32
33
def check_encode(self, rev_id):
33
34
rev_id_utf8 = rev_id.encode('utf-8')
34
self.failIf(rev_id in cache_utf8._unicode_to_utf8_map)
35
self.failIf(rev_id_utf8 in cache_utf8._utf8_to_unicode_map)
35
self.assertFalse(rev_id in cache_utf8._unicode_to_utf8_map)
36
self.assertFalse(rev_id_utf8 in cache_utf8._utf8_to_unicode_map)
37
38
# After a single encode, the mapping should exist for
39
40
self.assertEqual(rev_id_utf8, cache_utf8.encode(rev_id))
40
self.failUnless(rev_id in cache_utf8._unicode_to_utf8_map)
41
self.failUnless(rev_id_utf8 in cache_utf8._utf8_to_unicode_map)
41
self.assertTrue(rev_id in cache_utf8._unicode_to_utf8_map)
42
self.assertTrue(rev_id_utf8 in cache_utf8._utf8_to_unicode_map)
43
44
self.assertEqual(rev_id, cache_utf8.decode(rev_id_utf8))
45
46
cache_utf8.clear_encoding_cache()
46
self.failIf(rev_id in cache_utf8._unicode_to_utf8_map)
47
self.failIf(rev_id_utf8 in cache_utf8._utf8_to_unicode_map)
47
self.assertFalse(rev_id in cache_utf8._unicode_to_utf8_map)
48
self.assertFalse(rev_id_utf8 in cache_utf8._utf8_to_unicode_map)
49
50
def check_decode(self, rev_id):
50
51
rev_id_utf8 = rev_id.encode('utf-8')
51
self.failIf(rev_id in cache_utf8._unicode_to_utf8_map)
52
self.failIf(rev_id_utf8 in cache_utf8._utf8_to_unicode_map)
52
self.assertFalse(rev_id in cache_utf8._unicode_to_utf8_map)
53
self.assertFalse(rev_id_utf8 in cache_utf8._utf8_to_unicode_map)
54
55
# After a single decode, the mapping should exist for
56
57
self.assertEqual(rev_id, cache_utf8.decode(rev_id_utf8))
57
self.failUnless(rev_id in cache_utf8._unicode_to_utf8_map)
58
self.failUnless(rev_id_utf8 in cache_utf8._utf8_to_unicode_map)
58
self.assertTrue(rev_id in cache_utf8._unicode_to_utf8_map)
59
self.assertTrue(rev_id_utf8 in cache_utf8._utf8_to_unicode_map)
60
61
self.assertEqual(rev_id_utf8, cache_utf8.encode(rev_id))
61
62
cache_utf8.clear_encoding_cache()
63
self.failIf(rev_id in cache_utf8._unicode_to_utf8_map)
64
self.failIf(rev_id_utf8 in cache_utf8._utf8_to_unicode_map)
64
self.assertFalse(rev_id in cache_utf8._unicode_to_utf8_map)
65
self.assertFalse(rev_id_utf8 in cache_utf8._utf8_to_unicode_map)
66
67
def test_ascii(self):
67
68
self.check_decode(u'all_ascii_characters123123123')
72
73
self.check_decode(u'some_\xb5_unicode_\xe5_chars')
74
75
def test_cached_unicode(self):
75
x = u'\xb5yy' + u'\xe5zz'
76
y = u'\xb5yy' + u'\xe5zz'
76
# Note that this is intentionally split, to prevent Python from
77
# assigning x and y to the same object
81
self.assertIsNot(x, y)
78
82
xp = cache_utf8.get_cached_unicode(x)
79
83
yp = cache_utf8.get_cached_unicode(y)
92
96
self.assertIs(xp, yp)
94
98
def test_cached_ascii(self):
95
x = '%s %s' % ('simple', 'text')
96
y = '%s %s' % ('simple', 'text')
99
x = b'%s %s' % (b'simple', b'text')
100
y = b'%s %s' % (b'simple', b'text')
101
self.assertIsNot(x, y)
98
102
xp = cache_utf8.get_cached_ascii(x)
99
103
yp = cache_utf8.get_cached_ascii(y)
106
110
uni_x = cache_utf8.decode(x)
107
111
self.assertEqual(u'simple text', uni_x)
108
self.assertIsInstance(uni_x, unicode)
112
self.assertIsInstance(uni_x, text_type)
110
114
utf8_x = cache_utf8.encode(uni_x)
111
115
self.assertIs(utf8_x, x)
113
117
def test_decode_with_None(self):
114
118
self.assertEqual(None, cache_utf8._utf8_decode_with_None(None))
115
self.assertEqual(u'foo', cache_utf8._utf8_decode_with_None('foo'))
119
self.assertEqual(u'foo', cache_utf8._utf8_decode_with_None(b'foo'))
116
120
self.assertEqual(u'f\xb5',
117
cache_utf8._utf8_decode_with_None('f\xc2\xb5'))
121
cache_utf8._utf8_decode_with_None(b'f\xc2\xb5'))