Mercurial > cpython-lru_cache-weakref
annotate Lib/test/test_multibytecodec_support.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 | 53912b58eee6 |
| 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_support.py |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
4 # Common Unittest Routines for CJK codecs |
|
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 |
|
70178
83f4c270b27d
Issue #12057: Fix .hgeol and test_multibytecodec_support for the conversion of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
7 import codecs |
|
83f4c270b27d
Issue #12057: Fix .hgeol and test_multibytecodec_support for the conversion of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
8 import os |
|
83f4c270b27d
Issue #12057: Fix .hgeol and test_multibytecodec_support for the conversion of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
9 import re |
|
83f4c270b27d
Issue #12057: Fix .hgeol and test_multibytecodec_support for the conversion of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
10 import sys |
|
83f4c270b27d
Issue #12057: Fix .hgeol and test_multibytecodec_support for the conversion of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
11 import unittest |
|
57466
7b74abb80356
When open_urlresource() fails, HTTPException is another possible error
Antoine Pitrou <solipsis@pitrou.net>
parents:
57027
diff
changeset
|
12 from httplib import HTTPException |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
13 from test import test_support |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
14 from StringIO import StringIO |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
15 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
16 class TestBase: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
17 encoding = '' # codec name |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
18 codec = None # codec tuple (with 4 elements) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
19 tstring = '' # string to test StreamReader |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
20 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
21 codectests = None # must set. codec test tuple |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
22 roundtriptest = 1 # set if roundtrip is possible with unicode |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
23 has_iso10646 = 0 # set if this encoding contains whole iso10646 map |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
24 xmlcharnametest = None # string to test xmlcharrefreplace |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
25 unmappedunicode = u'\udeee' # a unicode codepoint that is not mapped. |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
26 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
27 def setUp(self): |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
28 if self.codec is None: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
29 self.codec = codecs.lookup(self.encoding) |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
30 self.encode = self.codec.encode |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
31 self.decode = self.codec.decode |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
32 self.reader = self.codec.streamreader |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
33 self.writer = self.codec.streamwriter |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
34 self.incrementalencoder = self.codec.incrementalencoder |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
35 self.incrementaldecoder = self.codec.incrementaldecoder |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
36 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
37 def test_chunkcoding(self): |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
38 for native, utf8 in zip(*[StringIO(f).readlines() |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
39 for f in self.tstring]): |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
40 u = self.decode(native)[0] |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
41 self.assertEqual(u, utf8.decode('utf-8')) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
42 if self.roundtriptest: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
43 self.assertEqual(native, self.encode(u)[0]) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
44 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
45 def test_errorhandle(self): |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
46 for source, scheme, expected in self.codectests: |
|
70630
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
47 if isinstance(source, bytes): |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
48 func = self.decode |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
49 else: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
50 func = self.encode |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
51 if expected: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
52 result = func(source, scheme)[0] |
|
70630
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
53 if func is self.decode: |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
54 self.assertTrue(type(result) is unicode, type(result)) |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
55 self.assertEqual(result, expected, |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
56 '%r.decode(%r, %r)=%r != %r' |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
57 % (source, self.encoding, scheme, result, |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
58 expected)) |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
59 else: |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
60 self.assertTrue(type(result) is bytes, type(result)) |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
61 self.assertEqual(result, expected, |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
62 '%r.encode(%r, %r)=%r != %r' |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
63 % (source, self.encoding, scheme, result, |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
64 expected)) |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
65 else: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
66 self.assertRaises(UnicodeError, func, source, scheme) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
67 |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
68 def test_xmlcharrefreplace(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
69 if self.has_iso10646: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
70 return |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
71 |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
72 s = u"\u0b13\u0b23\u0b60 nd eggs" |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
73 self.assertEqual( |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
74 self.encode(s, "xmlcharrefreplace")[0], |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
75 "ଓଣୠ nd eggs" |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
76 ) |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
77 |
|
37223
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
78 def test_customreplace_encode(self): |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
79 if self.has_iso10646: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
80 return |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
81 |
|
47177
454d38c1f47e
revert creation of the html.entities and html.parser modules
Fred Drake <fdrake@acm.org>
parents:
47143
diff
changeset
|
82 from htmlentitydefs import codepoint2name |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
83 |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
84 def xmlcharnamereplace(exc): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
85 if not isinstance(exc, UnicodeEncodeError): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
86 raise TypeError("don't know how to handle %r" % exc) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
