0PhA.dZddlZddlZddlZddlZddlmZddlmZddl m Z ddl m Z ddl mZddlZddlmZddlmZd d lmZmZmZmZd d lmZd d lmZd d lm Z m!Z!m"Z"m#Z#d dl$m%Z%d dl&m'Z'm(Z(m)Z)m*Z*ddl+m,Z,ddl-m.Z.gdZ/d(dZ0 d)dZ1dZ2dZ3dZ4dZ5GddZ6Gddee6edZ7d Z8Gd!d"e6eZ9d#Z:Gd$d%eeedZ;Gd&d'e9Zz)strip_accents_unicode..s(NNNa[5J15M5MNNNNr))encodeUnicodeEncodeErrorr:rjoin)s normalizeds r'r!r!ss},P *** PPP *6155 wwNN:NNNOOOOOPs>AAc~tjd|}|dddS)aTransform accentuated unicode symbols into ascii or nothing. Warning: this solution is only suited for languages that have a direct transliteration to ASCII symbols. Parameters ---------- s : str The string to strip. Returns ------- s : str The stripped string. See Also -------- strip_accents_unicode : Remove accentuated char for any unicode symbol. r7r4ignore)r:rr?decode)rB nkfd_forms r'r r s:(%fa00I   GX . . 5 5g > >>r)cjtjdtjd|S)a8Basic regexp based HTML / XML tag stripper function. For serious HTML/XML preprocessing you should rather use an external library such as lxml or BeautifulSoup. Parameters ---------- s : str The string to strip. Returns ------- s : str The stripped string. z <([^>]+)>)flags )recompileUNICODEsub)rBs r'r"r"s+ :l"* 5 5 5 9 9#q A AAr)c|dkrtSt|trtd|z|dSt |S)Nenglishznot a built-in stop list: %s)r isinstancestr ValueError frozenset)stops r'_check_stop_listrVsO y!! D#  7$>??? tr)ceZdZdZejdZdZddZdZ dZ dZ d Z d Z d Zd Zd ZdZdZdZdS)_VectorizerMixinz?Provides common code for text vectorizers (tokenization logic).z\s\s+cx|jdkr=t|d5}|}dddn #1swxYwYn|jdkr|}t|tr ||j|j}|tj urtd|S)a?Decode the input into a string of unicode symbols. The decoding strategy depends on the vectorizer parameters. Parameters ---------- doc : bytes or str The string to decode. Returns ------- doc: str A string of unicode symbols. filenamerbNfilez?np.nan is an invalid document, expected byte or unicode string.) inputopenreadrQbytesrFencoding decode_errornpnanrS)selfr%fhs r'rFz_VectorizerMixin.decodes : # #c4 Bggii               Z6 ! !((**C c5 ! ! ?**T]D,=>>C "&==Q  s=AANc |fd|D}|j\}}|dkr|}|dkrt|}|dz }ng}t|}|j}dj}t |t |dz|dzD]9} t || z dzD]!} |||| | | z":|S)zATurn tokens into a sequence of n-grams after stop words filteringNcg|]}|v| Sr+r+)r<wr1s r'r>z1_VectorizerMixin._word_ngrams..s#???A1J+>+>a+>+>+>r)rrJ) ngram_rangelistlenappendrArangemin) retokensr1min_nmax_noriginal_tokensn_original_tokens tokens_append space_joinnis ` r' _word_ngramsz_VectorizerMixin._word_ngramss  !???????F' u A::$Ozzo..  #O 4 4 #MMJ5#eai1BQ1F"G"GHH J J014q899JJA!M**_QQY-G"H"HIIIIJ r)c`|jd|}t|}|j\}}|dkrt |}|dz }ng}|j}t |t|dz|dzD]0}t ||z dzD]}|||||z1|S)z;Tokenize text_document into a sequence of character n-gramsrJr) _white_spacesrNrlrjrkrmrnro) re text_documenttext_lenrqrrr. ngrams_appendrwrxs r' _char_ngramsz_VectorizerMixin._char_ngrams s*..sMBB }%%' u A::-((F QJEEF uc%!)X\::;; 8 8A8a %> ! 4#566  ! # #,  $$r)c*t|jS)zBuild or fetch the effective stop words list. Returns ------- stop_words: list or None A list of stop words. )rVr1res r'get_stop_wordsz_VectorizerMixin.get_stop_wordsns 000r)ct|jt|ddkrdS t}|pdD]A}t |||}|D]}||vr||Bt|j|_|r$tjdt|z| S#t$rt|j|_YdSwxYw)aCheck if stop words are consistent Returns ------- is_consistent : True if stop words are consistent with the preprocessor and tokenizer, False if they are not, None if the check was previously performed, "error" if it could not be performed (e.g. because of the use of a custom preprocessor / tokenizer) _stop_words_idNr+z}Your stop_words may be inconsistent with your preprocessing. Tokenizing the stop words generated tokens %r not in stop_words.error) idr1getattrsetrkaddrwarningswarnsorted Exception)rer1 preprocesstokenize inconsistentrirptokens r'_check_stop_words_consistencyz._VectorizerMixin._check_stop_words_consistencyxs( do  '$0@$"G"G G G4 55L%2 0 0hhzz!}}5566#00EJ..$((///0#%T_"5"5D   "%+<$8$89 $# #   #%T_"5"5D 77  sBC#C('C(cRt|jr!tt|j|jS|}|jdkr"tt|j||jS|jdkr"tt|j||jS|jdkrc|}| }| |||tt|j |||j|Std|jz)a!Return a callable to process input data. The callable handles preprocessing, tokenization, and n-grams generation. Returns ------- analyzer: callable A function to handle preprocessing, tokenization and n-grams generation. )r,r0char)r.r/r0char_wbword)r.r-r/r0r1z.%s is not a valid tokenization scheme/analyzer) rr,rr2rFrrrrrrryrS)rerr1rs r'build_analyzerz_VectorizerMixin.build_analyzersB DM " " R8dmT[QQQ Q,,.. =F " "('  ]i ' '+'  ]f $ $,,..J++--H  . .z:x P P P("' %  @4=P r)c|j}|Et|trt|}t|tsHi}t |D]3\}}||||krd|z}t|4|}nt|}t|t|krtdtt|D])}||vr#dt||fz}t|*|stdd|_ t||_ dSd|_ dS)Nz Duplicate term in vocabulary: %rz%Vocabulary contains repeated indices.z/Vocabulary of size %d doesn't contain index %d.zempty vocabulary passed to fitTF) vocabularyrQrrr enumerate setdefaultrSvaluesrlrnfixed_vocabulary_dict vocabulary_)rervocabrxtmsgindicess r'_validate_vocabularyz%_VectorizerMixin._validate_vocabularys{_  !*c** 0#J// j'22 .