forked from linkedin/detext
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathtest_vocab.py
More file actions
152 lines (118 loc) · 4.97 KB
/
test_vocab.py
File metadata and controls
152 lines (118 loc) · 4.97 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
# -*- coding: utf-8 -*-
import gzip
import os
import six
import tensorflow as tf
from tensorflow.python.eager.context import context, EAGER_MODE, GRAPH_MODE
from detext.utils import vocab_utils
def switch_to(mode):
"""Switches tensorflow execution mode to mode (EAGER_MODE/GRAPH_MODE)."""
ctx = context()._eager_context
ctx.mode = mode
ctx.is_eager = mode == EAGER_MODE
class TestVocab(tf.test.TestCase):
"""Unit test for vocab_utils.py."""
def testSplit(self):
# Test case for ascii whitespace characters
s = 'a*($*#(*$& b c'
self.assertAllEqual(['a*($*#(*$&', 'b', 'c'], vocab_utils.split(s))
# This string is not empty, it contains a non-ascii whitespace character. Run print(len(s)) to check
s = u'
'
self.assertFalse(len(s) == 0)
self.assertAllEqual(vocab_utils.split(s), [s]) # Should not perform any vocab.split
def testStrip(self):
# Test case for ascii whitespace characters
s = '\na*($*#(*$& b c \r\n'
self.assertAllEqual('a*($*#(*$& b c', vocab_utils.strip(s))
# This string is not empty, it contains a non-ascii whitespace character. Run print(len(s)) to check
s = u'
'
self.assertFalse(len(s) == 0)
self.assertAllEqual(vocab_utils.strip(s), s) # Should not remove any characters
def testConvertToStr(self):
# str input
s = 'akf vna'
result = vocab_utils.convert_to_str(s)
self.assertTrue(isinstance(result, str))
self.assertAllEqual(result, 'akf vna')
# unicode input
s = u'akf vna'
result = vocab_utils.convert_to_str(s)
self.assertTrue(isinstance(result, str))
self.assertAllEqual(result, 'akf vna')
# bytes input
s = b'akf vna'
result = vocab_utils.convert_to_str(s)
self.assertTrue(isinstance(result, str))
self.assertAllEqual(result, 'akf vna')
def testConvertToBytes(self):
# str input
s = 'akf vna'
result = vocab_utils.convert_to_bytes(s)
self.assertTrue(isinstance(result, bytes))
self.assertAllEqual(result, b'akf vna')
# unicode input
s = u'akf vna'
result = vocab_utils.convert_to_bytes(s)
self.assertTrue(isinstance(result, bytes))
self.assertAllEqual(result, b'akf vna')
# bytes input
s = b'akf vna'
result = vocab_utils.convert_to_bytes(s)
self.assertTrue(isinstance(result, bytes))
self.assertAllEqual(result, b'akf vna')
def testConvertToUnicode(self):
# str input
s = 'akf vna'
result = vocab_utils.convert_to_unicode(s)
self.assertTrue(isinstance(result, six.text_type))
self.assertAllEqual(result, u'akf vna')
# unicode input
s = u'akf vna'
result = vocab_utils.convert_to_unicode(s)
self.assertTrue(isinstance(result, six.text_type))
self.assertAllEqual(result, u'akf vna')
# bytes input
s = b'akf vna'
result = vocab_utils.convert_to_unicode(s)
self.assertTrue(isinstance(result, six.text_type))
self.assertAllEqual(result, u'akf vna')
def testVocabLookUp(self):
"""Tests whether vocab lookup return the same result for str and unicode in python 2."""
if six.PY2:
# Switch to eager execution
switch_to(EAGER_MODE)
cur_dir = os.path.dirname(__file__)
vocab_file = os.path.join(cur_dir, '..', 'resources', 'multilingual_vocab.txt.gz')
vocab_table = vocab_utils.read_tf_vocab(vocab_file)
def get_index(some_string):
return vocab_table.lookup(tf.constant([some_string])).numpy()[0]
unk = get_index(vocab_utils.UNK)
# Non exist word
self.assertEqual(get_index('bj82149aksreuo'), unk)
# ascii characters
s = 'a'
s_u = u'a'
self.assertAllEqual(get_index(s), get_index(s_u))
self.assertTrue(get_index(s) != unk)
# Special non-ascii characters
s = '##' + '￥'
s_u = '##' + u'￥'
self.assertAllEqual(get_index(s), get_index(s_u))
self.assertTrue(get_index(s) != unk)
# Special non-ascii characters
s = '
'
s_u = u'
'
self.assertAllEqual(get_index(s), get_index(s_u))
self.assertTrue(get_index(s) != unk)
# Large-scale test
# Disable this if you want fast build
with gzip.GzipFile(fileobj=tf.gfile.Open(vocab_file, 'r')) as fin:
for line in fin:
line = line.strip('\n')
if line != vocab_utils.UNK:
self.assertAllEqual(get_index(line), get_index(vocab_utils.convert_to_unicode(line)))
self.assertTrue(get_index(line) != unk)
# Switch to graph execution
switch_to(GRAPH_MODE)
if __name__ == "__main__":
tf.test.main()