def test_getTokens_RemovesCarriageReturns(self):
tweet = "He hasn't gone Jogging \r or ruNniNg; \rand stick-BAll?! No! BASEBALL."
expected = ['gone', 'jogging', 'running', 'stick-ball', 'baseball']
actual = getTokens(tweet)
self.assertListEqual(expected, actual)
def test_getTokens_TweetWithAllConditions_PerformsAllOps(self):
tweet = "He hasn't gone Jogging or ruNniNg; and stick-BAll! & BASEBALL."
expected = ['gone', 'jogging', 'running', 'stick-ball', 'baseball']
actual = getTokens(tweet)
self.assertListEqual(expected, actual)
def test_getTokens_RemovesPunctuation(self):
tweet = "jogging, running: stick-ball?! baseball."
expected = ['jogging', 'running', 'stick-ball', 'baseball']
actual = getTokens(tweet)
self.assertListEqual(expected, actual)
def test_getTokens_LowersCase(self):
tweet = "Jogging, ruNniNg: stick-BAll?! BASEBALL."
expected = ['jogging', 'running', 'stick-ball', 'baseball']
actual = getTokens(tweet)
self.assertListEqual(expected, actual)
def test_getTokens_RemovesSpecialEntities(self):
tweet = "jogging <> running & stick-ball <> baseball"
expected = ['jogging', 'running', 'stick-ball', 'baseball']
actual = getTokens(tweet)
self.assertListEqual(expected, actual)