class SentimentTokenizer(object): def __init__(self, lang="nl"): self.corpora = Sentiment(lang) def __call__(self, t): if not hasattr(t, "__iter__"): t = WordTokenizer(self.corpora.lang)(t) for tt in t: p = self.corpora.polarity(tt[0]) if p != None: yield (tt[0], tt[1], tt[2], p)
def test_sentiment_corpus(self): s = Sentiment() w = [t for t in s.tokens()] self.assertEquals(len(w), 4971) def f(): Sentiment(lang="en") self.assertRaises(BaseException, f) self.assertEquals(s.polarity("zweetspoor"), -2) self.assertEquals(s.polarity("Zweetspoor"), -2) self.assertEquals(s.polarity("ZWEETSPOOR"), -2) self.assertEquals(s.polarity("zonneklaar"), 4) self.assertEquals(s.polarity("Zonneklaar"), 4) self.assertEquals(s.polarity("ZONNEKLAAR"), 4) self.assertEquals(s.polarity("ZONNeKLAAR"), 4) self.assertEquals(s.polarity("Not found token"), None)
def test_sentiment_corpus(self): s = Sentiment() w = [ t for t in s.tokens() ] self.assertEquals(len(w), 4971) def f(): Sentiment(lang="en") self.assertRaises(BaseException, f) self.assertEquals(s.polarity("zweetspoor"), -2) self.assertEquals(s.polarity("Zweetspoor"), -2) self.assertEquals(s.polarity("ZWEETSPOOR"), -2) self.assertEquals(s.polarity("zonneklaar"), 4) self.assertEquals(s.polarity("Zonneklaar"), 4) self.assertEquals(s.polarity("ZONNEKLAAR"), 4) self.assertEquals(s.polarity("ZONNeKLAAR"), 4) self.assertEquals(s.polarity("Not found token"), None)
def __init__(self, lang="nl"): self.corpora = Sentiment(lang)
def f(): Sentiment(lang="en")