def test_as_training_data_error(self):
with self.assertRaises(ValueError):
feature = IndexedPairFeature([
IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(4, [1, 2, 6])
], [IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(3, [5])], None)
feature.as_training_data()
with self.assertRaises(ValueError):
self.feature.as_training_data(mode="words+character")
def setUp(self):
super(TestIndexedPairFeature, self).setUp()
self.feature = IndexedPairFeature([
IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(2, [3, 4]),
IndexedFeatureWord(3, [5]),
IndexedFeatureWord(5, [1, 4, 1]),
IndexedFeatureWord(4, [1, 2, 6])
], [
IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(8, [3, 1, 2, 1]),
IndexedFeatureWord(2, [3, 4]),
IndexedFeatureWord(3, [5])
], [0, 1])
def test_less_than(self):
feature_1 = IndexedPairFeature(
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(4, [1, 2, 6])],
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(3, [5])], None)
feature_2 = IndexedPairFeature(
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(4, [1, 2, 6])],
[IndexedFeatureWord(2, [2, 2]),
IndexedFeatureWord(3, [5])], None)
feature_3 = IndexedPairFeature([IndexedFeatureWord(1, [1, 2])],
[IndexedFeatureWord(1, [2, 2])], None)
self.assertFalse(feature_1.__lt__(0))
self.assertFalse(feature_2.__lt__(feature_1))
self.assertLess(feature_1, feature_2)
self.assertLess(feature_3, feature_2)
def test_sort(self):
# lengths: 3, 4, 1, 2, 2
sorted_features = [IndexedPairFeature([IndexedFeatureWord(3, [1, 4, 1]),
IndexedFeatureWord(1, [1, 5])],
[IndexedFeatureWord(3, [1, 4, 1]),
IndexedFeatureWord(1, [1, 5]),
IndexedFeatureWord(3, [1, 4, 1]),
IndexedFeatureWord(2, [2, 1])],
[1, 0]),
IndexedPairFeature([IndexedFeatureWord(1, [1, 5]),
IndexedFeatureWord(2, [2, 1]),
IndexedFeatureWord(3, [1, 4, 1])],
[IndexedFeatureWord(2, [2, 1]),
IndexedFeatureWord(3, [1, 4, 1])],
[0, 1])]
self.assertNotEqual(sorted_features, self.indexed_dataset.features)
self.indexed_dataset.sort()
self.assertEquals(sorted_features, self.indexed_dataset.features)
def test_as_testing_data(self):
features = [IndexedPairFeature([IndexedFeatureWord(1, [1, 4, 4]),
IndexedFeatureWord(2, [2, 3]),
IndexedFeatureWord(3, [5, 1])],
[IndexedFeatureWord(2, [2, 3]),
IndexedFeatureWord(3, [5, 1])],
None),
IndexedPairFeature([IndexedFeatureWord(3, [5, 1]),
IndexedFeatureWord(1, [1, 4, 4])],
[IndexedFeatureWord(3, [5, 1]),
IndexedFeatureWord(1, [1, 4, 4]),
IndexedFeatureWord(3, [5, 1]),
IndexedFeatureWord(2, [2, 3])],
None)]
indexed_dataset = IndexedDataset(features)
indexed_dataset.pad_features(indexed_dataset.max_lengths())
inputs, labels = indexed_dataset.as_testing_data()
assert len(labels) == 0
first_sentence, second_sentence = inputs[0]
assert_allclose(first_sentence, np.array([1, 2, 3, 0]))
assert_allclose(second_sentence, np.array([2, 3, 0, 0]))
first_sentence, second_sentence = inputs[1]
assert_allclose(first_sentence, np.array([3, 1, 0, 0]))
assert_allclose(second_sentence, np.array([3, 1, 3, 2]))
inputs, labels = indexed_dataset.as_testing_data(mode="character")
assert len(labels) == 0
first_sentence, second_sentence = inputs[0]
assert_allclose(first_sentence, np.array([[1, 4, 4], [2, 3, 0],
[5, 1, 0], [0, 0, 0]]))
assert_allclose(second_sentence, np.array([[2, 3, 0], [5, 1, 0],
[0, 0, 0], [0, 0, 0]]))
first_sentence, second_sentence = inputs[1]
assert_allclose(first_sentence, np.array([[5, 1, 0], [1, 4, 4],
[0, 0, 0], [0, 0, 0]]))
assert_allclose(second_sentence, np.array([[5, 1, 0], [1, 4, 4],
[5, 1, 0], [2, 3, 0]]))
inputs, labels = indexed_dataset.as_testing_data(mode="word+character")
assert len(labels) == 0
(first_sentence_words, first_sentence_characters,
second_sentence_words, second_sentence_characters) = inputs[0]
assert_allclose(first_sentence_words, np.array([1, 2, 3, 0]))
assert_allclose(second_sentence_words, np.array([2, 3, 0, 0]))
assert_allclose(first_sentence_characters, np.array([[1, 4, 4], [2, 3, 0],
[5, 1, 0], [0, 0, 0]]))
assert_allclose(second_sentence_characters, np.array([[2, 3, 0], [5, 1, 0],
[0, 0, 0], [0, 0, 0]]))
(first_sentence_words, first_sentence_characters,
second_sentence_words, second_sentence_characters) = inputs[1]
assert_allclose(first_sentence_words, np.array([3, 1, 0, 0]))
assert_allclose(second_sentence_words, np.array([3, 1, 3, 2]))
assert_allclose(first_sentence_characters, np.array([[5, 1, 0], [1, 4, 4],
[0, 0, 0], [0, 0, 0]]))
assert_allclose(second_sentence_characters, np.array([[5, 1, 0], [1, 4, 4],
[5, 1, 0], [2, 3, 0]]))
with self.assertRaises(ValueError):
indexed_dataset.as_testing_data(mode="char")
def setUp(self):
super(TestIndexedDataset, self).setUp()
self.features = [IndexedPairFeature([IndexedFeatureWord(1, [1, 5]),
IndexedFeatureWord(2, [2, 1]),
IndexedFeatureWord(3, [1, 4, 1])],
[IndexedFeatureWord(2, [2, 1]),
IndexedFeatureWord(3, [1, 4, 1])],
[0, 1]),
IndexedPairFeature([IndexedFeatureWord(3, [1, 4, 1]),
IndexedFeatureWord(1, [1, 5])],
[IndexedFeatureWord(3, [1, 4, 1]),
IndexedFeatureWord(1, [1, 5]),
IndexedFeatureWord(3, [1, 4, 1]),
IndexedFeatureWord(2, [2, 1])],
[1, 0])]
self.indexed_dataset = IndexedDataset(self.features)
def test_equals(self):
feature_1 = IndexedPairFeature(
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(4, [1, 2, 6])],
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(3, [5])], None)
feature_2 = IndexedPairFeature(
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(4, [1, 2, 6])],
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(3, [5])], None)
feature_3 = IndexedPairFeature(
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(1, [2, 2])],
[IndexedFeatureWord(1, [1, 2]),
IndexedFeatureWord(3, [5])], None)
feature_4 = IndexedPairFeature([IndexedFeatureWord(1, [1, 2])],
[IndexedFeatureWord(1, [2, 2])], None)
self.assertNotEquals(feature_1, feature_4)
self.assertNotEquals(feature_1, feature_3)
self.assertFalse(feature_1.__eq__(0))
self.assertEquals(feature_1, feature_2)