annotate Lib/test/test_codecs.py @ 58765:2ba7677bfa95 legacy-trunk

testCopysign was defined twice in test_math; combine the definitions
author Mark Dickinson <dickinsm@gmail.com>
date Sat, 06 Feb 2010 23:11:25 +0000
parents 6ffcdafa0118
children 7e3c47354da3
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
24364
aa0d2d51fbde Get rid of relative imports in all unittests. Now anything that
Barry Warsaw <barry@python.org>
parents: 20013
diff changeset
1 from test import test_support
aa0d2d51fbde Get rid of relative imports in all unittests. Now anything that
Barry Warsaw <barry@python.org>
parents: 20013
diff changeset
2 import unittest
18163
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
3 import codecs
37077
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
4 import sys, StringIO, _testcapi
18163
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
5
33428
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
6 class Queue(object):
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
7 """
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
8 queue: write bytes at one end, read bytes from the other end
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
9 """
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
10 def __init__(self):
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
11 self._buffer = ""
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
12
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
13 def write(self, chars):
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
14 self._buffer += chars
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
15
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
16 def read(self, size=-1):
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
17 if size<0:
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
18 s = self._buffer
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
19 self._buffer = ""
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
20 return s
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
21 else:
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
22 s = self._buffer[:size]
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
23 self._buffer = self._buffer[size:]
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
24 return s
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
25
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
26 class ReadTest(unittest.TestCase):
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
27 def check_partial(self, input, partialresults):
33428
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
28 # get a StreamReader for the encoding and feed the bytestring version
43903
6aa8e135d9f8 Fix typo in comment.
Walter Dörwald <walter@livinglogic.de>
parents: 42649
diff changeset
29 # of input to the reader byte by byte. Read everything available from
33428
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
30 # the StreamReader and check that the results equal the appropriate
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
31 # entries from partialresults.
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
32 q = Queue()
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
33 r = codecs.getreader(self.encoding)(q)
33428
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
34 result = u""
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
35 for (c, partialresult) in zip(input.encode(self.encoding), partialresults):
33428
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
36 q.write(c)
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
37 result += r.read()
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
38 self.assertEqual(result, partialresult)
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
39 # check that there's nothing left in the buffers
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
40 self.assertEqual(r.read(), u"")
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
41 self.assertEqual(r.bytebuffer, "")
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
42 self.assertEqual(r.charbuffer, u"")
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
43
36996
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
44 # do the check again, this time using a incremental decoder
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
45 d = codecs.getincrementaldecoder(self.encoding)()
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
46 result = u""
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
47 for (c, partialresult) in zip(input.encode(self.encoding), partialresults):
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
48 result += d.decode(c)
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
49 self.assertEqual(result, partialresult)
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
50 # check that there's nothing left in the buffers
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
51 self.assertEqual(d.decode("", True), u"")
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
52 self.assertEqual(d.buffer, "")
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
53
43932
e8ac1ae4503c The incremental decoder for utf-7 must preserve its state between calls.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents: 43905
diff changeset
54 # Check whether the reset method works properly
36996
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
55 d.reset()
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
56 result = u""
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
57 for (c, partialresult) in zip(input.encode(self.encoding), partialresults):
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
58 result += d.decode(c)
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
59 self.assertEqual(result, partialresult)
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
60 # check that there's nothing left in the buffers
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
61 self.assertEqual(d.decode("", True), u"")
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
62 self.assertEqual(d.buffer, "")
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
63
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
64 # check iterdecode()
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
65 encoded = input.encode(self.encoding)
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
66 self.assertEqual(
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
67 input,
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
68 u"".join(codecs.iterdecode(encoded, self.encoding))
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
69 )
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
70
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
71 def test_readline(self):
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
72 def getreader(input):
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
73 stream = StringIO.StringIO(input.encode(self.encoding))
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
74 return codecs.getreader(self.encoding)(stream)
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
75
36855
204c307b01ce If size is specified, try to read at least size characters.
Walter Dörwald <walter@livinglogic.de>
parents: 36247
diff changeset
76 def readalllines(input, keepends=True, size=None):
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
77 reader = getreader(input)
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
78 lines = []
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
79 while True:
36855
204c307b01ce If size is specified, try to read at least size characters.
Walter Dörwald <walter@livinglogic.de>
parents: 36247
diff changeset
80 line = reader.readline(size=size, keepends=keepends)
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
81 if not line:
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
82 break
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
83 lines.append(line)
36855
204c307b01ce If size is specified, try to read at least size characters.
Walter Dörwald <walter@livinglogic.de>
parents: 36247
diff changeset
84 return "|".join(lines)
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
85
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
86 s = u"foo\nbar\r\nbaz\rspam\u2028eggs"
36855
204c307b01ce If size is specified, try to read at least size characters.
Walter Dörwald <walter@livinglogic.de>
parents: 36247
diff changeset
87 sexpected = u"foo\n|bar\r\n|baz\r|spam\u2028|eggs"
204c307b01ce If size is specified, try to read at least size characters.
Walter Dörwald <walter@livinglogic.de>
parents: 36247
diff changeset
88 sexpectednoends = u"foo|bar|baz|spam|eggs"
204c307b01ce If size is specified, try to read at least size characters.
Walter Dörwald <walter@livinglogic.de>
parents: 36247
diff changeset
89 self.assertEqual(readalllines(s, True), sexpected)
204c307b01ce If size is specified, try to read at least size characters.
Walter Dörwald <walter@livinglogic.de>
parents: 36247
diff changeset
90 self.assertEqual(readalllines(s, False), sexpectednoends)
204c307b01ce If size is specified, try to read at least size characters.
Walter Dörwald <walter@livinglogic.de>
parents: 36247
diff changeset
91 self.assertEqual(readalllines(s, True, 10), sexpected)
204c307b01ce If size is specified, try to read at least size characters.
Walter Dörwald <walter@livinglogic.de>
parents: 36247
diff changeset
92 self.assertEqual(readalllines(s, False, 10), sexpectednoends)
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
93
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
94 # Test long lines (multiple calls to read() in readline())
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
95 vw = []
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
96 vwo = []
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
97 for (i, lineend) in enumerate(u"\n \r\n \r \u2028".split()):
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
98 vw.append((i*200)*u"\3042" + lineend)
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
99 vwo.append((i*200)*u"\3042")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
100 self.assertEqual(readalllines("".join(vw), True), "".join(vw))
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
101 self.assertEqual(readalllines("".join(vw), False),"".join(vwo))
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
102
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
103 # Test lines where the first read might end with \r, so the
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
104 # reader has to look ahead whether this is a lone \r or a \r\n
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
105 for size in xrange(80):
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
106 for lineend in u"\n \r\n \r \u2028".split():
34865
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
107 s = 10*(size*u"a" + lineend + u"xxx\n")
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
108 reader = getreader(s)
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
109 for i in xrange(10):
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
110 self.assertEqual(
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
111 reader.readline(keepends=True),
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
112 size*u"a" + lineend,
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
113 )
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
114 reader = getreader(s)
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
115 for i in xrange(10):
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
116 self.assertEqual(
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
117 reader.readline(keepends=False),
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
118 size*u"a",
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
119 )
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
120
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
121 def test_bug1175396(self):
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
122 s = [
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
123 '<%!--===================================================\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
124 ' BLOG index page: show recent articles,\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
125 ' today\'s articles, or articles of a specific date.\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
126 '========================================================--%>\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
127 '<%@inputencoding="ISO-8859-1"%>\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
128 '<%@pagetemplate=TEMPLATE.y%>\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
129 '<%@import=import frog.util, frog%>\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
130 '<%@import=import frog.objects%>\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
131 '<%@import=from frog.storageerrors import StorageError%>\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
132 '<%\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
133 '\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
134 'import logging\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
135 'log=logging.getLogger("Snakelets.logger")\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
136 '\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
137 '\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
138 'user=self.SessionCtx.user\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
139 'storageEngine=self.SessionCtx.storageEngine\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
140 '\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
141 '\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
142 'def readArticlesFromDate(date, count=None):\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
143 ' entryids=storageEngine.listBlogEntries(date)\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
144 ' entryids.reverse() # descending\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
145 ' if count:\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
146 ' entryids=entryids[:count]\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
147 ' try:\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
148 ' return [ frog.objects.BlogEntry.load(storageEngine, date, Id) for Id in entryids ]\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
149 ' except StorageError,x:\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
150 ' log.error("Error loading articles: "+str(x))\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
151 ' self.abort("cannot load articles")\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
152 '\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
153 'showdate=None\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
154 '\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
155 'arg=self.Request.getArg()\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
156 'if arg=="today":\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
157 ' #-------------------- TODAY\'S ARTICLES\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
158 ' self.write("<h2>Today\'s articles</h2>")\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
159 ' showdate = frog.util.isodatestr() \r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
160 ' entries = readArticlesFromDate(showdate)\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
161 'elif arg=="active":\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
162 ' #-------------------- ACTIVE ARTICLES redirect\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
163 ' self.Yredirect("active.y")\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
164 'elif arg=="login":\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
165 ' #-------------------- LOGIN PAGE redirect\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
166 ' self.Yredirect("login.y")\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
167 'elif arg=="date":\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
168 ' #-------------------- ARTICLES OF A SPECIFIC DATE\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
169 ' showdate = self.Request.getParameter("date")\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
170 ' self.write("<h2>Articles written on %s</h2>"% frog.util.mediumdatestr(showdate))\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
171 ' entries = readArticlesFromDate(showdate)\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
172 'else:\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
173 ' #-------------------- RECENT ARTICLES\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
174 ' self.write("<h2>Recent articles</h2>")\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
175 ' dates=storageEngine.listBlogEntryDates()\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
176 ' if dates:\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
177 ' entries=[]\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
178 ' SHOWAMOUNT=10\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
179 ' for showdate in dates:\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
180 ' entries.extend( readArticlesFromDate(showdate, SHOWAMOUNT-len(entries)) )\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
181 ' if len(entries)>=SHOWAMOUNT:\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
182 ' break\r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
183 ' \r\n',
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
184 ]
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
185 stream = StringIO.StringIO("".join(s).encode(self.encoding))
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
186 reader = codecs.getreader(self.encoding)(stream)
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
187 for (i, line) in enumerate(reader):
b49d39111c48 Fix for SF bug #1175396: readline() will now read one more character, if
Walter Dörwald <walter@livinglogic.de>
parents: 34750
diff changeset
188 self.assertEqual(line, s[i])
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
189
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
190 def test_readlinequeue(self):
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
191 q = Queue()
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
192 writer = codecs.getwriter(self.encoding)(q)
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
193 reader = codecs.getreader(self.encoding)(q)
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
194
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
195 # No lineends
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
196 writer.write(u"foo\r")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
197 self.assertEqual(reader.readline(keepends=False), u"foo")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
198 writer.write(u"\nbar\r")
34925
1331c7799640 Update test to the current readline() behaviour.
Walter Dörwald <walter@livinglogic.de>
parents: 34865
diff changeset
199 self.assertEqual(reader.readline(keepends=False), u"")
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
200 self.assertEqual(reader.readline(keepends=False), u"bar")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
201 writer.write(u"baz")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
202 self.assertEqual(reader.readline(keepends=False), u"baz")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
203 self.assertEqual(reader.readline(keepends=False), u"")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
204
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
205 # Lineends
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
206 writer.write(u"foo\r")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
207 self.assertEqual(reader.readline(keepends=True), u"foo\r")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
208 writer.write(u"\nbar\r")
34925
1331c7799640 Update test to the current readline() behaviour.
