a Kb2@sdZddlZddlZddlZddlZddlZddlZddlmZddl m Z ddl m Z m Z ddlmZdagdZddd Zd d ZGd d d e ZGdddZedkrddlmZmZddZeeeZdS)z; Classifiers that make use of the external 'Weka' package. N)stdin) ClassifierI) config_javajava)DictionaryProbDist).z/usr/share/wekaz/usr/local/share/wekaz /usr/lib/wekaz/usr/local/lib/wekacCst|dur|atdurt}dtjvr:|dtjd|D]`}tjtj|dr>tj|dat t}|rt dtd|dn t dtt tq>tdurt ddS) NZWEKAHOMErzweka.jarz [Found Weka: z (version z)]z[Found Weka: %s]zUnable to find weka.jar! Use config_weka() or set the WEKAHOME environment variable. For more information about Weka, please see https://www.cs.waikato.ac.nz/ml/weka/) r_weka_classpath _weka_searchosenvironinsertpathexistsjoin_check_weka_versionprint LookupError) classpathZ searchpathr versionr1lib/python3.9/site-packages/nltk/classify/weka.py config_weka"s&   rc Cszt|}Wn$ttfy&YnYdS0zBz|dWW|StyjYW|dS0W|n |0dS)Nzweka/core/version.txt)zipfileZipFile SystemExitKeyboardInterruptreadcloseKeyError)ZjarzfrrrrCs   rc@sbeZdZddZddZddZddZd d Zd d Zd ddddddZ e dgdfddZ dS)WekaClassifiercCs||_||_dSN) _formatter_model)self formattermodel_filenamerrr__init__TszWekaClassifier.__init__cCs||gdS)N)-p0z -distribution_classify_manyr$ featuresetsrrrprob_classify_manyXsz!WekaClassifier.prob_classify_manycCs||ddgS)Nr(r)r*r,rrr classify_many[szWekaClassifier.classify_manyc Cs"tt}ztj|d}|j||dd|jd|g|}t |t t j t j d\}}|r~|s~d|vrrt dn t d|||tjd Wt|D]}ttj||qt|S]}ttj||qt|n2t|D]}ttj||qt|0dS) Nz test.arff!weka.classifiers.bayes.NaiveBayesz-lz-T)rstdoutstderrzIllegal options: -distributionzOThe installed version of weka does not support probability distribution output.z"Weka failed to generate output: %s )rtempfilemkdtempr r rr"writer#rr subprocessPIPE ValueErrorparse_weka_outputdecoderencodingsplitlistdirremovermdir) r$r-optionstemp_dirZ test_filenamecmdr1r2frrrr+^sH    zWekaClassifier._classify_manycCs2ddtd|D}tt|j|}t|S)NcSsg|]}|rt|qSr)stripfloat).0vrrr z:WekaClassifier.parse_weka_distribution..z[*,]+)rer=dictzipr"labelsr)r$sZprobsrrrparse_weka_distributionsz&WekaClassifier.parse_weka_distributioncst|D]&\}}|dr||d}q0q|dgdkrZdd|ddDS|dgdkrfdd|ddDStd |drd d|DS|dd D] }t|qtd |ddS) Ninst#r)rQactual predictederrorZ predictioncSs*g|]"}|r|dddqS):rEr=rGlinerrrrIrJz4WekaClassifier.parse_weka_output..rW)rQrRrSrTZ distributioncs&g|]}|r|dqS))rErPr=rYr$rrrIsz^0 \w+ [01]\.[0-9]* \?\s*$cSs g|]}|r|dqS)rWrXrYrrrrIrJ zRUnhandled output format -- your version of weka may not be supported. Header: %s) enumeraterE startswithr=rKmatchrr9)r$linesirZrr\rr:s&    z WekaClassifier.parse_weka_outputr0zweka.classifiers.trees.J48z#weka.classifiers.functions.Logisticzweka.classifiers.functions.SMOzweka.classifiers.lazy.KStarzweka.classifiers.rules.JRip) naivebayesC4.5Zlog_regressionZsvmZkstarZripperrcTc Cs8tt|}t}ztj|d}|||||j vrJ|j |} n ||j vr^|} n t d|| d|d|g} | t |7} |rt j} nd} t| t| dt||Wt|D]} ttj|| qt|S]} ttj|| qt|n2t|D]} ttj|| q t|0dS)Nz train.arffzUnknown classifier %sz-dz-t)rr1)rARFF_Formatter