def handleSeeWord(word, pos, defs, quotes): global g_seeWordDict word = convertEntities(word) defs = convertEntities(defs) add = True if -1 != defs.find("&"): add = False if -1 == defs.find("&?;"): print "------------" print pos + " " + word print defs for q in quotes: print convertEntities(q) else: print " *" if not add: return if defs.startswith("See "): defs = defs[4:].strip() else: print " Bad See <word> interpretation: %s" % defs # get word to see from def defs = defs.strip().strip(".").strip() # get only first seeWord = defs.split(",")[0].strip() g_seeWordDict[word] = seeWord
def setText(self, text, removeEntities=True): if removeEntities: text = convertEntities(text) try: text = text.encode("latin-1", "ignore") except: pass self.addParam(Param(paramTextValue, text))
def addWord(word, pos, defs, quotes): global g_wordsDict, g_dataTxt word = convertEntities(word) defs = convertEntities(defs) ind = defs.find(". See ") if ind != -1: defs = defs[: ind + 1] add = True if -1 != defs.find("&"): add = False if -1 == defs.find("&?;"): print "------------" print pos + " " + word print defs for q in quotes: print convertEntities(q) else: print " *" if not add: return # save word def textToSave = "" textToSave += "$" + pos + "\n" textToSave += "@" + defs + "\n" for q in quotes: q = convertEntities(q) textToSave += "#" + q + "\n" length = len(textToSave) offset = len(g_dataTxt) pair = [offset, length] g_dataTxt += textToSave try: g_wordsDict[word].append(pair) except: g_wordsDict[word] = [pair]
def addWord(word, pos, defs, quotes): global g_wordsDict, g_dataTxt word = convertEntities(word) defs = convertEntities(defs) ind = defs.find(". See ") if ind != -1: defs = defs[:ind + 1] add = True if -1 != defs.find("&"): add = False if -1 == defs.find("&?;"): print "------------" print pos + " " + word print defs for q in quotes: print convertEntities(q) else: print " *" if not add: return # save word def textToSave = "" textToSave += "$" + pos + "\n" textToSave += "@" + defs + "\n" for q in quotes: q = convertEntities(q) textToSave += "#" + q + "\n" length = len(textToSave) offset = len(g_dataTxt) pair = [offset, length] g_dataTxt += textToSave try: g_wordsDict[word].append(pair) except: g_wordsDict[word] = [pair]
def convertEntities(text): text = text.replace(" ", " ") text = entities.convertEntities(text) return text