Walter Dörwald <walter@livinglogic.de>
parents: 34865
diff changeset
209 self.assertEqual(reader.readline(keepends=True), u"\n")
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
210 self.assertEqual(reader.readline(keepends=True), u"bar\r")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
211 writer.write(u"baz")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
212 self.assertEqual(reader.readline(keepends=True), u"baz")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
213 self.assertEqual(reader.readline(keepends=True), u"")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
214 writer.write(u"foo\r\n")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
215 self.assertEqual(reader.readline(keepends=True), u"foo\r\n")
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
216
34384
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
217 def test_bug1098990_a(self):
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
218 s1 = u"xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy\r\n"
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
219 s2 = u"offending line: ladfj askldfj klasdj fskla dfzaskdj fasklfj laskd fjasklfzzzzaa%whereisthis!!!\r\n"
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
220 s3 = u"next line.\r\n"
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
221
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
222 s = (s1+s2+s3).encode(self.encoding)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
223 stream = StringIO.StringIO(s)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
224 reader = codecs.getreader(self.encoding)(stream)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
225 self.assertEqual(reader.readline(), s1)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
226 self.assertEqual(reader.readline(), s2)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
227 self.assertEqual(reader.readline(), s3)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
228 self.assertEqual(reader.readline(), u"")
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
229
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
230 def test_bug1098990_b(self):
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
231 s1 = u"aaaaaaaaaaaaaaaaaaaaaaaa\r\n"
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
232 s2 = u"bbbbbbbbbbbbbbbbbbbbbbbb\r\n"
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
233 s3 = u"stillokay:bbbbxx\r\n"
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
234 s4 = u"broken!!!!badbad\r\n"
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
235 s5 = u"againokay.\r\n"
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
236
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
237 s = (s1+s2+s3+s4+s5).encode(self.encoding)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
238 stream = StringIO.StringIO(s)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
239 reader = codecs.getreader(self.encoding)(stream)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
240 self.assertEqual(reader.readline(), s1)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
241 self.assertEqual(reader.readline(), s2)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
242 self.assertEqual(reader.readline(), s3)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
243 self.assertEqual(reader.readline(), s4)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
244 self.assertEqual(reader.readline(), s5)
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
245 self.assertEqual(reader.readline(), u"")
a216003cef04 Fix and test for SF bug #1098990: codec readline() splits lines apart.
Walter Dörwald <walter@livinglogic.de>
parents: 34279
diff changeset
246
42649
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
247 class UTF32Test(ReadTest):
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
248 encoding = "utf-32"
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
249
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
250 spamle = ('\xff\xfe\x00\x00'
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
251 's\x00\x00\x00p\x00\x00\x00a\x00\x00\x00m\x00\x00\x00'
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
252 's\x00\x00\x00p\x00\x00\x00a\x00\x00\x00m\x00\x00\x00')
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
253 spambe = ('\x00\x00\xfe\xff'
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
254 '\x00\x00\x00s\x00\x00\x00p\x00\x00\x00a\x00\x00\x00m'
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
255 '\x00\x00\x00s\x00\x00\x00p\x00\x00\x00a\x00\x00\x00m')
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
256
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
257 def test_only_one_bom(self):
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
258 _,_,reader,writer = codecs.lookup(self.encoding)
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
259 # encode some stream
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
260 s = StringIO.StringIO()
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
261 f = writer(s)
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
262 f.write(u"spam")
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
263 f.write(u"spam")
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
264 d = s.getvalue()
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
265 # check whether there is exactly one BOM in it
55002
6768b2350c57 convert usage of fail* to assert*
Benjamin Peterson <benjamin@python.org>
parents: 53855
diff changeset
266 self.assertTrue(d == self.spamle or d == self.spambe)
42649
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
267 # try to read it back
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
268 s = StringIO.StringIO(d)
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
269 f = reader(s)
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
270 self.assertEquals(f.read(), u"spamspam")
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
271
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
272 def test_badbom(self):
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
273 s = StringIO.StringIO(4*"\xff")
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
274 f = codecs.getreader(self.encoding)(s)
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
275 self.assertRaises(UnicodeError, f.read)
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
276
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
277 s = StringIO.StringIO(8*"\xff")
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
278 f = codecs.getreader(self.encoding)(s)
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
279 self.assertRaises(UnicodeError, f.read)
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
280
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
281 def test_partial(self):
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
282 self.check_partial(
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
283 u"\x00\xff\u0100\uffff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
284 [
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
285 u"", # first byte of BOM read
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
286 u"", # second byte of BOM read
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
287 u"", # third byte of BOM read
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
288 u"", # fourth byte of BOM read => byteorder known
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
289 u"",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
290 u"",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
291 u"",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
292 u"\x00",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
293 u"\x00",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
294 u"\x00",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
295 u"\x00",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
296 u"\x00\xff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
297 u"\x00\xff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
298 u"\x00\xff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
299 u"\x00\xff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
300 u"\x00\xff\u0100",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
301 u"\x00\xff\u0100",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
302 u"\x00\xff\u0100",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
303 u"\x00\xff\u0100",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
304 u"\x00\xff\u0100\uffff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
305 ]
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
306 )
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
307
56033
38a52b672429 Issue #6922: Fix an infinite loop when trying to decode an invalid
Georg Brandl <georg@python.org>
parents: 55002
diff changeset
308 def test_handlers(self):
38a52b672429 Issue #6922: Fix an infinite loop when trying to decode an invalid
Georg Brandl <georg@python.org>
parents: 55002
diff changeset
309 self.assertEqual((u'\ufffd', 1),
38a52b672429 Issue #6922: Fix an infinite loop when trying to decode an invalid
Georg Brandl <georg@python.org>
parents: 55002
diff changeset
310 codecs.utf_32_decode('\x01', 'replace', True))
38a52b672429 Issue #6922: Fix an infinite loop when trying to decode an invalid
Georg Brandl <georg@python.org>
parents: 55002
diff changeset
311 self.assertEqual((u'', 1),
38a52b672429 Issue #6922: Fix an infinite loop when trying to decode an invalid
Georg Brandl <georg@python.org>
parents: 55002
diff changeset
312 codecs.utf_32_decode('\x01', 'ignore', True))
38a52b672429 Issue #6922: Fix an infinite loop when trying to decode an invalid
Georg Brandl <georg@python.org>
parents: 55002
diff changeset
313
42649
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
314 def test_errors(self):
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
315 self.assertRaises(UnicodeDecodeError, codecs.utf_32_decode,
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
316 "\xff", "strict", True)
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
317
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
318 class UTF32LETest(ReadTest):
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
319 encoding = "utf-32-le"
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
320
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
321 def test_partial(self):
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
322 self.check_partial(
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
323 u"\x00\xff\u0100\uffff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
324 [
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
325 u"",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
326 u"",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
327 u"",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
328 u"\x00",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
329 u"\x00",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
330 u"\x00",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
331 u"\x00",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
332 u"\x00\xff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
333 u"\x00\xff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
334 u"\x00\xff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
335 u"\x00\xff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
336 u"\x00\xff\u0100",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
337 u"\x00\xff\u0100",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
338 u"\x00\xff\u0100",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
339 u"\x00\xff\u0100",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
340 u"\x00\xff\u0100\uffff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
341 ]
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
342 )
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
343
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
344 def test_simple(self):
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
345 self.assertEqual(u"\U00010203".encode(self.encoding), "\x03\x02\x01\x00")
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
346
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
347 def test_errors(self):
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
348 self.assertRaises(UnicodeDecodeError, codecs.utf_32_le_decode,
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
349 "\xff", "strict", True)
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
350
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
351 class UTF32BETest(ReadTest):
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
352 encoding = "utf-32-be"
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
353
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
354 def test_partial(self):
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
355 self.check_partial(
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
356 u"\x00\xff\u0100\uffff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
357 [
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
358 u"",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
359 u"",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
360 u"",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
361 u"\x00",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
362 u"\x00",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
363 u"\x00",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
364 u"\x00",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
365 u"\x00\xff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
366 u"\x00\xff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
367 u"\x00\xff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
368 u"\x00\xff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
369 u"\x00\xff\u0100",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
370 u"\x00\xff\u0100",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
371 u"\x00\xff\u0100",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
372 u"\x00\xff\u0100",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
373 u"\x00\xff\u0100\uffff",
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
374 ]
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
375 )
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
376
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
377 def test_simple(self):
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
378 self.assertEqual(u"\U00010203".encode(self.encoding), "\x00\x01\x02\x03")
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
379
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
380 def test_errors(self):
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
381 self.assertRaises(UnicodeDecodeError, codecs.utf_32_be_decode,
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
382 "\xff", "strict", True)
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
383
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
384 class UTF16Test(ReadTest):
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
385 encoding = "utf-16"
18163
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
386
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
387 spamle = '\xff\xfes\x00p\x00a\x00m\x00s\x00p\x00a\x00m\x00'
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
388 spambe = '\xfe\xff\x00s\x00p\x00a\x00m\x00s\x00p\x00a\x00m'
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
389
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
390 def test_only_one_bom(self):
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
391 _,_,reader,writer = codecs.lookup(self.encoding)
18163
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
392 # encode some stream
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
393 s = StringIO.StringIO()
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
394 f = writer(s)
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
395 f.write(u"spam")
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
396 f.write(u"spam")
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
397 d = s.getvalue()
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
398 # check whether there is exactly one BOM in it
55002
6768b2350c57 convert usage of fail* to assert*
Benjamin Peterson <benjamin@python.org>
parents: 53855
diff changeset
399 self.assertTrue(d == self.spamle or d == self.spambe)
18163
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
400 # try to read it back
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
401 s = StringIO.StringIO(d)
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
402 f = reader(s)
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
403 self.assertEquals(f.read(), u"spamspam")
f926eb1f5770 Test by Martin v. Loewis for the new UTF-16 codec handling of BOM
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
404
34538
37b2f11c1a00 Add a test for UTF-16 reading where the byte sequence doesn't start with
Walter Dörwald <walter@livinglogic.de>
parents: 34384
diff changeset
405 def test_badbom(self):
37b2f11c1a00 Add a test for UTF-16 reading where the byte sequence doesn't start with
Walter Dörwald <walter@livinglogic.de>
parents: 34384
diff changeset
406 s = StringIO.StringIO("\xff\xff")
34562
7a1277fad04e Fix stupid typo: Don't read from a writer.
Walter Dörwald <walter@livinglogic.de>
parents: 34538
diff changeset
407 f = codecs.getreader(self.encoding)(s)
34538
37b2f11c1a00 Add a test for UTF-16 reading where the byte sequence doesn't start with
Walter Dörwald <walter@livinglogic.de>
parents: 34384
diff changeset
408 self.assertRaises(UnicodeError, f.read)
37b2f11c1a00 Add a test for UTF-16 reading where the byte sequence doesn't start with
Walter Dörwald <walter@livinglogic.de>
parents: 34384
diff changeset
409
37b2f11c1a00 Add a test for UTF-16 reading where the byte sequence doesn't start with
Walter Dörwald <walter@livinglogic.de>
parents: 34384
diff changeset
410 s = StringIO.StringIO("\xff\xff\xff\xff")
34562
7a1277fad04e Fix stupid typo: Don't read from a writer.
