Mercurial > cpython-lru_cache-weakref
annotate Lib/test/test_normalization.py @ 48715:48dddfd2fbf3
RPC_WSTR is not available Visual Studio 2003 or earlier.
(Maybe I should have defined RPC_WSTR on old compiler,
but uuidcreate() is only place using it, so I simply replaced
with unsigned short *)
| author | Hirokazu Yamamoto <ocean-city@m2.ccsnet.ne.jp> |
|---|---|
| date | Sun, 17 Aug 2008 07:26:26 +0000 |
| parents | 9c930a32cebb |
| children | 25f14128a97c |
| rev | line source |
|---|---|
|
47201
9c930a32cebb
#2621 rename test.test_support to test.support
Benjamin Peterson <benjamin@python.org>
parents:
43089
diff
changeset
|
1 from test.support import run_unittest, open_urlresource |
|
41554
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
2 import unittest |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
3 |
|
26379
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
4 import sys |
|
26391
d207d1d78c5d
Fiddled things so that test_normalization is expected to be skipped if
Tim Peters <tim.peters@gmail.com>
parents:
26388
diff
changeset
|
5 import os |
|
26379
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
6 from unicodedata import normalize |
|
26391
d207d1d78c5d
Fiddled things so that test_normalization is expected to be skipped if
Tim Peters <tim.peters@gmail.com>
parents:
26388
diff
changeset
|
7 |
| 42622 | 8 TESTDATAFILE = "NormalizationTest.txt" |
|
36929
fab44c5c8df4
Update test data to 4.1; disable PRI #29 for now.
Martin v. Löwis <martin@v.loewis.de>
parents:
35998
diff
changeset
|
9 TESTDATAURL = "http://www.unicode.org/Public/4.1.0/ucd/" + TESTDATAFILE |
|
26379
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
10 |
|
37188
760ffb0fba37
Exceptions should inherit from Exception now.
Neal Norwitz <nnorwitz@gmail.com>
parents:
36937
diff
changeset
|
11 class RangeError(Exception): |
|
26379
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
12 pass |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
13 |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
14 def NFC(str): |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
15 return normalize("NFC", str) |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
16 |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
17 def NFKC(str): |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
18 return normalize("NFKC", str) |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
19 |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
20 def NFD(str): |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
21 return normalize("NFD", str) |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
22 |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
23 def NFKD(str): |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
24 return normalize("NFKD", str) |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
25 |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
26 def unistr(data): |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
27 data = [int(x, 16) for x in data.split(" ")] |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
28 for x in data: |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
29 if x > sys.maxunicode: |
|
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
30 raise RangeError |
|
41585
825d966338ea
Rename 'unicode' to 'str' in its tp_name field. Rename 'str' to 'str8'.
Guido van Rossum <guido@python.org>
parents:
41575
diff
changeset
|
31 return "".join([chr(x) for x in data]) |
|
26379
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
32 |
|
41554
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
33 class NormalizationTest(unittest.TestCase): |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
34 def test_main(self): |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
35 part1_data = {} |
|
43089
2e3c02c625de
Explicitly use UTF-8 as the encoding for the normalization file.
Martin v. Löwis <martin@v.loewis.de>
parents:
42622
diff
changeset
|
36 for line in open_urlresource(TESTDATAURL, encoding="utf-8"): |
|
41554
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
37 if '#' in line: |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
38 line = line.split('#')[0] |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
39 line = line.strip() |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
40 if not line: |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
41 continue |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
42 if line.startswith("@Part"): |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
43 part = line.split()[0] |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
44 continue |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
45 if part == "@Part3": |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
46 # XXX we don't support PRI #29 yet, so skip these tests for now |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
47 continue |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
48 try: |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
49 c1,c2,c3,c4,c5 = [unistr(x) for x in line.split(';')[:-1]] |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
50 except RangeError: |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
51 # Skip unsupported characters; |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
52 # try atleast adding c1 if we are in part1 |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
53 if part == "@Part1": |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
54 try: |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
55 c1 = unistr(line.split(';')[0]) |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
56 except RangeError: |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
57 pass |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
58 else: |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
59 part1_data[c1] = 1 |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
60 continue |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
61 |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
62 # Perform tests |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
63 self.failUnless(c2 == NFC(c1) == NFC(c2) == NFC(c3), line) |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
64 self.failUnless(c4 == NFC(c4) == NFC(c5), line) |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
65 self.failUnless(c3 == NFD(c1) == NFD(c2) == NFD(c3), line) |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
66 self.failUnless(c5 == NFD(c4) == NFD(c5), line) |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
67 self.failUnless(c4 == NFKC(c1) == NFKC(c2) == \ |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
68 NFKC(c3) == NFKC(c4) == NFKC(c5), |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
69 line) |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
70 self.failUnless(c5 == NFKD(c1) == NFKD(c2) == \ |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
71 NFKD(c3) == NFKD(c4) == NFKD(c5), |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
72 line) |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
73 |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
74 # Record part 1 data |
|
36929
fab44c5c8df4
Update test data to 4.1; disable PRI #29 for now.
Martin v. Löwis <martin@v.loewis.de>
parents:
35998
diff
changeset
|
75 if part == "@Part1": |
|
41554
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
76 part1_data[c1] = 1 |
|
26387
fcf9734e1021
Whitespace normalization.
Tim Peters <tim.peters@gmail.com>
parents:
26379
diff
changeset
|
77 |
|
41554
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
78 # Perform tests for all other data |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
79 for c in range(sys.maxunicode+1): |
|
41585
825d966338ea
Rename 'unicode' to 'str' in its tp_name field. Rename 'str' to 'str8'.
Guido van Rossum <guido@python.org>
parents:
41575
diff
changeset
|
80 X = chr(c) |
|
41554
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
81 if X in part1_data: |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
82 continue |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
83 self.failUnless(X == NFC(X) == NFD(X) == NFKC(X) == NFKD(X), c) |
|
26391
d207d1d78c5d
Fiddled things so that test_normalization is expected to be skipped if
Tim Peters <tim.peters@gmail.com>
parents:
26388
diff
changeset
|
84 |
|
41554
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
85 def test_bug_834676(self): |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
86 # Check for bug 834676 |
|
41575
beacf40b5960
Rip out all the u"..." literals and calls to unicode().
Guido van Rossum <guido@python.org>
parents:
41554
diff
changeset
|
87 normalize('NFC', '\ud55c\uae00') |
|
41554
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
88 |
|
26379
1885b2d6393a
Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff
changeset
|
89 |
|
41554
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
90 def test_main(): |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
91 # Hit the exception early |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
92 open_urlresource(TESTDATAURL) |
|
06145fbc7ab9
Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents:
40775
diff
changeset
|
93 run_unittest(NormalizationTest) |
|
30867
04ca4d1f751f
Overallocate target buffer for normalization more early. Fixes #834676.
Martin v. Löwis <martin@v.loewis.de>
parents:
30056
diff
changeset
|
94 |
|
26391
d207d1d78c5d
Fiddled things so that test_normalization is expected to be skipped if
Tim Peters <tim.peters@gmail.com>
parents:
26388
diff
changeset
|
95 if __name__ == "__main__": |
|
d207d1d78c5d
Fiddled things so that test_normalization is expected to be skipped if
Tim Peters <tim.peters@gmail.com>
parents:
26388
diff
changeset
|
96 test_main() |
