Mercurial > cpython-lru_cache-weakref
annotate Lib/test/test_multibytecodec.py @ 79675:c0a2294de787 2.7
Issue #16199: Backport 2to3 typo fix from 3.2.
| author | Chris Jerdonek <chris.jerdonek@gmail.com> |
|---|---|
| date | Thu, 11 Oct 2012 16:15:46 -0700 |
| parents | e789b4cda872 |
| children |
| rev | line source |
|---|---|
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
1 #!/usr/bin/env python |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
2 # |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
3 # test_multibytecodec.py |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
4 # Unit test for multibytecodec itself |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
5 # |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
6 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
7 from test import test_support |
|
39029
f93dbfcc663b
Add a testcase for r47086 which fixed a bug in codec_getstreamcodec().
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37925
diff
changeset
|
8 from test.test_support import TESTFN |
|
f93dbfcc663b
Add a testcase for r47086 which fixed a bug in codec_getstreamcodec().
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37925
diff
changeset
|
9 import unittest, StringIO, codecs, sys, os |
|
47989
efe706aab32b
#3305: self->stream can be NULL.
Georg Brandl <georg@python.org>
parents:
41951
diff
changeset
|
10 import _multibytecodec |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
11 |
|
39030
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
12 ALL_CJKENCODINGS = [ |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
13 # _codecs_cn |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
14 'gb2312', 'gbk', 'gb18030', 'hz', |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
15 # _codecs_hk |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
16 'big5hkscs', |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
17 # _codecs_jp |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
18 'cp932', 'shift_jis', 'euc_jp', 'euc_jisx0213', 'shift_jisx0213', |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
19 'euc_jis_2004', 'shift_jis_2004', |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
20 # _codecs_kr |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
21 'cp949', 'euc_kr', 'johab', |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
22 # _codecs_tw |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
23 'big5', 'cp950', |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
24 # _codecs_iso2022 |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
25 'iso2022_jp', 'iso2022_jp_1', 'iso2022_jp_2', 'iso2022_jp_2004', |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
26 'iso2022_jp_3', 'iso2022_jp_ext', 'iso2022_kr', |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
27 ] |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
28 |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
29 class Test_MultibyteCodec(unittest.TestCase): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
30 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
31 def test_nullcoding(self): |
|
39030
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
32 for enc in ALL_CJKENCODINGS: |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
33 self.assertEqual(''.decode(enc), u'') |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
34 self.assertEqual(unicode('', enc), u'') |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
35 self.assertEqual(u''.encode(enc), '') |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
36 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
37 def test_str_decode(self): |
|
39030
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
38 for enc in ALL_CJKENCODINGS: |
|
58c88a7c9326
Test using all CJK encodings for the testcases which don't require
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39029
diff
changeset
|
39 self.assertEqual('abcd'.encode(enc), 'abcd') |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
40 |
|
37223
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
41 def test_errorcallback_longindex(self): |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
42 dec = codecs.getdecoder('euc-kr') |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
43 myreplace = lambda exc: (u'', sys.maxint+1) |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
44 codecs.register_error('test.cjktest', myreplace) |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
45 self.assertRaises(IndexError, dec, |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
46 'apple\x92ham\x93spam', 'test.cjktest') |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
47 |
|
39029
f93dbfcc663b
Add a testcase for r47086 which fixed a bug in codec_getstreamcodec().
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37925
diff
changeset
|
48 def test_codingspec(self): |
|
65339
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
61843
diff
changeset
|
49 for enc in ALL_CJKENCODINGS: |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
61843
diff
changeset
|
50 code = '# coding: {}\n'.format(enc) |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
61843
diff
changeset
|
51 exec code |
|
39029
f93dbfcc663b
Add a testcase for r47086 which fixed a bug in codec_getstreamcodec().
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37925
diff
changeset
|
52 |
|
47989
efe706aab32b
#3305: self->stream can be NULL.
Georg Brandl <georg@python.org>
parents:
41951
diff
changeset
|
53 def test_init_segfault(self): |
|
efe706aab32b
#3305: self->stream can be NULL.