Walter Dörwald <walter@livinglogic.de>
parents: 34538
diff changeset
411 f = codecs.getreader(self.encoding)(s)
34538
37b2f11c1a00 Add a test for UTF-16 reading where the byte sequence doesn't start with
Walter Dörwald <walter@livinglogic.de>
parents: 34384
diff changeset
412 self.assertRaises(UnicodeError, f.read)
37b2f11c1a00 Add a test for UTF-16 reading where the byte sequence doesn't start with
Walter Dörwald <walter@livinglogic.de>
parents: 34384
diff changeset
413
33428
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
414 def test_partial(self):
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
415 self.check_partial(
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
416 u"\x00\xff\u0100\uffff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
417 [
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
418 u"", # first byte of BOM read
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
419 u"", # second byte of BOM read => byteorder known
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
420 u"",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
421 u"\x00",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
422 u"\x00",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
423 u"\x00\xff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
424 u"\x00\xff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
425 u"\x00\xff\u0100",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
426 u"\x00\xff\u0100",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
427 u"\x00\xff\u0100\uffff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
428 ]
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
429 )
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
430
56033
38a52b672429 Issue #6922: Fix an infinite loop when trying to decode an invalid
Georg Brandl <georg@python.org>
parents: 55002
diff changeset
431 def test_handlers(self):
38a52b672429 Issue #6922: Fix an infinite loop when trying to decode an invalid
Georg Brandl <georg@python.org>
parents: 55002
diff changeset
432 self.assertEqual((u'\ufffd', 1),
38a52b672429 Issue #6922: Fix an infinite loop when trying to decode an invalid
Georg Brandl <georg@python.org>
parents: 55002
diff changeset
433 codecs.utf_16_decode('\x01', 'replace', True))
38a52b672429 Issue #6922: Fix an infinite loop when trying to decode an invalid
Georg Brandl <georg@python.org>
parents: 55002
diff changeset
434 self.assertEqual((u'', 1),
38a52b672429 Issue #6922: Fix an infinite loop when trying to decode an invalid
Georg Brandl <georg@python.org>
parents: 55002
diff changeset
435 codecs.utf_16_decode('\x01', 'ignore', True))
38a52b672429 Issue #6922: Fix an infinite loop when trying to decode an invalid
Georg Brandl <georg@python.org>
parents: 55002
diff changeset
436
35882
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
437 def test_errors(self):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
438 self.assertRaises(UnicodeDecodeError, codecs.utf_16_decode, "\xff", "strict", True)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
439
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
440 class UTF16LETest(ReadTest):
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
441 encoding = "utf-16-le"
33428
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
442
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
443 def test_partial(self):
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
444 self.check_partial(
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
445 u"\x00\xff\u0100\uffff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
446 [
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
447 u"",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
448 u"\x00",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
449 u"\x00",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
450 u"\x00\xff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
451 u"\x00\xff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
452 u"\x00\xff\u0100",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
453 u"\x00\xff\u0100",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
454 u"\x00\xff\u0100\uffff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
455 ]
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
456 )
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
457
35882
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
458 def test_errors(self):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
459 self.assertRaises(UnicodeDecodeError, codecs.utf_16_le_decode, "\xff", "strict", True)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
460
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
461 class UTF16BETest(ReadTest):
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
462 encoding = "utf-16-be"
33428
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
463
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
464 def test_partial(self):
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
465 self.check_partial(
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
466 u"\x00\xff\u0100\uffff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
467 [
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
468 u"",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
469 u"\x00",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
470 u"\x00",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
471 u"\x00\xff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
472 u"\x00\xff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
473 u"\x00\xff\u0100",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
474 u"\x00\xff\u0100",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
475 u"\x00\xff\u0100\uffff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
476 ]
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
477 )
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
478
35882
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
479 def test_errors(self):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
480 self.assertRaises(UnicodeDecodeError, codecs.utf_16_be_decode, "\xff", "strict", True)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
481
34214
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
482 class UTF8Test(ReadTest):
abe3145dde4e Backport checkin:
Walter Dörwald <walter@livinglogic.de>
parents: 33855
diff changeset
483 encoding = "utf-8"
33428
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
484
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
485 def test_partial(self):
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
486 self.check_partial(
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
487 u"\x00\xff\u07ff\u0800\uffff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
488 [
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
489 u"\x00",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
490 u"\x00",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
491 u"\x00\xff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
492 u"\x00\xff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
493 u"\x00\xff\u07ff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
494 u"\x00\xff\u07ff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
495 u"\x00\xff\u07ff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
496 u"\x00\xff\u07ff\u0800",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
497 u"\x00\xff\u07ff\u0800",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
498 u"\x00\xff\u07ff\u0800",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
499 u"\x00\xff\u07ff\u0800\uffff",
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
500 ]
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
501 )
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
502
35882
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
503 class UTF7Test(ReadTest):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
504 encoding = "utf-7"
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
505
43932
e8ac1ae4503c The incremental decoder for utf-7 must preserve its state between calls.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents: 43905
diff changeset
506 def test_partial(self):
e8ac1ae4503c The incremental decoder for utf-7 must preserve its state between calls.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents: 43905
diff changeset
507 self.check_partial(
e8ac1ae4503c The incremental decoder for utf-7 must preserve its state between calls.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents: 43905
diff changeset
508 u"a+-b",
e8ac1ae4503c The incremental decoder for utf-7 must preserve its state between calls.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents: 43905
diff changeset
509 [
e8ac1ae4503c The incremental decoder for utf-7 must preserve its state between calls.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents: 43905
diff changeset
510 u"a",
e8ac1ae4503c The incremental decoder for utf-7 must preserve its state between calls.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents: 43905
diff changeset
511 u"a",
e8ac1ae4503c The incremental decoder for utf-7 must preserve its state between calls.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents: 43905
diff changeset
512 u"a+",
e8ac1ae4503c The incremental decoder for utf-7 must preserve its state between calls.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents: 43905
diff changeset
513 u"a+-",
e8ac1ae4503c The incremental decoder for utf-7 must preserve its state between calls.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents: 43905
diff changeset
514 u"a+-b",
e8ac1ae4503c The incremental decoder for utf-7 must preserve its state between calls.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents: 43905
diff changeset
515 ]
e8ac1ae4503c The incremental decoder for utf-7 must preserve its state between calls.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents: 43905
diff changeset
516 )
35882
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
517
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
518 class UTF16ExTest(unittest.TestCase):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
519
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
520 def test_errors(self):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
521 self.assertRaises(UnicodeDecodeError, codecs.utf_16_ex_decode, "\xff", "strict", 0, True)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
522
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
523 def test_bad_args(self):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
524 self.assertRaises(TypeError, codecs.utf_16_ex_decode)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
525
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
526 class ReadBufferTest(unittest.TestCase):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
527
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
528 def test_array(self):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
529 import array
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
530 self.assertEqual(
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
531 codecs.readbuffer_encode(array.array("c", "spam")),
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
532 ("spam", 4)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
533 )
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
534
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
535 def test_empty(self):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
536 self.assertEqual(codecs.readbuffer_encode(""), ("", 0))
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
537
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
538 def test_bad_args(self):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
539 self.assertRaises(TypeError, codecs.readbuffer_encode)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
540 self.assertRaises(TypeError, codecs.readbuffer_encode, 42)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
541
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
542 class CharBufferTest(unittest.TestCase):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
543
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
544 def test_string(self):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
545 self.assertEqual(codecs.charbuffer_encode("spam"), ("spam", 4))
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
546
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
547 def test_empty(self):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
548 self.assertEqual(codecs.charbuffer_encode(""), ("", 0))
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
549
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
550 def test_bad_args(self):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
551 self.assertRaises(TypeError, codecs.charbuffer_encode)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
552 self.assertRaises(TypeError, codecs.charbuffer_encode, 42)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
553
36247
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
554 class UTF8SigTest(ReadTest):
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
555 encoding = "utf-8-sig"
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
556
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
557 def test_partial(self):
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
558 self.check_partial(
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
559 u"\ufeff\x00\xff\u07ff\u0800\uffff",
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
560 [
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
561 u"",
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
562 u"",
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
563 u"", # First BOM has been read and skipped
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
564 u"",
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
565 u"",
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
566 u"\ufeff", # Second BOM has been read and emitted
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
567 u"\ufeff\x00", # "\x00" read and emitted
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
568 u"\ufeff\x00", # First byte of encoded u"\xff" read
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
569 u"\ufeff\x00\xff", # Second byte of encoded u"\xff" read
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
570 u"\ufeff\x00\xff", # First byte of encoded u"\u07ff" read
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
571 u"\ufeff\x00\xff\u07ff", # Second byte of encoded u"\u07ff" read
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
572 u"\ufeff\x00\xff\u07ff",
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
573 u"\ufeff\x00\xff\u07ff",
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
574 u"\ufeff\x00\xff\u07ff\u0800",
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
575 u"\ufeff\x00\xff\u07ff\u0800",
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
576 u"\ufeff\x00\xff\u07ff\u0800",
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
577 u"\ufeff\x00\xff\u07ff\u0800\uffff",
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
578 ]
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
579 )
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
580
40386
a3e0f1e11082 Change decode() so that it works with a buffer (i.e. unicode(..., 'utf-8-sig'))
Walter Dörwald <walter@livinglogic.de>
parents: 40284
diff changeset
581 def test_bug1601501(self):
a3e0f1e11082 Change decode() so that it works with a buffer (i.e. unicode(..., 'utf-8-sig'))
Walter Dörwald <walter@livinglogic.de>
parents: 40284
diff changeset
582 # SF bug #1601501: check that the codec works with a buffer
a3e0f1e11082 Change decode() so that it works with a buffer (i.e. unicode(..., 'utf-8-sig'))
Walter Dörwald <walter@livinglogic.de>
parents: 40284
diff changeset
583 unicode("\xef\xbb\xbf", "utf-8-sig")
a3e0f1e11082 Change decode() so that it works with a buffer (i.e. unicode(..., 'utf-8-sig'))
Walter Dörwald <walter@livinglogic.de>
parents: 40284
diff changeset
584
41445
8447314f500a Fix utf-8-sig incremental decoder, which didn't recognise a BOM when the
Walter Dörwald <walter@livinglogic.de>
parents: 40386
diff changeset
585 def test_bom(self):
8447314f500a Fix utf-8-sig incremental decoder, which didn't recognise a BOM when the
Walter Dörwald <walter@livinglogic.de>
parents: 40386
diff changeset
586 d = codecs.getincrementaldecoder("utf-8-sig")()
8447314f500a Fix utf-8-sig incremental decoder, which didn't recognise a BOM when the
Walter Dörwald <walter@livinglogic.de>
parents: 40386
diff changeset
587 s = u"spam"
8447314f500a Fix utf-8-sig incremental decoder, which didn't recognise a BOM when the
Walter Dörwald <walter@livinglogic.de>
parents: 40386
diff changeset
588 self.assertEqual(d.decode(s.encode("utf-8-sig")), s)
8447314f500a Fix utf-8-sig incremental decoder, which didn't recognise a BOM when the
Walter Dörwald <walter@livinglogic.de>
parents: 40386
diff changeset
589
43905
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
590 def test_stream_bom(self):
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
591 unistring = u"ABC\u00A1\u2200XYZ"
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
592 bytestring = codecs.BOM_UTF8 + "ABC\xC2\xA1\xE2\x88\x80XYZ"
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