87 l = [] |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
88 for c in exc.object[exc.start:exc.end]: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
89 if ord(c) in codepoint2name: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
90 l.append(u"&%s;" % codepoint2name[ord(c)]) |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
91 else: |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
92 l.append(u"&#%d;" % ord(c)) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
93 return (u"".join(l), exc.end) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
94 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
95 codecs.register_error("test.xmlcharnamereplace", xmlcharnamereplace) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
96 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
97 if self.xmlcharnametest: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
98 sin, sout = self.xmlcharnametest |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
99 else: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
100 sin = u"\xab\u211c\xbb = \u2329\u1234\u232a" |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
101 sout = "«ℜ» = ⟨ሴ⟩" |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
102 self.assertEqual(self.encode(sin, |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
103 "test.xmlcharnamereplace")[0], sout) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
104 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
105 def test_callback_wrong_objects(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
106 def myreplace(exc): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
107 return (ret, exc.end) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
108 codecs.register_error("test.cjktest", myreplace) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
109 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
110 for ret in ([1, 2, 3], [], None, object(), 'string', ''): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
111 self.assertRaises(TypeError, self.encode, self.unmappedunicode, |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
112 'test.cjktest') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
113 |
|
37223
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
114 def test_callback_long_index(self): |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
115 def myreplace(exc): |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
116 return (u'x', long(exc.end)) |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
117 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
|
118 self.assertEqual(self.encode(u'abcd' + self.unmappedunicode + u'efgh', |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
119 'test.cjktest'), ('abcdxefgh', 9)) |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
120 |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
121 def myreplace(exc): |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
122 return (u'x', 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
|
123 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
|
124 self.assertRaises(IndexError, self.encode, self.unmappedunicode, |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
125 'test.cjktest') |
|
c0818f21f65a
Allow long objects as a position value of error callbacks returned.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37218
diff
changeset
|
126 |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
127 def test_callback_None_index(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
128 def myreplace(exc): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
129 return (u'x', None) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
130 codecs.register_error("test.cjktest", myreplace) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
131 self.assertRaises(TypeError, self.encode, self.unmappedunicode, |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
132 'test.cjktest') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
133 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
134 def test_callback_backward_index(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
135 def myreplace(exc): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
136 if myreplace.limit > 0: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
137 myreplace.limit -= 1 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
138 return (u'REPLACED', 0) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
139 else: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
140 return (u'TERMINAL', exc.end) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
141 myreplace.limit = 3 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
142 codecs.register_error("test.cjktest", myreplace) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
143 self.assertEqual(self.encode(u'abcd' + self.unmappedunicode + u'efgh', |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
144 'test.cjktest'), |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
145 ('abcdREPLACEDabcdREPLACEDabcdREPLACEDabcdTERMINALefgh', 9)) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
146 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
147 def test_callback_forward_index(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
148 def myreplace(exc): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
149 return (u'REPLACED', exc.end + 2) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
150 codecs.register_error("test.cjktest", myreplace) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
151 self.assertEqual(self.encode(u'abcd' + self.unmappedunicode + u'efgh', |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
152 'test.cjktest'), ('abcdREPLACEDgh', 9)) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
153 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
154 def test_callback_index_outofbound(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
155 def myreplace(exc): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
156 return (u'TERM', 100) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
157 codecs.register_error("test.cjktest", myreplace) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
158 self.assertRaises(IndexError, self.encode, self.unmappedunicode, |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
159 'test.cjktest') |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
160 |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
161 def test_incrementalencoder(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
162 UTF8Reader = codecs.getreader('utf-8') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
163 for sizehint in [None] + range(1, 33) + \ |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
164 [64, 128, 256, 512, 1024]: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
165 istream = UTF8Reader(StringIO(self.tstring[1])) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
166 ostream = StringIO() |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
167 encoder = self.incrementalencoder() |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
168 while 1: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
169 if sizehint is not None: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
170 data = istream.read(sizehint) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
171 else: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
172 data = istream.read() |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
173 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
174 if not data: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
175 break |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
176 e = encoder.encode(data) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
177 ostream.write(e) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
178 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