Georg Brandl <georg@python.org>
parents:
41951
diff
changeset
|
54 # bug #3305: this used to segfault |
|
efe706aab32b
#3305: self->stream can be NULL.
Georg Brandl <georg@python.org>
parents:
41951
diff
changeset
|
55 self.assertRaises(AttributeError, |
|
efe706aab32b
#3305: self->stream can be NULL.
Georg Brandl <georg@python.org>
parents:
41951
diff
changeset
|
56 _multibytecodec.MultibyteStreamReader, None) |
|
efe706aab32b
#3305: self->stream can be NULL.
Georg Brandl <georg@python.org>
parents:
41951
diff
changeset
|
57 self.assertRaises(AttributeError, |
|
efe706aab32b
#3305: self->stream can be NULL.
Georg Brandl <georg@python.org>
parents:
41951
diff
changeset
|
58 _multibytecodec.MultibyteStreamWriter, None) |
|
efe706aab32b
#3305: self->stream can be NULL.
Georg Brandl <georg@python.org>
parents:
41951
diff
changeset
|
59 |
|
efe706aab32b
#3305: self->stream can be NULL.
Georg Brandl <georg@python.org>
parents:
41951
diff
changeset
|
60 |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
61 class Test_IncrementalEncoder(unittest.TestCase): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
62 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
63 def test_stateless(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
64 # cp949 encoder isn't stateful at all. |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
65 encoder = codecs.getincrementalencoder('cp949')() |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
66 self.assertEqual(encoder.encode(u'\ud30c\uc774\uc36c \ub9c8\uc744'), |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
67 '\xc6\xc4\xc0\xcc\xbd\xe3 \xb8\xb6\xc0\xbb') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
68 self.assertEqual(encoder.reset(), None) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
69 self.assertEqual(encoder.encode(u'\u2606\u223c\u2606', True), |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
70 '\xa1\xd9\xa1\xad\xa1\xd9') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
71 self.assertEqual(encoder.reset(), None) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
72 self.assertEqual(encoder.encode(u'', True), '') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
73 self.assertEqual(encoder.encode(u'', False), '') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
74 self.assertEqual(encoder.reset(), None) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
75 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
76 def test_stateful(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
77 # jisx0213 encoder is stateful for a few codepoints. eg) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
78 # U+00E6 => A9DC |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
79 # U+00E6 U+0300 => ABC4 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
80 # U+0300 => ABDC |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
81 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
82 encoder = codecs.getincrementalencoder('jisx0213')() |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
83 self.assertEqual(encoder.encode(u'\u00e6\u0300'), '\xab\xc4') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
84 self.assertEqual(encoder.encode(u'\u00e6'), '') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
85 self.assertEqual(encoder.encode(u'\u0300'), '\xab\xc4') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
86 self.assertEqual(encoder.encode(u'\u00e6', True), '\xa9\xdc') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
87 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
88 self.assertEqual(encoder.reset(), None) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
89 self.assertEqual(encoder.encode(u'\u0300'), '\xab\xdc') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
90 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
91 self.assertEqual(encoder.encode(u'\u00e6'), '') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
92 self.assertEqual(encoder.encode('', True), '\xa9\xdc') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
93 self.assertEqual(encoder.encode('', True), '') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
94 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
95 def test_stateful_keep_buffer(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
96 encoder = codecs.getincrementalencoder('jisx0213')() |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
97 self.assertEqual(encoder.encode(u'\u00e6'), '') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
98 self.assertRaises(UnicodeEncodeError, encoder.encode, u'\u0123') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
99 self.assertEqual(encoder.encode(u'\u0300\u00e6'), '\xab\xc4') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
100 self.assertRaises(UnicodeEncodeError, encoder.encode, u'\u0123') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
101 self.assertEqual(encoder.reset(), None) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
