jcgregorio@wpgntai-ubiq72.hot.corp.google.com | ed13252 | 2010-04-19 11:12:52 -0700 | [diff] [blame^] | 1 | from unittest import TestCase |
| 2 | |
| 3 | import simplejson as json |
| 4 | |
| 5 | class TestUnicode(TestCase): |
| 6 | def test_encoding1(self): |
| 7 | encoder = 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.assertEquals(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 = json.dumps(u, encoding='utf-8') |
| 18 | js = json.dumps(s, encoding='utf-8') |
| 19 | self.assertEquals(ju, js) |
| 20 | |
| 21 | def test_encoding3(self): |
| 22 | u = u'\N{GREEK SMALL LETTER ALPHA}\N{GREEK CAPITAL LETTER OMEGA}' |
| 23 | j = json.dumps(u) |
| 24 | self.assertEquals(j, '"\\u03b1\\u03a9"') |
| 25 | |
| 26 | def test_encoding4(self): |
| 27 | u = u'\N{GREEK SMALL LETTER ALPHA}\N{GREEK CAPITAL LETTER OMEGA}' |
| 28 | j = json.dumps([u]) |
| 29 | self.assertEquals(j, '["\\u03b1\\u03a9"]') |
| 30 | |
| 31 | def test_encoding5(self): |
| 32 | u = u'\N{GREEK SMALL LETTER ALPHA}\N{GREEK CAPITAL LETTER OMEGA}' |
| 33 | j = json.dumps(u, ensure_ascii=False) |
| 34 | self.assertEquals(j, u'"' + u + u'"') |
| 35 | |
| 36 | def test_encoding6(self): |
| 37 | u = u'\N{GREEK SMALL LETTER ALPHA}\N{GREEK CAPITAL LETTER OMEGA}' |
| 38 | j = json.dumps([u], ensure_ascii=False) |
| 39 | self.assertEquals(j, u'["' + u + u'"]') |
| 40 | |
| 41 | def test_big_unicode_encode(self): |
| 42 | u = u'\U0001d120' |
| 43 | self.assertEquals(json.dumps(u), '"\\ud834\\udd20"') |
| 44 | self.assertEquals(json.dumps(u, ensure_ascii=False), u'"\U0001d120"') |
| 45 | |
| 46 | def test_big_unicode_decode(self): |
| 47 | u = u'z\U0001d120x' |
| 48 | self.assertEquals(json.loads('"' + u + '"'), u) |
| 49 | self.assertEquals(json.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 | s = '"\\u%04x"' % (i,) |
| 56 | self.assertEquals(json.loads(s), u) |
| 57 | |
| 58 | def test_object_pairs_hook_with_unicode(self): |
| 59 | s = u'{"xkd":1, "kcw":2, "art":3, "hxm":4, "qrt":5, "pad":6, "hoy":7}' |
| 60 | p = [(u"xkd", 1), (u"kcw", 2), (u"art", 3), (u"hxm", 4), |
| 61 | (u"qrt", 5), (u"pad", 6), (u"hoy", 7)] |
| 62 | self.assertEqual(json.loads(s), eval(s)) |
| 63 | self.assertEqual(json.loads(s, object_pairs_hook=lambda x: x), p) |
| 64 | od = json.loads(s, object_pairs_hook=json.OrderedDict) |
| 65 | self.assertEqual(od, json.OrderedDict(p)) |
| 66 | self.assertEqual(type(od), json.OrderedDict) |
| 67 | # the object_pairs_hook takes priority over the object_hook |
| 68 | self.assertEqual(json.loads(s, |
| 69 | object_pairs_hook=json.OrderedDict, |
| 70 | object_hook=lambda x: None), |
| 71 | json.OrderedDict(p)) |
| 72 | |
| 73 | |
| 74 | def test_default_encoding(self): |
| 75 | self.assertEquals(json.loads(u'{"a": "\xe9"}'.encode('utf-8')), |
| 76 | {'a': u'\xe9'}) |
| 77 | |
| 78 | def test_unicode_preservation(self): |
| 79 | self.assertEquals(type(json.loads(u'""')), unicode) |
| 80 | self.assertEquals(type(json.loads(u'"a"')), unicode) |
| 81 | self.assertEquals(type(json.loads(u'["a"]')[0]), unicode) |
| 82 | |
| 83 | def test_ensure_ascii_false_returns_unicode(self): |
| 84 | # http://code.google.com/p/simplejson/issues/detail?id=48 |
| 85 | self.assertEquals(type(json.dumps([], ensure_ascii=False)), unicode) |
| 86 | self.assertEquals(type(json.dumps(0, ensure_ascii=False)), unicode) |
| 87 | self.assertEquals(type(json.dumps({}, ensure_ascii=False)), unicode) |
| 88 | self.assertEquals(type(json.dumps("", ensure_ascii=False)), unicode) |
| 89 | |
| 90 | def test_ensure_ascii_false_bytestring_encoding(self): |
| 91 | # http://code.google.com/p/simplejson/issues/detail?id=48 |
| 92 | doc1 = {u'quux': 'Arr\xc3\xaat sur images'} |
| 93 | doc2 = {u'quux': u'Arr\xeat sur images'} |
| 94 | doc_ascii = '{"quux": "Arr\\u00eat sur images"}' |
| 95 | doc_unicode = u'{"quux": "Arr\xeat sur images"}' |
| 96 | self.assertEquals(json.dumps(doc1), doc_ascii) |
| 97 | self.assertEquals(json.dumps(doc2), doc_ascii) |
| 98 | self.assertEquals(json.dumps(doc1, ensure_ascii=False), doc_unicode) |
| 99 | self.assertEquals(json.dumps(doc2, ensure_ascii=False), doc_unicode) |