def test_ragged(self):
source = tf.ragged.constant([['X', 'X'], ['X']])
expected = tf.constant([['<X>', '<X>'], ['<X>', '']])
result = wrap_with(source, '<', '>').to_tensor(default_value='')
expected, result = self.evaluate([expected, result])
self.assertAllEqual(expected, result)
def test_unicode(self):
expected = u'надо'
result = wrap_with(u'ад', u'н', u'о')
expected = tf.convert_to_tensor(expected, dtype=tf.string)
expected, result = self.evaluate([expected, result])
self.assertAllEqual(expected, result)
def test_inference_shape(self):
source = [
['1', '2', '3'],
['4', '5', '6'],
]
result = wrap_with(source, '<', '>')
self.assertAllEqual([2, 3], result.shape.as_list())
def test_actual_shape(self):
source = [
['1', '2', '3'],
['4', '5', '6'],
]
result = wrap_with(source, '<', '>')
result = tf.shape(result)
result = self.evaluate(result)
self.assertAllEqual([2, 3], result.tolist())
def ngram_features(input_words, minn, maxn):
input_words = normalize_unicode(input_words, 'NFKC')
input_words = replace_string( # accentuation
input_words,
[u'\u0060', u' \u0301', u'\u02CA', u'\u02CB', u'\u0300', u'\u0301'],
[''] * 6)
input_words = lower_case(input_words)
input_words = zero_digits(input_words)
input_words = wrap_with(input_words, '<', '>')
word_ngrams = char_ngrams(input_words, minn, maxn, itself='ALONE')
return word_ngrams
def test_skip(self):
result = wrap_with([['X', 'y']], '<', '>', skip=['y'])
result = self.evaluate(result)
self.assertAllEqual([[b'<X>', b'y']], result)
def test_2d(self):
result = wrap_with([['X']], '<', '>')
result = self.evaluate(result)
self.assertAllEqual([[b'<X>']], result)
def test_0d(self):
result = wrap_with('X', '<', '>')
result = self.evaluate(result)
self.assertAllEqual(b'<X>', result)
def test_empty_borders(self):
result = wrap_with('test', '', '')
result = self.evaluate(result)
self.assertAllEqual(b'test', result)
def test_empty(self):
result = wrap_with('', '<', '>')
result = self.evaluate(result)
self.assertAllEqual(b'<>', result)