102 self.assertEqual(encoder.encode(u'\u0300'), '\xab\xdc') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
103 self.assertEqual(encoder.encode(u'\u00e6'), '') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
104 self.assertRaises(UnicodeEncodeError, encoder.encode, u'\u0123') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
105 self.assertEqual(encoder.encode(u'', True), '\xa9\xdc') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
106 |
|
61845
14d23bddf793
Merged revisions 81454 via svnmerge from
Victor Stinner <victor.stinner@haypocalc.com>
parents:
47989
diff
changeset
|
107 def test_issue5640(self): |
|
14d23bddf793
Merged revisions 81454 via svnmerge from
Victor Stinner <victor.stinner@haypocalc.com>
parents:
47989
diff
changeset
|
108 encoder = codecs.getincrementalencoder('shift-jis')('backslashreplace') |
|
14d23bddf793
Merged revisions 81454 via svnmerge from
Victor Stinner <victor.stinner@haypocalc.com>
parents:
47989
diff
changeset
|
109 self.assertEqual(encoder.encode(u'\xff'), b'\\xff') |
|
14d23bddf793
Merged revisions 81454 via svnmerge from
Victor Stinner <victor.stinner@haypocalc.com>
parents:
47989
diff
changeset
|
110 self.assertEqual(encoder.encode(u'\n'), b'\n') |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
111 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
112 class Test_IncrementalDecoder(unittest.TestCase): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
113 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
114 def test_dbcs(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
115 # cp949 decoder is simple with only 1 or 2 bytes sequences. |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
116 decoder = codecs.getincrementaldecoder('cp949')() |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
117 self.assertEqual(decoder.decode('\xc6\xc4\xc0\xcc\xbd'), |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
118 u'\ud30c\uc774') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
119 self.assertEqual(decoder.decode('\xe3 \xb8\xb6\xc0\xbb'), |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
120 u'\uc36c \ub9c8\uc744') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
121 self.assertEqual(decoder.decode(''), u'') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
122 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
123 def test_dbcs_keep_buffer(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
124 decoder = codecs.getincrementaldecoder('cp949')() |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
125 self.assertEqual(decoder.decode('\xc6\xc4\xc0'), u'\ud30c') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
126 self.assertRaises(UnicodeDecodeError, decoder.decode, '', True) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
127 self.assertEqual(decoder.decode('\xcc'), u'\uc774') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
128 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
129 self.assertEqual(decoder.decode('\xc6\xc4\xc0'), u'\ud30c') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
130 self.assertRaises(UnicodeDecodeError, decoder.decode, '\xcc\xbd', True) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
131 self.assertEqual(decoder.decode('\xcc'), u'\uc774') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
132 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
133 def test_iso2022(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
134 decoder = codecs.getincrementaldecoder('iso2022-jp')() |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
135 ESC = '\x1b' |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
136 self.assertEqual(decoder.decode(ESC + '('), u'') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
137 self.assertEqual(decoder.decode('B', True), u'') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
138 self.assertEqual(decoder.decode(ESC + '$'), u'') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
139 self.assertEqual(decoder.decode('B@$'), u'\u4e16') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
140 self.assertEqual(decoder.decode('@$@'), u'\u4e16') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
141 self.assertEqual(decoder.decode('$', True), u'\u4e16') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
142 self.assertEqual(decoder.reset(), None) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
143 self.assertEqual(decoder.decode('@$'), u'@$') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
144 self.assertEqual(decoder.decode(ESC + '$'), u'') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
145 self.assertRaises(UnicodeDecodeError, decoder.decode, '', True) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
146 self.assertEqual(decoder.decode('B@$'), u'\u4e16') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
147 |
|
41949
c5cf9a4f4fa9
(Backport from r55770)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
40045
diff
changeset
|
148 class Test_StreamReader(unittest.TestCase): |
|
c5cf9a4f4fa9