593
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
594 reader = codecs.getreader("utf-8-sig")
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
595 for sizehint in [None] + range(1, 11) + \
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
596 [64, 128, 256, 512, 1024]:
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
597 istream = reader(StringIO.StringIO(bytestring))
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
598 ostream = StringIO.StringIO()
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
599 while 1:
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
600 if sizehint is not None:
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
601 data = istream.read(sizehint)
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
602 else:
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
603 data = istream.read()
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
604
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
605 if not data:
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
606 break
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
607 ostream.write(data)
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
608
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
609 got = ostream.getvalue()
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
610 self.assertEqual(got, unistring)
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
611
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
612 def test_stream_bare(self):
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
613 unistring = u"ABC\u00A1\u2200XYZ"
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
614 bytestring = "ABC\xC2\xA1\xE2\x88\x80XYZ"
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
615
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
616 reader = codecs.getreader("utf-8-sig")
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
617 for sizehint in [None] + range(1, 11) + \
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
618 [64, 128, 256, 512, 1024]:
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
619 istream = reader(StringIO.StringIO(bytestring))
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
620 ostream = StringIO.StringIO()
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
621 while 1:
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
622 if sizehint is not None:
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
623 data = istream.read(sizehint)
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
624 else:
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
625 data = istream.read()
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
626
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
627 if not data:
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
628 break
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
629 ostream.write(data)
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
630
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
631 got = ostream.getvalue()
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
632 self.assertEqual(got, unistring)
0e897905ee94 Fix for #1444: utf_8_sig.StreamReader was (indirectly through decode())
Walter Dörwald <walter@livinglogic.de>
parents: 43903
diff changeset
633
25100
bc590beadf0b Check whether a string resize is necessary at the end
Walter Dörwald <walter@livinglogic.de>
parents: 24364
diff changeset
634 class EscapeDecodeTest(unittest.TestCase):
35882
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
635 def test_empty(self):
25100
bc590beadf0b Check whether a string resize is necessary at the end
Walter Dörwald <walter@livinglogic.de>
parents: 24364
diff changeset
636 self.assertEquals(codecs.escape_decode(""), ("", 0))
bc590beadf0b Check whether a string resize is necessary at the end
Walter Dörwald <walter@livinglogic.de>
parents: 24364
diff changeset
637
27680
b19aaaeb6d64 Fix for [ 543344 ] Interpreter crashes when recoding; suggested
Marc-André Lemburg <mal@egenix.com>
parents: 25100
diff changeset
638 class RecodingTest(unittest.TestCase):
b19aaaeb6d64 Fix for [ 543344 ] Interpreter crashes when recoding; suggested
Marc-André Lemburg <mal@egenix.com>
parents: 25100
diff changeset
639 def test_recoding(self):
b19aaaeb6d64 Fix for [ 543344 ] Interpreter crashes when recoding; suggested
Marc-André Lemburg <mal@egenix.com>
parents: 25100
diff changeset
640 f = StringIO.StringIO()
b19aaaeb6d64 Fix for [ 543344 ] Interpreter crashes when recoding; suggested
Marc-André Lemburg <mal@egenix.com>
parents: 25100
diff changeset
641 f2 = codecs.EncodedFile(f, "unicode_internal", "utf-8")
b19aaaeb6d64 Fix for [ 543344 ] Interpreter crashes when recoding; suggested
Marc-André Lemburg <mal@egenix.com>
parents: 25100
diff changeset
642 f2.write(u"a")
b19aaaeb6d64 Fix for [ 543344 ] Interpreter crashes when recoding; suggested
Marc-André Lemburg <mal@egenix.com>
parents: 25100
diff changeset
643 f2.close()
b19aaaeb6d64 Fix for [ 543344 ] Interpreter crashes when recoding; suggested
Marc-André Lemburg <mal@egenix.com>
parents: 25100
diff changeset
644 # Python used to crash on this at exit because of a refcount
b19aaaeb6d64 Fix for [ 543344 ] Interpreter crashes when recoding; suggested
Marc-André Lemburg <mal@egenix.com>
parents: 25100
diff changeset
645 # bug in _codecsmodule.c
20013
e3184e46303b Change the PyUnit-based tests to use the test_main() approach. This
Fred Drake <fdrake@acm.org>
parents: 18163
diff changeset
646
28671
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
647 # From RFC 3492
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
648 punycode_testcases = [
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
649 # A Arabic (Egyptian):
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
650 (u"\u0644\u064A\u0647\u0645\u0627\u0628\u062A\u0643\u0644"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
651 u"\u0645\u0648\u0634\u0639\u0631\u0628\u064A\u061F",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
652 "egbpdaj6bu4bxfgehfvwxn"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
653 # B Chinese (simplified):
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
654 (u"\u4ED6\u4EEC\u4E3A\u4EC0\u4E48\u4E0D\u8BF4\u4E2D\u6587",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
655 "ihqwcrb4cv8a8dqg056pqjye"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
656 # C Chinese (traditional):
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
657 (u"\u4ED6\u5011\u7232\u4EC0\u9EBD\u4E0D\u8AAA\u4E2D\u6587",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
658 "ihqwctvzc91f659drss3x8bo0yb"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
659 # D Czech: Pro<ccaron>prost<ecaron>nemluv<iacute><ccaron>esky
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
660 (u"\u0050\u0072\u006F\u010D\u0070\u0072\u006F\u0073\u0074"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
661 u"\u011B\u006E\u0065\u006D\u006C\u0075\u0076\u00ED\u010D"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
662 u"\u0065\u0073\u006B\u0079",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
663 "Proprostnemluvesky-uyb24dma41a"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
664 # E Hebrew:
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
665 (u"\u05DC\u05DE\u05D4\u05D4\u05DD\u05E4\u05E9\u05D5\u05D8"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
666 u"\u05DC\u05D0\u05DE\u05D3\u05D1\u05E8\u05D9\u05DD\u05E2"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
667 u"\u05D1\u05E8\u05D9\u05EA",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
668 "4dbcagdahymbxekheh6e0a7fei0b"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
669 # F Hindi (Devanagari):
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
670 (u"\u092F\u0939\u0932\u094B\u0917\u0939\u093F\u0928\u094D"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
671 u"\u0926\u0940\u0915\u094D\u092F\u094B\u0902\u0928\u0939"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
672 u"\u0940\u0902\u092C\u094B\u0932\u0938\u0915\u0924\u0947"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
673 u"\u0939\u0948\u0902",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
674 "i1baa7eci9glrd9b2ae1bj0hfcgg6iyaf8o0a1dig0cd"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
675
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
676 #(G) Japanese (kanji and hiragana):
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
677 (u"\u306A\u305C\u307F\u3093\u306A\u65E5\u672C\u8A9E\u3092"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
678 u"\u8A71\u3057\u3066\u304F\u308C\u306A\u3044\u306E\u304B",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
679 "n8jok5ay5dzabd5bym9f0cm5685rrjetr6pdxa"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
680
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
681 # (H) Korean (Hangul syllables):
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
682 (u"\uC138\uACC4\uC758\uBAA8\uB4E0\uC0AC\uB78C\uB4E4\uC774"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
683 u"\uD55C\uAD6D\uC5B4\uB97C\uC774\uD574\uD55C\uB2E4\uBA74"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
684 u"\uC5BC\uB9C8\uB098\uC88B\uC744\uAE4C",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
685 "989aomsvi5e83db1d2a355cv1e0vak1dwrv93d5xbh15a0dt30a5j"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
686 "psd879ccm6fea98c"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
687
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
688 # (I) Russian (Cyrillic):
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
689 (u"\u043F\u043E\u0447\u0435\u043C\u0443\u0436\u0435\u043E"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
690 u"\u043D\u0438\u043D\u0435\u0433\u043E\u0432\u043E\u0440"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
691 u"\u044F\u0442\u043F\u043E\u0440\u0443\u0441\u0441\u043A"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
692 u"\u0438",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
693 "b1abfaaepdrnnbgefbaDotcwatmq2g4l"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
694
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
695 # (J) Spanish: Porqu<eacute>nopuedensimplementehablarenEspa<ntilde>ol
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
696 (u"\u0050\u006F\u0072\u0071\u0075\u00E9\u006E\u006F\u0070"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
697 u"\u0075\u0065\u0064\u0065\u006E\u0073\u0069\u006D\u0070"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
698 u"\u006C\u0065\u006D\u0065\u006E\u0074\u0065\u0068\u0061"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
699 u"\u0062\u006C\u0061\u0072\u0065\u006E\u0045\u0073\u0070"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
700 u"\u0061\u00F1\u006F\u006C",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
701 "PorqunopuedensimplementehablarenEspaol-fmd56a"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
702
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
703 # (K) Vietnamese:
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
704 # T<adotbelow>isaoh<odotbelow>kh<ocirc>ngth<ecirchookabove>ch\
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
705 # <ihookabove>n<oacute>iti<ecircacute>ngVi<ecircdotbelow>t
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
706 (u"\u0054\u1EA1\u0069\u0073\u0061\u006F\u0068\u1ECD\u006B"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
707 u"\u0068\u00F4\u006E\u0067\u0074\u0068\u1EC3\u0063\u0068"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
708 u"\u1EC9\u006E\u00F3\u0069\u0074\u0069\u1EBF\u006E\u0067"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
709 u"\u0056\u0069\u1EC7\u0074",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
710 "TisaohkhngthchnitingVit-kjcr8268qyxafd2f1b9g"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
711
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
712 #(L) 3<nen>B<gumi><kinpachi><sensei>
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
713 (u"\u0033\u5E74\u0042\u7D44\u91D1\u516B\u5148\u751F",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
714 "3B-ww4c5e180e575a65lsy2b"),
28774
c9bee11eaa8b Whitespace normalization.
Tim Peters <tim.peters@gmail.com>
parents: 28678
diff changeset
715
28671
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
716 # (M) <amuro><namie>-with-SUPER-MONKEYS
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
717 (u"\u5B89\u5BA4\u5948\u7F8E\u6075\u002D\u0077\u0069\u0074"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
718 u"\u0068\u002D\u0053\u0055\u0050\u0045\u0052\u002D\u004D"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
719 u"\u004F\u004E\u004B\u0045\u0059\u0053",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
720 "-with-SUPER-MONKEYS-pc58ag80a8qai00g7n9n"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
721
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
722 # (N) Hello-Another-Way-<sorezore><no><basho>
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
723 (u"\u0048\u0065\u006C\u006C\u006F\u002D\u0041\u006E\u006F"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
724 u"\u0074\u0068\u0065\u0072\u002D\u0057\u0061\u0079\u002D"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
725 u"\u305D\u308C\u305E\u308C\u306E\u5834\u6240",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
726 "Hello-Another-Way--fc4qua05auwb3674vfr0b"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
727
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
728 # (O) <hitotsu><yane><no><shita>2
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
729 (u"\u3072\u3068\u3064\u5C4B\u6839\u306E\u4E0B\u0032",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
730 "2-u9tlzr9756bt3uc0v"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
731
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
732 # (P) Maji<de>Koi<suru>5<byou><mae>
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
733 (u"\u004D\u0061\u006A\u0069\u3067\u004B\u006F\u0069\u3059"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
734 u"\u308B\u0035\u79D2\u524D",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
735 "MajiKoi5-783gue6qz075azm5e"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
736
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
737 # (Q) <pafii>de<runba>
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
738 (u"\u30D1\u30D5\u30A3\u30FC\u0064\u0065\u30EB\u30F3\u30D0",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
739 "de-jg4avhby1noc0d"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
740
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
741 # (R) <sono><supiido><de>
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
742 (u"\u305D\u306E\u30B9\u30D4\u30FC\u30C9\u3067",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
743 "d9juau41awczczp"),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
744
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
745 # (S) -> $1.00 <-
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
746 (u"\u002D\u003E\u0020\u0024\u0031\u002E\u0030\u0030\u0020"
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
747 u"\u003C\u002D",
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
748 "-> $1.00 <--")
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
749 ]
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
750
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
751 for i in punycode_testcases:
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
752 if len(i)!=2:
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
753 print repr(i)
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
754
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
755 class PunycodeTest(unittest.TestCase):
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
756 def test_encode(self):
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
757 for uni, puny in punycode_testcases:
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
758 # Need to convert both strings to lower case, since
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
759 # some of the extended encodings use upper case, but our
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
760 # code produces only lower case. Converting just puny to
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
761 # lower is also insufficient, since some of the input characters
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
762 # are upper case.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
763 self.assertEquals(uni.encode("punycode").lower(), puny.lower())
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
764
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
765 def test_decode(self):
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
766 for uni, puny in punycode_testcases:
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
767 self.assertEquals(uni, puny.decode("punycode"))
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
768
35484
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
769 class UnicodeInternalTest(unittest.TestCase):
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
770 def test_bug1251300(self):
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
771 # Decoding with unicode_internal used to not correctly handle "code
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
772 # points" above 0x10ffff on UCS-4 builds.