%j11..DAq''1--22@1D(oo-3# j//1122w<<3z??22$%LMMMs://..A''O OOS)oo- (  C !ABBB%)D "#J//D   %*D " " "r)ct|ds*||jstdt |jdkrt ddS)z4Check if vocabulary is empty or missing (not fitted)rz!Vocabulary not fitted or providedrzVocabulary is emptyN)hasattrrrrrlrrSrs r'_check_vocabularyz"_VectorizerMixin._check_vocabularysot]++ J  % % ' ' ') J$%HIII t A % %233 3 & %r)cn|j\}}||kr$tdt|jzdS)z'Check validity of ngram_range parameterzOInvalid value for ngram_range=%s lower boundary larger than the upper boundary.N)rjrSrR)rerqmax_ms r'_validate_ngram_rangez&_VectorizerMixin._validate_ngram_rangesJ' u 5==ACFtGWCXCXY  =r)c|j|jtjd|j(t |jrtjd|jdkr/|j(t |jrtjd|jdkst |jr^|jtjd|j|jdkrtjd|jtjd dSdSdS) NzMThe parameter 'token_pattern' will not be used since 'tokenizer' is not None'zKThe parameter 'preprocessor' will not be used since 'analyzer' is callable'rrzJThe parameter 'ngram_range' will not be used since 'analyzer' is callable'rzFThe parameter 'stop_words' will not be used since 'analyzer' != 'word' (?u)\b\w\w+\bzIThe parameter 'token_pattern' will not be used since 'analyzer' != 'word'zEThe parameter 'tokenizer' will not be used since 'analyzer' != 'word') r-rrrr/rr,rjr1rs r'_warn_for_unused_paramsz(_VectorizerMixin._warn_for_unused_paramssS > %$*<*H M2      (Xdm-D-D ( M1      & & ,''- M1    =F " "ht}&=&= "* 2 ".&*::: 2~) 2 # "*)r)N)__name__ __module__ __qualname____doc__rKrLr{rFryrrrrrrrrrrrr+r)r'rXrXsIIBJx((M@<.8YYY6%%%(111%%%N111f+++<444(((((r)rXc.eZdZUdZidehdgdegdehdgdedd hd egd d gd ed gded gdedhed gded gdegdehdegde e de j e j jdgdd gdeddhd gdd gddZeed <d!d"d#d d$d d d d%d&d'd(d)dd$e jd*d+Zed$,d3d-Zed$,d3d.Zd/Zd3d0Zd1Zfd2ZxZS)4raConvert a collection of text documents to a matrix of token occurrences. It turns a collection of text documents into a scipy.sparse matrix holding token occurrence counts (or binary occurrence information), possibly normalized as token frequencies if norm='l1' or projected on the euclidean unit sphere if norm='l2'. This text vectorizer implementation uses the hashing trick to find the token string name to feature integer index mapping. This strategy has several advantages: - it is very low memory scalable to large datasets as there is no need to store a vocabulary dictionary in memory. - it is fast to pickle and un-pickle as it holds no state besides the constructor parameters. - it can be used in a streaming (partial fit) or parallel pipeline as there is no state computed during fit. There are also a couple of cons (vs using a CountVectorizer with an in-memory vocabulary): - there is no way to compute the inverse transform (from feature indices to string feature names) which can be a problem when trying to introspect which features are most important to a model. - there can be collisions: distinct tokens can be mapped to the same feature index. However in practice this is rarely an issue if n_features is large enough (e.g. 2 ** 18 for text classification problems). - no IDF weighting as this would render the transformer stateful. The hash function employed is the signed 32-bit version of Murmurhash3. For an efficiency comparison of the different feature extractors, see :ref:`sphx_glr_auto_examples_text_plot_hashing_vs_dict_vectorizer.py`. For an example of document clustering and comparison with :class:`~sklearn.feature_extraction.text.TfidfVectorizer`, see :ref:`sphx_glr_auto_examples_text_plot_document_clustering.py`. Read more in the :ref:`User Guide `. Parameters ---------- input : {'filename', 'file', 'content'}, default='content' - If `'filename'`, the sequence passed as an argument to fit is expected to be a list of filenames that need reading to fetch the raw content to analyze. - If `'file'`, the sequence items must have a 'read' method (file-like object) that is called to fetch the bytes in memory. - If `'content'`, the input is expected to be a sequence of items that can be of type string or byte. encoding : str, default='utf-8' If bytes or files are given to analyze, this encoding is used to decode. decode_error : {'strict', 'ignore', 'replace'}, default='strict' Instruction on what to do if a byte sequence is given to analyze that contains characters not of the given `encoding`. By default, it is 'strict', meaning that a UnicodeDecodeError will be raised. Other values are 'ignore' and 'replace'. strip_accents : {'ascii', 'unicode'} or callable, default=None Remove accents and perform other character normalization during the preprocessing step. 'ascii' is a fast method that only works on characters that have a direct ASCII mapping. 'unicode' is a slightly slower method that works on any character. None (default) means no character normalization is performed. Both 'ascii' and 'unicode' use NFKD normalization from :func:`unicodedata.normalize`. lowercase : bool, default=True Convert all characters to lowercase before tokenizing. preprocessor : callable, default=None Override the preprocessing (string transformation) stage while preserving the tokenizing and n-grams generation steps. Only applies if ``analyzer`` is not callable. tokenizer : callable, default=None Override the string tokenization step while preserving the preprocessing and n-grams generation steps. Only applies if ``analyzer == 'word'``. stop_words : {'english'}, list, default=None If 'english', a built-in stop word list for English is used. There are several known issues with 'english' and you should consider an alternative (see :ref:`stop_words`). If a list, that list is assumed to contain stop words, all of which will be removed from the resulting tokens. Only applies if ``analyzer == 'word'``. token_pattern : str or None, default=r"(?u)\\b\\w\\w+\\b" Regular expression denoting what constitutes a "token", only used if ``analyzer == 'word'``. The default regexp selects tokens of 2 or more alphanumeric characters (punctuation is completely ignored and always treated as a token separator). If there is a capturing group in token_pattern then the captured group content, not the entire match, becomes the token. At most one capturing group is permitted. ngram_range : tuple (min_n, max_n), default=(1, 1) The lower and upper boundary of the range of n-values for different n-grams to be extracted. All values of n such that min_n <= n <= max_n will be used. For example an ``ngram_range`` of ``(1, 1)`` means only unigrams, ``(1, 2)`` means unigrams and bigrams, and ``(2, 2)`` means only bigrams. Only applies if ``analyzer`` is not callable. analyzer : {'word', 'char', 'char_wb'} or callable, default='word' Whether the feature should be made of word or character n-grams. Option 'char_wb' creates character n-grams only from text inside word boundaries; n-grams at the edges of words are padded with space. If a callable is passed it is used to extract the sequence of features out of the raw, unprocessed input. .. versionchanged:: 0.21 Since v0.21, if ``input`` is ``'filename'`` or ``'file'``, the data is first read from the file and then passed to the given callable analyzer. n_features : int, default=(2 ** 20) The number of features (columns) in the output matrices. Small numbers of features are likely to cause hash collisions, but large numbers will cause larger coefficient dimensions in linear learners. binary : bool, default=False If True, all non zero counts are set to 1. This is useful for discrete probabilistic models that model binary events rather than integer counts. norm : {'l1', 'l2'}, default='l2' Norm used to normalize term vectors. None for no normalization. alternate_sign : bool, default=True When True, an alternating sign is added to the features as to approximately conserve the inner product in the hashed space even for small n_features. This approach is similar to sparse random projection. .. versionadded:: 0.19 dtype : type, default=np.float64 Type of the matrix returned by fit_transform() or transform(). See Also -------- CountVectorizer : Convert a collection of text documents to a matrix of token counts. TfidfVectorizer : Convert a collection of raw documents to a matrix of TF-IDF features. Notes ----- This estimator is :term:`stateless` and does not need to be fitted. However, we recommend to call :meth:`fit_transform` instead of :meth:`transform`, as parameter validation is only performed in :meth:`fit`. Examples -------- >>> from sklearn.feature_extraction.text import HashingVectorizer >>> corpus = [ ... 'This is the first document.', ... 'This document is the second document.', ... 'And this is the third one.', ... 'Is this the first document?', ... ] >>> vectorizer = HashingVectorizer(n_features=2**4) >>> X = vectorizer.fit_transform(corpus) >>> print(X.shape) (4, 16) r]>r\contentrZrarb>rEr5replacerrrNrbooleanr/r-r1rPrrjr,>rrr n_featuresrleftclosedbinarynorml1l2alternate_signdtype no_validation_parameter_constraintsrutf-8r5TrrriF)r]rarbrrr/r-r1rrjr,rrrrrc||_||_||_||_||_||_| |_||_| |_||_ | |_ | |_ | |_ ||_ ||_||_dSr)r]rarbrr/r-r,rrr1rrjrrrr)rer]rarbrrr/r-r1rrjr,rrrrrs r'__init__zHashingVectorizer.__init__s~(   (*("  "*$$&  , r)prefer_skip_nested_validationc|S)Only validates estimator's parameters. This method allows to: (i) validate the estimator's parameters and (ii) be consistent with the scikit-learn transformer API. Parameters ---------- X : ndarray of shape [n_samples, n_features] Training data. y : Ignored Not used, present for API consistency by convention. Returns ------- self : object HashingVectorizer instance. r+reXys r' partial_fitzHashingVectorizer.partial_fit!s ( r)ct|trtd||||||S)rBIterable over raw text documents expected, string object received.)r)rQrRrSrr _get_hasherfitrs r'rzHashingVectorizer.fit7s{* a   T  $$&&& ""$$$ qA&&& r)ct|trtd|||fd|D}|jr|j d|j t||j d}|S)aTransform a sequence of documents to a document-term matrix. Parameters ---------- X : iterable over raw text documents, length = n_samples Samples. Each sample must be a text document (either bytes or unicode strings, file name or file object depending on the constructor argument) which will be tokenized and hashed. Returns ------- X : sparse matrix of shape (n_samples, n_features) Document-term matrix. rc3.K|]}|VdSrr+)r<r%r,s r' z.HashingVectorizer.transform..ns+(D(D3#(D(D(D(D(D(Dr)rNFrcopy) rQrRrSrrr transformrdatafillrr)rerr,s @r'rzHashingVectorizer.transformWs a   T  ""$$$&&((      ( ((D(D(D(D!(D(D(D D D ;  FKKNNN 9 !$)%888Ar)cT||||S)aTransform a sequence of documents to a document-term matrix. Parameters ---------- X : iterable over raw text documents, length = n_samples Samples. Each sample must be a text document (either bytes or unicode strings, file name or file object depending on the constructor argument) which will be tokenized and hashed. y : any Ignored. This parameter exists only for compatibility with sklearn.pipeline.Pipeline. Returns ------- X : sparse matrix of shape (n_samples, n_features) Document-term matrix. )rrrs r' fit_transformzHashingVectorizer.fit_transformus$$xx1~~''***r)cFt|jd|j|jS)Nstring)r input_typerr)rrrrrs r'rzHashingVectorizer._get_hashers,*.     r)cxt}d|j_d|j_|SNTFsuper__sklearn_tags__ input_tagsr two_d_arrayretags __class__s r'rz"HashingVectorizer.__sklearn_tags__1ww''))!%&+# r)r)rrrrrrRrrktuplerrrciinfoint32maxrr__annotations__float64rr rrrrrr __classcell__rs@r'rr.svvp$**<<<==>$SE$ $C$C$CDDE$ **gy%9::D(K $ i[ $ 4( $ h%$ zz9+..d;$ #t$ w$ ZZ ; ; ;<<++++(   r)r)auto_wrap_output_keysctj|r1|jdkr&tj|j|jdStj|jS)zACount the number of non-zero values for each feature in sparse X.csrr) minlength) spissparseformatrcbincountrshapediffindptr)rs r'_document_frequencyrsM {1~~!!