(Backport from r55770)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
40045
diff
changeset
|
149 def test_bug1728403(self): |
|
c5cf9a4f4fa9
(Backport from r55770)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
40045
diff
changeset
|
150 try: |
|
c5cf9a4f4fa9
(Backport from r55770)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
40045
diff
changeset
|
151 open(TESTFN, 'w').write('\xa1') |
|
c5cf9a4f4fa9
(Backport from r55770)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
40045
diff
changeset
|
152 f = codecs.open(TESTFN, encoding='cp949') |
|
c5cf9a4f4fa9
(Backport from r55770)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
40045
diff
changeset
|
153 self.assertRaises(UnicodeDecodeError, f.read, 2) |
|
c5cf9a4f4fa9
(Backport from r55770)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
40045
diff
changeset
|
154 finally: |
|
41952
01813841d64b
Fix for Windows: close a temporary file before trying to delete it.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
41949
diff
changeset
|
155 try: f.close() |
|
01813841d64b
Fix for Windows: close a temporary file before trying to delete it.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
41949
diff
changeset
|
156 except: pass |
|
41949
c5cf9a4f4fa9
(Backport from r55770)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
40045
diff
changeset
|
157 os.unlink(TESTFN) |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
158 |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
159 class Test_StreamWriter(unittest.TestCase): |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
160 if len(u'\U00012345') == 2: # UCS2 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
161 def test_gb18030(self): |
|
41947
b6c1a1a15a6c
Bug #1728403: Fix a bug that CJKCodecs StreamReader hangs when it
Hye-Shik Chang <hyeshik@gmail.com>
parents:
41534
diff
changeset
|
162 s = StringIO.StringIO() |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
163 c = codecs.getwriter('gb18030')(s) |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
164 c.write(u'123') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
165 self.assertEqual(s.getvalue(), '123') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
166 c.write(u'\U00012345') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
167 self.assertEqual(s.getvalue(), '123\x907\x959') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
168 c.write(u'\U00012345'[0]) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
169 self.assertEqual(s.getvalue(), '123\x907\x959') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
170 c.write(u'\U00012345'[1] + u'\U00012345' + u'\uac00\u00ac') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
171 self.assertEqual(s.getvalue(), |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
172 '123\x907\x959\x907\x959\x907\x959\x827\xcf5\x810\x851') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
173 c.write(u'\U00012345'[0]) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
174 self.assertEqual(s.getvalue(), |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
175 '123\x907\x959\x907\x959\x907\x959\x827\xcf5\x810\x851') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
176 self.assertRaises(UnicodeError, c.reset) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
177 self.assertEqual(s.getvalue(), |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
178 '123\x907\x959\x907\x959\x907\x959\x827\xcf5\x810\x851') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
179 |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
180 def test_utf_8(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
181 s= StringIO.StringIO() |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
182 c = codecs.getwriter('utf-8')(s) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
183 c.write(u'123') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
184 self.assertEqual(s.getvalue(), '123') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
185 c.write(u'\U00012345') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
186 self.assertEqual(s.getvalue(), '123\xf0\x92\x8d\x85') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
187 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
188 # Python utf-8 codec can't buffer surrogate pairs yet. |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
36960
diff
changeset
|
189 if 0: |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
190 c.write(u'\U00012345'[0]) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
191 self.assertEqual(s.getvalue(), '123\xf0\x92\x8d\x85') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
192 c.write(u'\U00012345'[1] + u'\U00012345' + u'\uac00\u00ac') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
193 self.assertEqual(s.getvalue(), |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
194 '123\xf0\x92\x8d\x85\xf0\x92\x8d\x85\xf0\x92\x8d\x85' |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
