def test_bow_build_vocab():
field = BoWField(min_freq=2, unk_token='<unk>')
assert field.vocab == {'<unk>': 0}
dummy = ["justo luctus Praesent luctus", "luctus praesent"]
field.setup(dummy)
vocab = {'<unk>': 0, 'luctus': 1}
assert field.vocab == vocab
def test_bow_text_process_scale():
with pytest.raises(ValueError):
field = BoWField(min_freq=2, scale_factor=10)
def test_bow_text_process_normalize():
field = BoWField(min_freq=2, normalize=True)
dummy = "justo praesent luctus justo praesent"
field.setup([dummy])
assert list(field.process(dummy)) == [0, 0.5, 0.5]
def test_bow_text_process_normalize_scale():
field = BoWField(min_freq=2, normalize=True, scale_factor=10)
dummy = "justo praesent luctus justo praesent"
field.setup([dummy])
assert list(field.process(dummy)) == [0, 5, 5]
def test_bow_text_process():
field = BoWField(min_freq=2)
dummy = "justo praesent luctus justo praesent"
field.setup([dummy])
assert list(field.process(dummy)) == [0, 2, 2]