from_trainr4r5r r rr6_CLASSIFIER_CLASSvaluesr9listr7r8rrr r>r?r@) clsr&r- classifierrAquietr%rBZtrain_filenameZ javaclassrCr1rDrrrtrains8          zWekaClassifier.trainN) __name__ __module__ __qualname__r'r.r/r+rPr:rg classmethodrmrrrrr Ss$,1 r c@sVeZdZdZddZddZddZdd Zed d Z d d Z dddZ ddZ dS)rez Converts featuresets and labeled featuresets to ARFF-formatted strings, appropriate for input into Weka. Features and classes can be specified manually in the constructor, or may be determined from data using ``from_train``. cCs||_||_dS)a) :param labels: A list of all class labels that can be generated. :param features: A list of feature specifications, where each feature specification is a tuple (fname, ftype); and ftype is an ARFF type string such as NUMERIC or STRING. N)_labels _features)r$rNfeaturesrrrr'szARFF_Formatter.__init__cCs|||S)zBReturns a string representation of ARFF output for the given data.)header_section data_section)r$tokensrrrformat szARFF_Formatter.formatcCs t|jS)zReturns the list of classes.)rirrr\rrrrNszARFF_Formatter.labelscCs0t|dst|d}||||dS)z.Writes ARFF data to a file for the given data.r6wN)hasattropenr6rxr)r$Zoutfilerwrrrr6s  zARFF_Formatter.writecCsdd|D}i}|D]\}}|D]\}}tt|trBd}nFtt|tttfr\d}n,tt|trpd}n|dur|q&n td|||||krtd||||<q&qt |}t ||S) z Constructs an ARFF_Formatter instance with class labels and feature types determined from the given data. Handles boolean, numeric and string (note: not nominal) types. cSsh|] \}}|qSrr)rGtoklabelrrr !rJz,ARFF_Formatter.from_train..z {True, False}ZNUMERICSTRINGNzUnsupported value type %rzInconsistent type for %s) items issubclasstypeboolintrFstrr9getsortedre)rwrNrtr|r}fnamefvalftyperrrrfs$     zARFF_Formatter.from_traincCsTddt}|d7}|jD]\}}|d||f7}q|ddd|jf7}|S)z#Returns an ARFF header as a string.z3% Weka ARFF file % Generated automatically by NLTK z%% %s z@RELATION rel z@ATTRIBUTE %-30r %s z@ATTRIBUTE %-30r {%s} z-label-,)timectimersrrr)r$rOrrrrrru9s zARFF_Formatter.header_sectionNc Cs|dur|ot|dttf}|s0dd|D}d}|D]B\}}|jD] \}}|d|||7}qF|d||7}q8|S)a Returns the ARFF data section for the given data. :param tokens: a list of featuresets (dicts) or labelled featuresets which are tuples (featureset, label). :param labeled: Indicates whether the given tokens are labeled or not. If None, then the tokens will be assumed to be labeled if the first token's value is a tuple or list. NrcSsg|] }|dfqSr!r)rGr|rrrrI]rJz/ARFF_Formatter.data_section..z @DATA z%s,z%s ) isinstancetuplerirs _fmt_arff_valr)r$rwZlabeledrOr|r}rrrrrrvNs  zARFF_Formatter.data_sectioncCs@|dur dSt|ttfr"d|St|tr4d|Sd|SdS)N?z%sz%r)rrrrF)r$rrrrrhs zARFF_Formatter._fmt_arff_val)N) rnrorp__doc__r'rxrNr6 staticmethodrfrurvrrrrrres   re__main__)binary_names_demo_features names_democCstd|dS)Nz/tmp/name.modelrd)r rm)r-rrrmake_classifiervsr)N)rr rKr7r4rrsysrZnltk.classify.apirZnltk.internalsrrZnltk.probabilityrrr rrr rernZnltk.classify.utilrrrrkrrrrs*    !$}