195 '\xea\xb0\x80\xc2\xac') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
196 c.write(u'\U00012345'[0]) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
197 self.assertEqual(s.getvalue(), |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
198 '123\xf0\x92\x8d\x85\xf0\x92\x8d\x85\xf0\x92\x8d\x85' |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
199 '\xea\xb0\x80\xc2\xac') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
200 c.reset() |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
201 self.assertEqual(s.getvalue(), |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
202 '123\xf0\x92\x8d\x85\xf0\x92\x8d\x85\xf0\x92\x8d\x85' |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
203 '\xea\xb0\x80\xc2\xac\xed\xa0\x88') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
204 c.write(u'\U00012345'[1]) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
205 self.assertEqual(s.getvalue(), |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
206 '123\xf0\x92\x8d\x85\xf0\x92\x8d\x85\xf0\x92\x8d\x85' |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
207 '\xea\xb0\x80\xc2\xac\xed\xa0\x88\xed\xbd\x85') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
208 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
209 else: # UCS4 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
210 pass |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
211 |
|
32727
69dadc2ca14d
Bring CJKCodecs 1.1 into trunk. This completely reorganizes source
Hye-Shik Chang <hyeshik@gmail.com>
parents:
31393
diff
changeset
|
212 def test_streamwriter_strwrite(self): |
|
69dadc2ca14d
Bring CJKCodecs 1.1 into trunk. This completely reorganizes source
Hye-Shik Chang <hyeshik@gmail.com>
parents:
31393
diff
changeset
|
213 s = StringIO.StringIO() |
|
69dadc2ca14d
Bring CJKCodecs 1.1 into trunk. This completely reorganizes source
Hye-Shik Chang <hyeshik@gmail.com>
parents:
31393
diff
changeset
|
214 wr = codecs.getwriter('gb18030')(s) |
|
69dadc2ca14d
Bring CJKCodecs 1.1 into trunk. This completely reorganizes source
Hye-Shik Chang <hyeshik@gmail.com>
parents:
31393
diff
changeset
|
215 wr.write('abcd') |
|
69dadc2ca14d
Bring CJKCodecs 1.1 into trunk. This completely reorganizes source
Hye-Shik Chang <hyeshik@gmail.com>
parents:
31393
diff
changeset
|
216 self.assertEqual(s.getvalue(), 'abcd') |
|
69dadc2ca14d
Bring CJKCodecs 1.1 into trunk. This completely reorganizes source
Hye-Shik Chang <hyeshik@gmail.com>
parents:
31393
diff
changeset
|
217 |
|
36960
3ad42f99a6db
Bug #1448490: Fix a bug that ISO-2022 codecs could not handle
Hye-Shik Chang <hyeshik@gmail.com>
parents:
32727
diff
changeset
|
218 class Test_ISO2022(unittest.TestCase): |
|
3ad42f99a6db
Bug #1448490: Fix a bug that ISO-2022 codecs could not handle
Hye-Shik Chang <hyeshik@gmail.com>
parents:
32727
diff
changeset
|
219 def test_g2(self): |
|
3ad42f99a6db
Bug #1448490: Fix a bug that ISO-2022 codecs could not handle
Hye-Shik Chang <hyeshik@gmail.com>
parents:
32727
diff
changeset
|
220 iso2022jp2 = '\x1b(B:hu4:unit\x1b.A\x1bNi de famille' |
|
3ad42f99a6db
Bug #1448490: Fix a bug that ISO-2022 codecs could not handle
Hye-Shik Chang <hyeshik@gmail.com>
parents:
32727
diff
changeset
|
221 uni = u':hu4:unit\xe9 de famille' |
|
3ad42f99a6db
Bug #1448490: Fix a bug that ISO-2022 codecs could not handle
Hye-Shik Chang <hyeshik@gmail.com>
parents:
32727
diff
changeset
|
222 self.assertEqual(iso2022jp2.decode('iso2022-jp-2'), uni) |
|
3ad42f99a6db
Bug #1448490: Fix a bug that ISO-2022 codecs could not handle
Hye-Shik Chang <hyeshik@gmail.com>
parents:
32727
diff
changeset
|
223 |
|
39760
15109781f016
Fix a few bugs on cjkcodecs found by Oren Tirosh:
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39030
diff
changeset
|
224 def test_iso2022_jp_g0(self): |
|
58460
17891566a478
use assert[Not]In where appropriate
Ezio Melotti <ezio.melotti@gmail.com>
parents:
55002
diff
changeset
|
225 self.assertNotIn('\x0e', u'\N{SOFT HYPHEN}'.encode('iso-2022-jp-2')) |
|
39760
15109781f016
Fix a few bugs on cjkcodecs found by Oren Tirosh:
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39030
diff
changeset
|
226 for encoding in ('iso-2022-jp-2004', 'iso-2022-jp-3'): |
|
15109781f016
Fix a few bugs on cjkcodecs found by Oren Tirosh:
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39030
diff
changeset
|
227 e = u'\u3406'.encode(encoding) |
|
55002
6768b2350c57
convert usage of fail* to assert*
Benjamin Peterson <benjamin@python.org>
parents:
47989
diff
changeset
|
228 self.assertFalse(filter(lambda x: x >= '\x80', e)) |
|
39760
15109781f016
Fix a few bugs on cjkcodecs found by Oren Tirosh:
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39030
diff
changeset
|
229 |
|
40045
347794391ce4
Bug #1572832: fix a bug in ISO-2022 codecs which may cause segfault
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39760
diff
changeset
|
230 def test_bug1572832(self): |
|
347794391ce4
Bug #1572832: fix a bug in ISO-2022 codecs which may cause segfault
