def _cleanUp(self, line):
a = factory.EngLowercase()
lines = a.lowercase(line)
stopset = set(stopwords.words('english')) | set(string.punctuation)
cleanup = " ".join(filter(lambda word: word not in stopset, lines.split()))
if (self.remove_digit):
cleanup = ''.join(i for i in cleanup if not i.isdigit())
if(self.remove_punctuation):
cleanup = cleanup.translate(self.table)
return cleanup
def clean_up(line):
a = factory.EngLowercase()
lines = a.lowercase(line)
stopset = set(stopwords.words('english')) | set(string.punctuation)
cleanup = " ".join(filter(lambda word: word not in stopset, lines.split()))
return cleanup