def test_mixed_unicode(self):
result = zero_digits(u'P.1, АБ1, ЯК12x, м²')
expected = tf.convert_to_tensor(u'P.0, АБ0, ЯК00x, м²',
dtype=tf.string)
expected, result = self.evaluate([expected, result])
self.assertEqual(expected, result)
def test_ragged(self):
source = tf.ragged.constant([['x1', '2x'], ['3x4']])
expected = tf.constant([['x0', '0x'], ['0x0', '']])
result = zero_digits(source).to_tensor(default_value='')
expected, result = self.evaluate([expected, result])
self.assertAllEqual(expected, result)
def test_inference_shape(self):
source = [
['1', '2', '3'],
['4', '5', '6'],
]
result = zero_digits(source)
self.assertAllEqual([2, 3], result.shape.as_list())
def test_actual_shape(self):
source = [
['1', '2', '3'],
['4', '5', '6'],
]
result = zero_digits(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 = zero_digits([['7', '8']], skip=['8'])
result = self.evaluate(result)
self.assertAllEqual([[b'0', b'8']], result)
def test_2d(self):
result = zero_digits([['7']])
result = self.evaluate(result)
self.assertAllEqual([[b'0']], result)
def test_0d(self):
result = zero_digits('7')
result = self.evaluate(result)
self.assertEqual(b'0', result)
def test_empty(self):
result = zero_digits('')
result = self.evaluate(result)
self.assertEqual(b'', result)