def test_pipeline_default(self):
s = pd.Series([
"Amazon! < br />< br /> If I was going to order any soft drink online, it would be Diet Coke with Lime",
pd.NA,
"-1234. Mère, Françoise, noël",
])
s_true = pd.Series([
"amazon going order soft drink online would diet coke lime",
"",
"mere francoise noel",
])
self.assertEqual(preprocessing.clean(s), s_true)
def test_pipeline_stopwords(self):
s = pd.Series("E-I-E-I-O\nAnd on")
s_true = pd.Series("e-i-e-i-o\n ")
pipeline = [preprocessing.lowercase, preprocessing.remove_stopwords]
self.assertEqual(preprocessing.clean(s, pipeline=pipeline), s_true)