fileid is specified, just the score for the two ngrams; otherwise, list of tuples of fileids and scores. ?cg|]}|dfS)r1r ).0fids r z7LinThesaurusCorpusReader.similarity..Qs<<.Zsg    &)=f)EEE!OC08@@!%    r )_fileidsrr)r"r7r8r's``` r similarityz#LinThesaurusCorpusReader.similarity?s V   =s<.usF07==??@r )rr=r9r"r-r's`` r scored_synonymsz(LinThesaurusCorpusReader.scored_synonymsfs^  ?6*517799 9"m r c||r%j|SfdjDS)a Returns a list of synonyms for the current ngram. :param ngram: ngram to lookup :type ngram: C{string} :param fileid: thesaurus fileid to search in. If None, search all fileids. :type fileid: C{string} :return: If fileid is specified, list of synonyms; otherwise, list of tuples of fileids and lists, where inner lists contain synonyms. c^g|])}|j|f*Sr )rkeysr>s r r5z5LinThesaurusCorpusReader.synonyms..sF07<<>>?r )rrCr9r?s`` r synonymsz!LinThesaurusCorpusReader.synonymszs^  ?6*516688 8"m r c:tfdjdS)z Determines whether or not the given ngram is in the thesaurus. :param ngram: ngram to lookup :type ngram: C{string} :return: whether the given ngram is in the thesaurus. c&|pj|vSN)r)accumr'r-r"s r z7LinThesaurusCorpusReader.__contains__..s%"MET_V5L,Lr F)rr9)r"r-s``r __contains__z%LinThesaurusCorpusReader.__contains__s/ M M M M M M    r )rrG)__name__ __module__ __qualname____doc__recompiler staticmethodrrr:r@rDrJ __classcell__)r/s@r rrsOObj<==G!!\!""""""H%%%%N(&        r rc4ddlm}d}d}td|zt||td|zt||td|zt||d td|zt||d td |d |d t|||dS) Nr) lin_thesaurusbusiness enterprisezGetting synonyms for zGetting scored synonyms for z5Getting synonyms from simN.lsp (noun subsection) for zsimN.lsp)r'zSimilarity score for z and :) nltk.corpusrTprintrDr@r:)thesword1word2s r demor]s$111111 E E !E )*** $--   (5 0111 $  u % %&&& AE IJJJ $--j- 1 1222 AE IJJJ $--j- 1 1222 6% 6 6e 6 6 6777 $//% ' '(((((r __main__) rO collectionsr functoolsrnltk.corpus.readerrrr]rKr r r rbs ######++++++J J J J J |J J J d)))* zDFFFFFr