annotate Lib/test/test_normalization.py @ 73798:6cde416ef03b

Credit Per Øyvind Karlsen for the initial implementation of the lzma module (issue #6715).
author Nadeem Vawda <nadeem.vawda@gmail.com>
date Thu, 01 Dec 2011 01:18:27 +0200
parents 3c9232a684ae
children 61227b4c169f 7d69d04522e3
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
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
57467
74230e5ab373 Merged revisions 76535 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents: 57051
diff changeset
4 from http.client import HTTPException
26379
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
5 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
6 import os
49133
f97779262bac Merged revisions 66369 via svnmerge from
Martin v. Löwis <martin@v.loewis.de>
parents: 49130
diff changeset
7 from unicodedata import normalize, unidata_version
26391
d207d1d78c5d Fiddled things so that test_normalization is expected to be skipped if
Tim Peters <tim.peters@gmail.com>
parents: 26388
diff changeset
8
42622
4d2aa4bfb2d6 Remove RISCOS support
Skip Montanaro <skip@pobox.com>
parents: 41585
diff changeset
9 TESTDATAFILE = "NormalizationTest.txt"
49133
f97779262bac Merged revisions 66369 via svnmerge from
Martin v. Löwis <martin@v.loewis.de>
parents: 49130
diff changeset
10 TESTDATAURL = "http://www.unicode.org/Public/" + unidata_version + "/ucd/" + TESTDATAFILE
f97779262bac Merged revisions 66369 via svnmerge from
Martin v. Löwis <martin@v.loewis.de>
parents: 49130
diff changeset
11
59720
8a28be45c242 Merged revisions 78982,78986 via svnmerge from
Florent Xicluna <florent.xicluna@gmail.com>
parents: 57467
diff changeset
12 def check_version(testfile):
8a28be45c242 Merged revisions 78982,78986 via svnmerge from
Florent Xicluna <florent.xicluna@gmail.com>
parents: 57467
diff changeset
13 hdr = testfile.readline()
8a28be45c242 Merged revisions 78982,78986 via svnmerge from
Florent Xicluna <florent.xicluna@gmail.com>
parents: 57467
diff changeset
14 return unidata_version in hdr
26379
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
15
37188
760ffb0fba37 Exceptions should inherit from Exception now.
Neal Norwitz <nnorwitz@gmail.com>
parents: 36937
diff changeset
16 class RangeError(Exception):
26379
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
17 pass
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
18
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
19 def NFC(str):
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
20 return normalize("NFC", str)
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
21
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
22 def NFKC(str):
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
23 return normalize("NFKC", str)
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
24
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
25 def NFD(str):
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
26 return normalize("NFD", str)
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
27
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
28 def NFKD(str):
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
29 return normalize("NFKD", str)
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
30
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
31 def unistr(data):
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
32 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
33 for x in data:
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
34 if x > sys.maxunicode:
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
35 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
36 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
37
41554
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
38 class NormalizationTest(unittest.TestCase):
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
39 def test_main(self):
59720
8a28be45c242 Merged revisions 78982,78986 via svnmerge from
Florent Xicluna <florent.xicluna@gmail.com>
parents: 57467
diff changeset
40 part = None
41554
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
41 part1_data = {}
57054
4580fe0719e7 Merged revisions 76036 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents: 55593
diff changeset
42 # Hit the exception early
4580fe0719e7 Merged revisions 76036 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents: 55593
diff changeset
43 try:
59720
8a28be45c242 Merged revisions 78982,78986 via svnmerge from
Florent Xicluna <florent.xicluna@gmail.com>
parents: 57467
diff changeset
44 testdata = open_urlresource(TESTDATAURL, encoding="utf-8",
8a28be45c242 Merged revisions 78982,78986 via svnmerge from
Florent Xicluna <florent.xicluna@gmail.com>
parents: 57467
diff changeset
45 check=check_version)
57467
74230e5ab373 Merged revisions 76535 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents: 57051
diff changeset
46 except (IOError, HTTPException):
57054
4580fe0719e7 Merged revisions 76036 via svnmerge from
Antoine Pitrou <solipsis@pitrou.net>
parents: 55593
diff changeset
47 self.skipTest("Could not retrieve " + TESTDATAURL)
67087
89a416fd7c29 Fix ResourceWarning in test_normalization
