def storeWords(self): self.wl = WordList() xReader = XMLReader() xParser = XMLParser() if xReader.checkIfExistsQuiet('xml/words.xml'): tree = xReader.getTree('xml/words.xml') wordAvg, avgRatio = xParser.getGeneralFromWords(tree) self.wl = xParser.getWords(tree) usf = 0 usl = 0 if self.vote == "up": usf = 1 else: usl = 1 for ind, obj in enumerate(self.XMLInspections): if obj.ID != self.voteId: continue pl = PageLoader(obj.fil) if not pl.isReadable(): print('Abort. File not readable:', obj.fil) exit() pl.read() patt = "^[a-zA-Z0-9]*$" pl.linkWords = self.removeListElesNotPatterned(patt, pl.linkWords) pl.titleWords = self.removeListElesNotPatterned(patt, pl.titleWords) pl.headerWords = self.removeListElesNotPatterned(patt, pl.headerWords) pl.specialWords = self.removeListElesNotPatterned(patt, pl.specialWords) pl.normalWords = self.removeListElesNotPatterned(patt, pl.normalWords) for word in pl.linkWords: self.wl.append(word, usf, usl) for word in pl.titleWords: self.wl.append(word, usf, usl) for word in pl.headerWords: self.wl.append(word, usf, usl) for word in pl.specialWords: self.wl.append(word, usf, usl) for word in pl.normalWords: self.wl.append(word, usf, usl) return
def getScore(self): lt = ListTool() for data in self.xmlInsp: pl = PageLoader(data.fil) if not pl.isReadable(): print('Abort. XMLInspections data corrupted. File not readable:', data.fil) return False pl.read() lWords = lt.getNonUniques(self.keyWords, pl.linkWords) tWords = lt.getNonUniques(self.keyWords, pl.titleWords) hWords = lt.getNonUniques(self.keyWords, pl.headerWords) sWords = lt.getNonUniques(self.keyWords, pl.specialWords) nWords = lt.getNonUniques(self.keyWords, pl.normalWords) score = len(lWords) * self.lMulti score += len(tWords) * self.tMulti score += len(hWords) * self.hMulti score += len(sWords) * self.sMulti score += len(nWords) * self.nMulti data.score = score return self.xmlInsp