Commit | Line | Data |
---|---|---|
86530b38 AT |
1 | #!/usr/bin/env python |
2 | # | |
3 | # test_codecencodings_jp.py | |
4 | # Codec encoding tests for Japanese encodings. | |
5 | # | |
6 | # $CJKCodecs: test_codecencodings_jp.py,v 1.3 2004/06/19 06:09:55 perky Exp $ | |
7 | ||
8 | from test import test_support | |
9 | from test import test_multibytecodec_support | |
10 | import unittest | |
11 | ||
12 | class Test_CP932(test_multibytecodec_support.TestBase, unittest.TestCase): | |
13 | encoding = 'cp932' | |
14 | tstring = test_multibytecodec_support.load_teststring('shift_jis') | |
15 | codectests = ( | |
16 | # invalid bytes | |
17 | ("abc\x81\x00\x81\x00\x82\x84", "strict", None), | |
18 | ("abc\xf8", "strict", None), | |
19 | ("abc\x81\x00\x82\x84", "replace", u"abc\ufffd\uff44"), | |
20 | ("abc\x81\x00\x82\x84\x88", "replace", u"abc\ufffd\uff44\ufffd"), | |
21 | ("abc\x81\x00\x82\x84", "ignore", u"abc\uff44"), | |
22 | # sjis vs cp932 | |
23 | ("\\\x7e", "replace", u"\\\x7e"), | |
24 | ("\x81\x5f\x81\x61\x81\x7c", "replace", u"\uff3c\u2225\uff0d"), | |
25 | ) | |
26 | ||
27 | class Test_EUC_JISX0213(test_multibytecodec_support.TestBase, | |
28 | unittest.TestCase): | |
29 | encoding = 'euc_jisx0213' | |
30 | tstring = test_multibytecodec_support.load_teststring('euc_jisx0213') | |
31 | codectests = ( | |
32 | # invalid bytes | |
33 | ("abc\x80\x80\xc1\xc4", "strict", None), | |
34 | ("abc\xc8", "strict", None), | |
35 | ("abc\x80\x80\xc1\xc4", "replace", u"abc\ufffd\u7956"), | |
36 | ("abc\x80\x80\xc1\xc4\xc8", "replace", u"abc\ufffd\u7956\ufffd"), | |
37 | ("abc\x80\x80\xc1\xc4", "ignore", u"abc\u7956"), | |
38 | ("abc\x8f\x83\x83", "replace", u"abc\ufffd"), | |
39 | ("\xc1\x64", "strict", None), | |
40 | ("\xa1\xc0", "strict", u"\uff3c"), | |
41 | ) | |
42 | xmlcharnametest = ( | |
43 | u"\xab\u211c\xbb = \u2329\u1234\u232a", | |
44 | "\xa9\xa8ℜ\xa9\xb2 = ⟨ሴ⟩" | |
45 | ) | |
46 | ||
47 | eucjp_commontests = ( | |
48 | ("abc\x80\x80\xc1\xc4", "strict", None), | |
49 | ("abc\xc8", "strict", None), | |
50 | ("abc\x80\x80\xc1\xc4", "replace", u"abc\ufffd\u7956"), | |
51 | ("abc\x80\x80\xc1\xc4\xc8", "replace", u"abc\ufffd\u7956\ufffd"), | |
52 | ("abc\x80\x80\xc1\xc4", "ignore", u"abc\u7956"), | |
53 | ("abc\x8f\x83\x83", "replace", u"abc\ufffd"), | |
54 | ("\xc1\x64", "strict", None), | |
55 | ) | |
56 | ||
57 | class Test_EUC_JP_COMPAT(test_multibytecodec_support.TestBase, | |
58 | unittest.TestCase): | |
59 | encoding = 'euc_jp' | |
60 | tstring = test_multibytecodec_support.load_teststring('euc_jp') | |
61 | codectests = eucjp_commontests + ( | |
62 | ("\xa1\xc0\\", "strict", u"\uff3c\\"), | |
63 | (u"\xa5", "strict", "\x5c"), | |
64 | (u"\u203e", "strict", "\x7e"), | |
65 | ) | |
66 | ||
67 | shiftjis_commonenctests = ( | |
68 | ("abc\x80\x80\x82\x84", "strict", None), | |
69 | ("abc\xf8", "strict", None), | |
70 | ("abc\x80\x80\x82\x84", "replace", u"abc\ufffd\uff44"), | |
71 | ("abc\x80\x80\x82\x84\x88", "replace", u"abc\ufffd\uff44\ufffd"), | |
72 | ("abc\x80\x80\x82\x84def", "ignore", u"abc\uff44def"), | |
73 | ) | |
74 | ||
75 | class Test_SJIS_COMPAT(test_multibytecodec_support.TestBase, unittest.TestCase): | |
76 | encoding = 'shift_jis' | |
77 | tstring = test_multibytecodec_support.load_teststring('shift_jis') | |
78 | codectests = shiftjis_commonenctests + ( | |
79 | ("\\\x7e", "strict", u"\\\x7e"), | |
80 | ("\x81\x5f\x81\x61\x81\x7c", "strict", u"\uff3c\u2016\u2212"), | |
81 | ) | |
82 | ||
83 | class Test_SJISX0213(test_multibytecodec_support.TestBase, unittest.TestCase): | |
84 | encoding = 'shift_jisx0213' | |
85 | tstring = test_multibytecodec_support.load_teststring('shift_jisx0213') | |
86 | codectests = ( | |
87 | # invalid bytes | |
88 | ("abc\x80\x80\x82\x84", "strict", None), | |
89 | ("abc\xf8", "strict", None), | |
90 | ("abc\x80\x80\x82\x84", "replace", u"abc\ufffd\uff44"), | |
91 | ("abc\x80\x80\x82\x84\x88", "replace", u"abc\ufffd\uff44\ufffd"), | |
92 | ("abc\x80\x80\x82\x84def", "ignore", u"abc\uff44def"), | |
93 | # sjis vs cp932 | |
94 | ("\\\x7e", "replace", u"\xa5\u203e"), | |
95 | ("\x81\x5f\x81\x61\x81\x7c", "replace", u"\x5c\u2016\u2212"), | |
96 | ) | |
97 | xmlcharnametest = ( | |
98 | u"\xab\u211c\xbb = \u2329\u1234\u232a", | |
99 | "\x85Gℜ\x85Q = ⟨ሴ⟩" | |
100 | ) | |
101 | ||
102 | def test_main(): | |
103 | suite = unittest.TestSuite() | |
104 | suite.addTest(unittest.makeSuite(Test_CP932)) | |
105 | suite.addTest(unittest.makeSuite(Test_EUC_JISX0213)) | |
106 | suite.addTest(unittest.makeSuite(Test_EUC_JP_COMPAT)) | |
107 | suite.addTest(unittest.makeSuite(Test_SJIS_COMPAT)) | |
108 | suite.addTest(unittest.makeSuite(Test_SJISX0213)) | |
109 | test_support.run_suite(suite) | |
110 | ||
111 | if __name__ == "__main__": | |
112 | test_main() |