def test_simple(alignments):
# various tests on the class OneHotProcessor
ali1 = alignments['S01F1522_0001']
phn1 = ali1.get_tokens_inventory()
all_tokens = alignments.get_tokens_inventory()
# no tokens_set specification, used the ones from ali1
proc = onehot.OneHotProcessor(tokens=phn1)
feat1 = proc.process(ali1)
assert proc.ndims == feat1.ndims == len(phn1)
assert feat1.shape == (ali1.tokens.shape[0], len(phn1))
assert all(feat1.data.sum(axis=1) == 1)
assert np.array_equal(feat1.times, ali1.times)
assert set(feat1.properties['onehot'].keys()) == set(
['token2index', 'tokens'])
# tokens_set used is the one from the whole alignment collection
feat2 = onehot.OneHotProcessor(tokens=all_tokens).process(ali1)
assert feat2.shape == (
ali1.tokens.shape[0], len(all_tokens))
assert all(feat2.data.sum(axis=1) != 0)
assert np.array_equal(feat1.times, feat2.times)
# another alignment with the whole tokens_set
ali2 = alignments['S01F1522_0002']
feat3 = onehot.OneHotProcessor(tokens=all_tokens).process(ali2)
assert feat3.shape == (ali2.tokens.shape[0], len(all_tokens))
assert all(feat3.data.sum(axis=1) == 1)
assert feat2.shape[1] == feat3.shape[1]
assert feat1.shape[1] < feat3.shape[1]
def test_params(params):
proc = onehot.OneHotProcessor(**params)
assert params == proc.get_params()
proc = onehot.OneHotProcessor()
proc.set_params(**params)
assert params == proc.get_params()
def test_bad_tokens(alignments):
phn = alignments.get_tokens_inventory()
phn.remove('SIL')
# a token missing in the provided inventory
proc = onehot.OneHotProcessor(tokens=phn)
with pytest.raises(ValueError):
proc.process(alignments['S01F1522_0001'])