Ejemplo n.º 1
0
def test_get_stop_words():
    stop = tokenization._get_stop_words("nltk")
    assert stop == tokenization.nltk_stop_words()
    stop = tokenization._get_stop_words("sklearn")
    assert stop == tokenization.sklearn_stop_words()
    stop = tokenization._get_stop_words(["sklearn", "nltk"])
    assert stop == {"sklearn", "nltk"}
    with pytest.raises(ValueError):
        tokenization._get_stop_words("other")
Ejemplo n.º 2
0
def test_sklearn_stop_words():
    stop_words = tokenization.sklearn_stop_words()
    assert "the" in stop_words
    assert len(stop_words) > 100