179 self.assertEqual(ostream.getvalue(), self.tstring[0]) |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
180 |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
181 def test_incrementaldecoder(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
182 UTF8Writer = codecs.getwriter('utf-8') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
183 for sizehint in [None, -1] + range(1, 33) + \ |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
184 [64, 128, 256, 512, 1024]: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
185 istream = StringIO(self.tstring[0]) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
186 ostream = UTF8Writer(StringIO()) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
187 decoder = self.incrementaldecoder() |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
188 while 1: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
189 data = istream.read(sizehint) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
190 if not data: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
191 break |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
192 else: |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
193 u = decoder.decode(data) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
194 ostream.write(u) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
195 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
196 self.assertEqual(ostream.getvalue(), self.tstring[1]) |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
197 |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
198 def test_incrementalencoder_error_callback(self): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
199 inv = self.unmappedunicode |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
200 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
201 e = self.incrementalencoder() |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
202 self.assertRaises(UnicodeEncodeError, e.encode, inv, True) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
203 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
204 e.errors = 'ignore' |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
205 self.assertEqual(e.encode(inv, True), '') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
206 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
207 e.reset() |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
208 def tempreplace(exc): |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
209 return (u'called', exc.end) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
210 codecs.register_error('test.incremental_error_callback', tempreplace) |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
211 e.errors = 'test.incremental_error_callback' |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
212 self.assertEqual(e.encode(inv, True), 'called') |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
213 |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
214 # again |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
215 e.errors = 'ignore' |
|
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
216 self.assertEqual(e.encode(inv, True), '') |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
217 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
218 def test_streamreader(self): |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
219 UTF8Writer = codecs.getwriter('utf-8') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
220 for name in ["read", "readline", "readlines"]: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
221 for sizehint in [None, -1] + range(1, 33) + \ |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
222 [64, 128, 256, 512, 1024]: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
223 istream = self.reader(StringIO(self.tstring[0])) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
224 ostream = UTF8Writer(StringIO()) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
225 func = getattr(istream, name) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
226 while 1: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
227 data = func(sizehint) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
228 if not data: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
229 break |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
230 if name == "readlines": |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
231 ostream.writelines(data) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
232 else: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
233 ostream.write(data) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
234 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
235 self.assertEqual(ostream.getvalue(), self.tstring[1]) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
236 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
237 def test_streamwriter(self): |
|
37218
e234ff8beaa2
Patch #1443155: Add the incremental codecs support for CJK codecs.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
35998
diff
changeset
|
238 readfuncs = ('read', 'readline', 'readlines') |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
239 UTF8Reader = codecs.getreader('utf-8') |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
240 for name in readfuncs: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
241 for sizehint in [None] + range(1, 33) + \ |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
242 [64, 128, 256, 512, 1024]: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
243 istream = UTF8Reader(StringIO(self.tstring[1])) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
244 ostream = self.writer(StringIO()) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
245 func = getattr(istream, name) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
246 while 1: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
247 if sizehint is not None: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
248 data = func(sizehint) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
249 else: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
250 data = func() |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
251 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
252 if not data: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
253 break |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
254 if name == "readlines": |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
255 ostream.writelines(data) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
256 else: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
257 ostream.write(data) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
258 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
259 self.assertEqual(ostream.getvalue(), self.tstring[0]) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
260 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
261 class TestBase_Mapping(unittest.TestCase): |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
262 pass_enctest = [] |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
263 pass_dectest = [] |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
264 supmaps = [] |
|
70630
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
265 codectests = [] |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
266 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
267 def __init__(self, *args, **kw): |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
268 unittest.TestCase.__init__(self, *args, **kw) |
| 57029 | 269 try: |
|
65339
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
270 self.open_mapping_file().close() # test it to report the error early |
|
57466
7b74abb80356
When open_urlresource() fails, HTTPException is another possible error
Antoine Pitrou <solipsis@pitrou.net>
parents:
57027
diff
changeset
|
271 except (IOError, HTTPException): |
|
57027
ff8e359e2808
Hum, test skipping when the URL isn't reachable hadn't been applied to trunk.