(e++{19 ;;;;wqx   r)ceZdZUdZdejiZdejiZidehdgde gdehdgded d hd e gd d gde d gde d gdedhe d gde d gde gdehde gde eddde edd dgde eddde edd dgde edd dd gdeedd gd d gd!d"Zeed#<d$d%d&d d'd d d d(d)d*d+dd d d,ejd-d.Zd/Zd9d0Zd1Zd:d2Zed'3d:d4Zd5Zd6Zd:d7Z fd8Z!xZ"S);ra Convert a collection of text documents to a matrix of token counts. This implementation produces a sparse representation of the counts using scipy.sparse.csr_matrix. If you do not provide an a-priori dictionary and you do not use an analyzer that does some kind of feature selection then the number of features will be equal to the vocabulary size found by analyzing the data. For an efficiency comparison of the different feature extractors, see :ref:`sphx_glr_auto_examples_text_plot_hashing_vs_dict_vectorizer.py`. Read more in the :ref:`User Guide `. Parameters ---------- input : {'filename', 'file', 'content'}, default='content' - If `'filename'`, the sequence passed as an argument to fit is expected to be a list of filenames that need reading to fetch the raw content to analyze. - If `'file'`, the sequence items must have a 'read' method (file-like object) that is called to fetch the bytes in memory. - If `'content'`, the input is expected to be a sequence of items that can be of type string or byte. encoding : str, default='utf-8' If bytes or files are given to analyze, this encoding is used to decode. decode_error : {'strict', 'ignore', 'replace'}, default='strict' Instruction on what to do if a byte sequence is given to analyze that contains characters not of the given `encoding`. By default, it is 'strict', meaning that a UnicodeDecodeError will be raised. Other values are 'ignore' and 'replace'. strip_accents : {'ascii', 'unicode'} or callable, default=None Remove accents and perform other character normalization during the preprocessing step. 'ascii' is a fast method that only works on characters that have a direct ASCII mapping. 'unicode' is a slightly slower method that works on any characters. None (default) means no character normalization is performed. Both 'ascii' and 'unicode' use NFKD normalization from :func:`unicodedata.normalize`. lowercase : bool, default=True Convert all characters to lowercase before tokenizing. preprocessor : callable, default=None Override the preprocessing (strip_accents and lowercase) stage while preserving the tokenizing and n-grams generation steps. Only applies if ``analyzer`` is not callable. tokenizer : callable, default=None Override the string tokenization step while preserving the preprocessing and n-grams generation steps. Only applies if ``analyzer == 'word'``. stop_words : {'english'}, list, default=None If 'english', a built-in stop word list for English is used. There are several known issues with 'english' and you should consider an alternative (see :ref:`stop_words`). If a list, that list is assumed to contain stop words, all of which will be removed from the resulting tokens. Only applies if ``analyzer == 'word'``. If None, no stop words will be used. In this case, setting `max_df` to a higher value, such as in the range (0.7, 1.0), can automatically detect and filter stop words based on intra corpus document frequency of terms. token_pattern : str or None, default=r"(?u)\\b\\w\\w+\\b" Regular expression denoting what constitutes a "token", only used if ``analyzer == 'word'``. The default regexp select tokens of 2 or more alphanumeric characters (punctuation is completely ignored and always treated as a token separator). If there is a capturing group in token_pattern then the captured group content, not the entire match, becomes the token. At most one capturing group is permitted. ngram_range : tuple (min_n, max_n), default=(1, 1) The lower and upper boundary of the range of n-values for different word n-grams or char n-grams to be extracted. All values of n such such that min_n <= n <= max_n will be used. For example an ``ngram_range`` of ``(1, 1)`` means only unigrams, ``(1, 2)`` means unigrams and bigrams, and ``(2, 2)`` means only bigrams. Only applies if ``analyzer`` is not callable. analyzer : {'word', 'char', 'char_wb'} or callable, default='word' Whether the feature should be made of word n-gram or character n-grams. Option 'char_wb' creates character n-grams only from text inside word boundaries; n-grams at the edges of words are padded with space. If a callable is passed it is used to extract the sequence of features out of the raw, unprocessed input. .. versionchanged:: 0.21 Since v0.21, if ``input`` is ``filename`` or ``file``, the data is first read from the file and then passed to the given callable analyzer. max_df : float in range [0.0, 1.0] or int, default=1.0 When building the vocabulary ignore terms that have a document frequency strictly higher than the given threshold (corpus-specific stop words). If float, the parameter represents a proportion of documents, integer absolute counts. This parameter is ignored if vocabulary is not None. min_df : float in range [0.0, 1.0] or int, default=1 When building the vocabulary ignore terms that have a document frequency strictly lower than the given threshold. This value is also called cut-off in the literature. If float, the parameter represents a proportion of documents, integer absolute counts. This parameter is ignored if vocabulary is not None. max_features : int, default=None If not None, build a vocabulary that only consider the top `max_features` ordered by term frequency