Antoine Pitrou <solipsis@pitrou.net>
parents: 59720
diff changeset
48 self.addCleanup(testdata.close)
59720
8a28be45c242 Merged revisions 78982,78986 via svnmerge from
Florent Xicluna <florent.xicluna@gmail.com>
parents: 57467
diff changeset
49 for line in testdata:
41554
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
50 if '#' in line:
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
51 line = line.split('#')[0]
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
52 line = line.strip()
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
53 if not line:
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
54 continue
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
55 if line.startswith("@Part"):
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
56 part = line.split()[0]
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
57 continue
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
58 try:
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
59 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
60 except RangeError:
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
61 # Skip unsupported characters;
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
62 # 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
63 if part == "@Part1":
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
64 try:
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
65 c1 = unistr(line.split(';')[0])
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
66 except RangeError:
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
67 pass
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
68 else:
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
69 part1_data[c1] = 1
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
70 continue
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
71
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
72 # Perform tests
55593
0bb232078b8c Merged revisions 73715 via svnmerge from
Georg Brandl <georg@python.org>
parents: 50203
diff changeset
73 self.assertTrue(c2 == NFC(c1) == NFC(c2) == NFC(c3), line)
0bb232078b8c Merged revisions 73715 via svnmerge from
Georg Brandl <georg@python.org>
parents: 50203
diff changeset
74 self.assertTrue(c4 == NFC(c4) == NFC(c5), line)
0bb232078b8c Merged revisions 73715 via svnmerge from
Georg Brandl <georg@python.org>
parents: 50203
diff changeset
75 self.assertTrue(c3 == NFD(c1) == NFD(c2) == NFD(c3), line)
0bb232078b8c Merged revisions 73715 via svnmerge from
Georg Brandl <georg@python.org>
parents: 50203
diff changeset
76 self.assertTrue(c5 == NFD(c4) == NFD(c5), line)
0bb232078b8c Merged revisions 73715 via svnmerge from
Georg Brandl <georg@python.org>
parents: 50203
diff changeset
77 self.assertTrue(c4 == NFKC(c1) == NFKC(c2) == \
41554
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
78 NFKC(c3) == NFKC(c4) == NFKC(c5),
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
79 line)
55593
0bb232078b8c Merged revisions 73715 via svnmerge from
Georg Brandl <georg@python.org>
parents: 50203
diff changeset
80 self.assertTrue(c5 == NFKD(c1) == NFKD(c2) == \
41554
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
81 NFKD(c3) == NFKD(c4) == NFKD(c5),
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
82 line)
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
83
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
84 # 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
85 if part == "@Part1":
41554
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
86 part1_data[c1] = 1
26387
fcf9734e1021 Whitespace normalization.
Tim Peters <tim.peters@gmail.com>
parents: 26379
diff changeset
87
41554
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
88 # Perform tests for all other data
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
89 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
90 X = chr(c)
41554
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
91 if X in part1_data:
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
92 continue
55593
0bb232078b8c Merged revisions 73715 via svnmerge from
Georg Brandl <georg@python.org>
parents: 50203
diff changeset
93 self.assertTrue(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
94
41554
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
95 def test_bug_834676(self):
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
96 # 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
97 normalize('NFC', '\ud55c\uae00')
41554
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
98
26379
1885b2d6393a Patch #626485: Support Unicode normalization.
Martin v. Löwis <martin@v.loewis.de>
parents:
diff changeset
99
41554
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
100 def test_main():
06145fbc7ab9 Merged revisions 53952-54987 via svnmerge from
Guido van Rossum <guido@python.org>
parents: 40775
diff changeset
101 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
102
26391
d207d1d78c5d Fiddled things so that test_normalization is expected to be skipped if
Tim Peters <tim.peters@gmail.com>
parents: 26388
diff changeset
103 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
104 test_main()