def get_word_frequency_for_viewpoint(self, viewpoint):
self.db_cursor.execute("SELECT tweet_text "
"FROM tweet "
"WHERE viewpoint = %s",
(viewpoint,))
counter = Counter()
[counter.update(WordCloud.parse_tweet(res.tweet_text)) for res in self.db_cursor]
return counter
def test_parse_tweet_with_punctuation(self):
words = WordCloud.parse_tweet(
"alpha.. “bravo?” charlie’s delta! echo/2")
self.assertEqual(["alpha", "bravo", "charlie's", "delta", "echo/2"],
words)
def test_parse_tweet_upper_case(self):
words = WordCloud.parse_tweet("ALPHA bravo cHaRlIe")
self.assertEqual(["alpha", "bravo", "charlie"], words)
def test_parse_tweet_base_case(self):
words = WordCloud.parse_tweet("alpha bravo charlie")
self.assertEqual(["alpha", "bravo", "charlie"], words)
def test_parse_tweet_with_short_words(self):
words = WordCloud.parse_tweet("I am 3 or 4 years")
self.assertEqual(["am", "or", "years"], words)
def test_parse_tweet_with_links(self):
words = WordCloud.parse_tweet("alpha https://t.co/randomletters bravo")
self.assertEqual(["alpha", "https", "bravo"], words)
def test_parse_tweet_with_angle_brackets(self):
words = WordCloud.parse_tweet("alpha <3 bravo")
self.assertEqual(["alpha", "<3", "bravo"], words)