across the corpus. Otherwise, all features are used. This parameter is ignored if vocabulary is not None. vocabulary : Mapping or iterable, default=None Either a Mapping (e.g., a dict) where keys are terms and values are indices in the feature matrix, or an iterable over terms. If not given, a vocabulary is determined from the input documents. Indices in the mapping should not be repeated and should not have any gap between 0 and the largest index. binary : bool, default=False If True, all non zero counts are set to 1. This is useful for discrete probabilistic models that model binary events rather than integer counts. dtype : dtype, default=np.int64 Type of the matrix returned by fit_transform() or transform(). Attributes ---------- vocabulary_ : dict A mapping of terms to feature indices. fixed_vocabulary_ : bool True if a fixed vocabulary of term to indices mapping is provided by the user. See Also -------- HashingVectorizer : Convert a collection of text documents to a matrix of token counts. TfidfVectorizer : Convert a collection of raw documents to a matrix of TF-IDF features. Examples -------- >>> from sklearn.feature_extraction.text import CountVectorizer >>> corpus = [ ... 'This is the first document.', ... 'This document is the second document.', ... 'And this is the third one.', ... 'Is this the first document?', ... ] >>> vectorizer = CountVectorizer() >>> X = vectorizer.fit_transform(corpus) >>> vectorizer.get_feature_names_out() array(['and', 'document', 'first', 'is', 'one', 'second', 'the', 'third', 'this'], ...) >>> print(X.toarray()) [[0 1 1 1 0 0 1 0 1] [0 2 0 1 0 1 1 0 1] [1 0 0 1 1 0 1 1 1] [0 1 1 1 0 0 1 0 1]] >>> vectorizer2 = CountVectorizer(analyzer='word', ngram_range=(2, 2)) >>> X2 = vectorizer2.fit_transform(corpus) >>> vectorizer2.get_feature_names_out() array(['and this', 'document is', 'first document', 'is the', 'is this', 'second document', 'the first', 'the second', 'the third', 'third one', 'this document', 'this is', 'this the'], ...) >>> print(X2.toarray()) [[0 0 1 1 0 0 1 0 0 0 0 1 0] [0 1 0 1 0 1 0 1 0 0 1 0 0] [1 0 0 1 0 0 0 0 1 1 0 1 0] [0 0 1 0 1 0 1 0 0 0 0 0 1]] raw_documentsr]>r\rrZrarb>rEr5rrrrNrrr/r-r1rPrrjr,>rrrmax_dfrrbothrrmin_df max_featuresr__iter__rrrrrrr5Trrr?F)r]rarbrrr/r-r1rrjr,rrrrrrc||_||_||_||_||_||_| |_||_| |_||_ | |_ | |_ ||_ | |_ ||_||_||_dSr)r]rarbrr/r-r,rrr1rrrrjrrr)rer]rarbrrr/r-r1rrjr,rrrrrrs r'rzCountVectorizer.__init__s*   (*("  "*$  (&$  r)c(t|}tjt ||jj}t|D]\}\}}|||<|||<||jd|_|S)zgSort features by name Returns a reordered matrix and modifies the vocabulary in place rclip)mode) ritemsrcemptyrlrrrtake)rerrsorted_features map_indexnew_valtermold_vals r'_sort_featureszCountVectorizer._sort_featuress !!1!1!3!344HS11III (1/(B(B ) ) $G_dG&Jt !(Ig  NN196N:: r)c||||tfSt|}tjt |t }| |||kz}| |||kz}|||krtj|d}|| d|} tj t |t } d| tj |d| <| }tj |dz } t|D]\} } || r | | || <|| =tj |d}t |dkrtd|dd|fS)a*Remove too rare or too common features. Prune features that are non zero in more samples than high or less documents than low, modifying the vocabulary, and restricting it to at most the limit most frequent. This does not prune samples with zero features. Nrr)axisTrzFAfter pruning, no terms remain. Try a lower min_df or a higher max_df.)rrrconesrlboolsumasarrayravelargsortzeroswherecumsumrkrrS)rerrhighlowlimitdfsmasktfs mask_indsnew_mask new_indicesr# old_index kept_indicess r'_limit_featureszCountVectorizer._limit_featuress >> F-888vRW555 M Y 'v;;?C OO4*    1}s4"B B$#B$c0|||S)anLearn a vocabulary dictionary of all tokens in the raw documents. Parameters ---------- raw_documents : iterable An iterable which generates either str, unicode or file objects. y : None This parameter is ignored. Returns ------- self : object Fitted vectorizer. )r)rerrs r'rzCountVectorizer.fits =))) r)rczt|trtd||||j}|j}|j}|j rN|j rG|j D]?}tttj|rtjdn@|||j \}}|jr|jd|j s|jd} t|t,r|n|| z} t|t,r|n|| z} | | krtd||||}|||| | |}||||}||_|S)aLearn the vocabulary dictionary and return document-term matrix. This is equivalent to fit followed by transform, but more efficiently implemented. Parameters ---------- raw_documents : iterable An iterable which generates either str, unicode or file objects. y : None This parameter is ignored. Returns ------- X : array of shape (n_samples, n_features) Document-term matrix. rzyUpper case characters found in vocabulary while 'lowercase' is True. These entries will not be matched with any documentsrrz-max_df corresponds to < documents than min_df)rQrRrSrrrrrrrrranymapisupperrrrQrrrr rr%r<r) rerrrrrr#rrn_doc max_doc_count min_doc_counts r'rzCountVectorizer.fit_transform2s. mS ) ) T  ""$$$ $$&&& !!###(  ! dn   s3;--..M9 E))-9OPP A ;  FKKNNN% *GAJE&0&B&BVFFQVM&0&B&BVFFQVM},, !PQQQ''':66$$:}m\A#'':66)D r)ct|trtd|||d\}}|jr|jd|S)aTransform documents to document-term matrix. Extract token counts out of raw text documents using the vocabulary fitted with fit or the one provided to the constructor. Parameters ---------- raw_documents : iterable An iterable which generates either str, unicode or file objects. Returns ------- X : sparse matrix of shape (n_samples, n_features) Document-term matrix. rT)rJr)rQrRrSrrQrrr)rer_rs r'rzCountVectorizer.transformvs} mS ) ) T       D AA1 ;  FKKNNNr)c|tdjd}tjt |j}tjt |j}|tj |tj rfdt|DSfdt|DS)aJReturn terms per document with nonzero entries in X. Parameters ---------- X : {array-like, sparse matrix} of shape (n_samples, n_features) Document-term matrix. Returns ------- X_inv : list of arrays of shape (n_samples,) List of arrays of terms. r) accept_sparsercg|]>}|ddfd?S)Nr)nonzeror,r<rxrinverse_vocabularys r'r>z5CountVectorizer.inverse_transform..sU#1QT7??