Add test for English time exceptions ("1a.m." etc.)

This commit is contained in:
ines 2017-03-12 13:44:20 +01:00
parent ce9568af84
commit c89e30d1a3

View File

@ -18,3 +18,10 @@ def test_tokenizer_handles_exc_in_text(en_tokenizer):
tokens = en_tokenizer(text)
assert len(tokens) == 6
assert tokens[3].text == "i.e."
@pytest.mark.parametrize('text', ["1am", "12a.m.", "11p.m.", "4pm"])
def test_tokenizer_handles_times(en_tokenizer, text):
tokens = en_tokenizer(text)
assert len(tokens) == 2
assert tokens[1].lemma_ in ["a.m.", "p.m."]