def getStopWords(): stopWordFile = FileHandler.readFile("stopwords.txt").split("\n") stopWords = [] for w in stopWordFile: stopWords.append(w[0:-1]) stopWords = stopWords[0:-1] return stopWords
def analizeDocuments(pDocumentsPaths): global _TermsFrequences,_DocumentsPaths _DocumentsPaths = pDocumentsPaths for iDocument in range(len(_DocumentsPaths)): documentPath = _DocumentsPaths[iDocument] text = FileHandler.readFile(documentPath) text = removeTags(text) text = removeStopWords(text) stemmed_words = steamS(text) addWords(stemmed_words, _TermsFrequences,iDocument) print("Document "+ str(iDocument) + " analized") return _TermsFrequences