def test_validate(self):
ds = resources.create_single_label_corpus()
utt4_ll = annotations.LabelList(idx='default',
labels=[
annotations.Label('a',
start=0.0,
end=1.44),
annotations.Label('a',
start=1.89,
end=10.0),
])
ds.utterances['utt-4'].set_label_list(utt4_ll)
utt6_ll = annotations.LabelList(idx='default',
labels=[
annotations.Label('a',
start=1.33,
end=5.9),
annotations.Label('a',
start=5.9,
end=14.7),
])
ds.utterances['utt-6'].set_label_list(utt6_ll)
val = validation.LabelCoverageValidator('default')
result = val.validate(ds)
assert not result.passed
assert set(result.uncovered_segments.keys()) == {'utt-4', 'utt-6'}
assert result.uncovered_segments['utt-4'] == [(1.44, 1.89)]
assert result.uncovered_segments['utt-6'] == [(0.0, 1.33),
(14.7, 15.0)]
def test_validate_returns_part_of_overlapping_label(self):
ds = resources.create_single_label_corpus()
utt4_ll = annotations.LabelList(idx='default',
labels=[
annotations.Label('a',
start=0.0,
end=9.0),
annotations.Label('b',
start=9.0,
end=13.0),
])
ds.utterances['utt-4'].set_label_list(utt4_ll)
utt6_ll = annotations.LabelList(idx='default',
labels=[
annotations.Label('a',
start=-2.0,
end=5.9),
annotations.Label('b',
start=5.9,
end=14.7),
])
ds.utterances['utt-6'].set_label_list(utt6_ll)
val = validation.LabelOverflowValidator('default')
result = val.validate(ds)
assert not result.passed
assert set(result.overflow_segments.keys()) == {'utt-4', 'utt-6'}
assert result.overflow_segments['utt-4'] == [(10.0, 13.0, 'b')]
assert result.overflow_segments['utt-6'] == [(-2.0, 0.0, 'a')]
def corpus_with_more_labels():
"""
Corpus with an extra label-list.
"""
corpus = resources.create_single_label_corpus()
corpus.utterances['utt-1'].set_label_list(
assets.LabelList(idx='radio', labels=[assets.Label('alpha')]))
corpus.utterances['utt-2'].set_label_list(
assets.LabelList(idx='radio', labels=[assets.Label('alpha')]))
corpus.utterances['utt-3'].set_label_list(
assets.LabelList(idx='radio', labels=[assets.Label('beta')]))
corpus.utterances['utt-4'].set_label_list(
assets.LabelList(idx='radio', labels=[assets.Label('beta')]))
corpus.utterances['utt-5'].set_label_list(
assets.LabelList(idx='radio', labels=[assets.Label('beta')]))
corpus.utterances['utt-6'].set_label_list(
assets.LabelList(idx='radio', labels=[assets.Label('beta')]))
corpus.utterances['utt-7'].set_label_list(
assets.LabelList(idx='radio', labels=[assets.Label('beta')]))
corpus.utterances['utt-8'].set_label_list(
assets.LabelList(idx='radio', labels=[assets.Label('beta')]))
return corpus
def test_validate_passes(self):
ds = resources.create_single_label_corpus()
val = validation.LabelCoverageValidator('default')
result = val.validate(ds)
assert result.passed
assert len(result.uncovered_segments) == 0
def test_encode_corpus(self, tmpdir):
ds = resources.create_single_label_corpus()
target_path = os.path.join(tmpdir.strpath, 'data.hdf5')
encoder = EncoderMock()
container = encoder.encode_corpus(ds, target_path)
with container as ct:
assert ct.path == target_path
assert set(ct.keys()) == set(ds.utterances.keys())
for utterance_idx in ds.utterances:
assert np.array_equal(ct.get(utterance_idx, mem_map=False), np.array([1, 2, 3]))
def corpus():
return resources.create_single_label_corpus()