Antoine Pitrou <solipsis@pitrou.net>
parents:
47177
diff
changeset
|
272 self.skipTest("Could not retrieve "+self.mapfileurl) |
|
35998
96d1fa36964d
Patch #1276356: Implement new resource "urlfetch" for regrtest.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
32727
diff
changeset
|
273 |
|
96d1fa36964d
Patch #1276356: Implement new resource "urlfetch" for regrtest.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
32727
diff
changeset
|
274 def open_mapping_file(self): |
|
96d1fa36964d
Patch #1276356: Implement new resource "urlfetch" for regrtest.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
32727
diff
changeset
|
275 return test_support.open_urlresource(self.mapfileurl) |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
276 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
277 def test_mapping_file(self): |
|
42447
3fe57daaa608
Fix gb18030 codec's bug that doesn't map two-byte characters on
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37223
diff
changeset
|
278 if self.mapfileurl.endswith('.xml'): |
|
3fe57daaa608
Fix gb18030 codec's bug that doesn't map two-byte characters on
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37223
diff
changeset
|
279 self._test_mapping_file_ucm() |
|
3fe57daaa608
Fix gb18030 codec's bug that doesn't map two-byte characters on
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37223
diff
changeset
|
280 else: |
|
3fe57daaa608
Fix gb18030 codec's bug that doesn't map two-byte characters on
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37223
diff
changeset
|
281 self._test_mapping_file_plain() |
|
3fe57daaa608
Fix gb18030 codec's bug that doesn't map two-byte characters on
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37223
diff
changeset
|
282 |
|
3fe57daaa608
Fix gb18030 codec's bug that doesn't map two-byte characters on
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37223
diff
changeset
|
283 def _test_mapping_file_plain(self): |
|
59184
86402fa0c5b8
Merged revisions 78497 via svnmerge from
Florent Xicluna <florent.xicluna@gmail.com>
parents:
57029
diff
changeset
|
284 _unichr = lambda c: eval("u'\\U%08x'" % int(c, 16)) |
|
86402fa0c5b8
Merged revisions 78497 via svnmerge from
Florent Xicluna <florent.xicluna@gmail.com>
parents:
57029
diff
changeset
|
285 unichrs = lambda s: u''.join(_unichr(c) for c in s.split('+')) |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
286 urt_wa = {} |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
287 |
|
65339
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
288 with self.open_mapping_file() as f: |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
289 for line in f: |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
290 if not line: |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
291 break |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
292 data = line.split('#')[0].strip().split() |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
293 if len(data) != 2: |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
294 continue |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
295 |
|
65339
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
296 csetval = eval(data[0]) |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
297 if csetval <= 0x7F: |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
298 csetch = chr(csetval & 0xff) |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
299 elif csetval >= 0x1000000: |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
300 csetch = chr(csetval >> 24) + chr((csetval >> 16) & 0xff) + \ |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
301 chr((csetval >> 8) & 0xff) + chr(csetval & 0xff) |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
302 elif csetval >= 0x10000: |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
303 csetch = chr(csetval >> 16) + \ |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
304 chr((csetval >> 8) & 0xff) + chr(csetval & 0xff) |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
305 elif csetval >= 0x100: |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
306 csetch = chr(csetval >> 8) + chr(csetval & 0xff) |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
307 else: |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
308 continue |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
309 |
|
65339
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
310 unich = unichrs(data[1]) |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
311 if unich == u'\ufffd' or unich in urt_wa: |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
312 continue |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
313 urt_wa[unich] = csetch |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
314 |
|
65339
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
315 self._testpoint(csetch, unich) |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
316 |
|
42447
3fe57daaa608
Fix gb18030 codec's bug that doesn't map two-byte characters on
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37223
diff
changeset
|
317 def _test_mapping_file_ucm(self): |
|
65339
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
318 with self.open_mapping_file() as f: |
|
fea8f21b199d
Merged revisions 85503 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents:
59183
diff
changeset
|
319 ucmdata = f.read() |
|
42447
3fe57daaa608
Fix gb18030 codec's bug that doesn't map two-byte characters on
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37223
diff
changeset
|
320 uc = re.findall('<a u="([A-F0-9]{4})" b="([0-9A-F ]+)"/>', ucmdata) |
|
3fe57daaa608
Fix gb18030 codec's bug that doesn't map two-byte characters on
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37223
diff
changeset
|
321 for uni, coded in uc: |
|
3fe57daaa608
Fix gb18030 codec's bug that doesn't map two-byte characters on
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37223
diff
changeset
|
322 unich = unichr(int(uni, 16)) |
|
3fe57daaa608
Fix gb18030 codec's bug that doesn't map two-byte characters on
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37223
diff
changeset
|
323 codech = ''.join(chr(int(c, 16)) for c in coded.split()) |
|
3fe57daaa608
Fix gb18030 codec's bug that doesn't map two-byte characters on
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37223
diff
changeset
|
324 self._testpoint(codech, unich) |
|
3fe57daaa608
Fix gb18030 codec's bug that doesn't map two-byte characters on