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39760
diff
changeset
|
231 if sys.maxunicode >= 0x10000: |
|
347794391ce4
Bug #1572832: fix a bug in ISO-2022 codecs which may cause segfault
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39760
diff
changeset
|
232 myunichr = unichr |
|
347794391ce4
Bug #1572832: fix a bug in ISO-2022 codecs which may cause segfault
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39760
diff
changeset
|
233 else: |
|
347794391ce4
Bug #1572832: fix a bug in ISO-2022 codecs which may cause segfault
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39760
diff
changeset
|
234 myunichr = lambda x: unichr(0xD7C0+(x>>10)) + unichr(0xDC00+(x&0x3FF)) |
|
347794391ce4
Bug #1572832: fix a bug in ISO-2022 codecs which may cause segfault
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39760
diff
changeset
|
235 |
|
347794391ce4
Bug #1572832: fix a bug in ISO-2022 codecs which may cause segfault
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39760
diff
changeset
|
236 for x in xrange(0x10000, 0x110000): |
|
347794391ce4
Bug #1572832: fix a bug in ISO-2022 codecs which may cause segfault
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39760
diff
changeset
|
237 # Any ISO 2022 codec will cause the segfault |
|
347794391ce4
Bug #1572832: fix a bug in ISO-2022 codecs which may cause segfault
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39760
diff
changeset
|
238 myunichr(x).encode('iso_2022_jp', 'ignore') |
|
347794391ce4
Bug #1572832: fix a bug in ISO-2022 codecs which may cause segfault
Hye-Shik Chang <hyeshik@gmail.com>
parents:
39760
diff
changeset
|
239 |
|
70350
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
240 class TestStateful(unittest.TestCase): |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
241 text = u'\u4E16\u4E16' |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
242 encoding = 'iso-2022-jp' |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
243 expected = b'\x1b$B@$@$' |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
244 expected_reset = b'\x1b$B@$@$\x1b(B' |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
245 |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
246 def test_encode(self): |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
247 self.assertEqual(self.text.encode(self.encoding), self.expected_reset) |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
248 |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
249 def test_incrementalencoder(self): |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
250 encoder = codecs.getincrementalencoder(self.encoding)() |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
251 output = b''.join( |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
252 encoder.encode(char) |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
253 for char in self.text) |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
254 self.assertEqual(output, self.expected) |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
255 |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
256 def test_incrementalencoder_final(self): |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
257 encoder = codecs.getincrementalencoder(self.encoding)() |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
258 last_index = len(self.text) - 1 |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
259 output = b''.join( |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
260 encoder.encode(char, index == last_index) |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
261 for index, char in enumerate(self.text)) |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
262 self.assertEqual(output, self.expected_reset) |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
263 |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
264 class TestHZStateful(TestStateful): |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
265 text = u'\u804a\u804a' |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
266 encoding = 'hz' |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
267 expected = b'~{ADAD' |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
268 expected_reset = b'~{ADAD~}' |
|
e789b4cda872
Issue #12100: Don't reset incremental encoders of CJK codecs at each call to
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
269 |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
270 def test_main(): |
|
41534
6c4d7678af5e
Standardize on test.test_support.run_unittest() (as opposed to a mix of run_unittest() and run_suite()). Also, add functionality to run_unittest() that admits usage of unittest.TestLoader.loadTestsFromModule().
Collin Winter <collinw@gmail.com>
parents:
40045
diff
changeset
|
271 test_support.run_unittest(__name__) |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
272 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
273 if __name__ == "__main__": |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
274 test_main() |