#4#4Q#78>>@@r)c |g|]8}tj|ddf9Sr)rc flatnonzeror,r`s r'r>z5CountVectorizer.inverse_transform..sP#2>!AqqqD'#:#:;AACCr)) rrr rcarrayrkrrDrr-rr rn)rer n_samplestermsrras ` @r'inverse_transformz!CountVectorizer.inverse_transforms     / / /GAJ d.33556677(4 0 7 7 9 9::;;"2:g#6#67 ;q>> y))  y)) r)c |tjdt|jt dDtS)a^Get output feature names for transformation. Parameters ---------- input_features : array-like of str or None, default=None Not used, present here for API consistency by convention. Returns ------- feature_names_out : ndarray of str objects Transformed feature names. cg|]\}}|Sr+r+)r<rrxs r'r>z9CountVectorizer.get_feature_names_out..s O O O41aQ O O Or)r)keyr)rrcr+rrrrobject)reinput_featuress r'get_feature_names_outz%CountVectorizer.get_feature_names_outse    z O O6$"2"8"8":": 1 NNN O O O    r)cxt}d|j_d|j_|Srrrs r'rz CountVectorizer.__sklearn_tags__rr))NNNr)#rrrrrUNUSED'_CountVectorizer__metadata_request__fit-_CountVectorizer__metadata_request__transformrrRrrkrrrrrrrrrrcrErr%r<rQrr rrrgrmrrrs@r'rrsK@@H /0@0GH%46F6M$N!$**<<<==>$SE$ $C$C$CDDE$ **gy%9::D(K $ i[ $ 4( $ h%$ zz9+..d;$ #t$ w$ ZZ ; ; ;<`. Parameters ---------- norm : {'l1', 'l2'} or None, default='l2' Each output row will have unit norm, either: - 'l2': Sum of squares of vector elements is 1. The cosine similarity between two vectors is their dot product when l2 norm has been applied. - 'l1': Sum of absolute values of vector elements is 1. See :func:`~sklearn.preprocessing.normalize`. - None: No normalization. use_idf : bool, default=True Enable inverse-document-frequency reweighting. If False, idf(t) = 1. smooth_idf : bool, default=True Smooth idf weights by adding one to document frequencies, as if an extra document was seen containing every term in the collection exactly once. Prevents zero divisions. sublinear_tf : bool, default=False Apply sublinear tf scaling, i.e. replace tf with 1 + log(tf). Attributes ---------- idf_ : array of shape (n_features) The inverse document frequency (IDF) vector; only defined if ``use_idf`` is True. .. versionadded:: 0.20 n_features_in_ : int Number of features seen during :term:`fit`. .. versionadded:: 1.0 feature_names_in_ : ndarray of shape (`n_features_in_`,) Names of features seen during :term:`fit`. Defined only when `X` has feature names that are all strings. .. versionadded:: 1.0 See Also -------- CountVectorizer : Transforms text into a sparse matrix of n-gram counts. TfidfVectorizer : Convert a collection of raw documents to a matrix of TF-IDF features. HashingVectorizer : Convert a collection of text documents to a matrix of token occurrences. References ---------- .. [Yates2011] R. Baeza-Yates and B. Ribeiro-Neto (2011). Modern Information Retrieval. Addison Wesley, pp. 68-74. .. [MRS2008] C.D. Manning, P. Raghavan and H. Schütze (2008). Introduction to Information Retrieval. Cambridge University Press, pp. 118-120. Examples -------- >>> from sklearn.feature_extraction.text import TfidfTransformer >>> from sklearn.feature_extraction.text import CountVectorizer >>> from sklearn.pipeline import Pipeline >>> corpus = ['this is the first document', ... 'this document is the second document', ... 'and this is the third one', ... 'is this the first document'] >>> vocabulary = ['this', 'document', 'first', 'is', 'second', 'the', ... 'and', 'one'] >>> pipe = Pipeline([('count', CountVectorizer(vocabulary=vocabulary)), ... ('tfid', TfidfTransformer())]).fit(corpus) >>> pipe['count'].transform(corpus).toarray() array([[1, 1, 1, 1, 0, 1, 0, 0], [1, 2, 0, 1, 1, 1, 0, 0], [1, 0, 0, 1, 0, 1, 1, 1], [1, 1, 1, 1, 0, 1, 0, 0]]) >>> pipe['tfid'].idf_ array([1. , 1.22314355, 1.51082562, 1. , 1.91629073, 1. , 1.91629073, 1.91629073]) >>> pipe.transform(corpus).shape (4, 8) rrNrruse_idf smooth_idf sublinear_tfrTFc>||_||_||_||_dSrrt)rerrurvrws r'rzTfidfTransformer.__init__\s%  $(r)rcnt||dt }tj|stj|}|jt jt jfvr|jn t j}|j r|j \}}t|}| |d}|t|jz }|t|jz }t j||||_|xj|zc_t j|j|j|xjdz c_|S)afLearn the idf vector (global term weights). Parameters ---------- X : sparse matrix of shape (n_samples, n_features) A matrix of term/token counts. y : None This parameter is not needed to compute tf-idf. Returns ------- self : object Fitted transformer. )rcsc)r]accept_large_sparseFr) fill_valuer)outr)rrrr rHrrcrfloat32rur rastypefloatrvint full_likeidf_log)rerrrrer[dfs r'rzTfidfTransformer.fitbs(  !>9}   {1~~ ! a  A7rz2:&>>>BJ < 7LIq$Q''B5u--B %(( (B T_-- -I  RIUKKKDI IIOII F49$) , , , , II II r)ct|t||dtjtjg|d}t j|st j||j}|j r/tj |j |j |xj dz c_ t|dr |xj |j |jzc_ |jt!||jd}|S) aTransform a count matrix to a tf or tf-idf representation. Parameters ---------- X : sparse matrix of (n_samples, n_features) A matrix of term/token counts. copy : bool, default=True Whether to copy X and operate on the copy or perform in-place operations. `copy=False` will only be effective with CSR sparse matrix. Returns ------- vectors : sparse matrix of shape (n_samples, n_features) Tf-idf-weighted document-term matrix. rF)r]rrresetrrrNr)rrrcrrrr rHrrwrrrrrrr)rerrs r'rzTfidfTransformer.transforms"    :rz*    {1~~ 0 aqw///A    F1616 " " " FFcMFF 4  + FFdi * *FF 9 !$)%888Ar)c|t}d|j_ddg|j_|S)NTrr)rrrsparsetransformer_tagspreserves_dtypers r'rz!TfidfTransformer.__sklearn_tags__s9ww''))!