1 | from collections import OrderedDict
|
---|
2 | from json.tests import PyTest, CTest
|
---|
3 |
|
---|
4 |
|
---|
5 | class TestUnicode(object):
|
---|
6 | def test_encoding1(self):
|
---|
7 | encoder = self.json.JSONEncoder(encoding='utf-8')
|
---|
8 | u = u'\N{GREEK SMALL LETTER ALPHA}\N{GREEK CAPITAL LETTER OMEGA}'
|
---|
9 | s = u.encode('utf-8')
|
---|
10 | ju = encoder.encode(u)
|
---|
11 | js = encoder.encode(s)
|
---|
12 | self.assertEqual(ju, js)
|
---|
13 |
|
---|
14 | def test_encoding2(self):
|
---|
15 | u = u'\N{GREEK SMALL LETTER ALPHA}\N{GREEK CAPITAL LETTER OMEGA}'
|
---|
16 | s = u.encode('utf-8')
|
---|
17 | ju = self.dumps(u, encoding='utf-8')
|
---|
18 | js = self.dumps(s, encoding='utf-8')
|
---|
19 | self.assertEqual(ju, js)
|
---|
20 |
|
---|
21 | def test_encoding3(self):
|
---|
22 | u = u'\N{GREEK SMALL LETTER ALPHA}\N{GREEK CAPITAL LETTER OMEGA}'
|
---|
23 | j = self.dumps(u)
|
---|
24 | self.assertEqual(j, '"\\u03b1\\u03a9"')
|
---|
25 |
|
---|
26 | def test_encoding4(self):
|
---|
27 | u = u'\N{GREEK SMALL LETTER ALPHA}\N{GREEK CAPITAL LETTER OMEGA}'
|
---|
28 | j = self.dumps([u])
|
---|
29 | self.assertEqual(j, '["\\u03b1\\u03a9"]')
|
---|
30 |
|
---|
31 | def test_encoding5(self):
|
---|
32 | u = u'\N{GREEK SMALL LETTER ALPHA}\N{GREEK CAPITAL LETTER OMEGA}'
|
---|
33 | j = self.dumps(u, ensure_ascii=False)
|
---|
34 | self.assertEqual(j, u'"{0}"'.format(u))
|
---|
35 |
|
---|
36 | def test_encoding6(self):
|
---|
37 | u = u'\N{GREEK SMALL LETTER ALPHA}\N{GREEK CAPITAL LETTER OMEGA}'
|
---|
38 | j = self.dumps([u], ensure_ascii=False)
|
---|
39 | self.assertEqual(j, u'["{0}"]'.format(u))
|
---|
40 |
|
---|
41 | def test_big_unicode_encode(self):
|
---|
42 | u = u'\U0001d120'
|
---|
43 | self.assertEqual(self.dumps(u), '"\\ud834\\udd20"')
|
---|
44 | self.assertEqual(self.dumps(u, ensure_ascii=False), u'"\U0001d120"')
|
---|
45 |
|
---|
46 | def test_big_unicode_decode(self):
|
---|
47 | u = u'z\U0001d120x'
|
---|
48 | self.assertEqual(self.loads('"' + u + '"'), u)
|
---|
49 | self.assertEqual(self.loads('"z\\ud834\\udd20x"'), u)
|
---|
50 |
|
---|
51 | def test_unicode_decode(self):
|
---|
52 | for i in range(0, 0xd7ff):
|
---|
53 | u = unichr(i)
|
---|
54 | s = '"\\u{0:04x}"'.format(i)
|
---|
55 | self.assertEqual(self.loads(s), u)
|
---|
56 |
|
---|
57 | def test_object_pairs_hook_with_unicode(self):
|
---|
58 | s = u'{"xkd":1, "kcw":2, "art":3, "hxm":4, "qrt":5, "pad":6, "hoy":7}'
|
---|
59 | p = [(u"xkd", 1), (u"kcw", 2), (u"art", 3), (u"hxm", 4),
|
---|
60 | (u"qrt", 5), (u"pad", 6), (u"hoy", 7)]
|
---|
61 | self.assertEqual(self.loads(s), eval(s))
|
---|
62 | self.assertEqual(self.loads(s, object_pairs_hook = lambda x: x), p)
|
---|
63 | od = self.loads(s, object_pairs_hook = OrderedDict)
|
---|
64 | self.assertEqual(od, OrderedDict(p))
|
---|
65 | self.assertEqual(type(od), OrderedDict)
|
---|
66 | # the object_pairs_hook takes priority over the object_hook
|
---|
67 | self.assertEqual(self.loads(s,
|
---|
68 | object_pairs_hook = OrderedDict,
|
---|
69 | object_hook = lambda x: None),
|
---|
70 | OrderedDict(p))
|
---|
71 |
|
---|
72 | def test_default_encoding(self):
|
---|
73 | self.assertEqual(self.loads(u'{"a": "\xe9"}'.encode('utf-8')),
|
---|
74 | {'a': u'\xe9'})
|
---|
75 |
|
---|
76 | def test_unicode_preservation(self):
|
---|
77 | self.assertEqual(type(self.loads(u'""')), unicode)
|
---|
78 | self.assertEqual(type(self.loads(u'"a"')), unicode)
|
---|
79 | self.assertEqual(type(self.loads(u'["a"]')[0]), unicode)
|
---|
80 | # Issue 10038.
|
---|
81 | self.assertEqual(type(self.loads('"foo"')), unicode)
|
---|
82 |
|
---|
83 | def test_bad_encoding(self):
|
---|
84 | self.assertRaises(UnicodeEncodeError, self.loads, '"a"', u"rat\xe9")
|
---|
85 | self.assertRaises(TypeError, self.loads, '"a"', 1)
|
---|
86 |
|
---|
87 |
|
---|
88 | class TestPyUnicode(TestUnicode, PyTest): pass
|
---|
89 | class TestCUnicode(TestUnicode, CTest): pass
|
---|