Hye-Shik Chang <hyeshik@gmail.com>
parents:
37223
diff
changeset
|
325 |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
326 def test_mapping_supplemental(self): |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
327 for mapping in self.supmaps: |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
328 self._testpoint(*mapping) |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
329 |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
330 def _testpoint(self, csetch, unich): |
|
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
331 if (csetch, unich) not in self.pass_enctest: |
|
45235
eca90b8aa7fd
Update big5hkscs codec to conform to the HKSCS:2004 revision.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
42447
diff
changeset
|
332 try: |
|
eca90b8aa7fd
Update big5hkscs codec to conform to the HKSCS:2004 revision.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
42447
diff
changeset
|
333 self.assertEqual(unich.encode(self.encoding), csetch) |
|
eca90b8aa7fd
Update big5hkscs codec to conform to the HKSCS:2004 revision.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
42447
diff
changeset
|
334 except UnicodeError, exc: |
|
eca90b8aa7fd
Update big5hkscs codec to conform to the HKSCS:2004 revision.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
42447
diff
changeset
|
335 self.fail('Encoding failed while testing %s -> %s: %s' % ( |
|
eca90b8aa7fd
Update big5hkscs codec to conform to the HKSCS:2004 revision.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
42447
diff
changeset
|
336 repr(unich), repr(csetch), exc.reason)) |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
337 if (csetch, unich) not in self.pass_dectest: |
|
45235
eca90b8aa7fd
Update big5hkscs codec to conform to the HKSCS:2004 revision.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
42447
diff
changeset
|
338 try: |
|
eca90b8aa7fd
Update big5hkscs codec to conform to the HKSCS:2004 revision.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
42447
diff
changeset
|
339 self.assertEqual(csetch.decode(self.encoding), unich) |
|
eca90b8aa7fd
Update big5hkscs codec to conform to the HKSCS:2004 revision.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
42447
diff
changeset
|
340 except UnicodeError, exc: |
|
eca90b8aa7fd
Update big5hkscs codec to conform to the HKSCS:2004 revision.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
42447
diff
changeset
|
341 self.fail('Decoding failed while testing %s -> %s: %s' % ( |
|
eca90b8aa7fd
Update big5hkscs codec to conform to the HKSCS:2004 revision.
Hye-Shik Chang <hyeshik@gmail.com>
parents:
42447
diff
changeset
|
342 repr(csetch), repr(unich), exc.reason)) |
|
31386
887ce39f95f2
Add CJK codecs support as discussed on python-dev. (SF #873597)
Hye-Shik Chang <hyeshik@gmail.com>
parents:
diff
changeset
|
343 |
|
70630
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
344 def test_errorhandle(self): |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
345 for source, scheme, expected in self.codectests: |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
346 if isinstance(source, bytes): |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
347 func = source.decode |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
348 else: |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
349 func = source.encode |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
350 if expected: |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
351 if isinstance(source, bytes): |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
352 result = func(self.encoding, scheme) |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
353 self.assertTrue(type(result) is unicode, type(result)) |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
354 self.assertEqual(result, expected, |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
355 '%r.decode(%r, %r)=%r != %r' |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
356 % (source, self.encoding, scheme, result, |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
357 expected)) |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
358 else: |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
359 result = func(self.encoding, scheme) |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
360 self.assertTrue(type(result) is bytes, type(result)) |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
361 self.assertEqual(result, expected, |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
362 '%r.encode(%r, %r)=%r != %r' |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
363 % (source, self.encoding, scheme, result, |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
364 expected)) |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
365 else: |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
366 self.assertRaises(UnicodeError, func, self.encoding, scheme) |
|
53912b58eee6
Issue #12016: Add test_errorhandle() to TestBase_Mapping of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
70540
diff
changeset
|
367 |
|
70178
83f4c270b27d
Issue #12057: Fix .hgeol and test_multibytecodec_support for the conversion of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
368 def load_teststring(name): |
|
83f4c270b27d
Issue #12057: Fix .hgeol and test_multibytecodec_support for the conversion of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
369 dir = os.path.join(os.path.dirname(__file__), 'cjkencodings') |
|
83f4c270b27d
Issue #12057: Fix .hgeol and test_multibytecodec_support for the conversion of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
370 with open(os.path.join(dir, name + '.txt'), 'rb') as f: |
|
83f4c270b27d
Issue #12057: Fix .hgeol and test_multibytecodec_support for the conversion of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
371 encoded = f.read() |
|
83f4c270b27d
Issue #12057: Fix .hgeol and test_multibytecodec_support for the conversion of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
372 with open(os.path.join(dir, name + '-utf8.txt'), 'rb') as f: |
|
83f4c270b27d
Issue #12057: Fix .hgeol and test_multibytecodec_support for the conversion of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
373 utf8 = f.read() |
|
83f4c270b27d
Issue #12057: Fix .hgeol and test_multibytecodec_support for the conversion of
Victor Stinner <victor.stinner@haypocalc.com>
parents:
68256
diff
changeset
|
374 return encoded, utf8 |