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
773 if sys.maxunicode > 0xffff:
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
774 ok = [
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
775 ("\x00\x10\xff\xff", u"\U0010ffff"),
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
776 ("\x00\x00\x01\x01", u"\U00000101"),
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
777 ("", u""),
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
778 ]
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
779 not_ok = [
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
780 "\x7f\xff\xff\xff",
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
781 "\x80\x00\x00\x00",
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
782 "\x81\x00\x00\x00",
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
783 "\x00",
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
784 "\x00\x00\x00\x00\x00",
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
785 ]
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
786 for internal, uni in ok:
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
787 if sys.byteorder == "little":
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
788 internal = "".join(reversed(internal))
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
789 self.assertEquals(uni, internal.decode("unicode_internal"))
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
790 for internal in not_ok:
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
791 if sys.byteorder == "little":
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
792 internal = "".join(reversed(internal))
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
793 self.assertRaises(UnicodeDecodeError, internal.decode,
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
794 "unicode_internal")
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
795
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
796 def test_decode_error_attributes(self):
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
797 if sys.maxunicode > 0xffff:
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
798 try:
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
799 "\x00\x00\x00\x00\x00\x11\x11\x00".decode("unicode_internal")
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
800 except UnicodeDecodeError, ex:
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
801 self.assertEquals("unicode_internal", ex.encoding)
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
802 self.assertEquals("\x00\x00\x00\x00\x00\x11\x11\x00", ex.object)
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
803 self.assertEquals(4, ex.start)
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
804 self.assertEquals(8, ex.end)
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
805 else:
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
806 self.fail()
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
807
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
808 def test_decode_callback(self):
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
809 if sys.maxunicode > 0xffff:
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
810 codecs.register_error("UnicodeInternalTest", codecs.ignore_errors)
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
811 decoder = codecs.getdecoder("unicode_internal")
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
812 ab = u"ab".encode("unicode_internal")
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
813 ignored = decoder("%s\x22\x22\x22\x22%s" % (ab[:4], ab[4:]),
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
814 "UnicodeInternalTest")
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
815 self.assertEquals((u"ab", 12), ignored)
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
816
53855
7dca250343ea Issue 3739: The unicode-internal encoder now reports the number of *characters*
Walter Dörwald <walter@livinglogic.de>
parents: 43932
diff changeset
817 def test_encode_length(self):
7dca250343ea Issue 3739: The unicode-internal encoder now reports the number of *characters*
Walter Dörwald <walter@livinglogic.de>
parents: 43932
diff changeset
818 # Issue 3739
7dca250343ea Issue 3739: The unicode-internal encoder now reports the number of *characters*
Walter Dörwald <walter@livinglogic.de>
parents: 43932
diff changeset
819 encoder = codecs.getencoder("unicode_internal")
7dca250343ea Issue 3739: The unicode-internal encoder now reports the number of *characters*
Walter Dörwald <walter@livinglogic.de>
parents: 43932
diff changeset
820 self.assertEquals(encoder(u"a")[1], 1)
7dca250343ea Issue 3739: The unicode-internal encoder now reports the number of *characters*
Walter Dörwald <walter@livinglogic.de>
parents: 43932
diff changeset
821 self.assertEquals(encoder(u"\xe9\u0142")[1], 2)
7dca250343ea Issue 3739: The unicode-internal encoder now reports the number of *characters*
Walter Dörwald <walter@livinglogic.de>
parents: 43932
diff changeset
822
28671
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
823 # From http://www.gnu.org/software/libidn/draft-josefsson-idn-test-vectors.html
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
824 nameprep_tests = [
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
825 # 3.1 Map to nothing.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
826 ('foo\xc2\xad\xcd\x8f\xe1\xa0\x86\xe1\xa0\x8bbar'
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
827 '\xe2\x80\x8b\xe2\x81\xa0baz\xef\xb8\x80\xef\xb8\x88\xef'
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
828 '\xb8\x8f\xef\xbb\xbf',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
829 'foobarbaz'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
830 # 3.2 Case folding ASCII U+0043 U+0041 U+0046 U+0045.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
831 ('CAFE',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
832 'cafe'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
833 # 3.3 Case folding 8bit U+00DF (german sharp s).
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
834 # The original test case is bogus; it says \xc3\xdf
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
835 ('\xc3\x9f',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
836 'ss'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
837 # 3.4 Case folding U+0130 (turkish capital I with dot).
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
838 ('\xc4\xb0',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
839 'i\xcc\x87'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
840 # 3.5 Case folding multibyte U+0143 U+037A.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
841 ('\xc5\x83\xcd\xba',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
842 '\xc5\x84 \xce\xb9'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
843 # 3.6 Case folding U+2121 U+33C6 U+1D7BB.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
844 # XXX: skip this as it fails in UCS-2 mode
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
845 #('\xe2\x84\xa1\xe3\x8f\x86\xf0\x9d\x9e\xbb',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
846 # 'telc\xe2\x88\x95kg\xcf\x83'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
847 (None, None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
848 # 3.7 Normalization of U+006a U+030c U+00A0 U+00AA.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
849 ('j\xcc\x8c\xc2\xa0\xc2\xaa',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
850 '\xc7\xb0 a'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
851 # 3.8 Case folding U+1FB7 and normalization.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
852 ('\xe1\xbe\xb7',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
853 '\xe1\xbe\xb6\xce\xb9'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
854 # 3.9 Self-reverting case folding U+01F0 and normalization.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
855 # The original test case is bogus, it says `\xc7\xf0'
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
856 ('\xc7\xb0',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
857 '\xc7\xb0'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
858 # 3.10 Self-reverting case folding U+0390 and normalization.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
859 ('\xce\x90',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
860 '\xce\x90'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
861 # 3.11 Self-reverting case folding U+03B0 and normalization.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
862 ('\xce\xb0',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
863 '\xce\xb0'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
864 # 3.12 Self-reverting case folding U+1E96 and normalization.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
865 ('\xe1\xba\x96',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
866 '\xe1\xba\x96'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
867 # 3.13 Self-reverting case folding U+1F56 and normalization.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
868 ('\xe1\xbd\x96',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
869 '\xe1\xbd\x96'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
870 # 3.14 ASCII space character U+0020.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
871 (' ',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
872 ' '),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
873 # 3.15 Non-ASCII 8bit space character U+00A0.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
874 ('\xc2\xa0',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
875 ' '),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
876 # 3.16 Non-ASCII multibyte space character U+1680.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
877 ('\xe1\x9a\x80',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
878 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
879 # 3.17 Non-ASCII multibyte space character U+2000.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
880 ('\xe2\x80\x80',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
881 ' '),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
882 # 3.18 Zero Width Space U+200b.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
883 ('\xe2\x80\x8b',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
884 ''),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
885 # 3.19 Non-ASCII multibyte space character U+3000.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
886 ('\xe3\x80\x80',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
887 ' '),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
888 # 3.20 ASCII control characters U+0010 U+007F.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
889 ('\x10\x7f',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
890 '\x10\x7f'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
891 # 3.21 Non-ASCII 8bit control character U+0085.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
892 ('\xc2\x85',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
893 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
894 # 3.22 Non-ASCII multibyte control character U+180E.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
895 ('\xe1\xa0\x8e',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
896 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
897 # 3.23 Zero Width No-Break Space U+FEFF.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
898 ('\xef\xbb\xbf',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
899 ''),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
900 # 3.24 Non-ASCII control character U+1D175.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
901 ('\xf0\x9d\x85\xb5',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
902 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
903 # 3.25 Plane 0 private use character U+F123.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
904 ('\xef\x84\xa3',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
905 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
906 # 3.26 Plane 15 private use character U+F1234.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
907 ('\xf3\xb1\x88\xb4',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
908 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
909 # 3.27 Plane 16 private use character U+10F234.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
910 ('\xf4\x8f\x88\xb4',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
911 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
912 # 3.28 Non-character code point U+8FFFE.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
913 ('\xf2\x8f\xbf\xbe',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
914 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
915 # 3.29 Non-character code point U+10FFFF.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
916 ('\xf4\x8f\xbf\xbf',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
917 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
918 # 3.30 Surrogate code U+DF42.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
919 ('\xed\xbd\x82',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
920 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
921 # 3.31 Non-plain text character U+FFFD.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
922 ('\xef\xbf\xbd',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
923 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
924 # 3.32 Ideographic description character U+2FF5.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
925 ('\xe2\xbf\xb5',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
926 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
927 # 3.33 Display property character U+0341.
28774
c9bee11eaa8b Whitespace normalization.
Tim Peters <tim.peters@gmail.com>
parents: 28678
diff changeset
928 ('\xcd\x81',
28671
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
929 '\xcc\x81'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
930 # 3.34 Left-to-right mark U+200E.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
931 ('\xe2\x80\x8e',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
932 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
933 # 3.35 Deprecated U+202A.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
934 ('\xe2\x80\xaa',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
935 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
936 # 3.36 Language tagging character U+E0001.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
937 ('\xf3\xa0\x80\x81',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
938 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
939 # 3.37 Language tagging character U+E0042.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
940 ('\xf3\xa0\x81\x82',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
941 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
942 # 3.38 Bidi: RandALCat character U+05BE and LCat characters.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
943 ('foo\xd6\xbebar',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
944 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
945 # 3.39 Bidi: RandALCat character U+FD50 and LCat characters.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
946 ('foo\xef\xb5\x90bar',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
947 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
948 # 3.40 Bidi: RandALCat character U+FB38 and LCat characters.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
949 ('foo\xef\xb9\xb6bar',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
950 'foo \xd9\x8ebar'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
951 # 3.41 Bidi: RandALCat without trailing RandALCat U+0627 U+0031.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
952 ('\xd8\xa71',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
953 None),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
954 # 3.42 Bidi: RandALCat character U+0627 U+0031 U+0628.
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
955 ('\xd8\xa71\xd8\xa8',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
956 '\xd8\xa71\xd8\xa8'),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
957 # 3.43 Unassigned code point U+E0002.
28678
31cb8e0260db Skip nameprep test 3.43, as we do allow unassigned characters. The test
Martin v. Löwis <martin@v.loewis.de>
parents: 28671
diff changeset
958 # Skip this test as we allow unassigned
31cb8e0260db Skip nameprep test 3.43, as we do allow unassigned characters. The test
Martin v. Löwis <martin@v.loewis.de>
parents: 28671
diff changeset
959 #('\xf3\xa0\x80\x82',
31cb8e0260db Skip nameprep test 3.43, as we do allow unassigned characters. The test
Martin v. Löwis <martin@v.loewis.de>
parents: 28671
diff changeset
960 # None),
31cb8e0260db Skip nameprep test 3.43, as we do allow unassigned characters. The test
Martin v. Löwis <martin@v.loewis.de>
parents: 28671
diff changeset
961 (None, None),
28671
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
962 # 3.44 Larger test (shrinking).
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
963 # Original test case reads \xc3\xdf
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
964 ('X\xc2\xad\xc3\x9f\xc4\xb0\xe2\x84\xa1j\xcc\x8c\xc2\xa0\xc2'
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
965 '\xaa\xce\xb0\xe2\x80\x80',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
966 'xssi\xcc\x87tel\xc7\xb0 a\xce\xb0 '),
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
967 # 3.45 Larger test (expanding).
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
968 # Original test case reads \xc3\x9f
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
969 ('X\xc3\x9f\xe3\x8c\x96\xc4\xb0\xe2\x84\xa1\xe2\x92\x9f\xe3\x8c'
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
970 '\x80',
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
971 'xss\xe3\x82\xad\xe3\x83\xad\xe3\x83\xa1\xe3\x83\xbc\xe3'
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
972 '\x83\x88\xe3\x83\xabi\xcc\x87tel\x28d\x29\xe3\x82'
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
973 '\xa2\xe3\x83\x91\xe3\x83\xbc\xe3\x83\x88')
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
974 ]
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
975
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
976
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
977 class NameprepTest(unittest.TestCase):
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
978 def test_nameprep(self):
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
979 from encodings.idna import nameprep
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
980 for pos, (orig, prepped) in enumerate(nameprep_tests):
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
981 if orig is None:
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
982 # Skipped
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
983 continue
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
984 # The Unicode strings are given in UTF-8
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
985 orig = unicode(orig, "utf-8")
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
986 if prepped is None:
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
987 # Input contains prohibited characters
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
988 self.assertRaises(UnicodeError, nameprep, orig)
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
989 else:
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
990 prepped = unicode(prepped, "utf-8")
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
991 try:
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
992 self.assertEquals(nameprep(orig), prepped)
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
993 except Exception,e:
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
994 raise test_support.TestFailed("Test 3.%d: %s" % (pos+1, str(e)))
d41d8de9768f Implement IDNA (Internationalized Domain Names in Applications).
