def wordInDocFrequency(databaseName, word, docID): wordscount = documentInfo(databaseName, docID)['words'] return getIndex(databaseName).term_frequency(createStem(word, 'cs'), int(docID), wordscount)
def documentFrequency(databaseName, word): return getIndex(databaseName).document_frequency(createStem(word, 'cs'))
def getWordCountInDoc(databaseName, word, docID): return getIndex(databaseName).getTermCountInDoc(createStem(word, 'cs'), int(docID))
def wordFrequency(databaseName, word): return getIndex(databaseName).totalTermFrequency(createStem(word, 'cs'))
def getstem(word, lang): word = normalize_text(word) stem = createStem(word, lang) stem = strip_accents(stem) return stem
def tostems(words, lang): return map(lambda x: createStem(x, lang, save=True), words)