%2;I0F- r)r)T)rrrrrrrrrr rrrrrs@r'rrs~~BT4L))40; k" $$D $TdQV))))) \555---65-^))))Vr)rcXeZdZUdZiejZeed<ee ddhdgdgdgdgddd d dd ddd dd dddddde j dd d ddfd Z e dZejdZdZed dfd Zdfd ZfdZfdZxZS)ra"Convert a collection of raw documents to a matrix of TF-IDF features. Equivalent to :class:`CountVectorizer` followed by :class:`TfidfTransformer`. For an example of usage, see :ref:`sphx_glr_auto_examples_text_plot_document_classification_20newsgroups.py`. For an efficiency comparison of the different feature extractors, see :ref:`sphx_glr_auto_examples_text_plot_hashing_vs_dict_vectorizer.py`. For an example of document clustering and comparison with :class:`~sklearn.feature_extraction.text.HashingVectorizer`, see :ref:`sphx_glr_auto_examples_text_plot_document_clustering.py`. Read more in the :ref:`User Guide `. Parameters ---------- input : {'filename', 'file', 'content'}, default='content' - If `'filename'`, the sequence passed as an argument to fit is expected to be a list of filenames that need reading to fetch the raw content to analyze. - If `'file'`, the sequence items must have a 'read' method (file-like object) that is called to fetch the bytes in memory. - If `'content'`, the input is expected to be a sequence of items that can be of type string or byte. encoding : str, default='utf-8' If bytes or files are given to analyze, this encoding is used to decode. decode_error : {'strict', 'ignore', 'replace'}, default='strict' Instruction on what to do if a byte sequence is given to analyze that contains characters not of the given `encoding`. By default, it is 'strict', meaning that a UnicodeDecodeError will be raised. Other values are 'ignore' and 'replace'. strip_accents : {'ascii', 'unicode'} or callable, default=None Remove accents and perform other character normalization during the preprocessing step. 'ascii' is a fast method that only works on characters that have a direct ASCII mapping. 'unicode' is a slightly slower method that works on any characters. None (default) means no character normalization is performed. Both 'ascii' and 'unicode' use NFKD normalization from :func:`unicodedata.normalize`. lowercase : bool, default=True Convert all characters to lowercase before tokenizing. preprocessor : callable, default=None Override the preprocessing (string transformation) stage while preserving the tokenizing and n-grams generation steps. Only applies if ``analyzer`` is not callable. tokenizer : callable, default=None Override the string tokenization step while preserving the preprocessing and n-grams generation steps. Only applies if ``analyzer == 'word'``. analyzer : {'word', 'char', 'char_wb'} or callable, default='word' Whether the feature should be made of word or character n-grams. Option 'char_wb' creates character n-grams only from text inside word boundaries; n-grams at the edges of words are padded with space. If a callable is passed it is used to extract the sequence of features out of the raw, unprocessed input. .. versionchanged:: 0.21 Since v0.21, if ``input`` is ``'filename'`` or ``'file'``, the data is first read from the file and then passed to the given callable analyzer. stop_words : {'english'}, list, default=None If a string, it is passed to _check_stop_list and the appropriate stop list is returned. 'english' is currently the only supported string value. There are several known issues with 'english' and you should consider an alternative (see :ref:`stop_words`). If a list, that list is assumed to contain stop words, all of which will be removed from the resulting tokens. Only applies if ``analyzer == 'word'``. If None, no stop words will be used. In this case, setting `max_df` to a higher value, such as in the range (0.7, 1.0), can automatically detect and filter stop words based on intra corpus document frequency of terms. token_pattern : str, default=r"(?u)\\b\\w\\w+\\b" Regular expression denoting what constitutes a "token", only used if ``analyzer == 'word'``. The default regexp selects tokens of 2 or more alphanumeric characters (punctuation is completely ignored and always treated as a token separator). If there is a capturing group in token_pattern then the captured group content, not the entire match, becomes the token. At most one capturing group is permitted. ngram_range : tuple (min_n, max_n), default=(1, 1) The lower and upper boundary of the range of n-values for different n-grams to be extracted. All values of n such that min_n <= n <= max_n will be used. For example an ``ngram_range`` of ``(1, 1)`` means only unigrams, ``(1, 2)`` means unigrams and bigrams, and ``(2, 2)`` means only bigrams. Only applies if ``analyzer`` is not callable. max_df : float or int, default=1.0 When building the vocabulary ignore terms that have a document frequency strictly higher than the given threshold (corpus-specific stop words). If float in range [0.0, 1.0], the parameter represents a proportion of documents, integer absolute counts. This parameter is ignored if vocabulary is not None. min_df : float or int, default=1 When building the vocabulary ignore terms that have a document frequency strictly lower than the given threshold. This value is also called cut-off in the literature. If float in range of [0.0, 1.0], the parameter represents a proportion of documents, integer absolute counts. This parameter is ignored if vocabulary is not None. max_features : int, default=None If not None, build a vocabulary that only consider the top `max_features` ordered by term frequency across the corpus. Otherwise, all features are used. This parameter is ignored if vocabulary is not None. vocabulary : Mapping or iterable, default=None Either a Mapping (e.g., a dict) where keys are terms and values are indices in the feature matrix, or an iterable over terms. If not given, a vocabulary is determined from the input documents. binary : bool, default=False If True, all non-zero term counts are set to 1. This does not mean outputs will have only 0/1 values, only that the tf term in tf-idf is binary. (Set `binary` to True, `use_idf` to False and `norm` to None to get 0/1 outputs). dtype : dtype, default=float64 Type of the matrix returned by fit_transform() or transform(). norm : {'l1', 'l2'} or None, default='l2' Each output row will have unit norm, either: - 'l2': Sum of squares of vector elements is 1. The cosine similarity between two vectors is their dot product when l2 norm has been applied. - 'l1': Sum of absolute values of vector elements is 1. See :func:`~sklearn.preprocessing.normalize`. - None: No normalization. use_idf : bool, default=True Enable inverse-document-frequency reweighting. If False, idf(t) = 1. smooth_idf : bool, default=True Smooth idf weights by adding one to document frequencies, as if an extra document was seen containing every term in the collection exactly once. Prevents zero divisions. sublinear_tf : bool, default=False Apply sublinear tf scaling, i.e. replace tf with 1 + log(tf). Attributes ---------- vocabulary_ : dict A mapping of terms to feature indices. fixed_vocabulary_ : bool True if a fixed vocabulary of term to indices mapping is provided by the user. idf_ : array of shape (n_features,) The inverse document frequency (IDF) vector; only defined if ``use_idf`` is True. See Also -------- CountVectorizer : Transforms text into a sparse matrix of n-gram counts. TfidfTransformer : Performs the TF-IDF transformation from a provided matrix of counts. Examples -------- >>> from sklearn.feature_extraction.text import TfidfVectorizer >>> corpus = [ ... 'This is the first document.', ... 'This document is the second document.', ... 'And this is the third one.', ... 'Is this the first document?', ... ] >>> vectorizer = TfidfVectorizer() >>> X = vectorizer.fit_transform(corpus) >>> vectorizer.get_feature_names_out() array(['and', 'document', 'first', 'is', 'one', 'second', 'the', 'third', 'this'], ...) >>> print(X.shape) (4, 9) rrrNrrtrrr5TrrrrrF)r]rarbrrr/r-r,r1rrjrrrrrrrrurvrwct||||||||| | | | | ||||||_||_||_||_dS)N)r]rarbrrr/r-r,r1rrjrrrrrr)rrrrurvrw)rer]rarbrrr/r-r,r1rrjrrrrrrrrurvrwrs r'rzTfidfVectorizer.__init__s~2 %'%!'#%!#    &  $(r)crt|dst|jjd|jjS)zInverse document frequency vector, only defined if `use_idf=True`. Returns ------- ndarray of shape (n_features,) _tfidfzV is not fitted yet. Call 'fit' with appropriate arguments before using this attribute.)rrrrrrrs r'rzTfidfVectorizer.idf_sOtX&&  >*EEE {r)c|jstdt|ds,t|j|j|j|j|_|t|drXt|j t|kr3tdt|t|j fz||j_ dS)Nz+`idf_` cannot be set when `user_idf=False`.rrtrz5idf length = %d must be equal to vocabulary size = %d) rurSrrrrvrwrrrlrrr)revalues r'rzTfidfVectorizer.idf_s| LJKK KtX&& +Y ?!. DK !!### 4 ' ' 4#$$E 22 K5zz3t#7#789! r)c|jtvr:tjdt|jt dSdS)NzKOnly {} 'dtype' should be used. {} 'dtype' will be converted to np.float64.)rrrrr  UserWarningrs r' _check_paramszTfidfVectorizer._check_paramssN :\ ) ) M..4f\4:.N.N      * )r)rc&||t|j|j|j|j|_t |}|j ||S)alLearn vocabulary and idf from training set. Parameters ---------- raw_documents : iterable An iterable which generates either str, unicode or file objects. y : None This parameter is not needed to compute tfidf. Returns ------- self : object Fitted vectorizer. rt) rrrrrurvrwrrrrrerrrrs r'rzTfidfVectorizer.fits"  $$&&&&L*     GG ! !- 0 0  r)c2|t|j|j|j|j|_t|}|j ||j |dS)aLearn vocabulary and idf, return document-term matrix. This is equivalent to fit followed by transform, but more efficiently implemented. Parameters ---------- raw_documents : iterable An iterable which generates either str, unicode or file objects. y : None This parameter is ignored. Returns ------- X : sparse matrix of (n_samples, n_features) Tf-idf-weighted document-term matrix. rtFr|) rrrrurvrwrrrrrrs r'rzTfidfVectorizer.fit_transforms& &L*     GG ! !- 0 0 {$$QU$333r)ct|dt|}|j|dS)aTransform documents to document-term matrix. Uses the vocabulary and document frequencies (df) learned by fit (or fit_transform). Parameters ---------- raw_documents : iterable An iterable which generates either str, unicode or file objects. Returns ------- X : sparse matrix of (n_samples, n_features) Tf-idf-weighted document-term matrix. z#The TF-IDF vectorizer is not fitted)rFr|)rrrr)rerrrs r'rzTfidfVectorizer.transform>sK "GHHHH GG  m , ,{$$QU$333r)ct}d|j_d|j_d|_|Sr)rrrrr _skip_testrs r'rz TfidfVectorizer.__sklearn_tags__Ss8ww''))!%&+# r)r)rrrrrrrrupdaterrcrrpropertyrsetterrr rrrrrrs@r'rrsLL\$No&L#MDMMM!!Zt --t4!{$+&K   &j //)/)/)/)/)/)/)h   X   [!![!,\555658444444@44444*r)r)NF)NNNNNN)=rrdrKr:r collectionsrcollections.abcr functoolsrnumbersroperatorrnumpyrc scipy.sparserr sklearn.utilsrbaser r r r exceptionsr preprocessingrutils._param_validationrrrr utils.fixesrutils.validationrrrr_hashr _stop_wordsr__all__r(r2r!r r"rVrXrrrrArrr+r)r'rs[==  ############******VVVVVVVVVVVV''''''%%%%%%RRRRRRRRRRRR######XXXXXXXXXXXX ++++++   8  ,,,,^PPP@???0BBB&bbbbbbbbJ ggggg& TggggT !!!jjjjj& jjjZ!!! qqqqq*MQUqqqqhRRRRRoRRRRRr)