Martin v. Löwis <martin@v.loewis.de>
parents: 27680
diff changeset
995
37719
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
996 class IDNACodecTest(unittest.TestCase):
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
997 def test_builtin_decode(self):
31795
bf37429f6d9f Add test case for unicode(somestring, "idna").
Martin v. Löwis <martin@v.loewis.de>
parents: 28917
diff changeset
998 self.assertEquals(unicode("python.org", "idna"), u"python.org")
37719
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
999 self.assertEquals(unicode("python.org.", "idna"), u"python.org.")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1000 self.assertEquals(unicode("xn--pythn-mua.org", "idna"), u"pyth\xf6n.org")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1001 self.assertEquals(unicode("xn--pythn-mua.org.", "idna"), u"pyth\xf6n.org.")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1002
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1003 def test_builtin_encode(self):
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1004 self.assertEquals(u"python.org".encode("idna"), "python.org")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1005 self.assertEquals("python.org.".encode("idna"), "python.org.")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1006 self.assertEquals(u"pyth\xf6n.org".encode("idna"), "xn--pythn-mua.org")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1007 self.assertEquals(u"pyth\xf6n.org.".encode("idna"), "xn--pythn-mua.org.")
31795
bf37429f6d9f Add test case for unicode(somestring, "idna").
Martin v. Löwis <martin@v.loewis.de>
parents: 28917
diff changeset
1008
35435
9d343bc81863 Make IDNA return an empty string when the input is empty. Fixes #1163178.
Martin v. Löwis <martin@v.loewis.de>
parents: 35269
diff changeset
1009 def test_stream(self):
9d343bc81863 Make IDNA return an empty string when the input is empty. Fixes #1163178.
Martin v. Löwis <martin@v.loewis.de>
parents: 35269
diff changeset
1010 import StringIO
9d343bc81863 Make IDNA return an empty string when the input is empty. Fixes #1163178.
Martin v. Löwis <martin@v.loewis.de>
parents: 35269
diff changeset
1011 r = codecs.getreader("idna")(StringIO.StringIO("abc"))
9d343bc81863 Make IDNA return an empty string when the input is empty. Fixes #1163178.
Martin v. Löwis <martin@v.loewis.de>
parents: 35269
diff changeset
1012 r.read(3)
9d343bc81863 Make IDNA return an empty string when the input is empty. Fixes #1163178.
Martin v. Löwis <martin@v.loewis.de>
parents: 35269
diff changeset
1013 self.assertEquals(r.read(), u"")
9d343bc81863 Make IDNA return an empty string when the input is empty. Fixes #1163178.
Martin v. Löwis <martin@v.loewis.de>
parents: 35269
diff changeset
1014
37719
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1015 def test_incremental_decode(self):
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1016 self.assertEquals(
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1017 "".join(codecs.iterdecode("python.org", "idna")),
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1018 u"python.org"
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1019 )
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1020 self.assertEquals(
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1021 "".join(codecs.iterdecode("python.org.", "idna")),
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1022 u"python.org."
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1023 )
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1024 self.assertEquals(
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1025 "".join(codecs.iterdecode("xn--pythn-mua.org.", "idna")),
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1026 u"pyth\xf6n.org."
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1027 )
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1028 self.assertEquals(
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1029 "".join(codecs.iterdecode("xn--pythn-mua.org.", "idna")),
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1030 u"pyth\xf6n.org."
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1031 )
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1032
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1033 decoder = codecs.getincrementaldecoder("idna")()
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1034 self.assertEquals(decoder.decode("xn--xam", ), u"")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1035 self.assertEquals(decoder.decode("ple-9ta.o", ), u"\xe4xample.")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1036 self.assertEquals(decoder.decode(u"rg"), u"")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1037 self.assertEquals(decoder.decode(u"", True), u"org")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1038
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1039 decoder.reset()
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1040 self.assertEquals(decoder.decode("xn--xam", ), u"")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1041 self.assertEquals(decoder.decode("ple-9ta.o", ), u"\xe4xample.")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1042 self.assertEquals(decoder.decode("rg."), u"org.")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1043 self.assertEquals(decoder.decode("", True), u"")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1044
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1045 def test_incremental_encode(self):
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1046 self.assertEquals(
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1047 "".join(codecs.iterencode(u"python.org", "idna")),
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1048 "python.org"
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1049 )
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1050 self.assertEquals(
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1051 "".join(codecs.iterencode(u"python.org.", "idna")),
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1052 "python.org."
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1053 )
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1054 self.assertEquals(
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1055 "".join(codecs.iterencode(u"pyth\xf6n.org.", "idna")),
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1056 "xn--pythn-mua.org."
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1057 )
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1058 self.assertEquals(
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1059 "".join(codecs.iterencode(u"pyth\xf6n.org.", "idna")),
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1060 "xn--pythn-mua.org."
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1061 )
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1062
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1063 encoder = codecs.getincrementalencoder("idna")()
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1064 self.assertEquals(encoder.encode(u"\xe4x"), "")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1065 self.assertEquals(encoder.encode(u"ample.org"), "xn--xample-9ta.")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1066 self.assertEquals(encoder.encode(u"", True), "org")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1067
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1068 encoder.reset()
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1069 self.assertEquals(encoder.encode(u"\xe4x"), "")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1070 self.assertEquals(encoder.encode(u"ample.org."), "xn--xample-9ta.org.")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1071 self.assertEquals(encoder.encode(u"", True), "")
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1072
32572
8ea2cb1ec598 Add generic codecs.encode() and .decode() APIs that don't impose
Marc-André Lemburg <mal@egenix.com>
parents: 32538
diff changeset
1073 class CodecsModuleTest(unittest.TestCase):
8ea2cb1ec598 Add generic codecs.encode() and .decode() APIs that don't impose
Marc-André Lemburg <mal@egenix.com>
parents: 32538
diff changeset
1074
8ea2cb1ec598 Add generic codecs.encode() and .decode() APIs that don't impose
Marc-André Lemburg <mal@egenix.com>
parents: 32538
diff changeset
1075 def test_decode(self):
8ea2cb1ec598 Add generic codecs.encode() and .decode() APIs that don't impose
Marc-André Lemburg <mal@egenix.com>
parents: 32538
diff changeset
1076 self.assertEquals(codecs.decode('\xe4\xf6\xfc', 'latin-1'),
8ea2cb1ec598 Add generic codecs.encode() and .decode() APIs that don't impose
Marc-André Lemburg <mal@egenix.com>
parents: 32538
diff changeset
1077 u'\xe4\xf6\xfc')
33855
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1078 self.assertRaises(TypeError, codecs.decode)
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1079 self.assertEquals(codecs.decode('abc'), u'abc')
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1080 self.assertRaises(UnicodeDecodeError, codecs.decode, '\xff', 'ascii')
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1081
32572
8ea2cb1ec598 Add generic codecs.encode() and .decode() APIs that don't impose
Marc-André Lemburg <mal@egenix.com>
parents: 32538
diff changeset
1082 def test_encode(self):
8ea2cb1ec598 Add generic codecs.encode() and .decode() APIs that don't impose
Marc-André Lemburg <mal@egenix.com>
parents: 32538
diff changeset
1083 self.assertEquals(codecs.encode(u'\xe4\xf6\xfc', 'latin-1'),
8ea2cb1ec598 Add generic codecs.encode() and .decode() APIs that don't impose
Marc-André Lemburg <mal@egenix.com>
parents: 32538
diff changeset
1084 '\xe4\xf6\xfc')
33855
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1085 self.assertRaises(TypeError, codecs.encode)
35889
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1086 self.assertRaises(LookupError, codecs.encode, "foo", "__spam__")
33855
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1087 self.assertEquals(codecs.encode(u'abc'), 'abc')
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1088 self.assertRaises(UnicodeEncodeError, codecs.encode, u'\xffff', 'ascii')
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1089
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1090 def test_register(self):
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1091 self.assertRaises(TypeError, codecs.register)
35889
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1092 self.assertRaises(TypeError, codecs.register, 42)
33855
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1093
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1094 def test_lookup(self):
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1095 self.assertRaises(TypeError, codecs.lookup)
1cd19daaa9d4 Trigger a few error cases in Modules/_codecsmodule.c.
Walter Dörwald <walter@livinglogic.de>
parents: 33782
diff changeset
1096 self.assertRaises(LookupError, codecs.lookup, "__spam__")
35889
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1097 self.assertRaises(LookupError, codecs.lookup, " ")
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1098
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1099 def test_getencoder(self):
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1100 self.assertRaises(TypeError, codecs.getencoder)
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1101 self.assertRaises(LookupError, codecs.getencoder, "__spam__")
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1102
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1103 def test_getdecoder(self):
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1104 self.assertRaises(TypeError, codecs.getdecoder)
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1105 self.assertRaises(LookupError, codecs.getdecoder, "__spam__")
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1106
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1107 def test_getreader(self):
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1108 self.assertRaises(TypeError, codecs.getreader)
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1109 self.assertRaises(LookupError, codecs.getreader, "__spam__")
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1110
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1111 def test_getwriter(self):
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1112 self.assertRaises(TypeError, codecs.getwriter)
e2473c4f2ee0 Add tests to increase code coverage in Python/codecs.c and Python/exceptions.c.
Walter Dörwald <walter@livinglogic.de>
parents: 35882
diff changeset
1113 self.assertRaises(LookupError, codecs.getwriter, "__spam__")
32572
8ea2cb1ec598 Add generic codecs.encode() and .decode() APIs that don't impose
Marc-André Lemburg <mal@egenix.com>
parents: 32538
diff changeset
1114
33782
a52de911b284 SF #1048865: Fix a trivial typo that breaks StreamReader.readlines()
Hye-Shik Chang <hyeshik@gmail.com>
parents: 33428
diff changeset
1115 class StreamReaderTest(unittest.TestCase):
a52de911b284 SF #1048865: Fix a trivial typo that breaks StreamReader.readlines()
Hye-Shik Chang <hyeshik@gmail.com>
parents: 33428
diff changeset
1116
a52de911b284 SF #1048865: Fix a trivial typo that breaks StreamReader.readlines()
Hye-Shik Chang <hyeshik@gmail.com>
parents: 33428
diff changeset
1117 def setUp(self):
a52de911b284 SF #1048865: Fix a trivial typo that breaks StreamReader.readlines()
Hye-Shik Chang <hyeshik@gmail.com>
parents: 33428
diff changeset
1118 self.reader = codecs.getreader('utf-8')
a52de911b284 SF #1048865: Fix a trivial typo that breaks StreamReader.readlines()
Hye-Shik Chang <hyeshik@gmail.com>
parents: 33428
diff changeset
1119 self.stream = StringIO.StringIO('\xed\x95\x9c\n\xea\xb8\x80')
a52de911b284 SF #1048865: Fix a trivial typo that breaks StreamReader.readlines()
Hye-Shik Chang <hyeshik@gmail.com>
parents: 33428
diff changeset
1120
a52de911b284 SF #1048865: Fix a trivial typo that breaks StreamReader.readlines()
Hye-Shik Chang <hyeshik@gmail.com>
parents: 33428
diff changeset
1121 def test_readlines(self):
a52de911b284 SF #1048865: Fix a trivial typo that breaks StreamReader.readlines()
Hye-Shik Chang <hyeshik@gmail.com>
parents: 33428
diff changeset
1122 f = self.reader(self.stream)
a52de911b284 SF #1048865: Fix a trivial typo that breaks StreamReader.readlines()
Hye-Shik Chang <hyeshik@gmail.com>
parents: 33428
diff changeset
1123 self.assertEquals(f.readlines(), [u'\ud55c\n', u'\uae00'])
a52de911b284 SF #1048865: Fix a trivial typo that breaks StreamReader.readlines()
Hye-Shik Chang <hyeshik@gmail.com>
parents: 33428
diff changeset
1124
40239
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1125 class EncodedFileTest(unittest.TestCase):
40284
ff678669aa6e Whitespace normalization.
Tim Peters <tim.peters@gmail.com>
parents: 40274
diff changeset
1126
40239
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1127 def test_basic(self):
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1128 f = StringIO.StringIO('\xed\x95\x9c\n\xea\xb8\x80')
40249
9663100c95ca Fix the new EncodedFile test to work with big endian platforms.
Georg Brandl <georg@python.org>
parents: 40239
diff changeset
1129 ef = codecs.EncodedFile(f, 'utf-16-le', 'utf-8')
9663100c95ca Fix the new EncodedFile test to work with big endian platforms.
Georg Brandl <georg@python.org>
parents: 40239
diff changeset
1130 self.assertEquals(ef.read(), '\\\xd5\n\x00\x00\xae')
40239
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1131
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1132 f = StringIO.StringIO()
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1133 ef = codecs.EncodedFile(f, 'utf-8', 'latin1')
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1134 ef.write('\xc3\xbc')
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1135 self.assertEquals(f.getvalue(), '\xfc')
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1136
35269
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1137 class Str2StrTest(unittest.TestCase):
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1138
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1139 def test_read(self):
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1140 sin = "\x80".encode("base64_codec")
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1141 reader = codecs.getreader("base64_codec")(StringIO.StringIO(sin))
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1142 sout = reader.read()
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1143 self.assertEqual(sout, "\x80")
58472
6ffcdafa0118 use assert[Not]IsInstance where appropriate
Ezio Melotti <ezio.melotti@gmail.com>
parents: 56033
diff changeset
1144 self.assertIsInstance(sout, str)
35269
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1145
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1146 def test_readline(self):
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1147 sin = "\x80".encode("base64_codec")
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1148 reader = codecs.getreader("base64_codec")(StringIO.StringIO(sin))
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1149 sout = reader.readline()
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1150 self.assertEqual(sout, "\x80")
58472
6ffcdafa0118 use assert[Not]IsInstance where appropriate
Ezio Melotti <ezio.melotti@gmail.com>
parents: 56033
diff changeset
1151 self.assertIsInstance(sout, str)
35269
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1152
34279
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1153 all_unicode_encodings = [
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1154 "ascii",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1155 "base64_codec",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1156 "big5",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1157 "big5hkscs",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1158 "charmap",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1159 "cp037",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1160 "cp1006",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1161 "cp1026",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1162 "cp1140",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1163 "cp1250",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1164 "cp1251",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1165 "cp1252",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1166 "cp1253",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1167 "cp1254",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1168 "cp1255",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1169 "cp1256",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1170 "cp1257",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1171 "cp1258",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1172 "cp424",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1173 "cp437",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1174 "cp500",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1175 "cp737",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1176 "cp775",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1177 "cp850",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1178 "cp852",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1179 "cp855",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1180 "cp856",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1181 "cp857",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1182 "cp860",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1183 "cp861",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1184 "cp862",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1185 "cp863",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1186 "cp864",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1187 "cp865",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1188 "cp866",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1189 "cp869",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1190 "cp874",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1191 "cp875",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1192 "cp932",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1193 "cp949",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1194 "cp950",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1195 "euc_jis_2004",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1196 "euc_jisx0213",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1197 "euc_jp",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1198 "euc_kr",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1199 "gb18030",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1200 "gb2312",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1201 "gbk",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1202 "hex_codec",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1203 "hp_roman8",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1204 "hz",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1205 "idna",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1206 "iso2022_jp",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1207 "iso2022_jp_1",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1208 "iso2022_jp_2",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1209 "iso2022_jp_2004",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1210 "iso2022_jp_3",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1211 "iso2022_jp_ext",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1212 "iso2022_kr",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1213 "iso8859_1",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1214 "iso8859_10",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1215 "iso8859_11",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1216 "iso8859_13",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1217 "iso8859_14",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1218 "iso8859_15",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1219 "iso8859_16",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1220 "iso8859_2",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1221 "iso8859_3",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1222 "iso8859_4",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1223 "iso8859_5",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1224 "iso8859_6",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1225 "iso8859_7",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1226 "iso8859_8",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1227 "iso8859_9",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1228 "johab",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1229 "koi8_r",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1230 "koi8_u",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1231 "latin_1",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1232 "mac_cyrillic",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1233 "mac_greek",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1234 "mac_iceland",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1235 "mac_latin2",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1236 "mac_roman",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1237 "mac_turkish",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1238 "palmos",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1239 "ptcp154",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1240 "punycode",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1241 "raw_unicode_escape",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1242 "rot_13",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1243 "shift_jis",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1244 "shift_jis_2004",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1245 "shift_jisx0213",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1246 "tis_620",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1247 "unicode_escape",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1248 "unicode_internal",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1249 "utf_16",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1250 "utf_16_be",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1251 "utf_16_le",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1252 "utf_7",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1253 "utf_8",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1254 ]
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1255
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1256 if hasattr(codecs, "mbcs_encode"):
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1257 all_unicode_encodings.append("mbcs")
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1258
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1259 # The following encodings work only with str, not unicode
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1260 all_string_encodings = [
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1261 "quopri_codec",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1262 "string_escape",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1263 "uu_codec",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1264 ]
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1265
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1266 # The following encoding is not tested, because it's not supposed
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1267 # to work:
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1268 # "undefined"
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1269
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1270 # The following encodings don't work in stateful mode
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1271 broken_unicode_with_streams = [
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1272 "base64_codec",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1273 "hex_codec",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1274 "punycode",
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1275 "unicode_internal"
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1276 ]
40251
e16800ac1eea Bug #1586613: fix zlib and bz2 codecs' incremental en/decoders.
Georg Brandl <georg@python.org>
parents: 40249
diff changeset
1277 broken_incremental_coders = broken_unicode_with_streams[:]
34279
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1278
40271
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1279 # The following encodings only support "strict" mode
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1280 only_strict_mode = [
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1281 "idna",
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1282 "zlib_codec",
40274
bc8ffc3e3cdb I'm assuming this is correct, it fixes the tests so they pass again
Neal Norwitz <nnorwitz@gmail.com>
parents: 40271
diff changeset
1283 "bz2_codec",
40271
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1284 ]
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1285
34279
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1286 try:
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1287 import bz2
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1288 except ImportError:
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1289 pass
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1290 else:
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1291 all_unicode_encodings.append("bz2_codec")
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1292 broken_unicode_with_streams.append("bz2_codec")
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1293
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1294 try:
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1295 import zlib
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1296 except ImportError:
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1297 pass
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1298 else:
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1299 all_unicode_encodings.append("zlib_codec")
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1300 broken_unicode_with_streams.append("zlib_codec")
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1301
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1302 class BasicUnicodeTest(unittest.TestCase):
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1303 def test_basics(self):
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1304 s = u"abc123" # all codecs should be able to encode these
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1305 for encoding in all_unicode_encodings:
36996
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1306 name = codecs.lookup(encoding).name
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1307 if encoding.endswith("_codec"):
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1308 name += "_codec"
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1309 elif encoding == "latin_1":
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1310 name = "latin_1"
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1311 self.assertEqual(encoding.replace("_", "-"), name.replace("_", "-"))
34279
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1312 (bytes, size) = codecs.getencoder(encoding)(s)
53855
7dca250343ea Issue 3739: The unicode-internal encoder now reports the number of *characters*
Walter Dörwald <walter@livinglogic.de>
parents: 43932
diff changeset
1313 self.assertEqual(size, len(s), "%r != %r (encoding=%r)" % (size, len(s), encoding))
34279
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1314 (chars, size) = codecs.getdecoder(encoding)(bytes)
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1315 self.assertEqual(chars, s, "%r != %r (encoding=%r)" % (chars, s, encoding))
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1316
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1317 if encoding not in broken_unicode_with_streams:
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1318 # check stream reader/writer
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1319 q = Queue()
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1320 writer = codecs.getwriter(encoding)(q)
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1321 encodedresult = ""
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1322 for c in s:
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1323 writer.write(c)
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1324 encodedresult += q.read()
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1325 q = Queue()
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1326 reader = codecs.getreader(encoding)(q)
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1327 decodedresult = u""
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1328 for c in encodedresult:
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1329 q.write(c)
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1330 decodedresult += reader.read()
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1331 self.assertEqual(decodedresult, s, "%r != %r (encoding=%r)" % (decodedresult, s, encoding))
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1332
40251
e16800ac1eea Bug #1586613: fix zlib and bz2 codecs' incremental en/decoders.
Georg Brandl <georg@python.org>
parents: 40249
diff changeset
1333 if encoding not in broken_incremental_coders:
37077
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
1334 # check incremental decoder/encoder (fetched via the Python
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
1335 # and C API) and iterencode()/iterdecode()
36996
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1336 try:
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1337 encoder = codecs.getincrementalencoder(encoding)()
37077
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
1338 cencoder = _testcapi.codec_incrementalencoder(encoding)
36996
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1339 except LookupError: # no IncrementalEncoder
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1340 pass
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1341 else:
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1342 # check incremental decoder/encoder
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1343 encodedresult = ""
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1344 for c in s:
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1345 encodedresult += encoder.encode(c)
37708
230fff6e0618 Call encode()/decode() with final==True as the last call in the
Walter Dörwald <walter@livinglogic.de>
parents: 37077
diff changeset
1346 encodedresult += encoder.encode(u"", True)
36996
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1347 decoder = codecs.getincrementaldecoder(encoding)()
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1348 decodedresult = u""
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1349 for c in encodedresult:
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1350 decodedresult += decoder.decode(c)
37708
230fff6e0618 Call encode()/decode() with final==True as the last call in the
Walter Dörwald <walter@livinglogic.de>
parents: 37077
diff changeset
1351 decodedresult += decoder.decode("", True)
36996
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1352 self.assertEqual(decodedresult, s, "%r != %r (encoding=%r)" % (decodedresult, s, encoding))
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1353
37077
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
1354 # check C API
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
1355 encodedresult = ""
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
1356 for c in s:
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
1357 encodedresult += cencoder.encode(c)
37708
230fff6e0618 Call encode()/decode() with final==True as the last call in the
Walter Dörwald <walter@livinglogic.de>
parents: 37077
diff changeset
1358 encodedresult += cencoder.encode(u"", True)
37077
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
1359 cdecoder = _testcapi.codec_incrementaldecoder(encoding)
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
1360 decodedresult = u""
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
1361 for c in encodedresult:
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
1362 decodedresult += cdecoder.decode(c)
37708
230fff6e0618 Call encode()/decode() with final==True as the last call in the
Walter Dörwald <walter@livinglogic.de>
parents: 37077
diff changeset
1363 decodedresult += cdecoder.decode("", True)
37077
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
1364 self.assertEqual(decodedresult, s, "%r != %r (encoding=%r)" % (decodedresult, s, encoding))
38400b662404 Add tests for the C APIs PyCodec_IncrementalEncoder() and
Walter Dörwald <walter@livinglogic.de>
parents: 36996
diff changeset
1365
36996
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1366 # check iterencode()/iterdecode()
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1367 result = u"".join(codecs.iterdecode(codecs.iterencode(s, encoding), encoding))
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1368 self.assertEqual(result, s, "%r != %r (encoding=%r)" % (result, s, encoding))
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1369
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1370 # check iterencode()/iterdecode() with empty string
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1371 result = u"".join(codecs.iterdecode(codecs.iterencode(u"", encoding), encoding))
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1372 self.assertEqual(result, u"")
7fa4c29d7030 Patch #1436130: codecs.lookup() now returns a CodecInfo object (a subclass
Walter Dörwald <walter@livinglogic.de>
parents: 36855
diff changeset
1373
40271
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1374 if encoding not in only_strict_mode:
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1375 # check incremental decoder/encoder with errors argument
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1376 try:
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1377 encoder = codecs.getincrementalencoder(encoding)("ignore")
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1378 cencoder = _testcapi.codec_incrementalencoder(encoding, "ignore")
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1379 except LookupError: # no IncrementalEncoder
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1380 pass
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1381 else:
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1382 encodedresult = "".join(encoder.encode(c) for c in s)
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1383 decoder = codecs.getincrementaldecoder(encoding)("ignore")
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1384 decodedresult = u"".join(decoder.decode(c) for c in encodedresult)
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1385 self.assertEqual(decodedresult, s, "%r != %r (encoding=%r)" % (decodedresult, s, encoding))
40284
ff678669aa6e Whitespace normalization.
Tim Peters <tim.peters@gmail.com>
parents: 40274
diff changeset
1386
40271
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1387 encodedresult = "".join(cencoder.encode(c) for c in s)
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1388 cdecoder = _testcapi.codec_incrementaldecoder(encoding, "ignore")
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1389 decodedresult = u"".join(cdecoder.decode(c) for c in encodedresult)
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1390 self.assertEqual(decodedresult, s, "%r != %r (encoding=%r)" % (decodedresult, s, encoding))
c5a2f431c698 Add tests for incremental codecs with an errors
Walter Dörwald <walter@livinglogic.de>
parents: 40251
diff changeset
1391
34750
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1392 def test_seek(self):
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1393 # all codecs should be able to encode these
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1394 s = u"%s\n%s\n" % (100*u"abc123", 100*u"def456")
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1395 for encoding in all_unicode_encodings:
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1396 if encoding == "idna": # FIXME: See SF bug #1163178
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1397 continue
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1398 if encoding in broken_unicode_with_streams:
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1399 continue
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1400 reader = codecs.getreader(encoding)(StringIO.StringIO(s.encode(encoding)))
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1401 for t in xrange(5):
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1402 # Test that calling seek resets the internal codec state and buffers
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1403 reader.seek(0, 0)
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1404 line = reader.readline()
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1405 self.assertEqual(s[:len(line)], line)
679c16115613 Reset internal buffers when seek() is called. This fixes SF bug #1156259.
Walter Dörwald <walter@livinglogic.de>
parents: 34562
diff changeset
1406
35882
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1407 def test_bad_decode_args(self):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1408 for encoding in all_unicode_encodings:
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1409 decoder = codecs.getdecoder(encoding)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1410 self.assertRaises(TypeError, decoder)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1411 if encoding not in ("idna", "punycode"):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1412 self.assertRaises(TypeError, decoder, 42)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1413
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1414 def test_bad_encode_args(self):
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1415 for encoding in all_unicode_encodings:
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1416 encoder = codecs.getencoder(encoding)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1417 self.assertRaises(TypeError, encoder)
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1418
38787
70b166ca1c49 Verify the crash due to EncodingMap not initialized does not return
Neal Norwitz <nnorwitz@gmail.com>
parents: 37719
diff changeset
1419 def test_encoding_map_type_initialized(self):
70b166ca1c49 Verify the crash due to EncodingMap not initialized does not return
Neal Norwitz <nnorwitz@gmail.com>
parents: 37719
diff changeset
1420 from encodings import cp1140
70b166ca1c49 Verify the crash due to EncodingMap not initialized does not return
Neal Norwitz <nnorwitz@gmail.com>
parents: 37719
diff changeset
1421 # This used to crash, we are only verifying there's no crash.
70b166ca1c49 Verify the crash due to EncodingMap not initialized does not return
Neal Norwitz <nnorwitz@gmail.com>
parents: 37719
diff changeset
1422 table_type = type(cp1140.encoding_table)
70b166ca1c49 Verify the crash due to EncodingMap not initialized does not return
Neal Norwitz <nnorwitz@gmail.com>
parents: 37719
diff changeset
1423 self.assertEqual(table_type, table_type)
70b166ca1c49 Verify the crash due to EncodingMap not initialized does not return
Neal Norwitz <nnorwitz@gmail.com>
parents: 37719
diff changeset
1424
34279
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1425 class BasicStrTest(unittest.TestCase):
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1426 def test_basics(self):
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1427 s = "abc123"
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1428 for encoding in all_string_encodings:
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1429 (bytes, size) = codecs.getencoder(encoding)(s)
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1430 self.assertEqual(size, len(s))
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1431 (chars, size) = codecs.getdecoder(encoding)(bytes)
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1432 self.assertEqual(chars, s, "%r != %r (encoding=%r)" % (chars, s, encoding))
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1433
35702
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1434 class CharmapTest(unittest.TestCase):
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1435 def test_decode_with_string_map(self):
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1436 self.assertEquals(
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1437 codecs.charmap_decode("\x00\x01\x02", "strict", u"abc"),
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1438 (u"abc", 3)
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1439 )
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1440
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1441 self.assertEquals(
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1442 codecs.charmap_decode("\x00\x01\x02", "replace", u"ab"),
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1443 (u"ab\ufffd", 3)
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1444 )
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1445
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1446 self.assertEquals(
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1447 codecs.charmap_decode("\x00\x01\x02", "replace", u"ab\ufffe"),
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1448 (u"ab\ufffd", 3)
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1449 )
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1450
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1451 self.assertEquals(
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1452 codecs.charmap_decode("\x00\x01\x02", "ignore", u"ab"),
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1453 (u"ab", 3)
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1454 )
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1455
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1456 self.assertEquals(
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1457 codecs.charmap_decode("\x00\x01\x02", "ignore", u"ab\ufffe"),
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1458 (u"ab", 3)
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1459 )
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1460
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1461 allbytes = "".join(chr(i) for i in xrange(256))
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1462 self.assertEquals(
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1463 codecs.charmap_decode(allbytes, "ignore", u""),
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1464 (u"", len(allbytes))
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1465 )
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1466
40239
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1467 class WithStmtTest(unittest.TestCase):
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1468 def test_encodedfile(self):
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1469 f = StringIO.StringIO("\xc3\xbc")
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1470 with codecs.EncodedFile(f, "latin-1", "utf-8") as ef:
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1471 self.assertEquals(ef.read(), "\xfc")
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1472
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1473 def test_streamreaderwriter(self):
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1474 f = StringIO.StringIO("\xc3\xbc")
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1475 info = codecs.lookup("utf-8")
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1476 with codecs.StreamReaderWriter(f, info.streamreader,
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1477 info.streamwriter, 'strict') as srw:
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1478 self.assertEquals(srw.read(), u"\xfc")
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1479
35702
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1480
20013
e3184e46303b Change the PyUnit-based tests to use the test_main() approach. This
Fred Drake <fdrake@acm.org>
parents: 18163
diff changeset
1481 def test_main():
28917
1741af48fd9f Combine the functionality of test_support.run_unittest()
Walter Dörwald <walter@livinglogic.de>
parents: 28774
diff changeset
1482 test_support.run_unittest(
42649
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
1483 UTF32Test,
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
1484 UTF32LETest,
c916731321c0 Backport r57105 and r57145 from the py3k branch: UTF-32 codecs.
Walter Dörwald <walter@livinglogic.de>
parents: 41445
diff changeset
1485 UTF32BETest,
28917
1741af48fd9f Combine the functionality of test_support.run_unittest()
Walter Dörwald <walter@livinglogic.de>
parents: 28774
diff changeset
1486 UTF16Test,
33428
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
1487 UTF16LETest,
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
1488 UTF16BETest,
57478a13fbf6 SF patch #998993: The UTF-8 and the UTF-16 stateful decoders now support
Walter Dörwald <walter@livinglogic.de>
parents: 32572
diff changeset
1489 UTF8Test,
36247
3ba9ff79bb29 Patch #1177307: UTF-8-Sig codec.
Martin v. Löwis <martin@v.loewis.de>
parents: 35889
diff changeset
1490 UTF8SigTest,
35882
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1491 UTF7Test,
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1492 UTF16ExTest,
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1493 ReadBufferTest,
3236b5fb1fb7 Add tests for various error cases and for readbuffer_encode() and
Walter Dörwald <walter@livinglogic.de>
parents: 35702
diff changeset
1494 CharBufferTest,
28917
1741af48fd9f Combine the functionality of test_support.run_unittest()
Walter Dörwald <walter@livinglogic.de>
parents: 28774
diff changeset
1495 EscapeDecodeTest,
1741af48fd9f Combine the functionality of test_support.run_unittest()
Walter Dörwald <walter@livinglogic.de>
parents: 28774
diff changeset
1496 RecodingTest,
1741af48fd9f Combine the functionality of test_support.run_unittest()
Walter Dörwald <walter@livinglogic.de>
parents: 28774
diff changeset
1497 PunycodeTest,
35484
4fc7a7edcb4b SF bug #1251300: On UCS-4 builds the "unicode-internal" codec will now complain
Walter Dörwald <walter@livinglogic.de>
parents: 35435
diff changeset
1498 UnicodeInternalTest,
31795
bf37429f6d9f Add test case for unicode(somestring, "idna").
Martin v. Löwis <martin@v.loewis.de>
parents: 28917
diff changeset
1499 NameprepTest,
37719
cf71c3814b0b Add a BufferedIncrementalEncoder class that can be used for implementing
Walter Dörwald <walter@livinglogic.de>
parents: 37708
diff changeset
1500 IDNACodecTest,
33782
a52de911b284 SF #1048865: Fix a trivial typo that breaks StreamReader.readlines()
Hye-Shik Chang <hyeshik@gmail.com>
parents: 33428
diff changeset
1501 CodecsModuleTest,
34279
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1502 StreamReaderTest,
40239
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1503 EncodedFileTest,
35269
2b117eb18944 Make attributes and local variables in the StreamReader str objects instead
Walter Dörwald <walter@livinglogic.de>
parents: 34925
diff changeset
1504 Str2StrTest,
34279
b3af7c6db1c6 Add a test that checks the basic functionality of every encoding.
Walter Dörwald <walter@livinglogic.de>
parents: 34214
diff changeset
1505 BasicUnicodeTest,
35702
22a720b35314 Part of SF patch #1313939: Speedup charmap decoding by extending
Walter Dörwald <walter@livinglogic.de>
parents: 35484
diff changeset
1506 BasicStrTest,
40239
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1507 CharmapTest,
7c4058681224 Fix codecs.EncodedFile which did not use file_encoding in 2.5.0, and
Georg Brandl <georg@python.org>
parents: 38787
diff changeset
1508 WithStmtTest,
28917
1741af48fd9f Combine the functionality of test_support.run_unittest()
Walter Dörwald <walter@livinglogic.de>
parents: 28774
diff changeset
1509 )
20013
e3184e46303b Change the PyUnit-based tests to use the test_main() approach. This
Fred Drake <fdrake@acm.org>
parents: 18163
diff changeset
1510
e3184e46303b Change the PyUnit-based tests to use the test_main() approach. This
Fred Drake <fdrake@acm.org>
parents: 18163
diff changeset
1511
e3184e46303b Change the PyUnit-based tests to use the test_main() approach. This
Fred Drake <fdrake@acm.org>
parents: 18163
diff changeset
1512 if __name__ == "__main__":
e3184e46303b Change the PyUnit-based tests to use the test_main() approach. This
Fred Drake <fdrake@acm.org>
parents: 18163
diff changeset
1513 test_main()