annotate Tools/unicode/comparecodecs.py @ 70817:dd775b1561ba

Fix markup.
author Georg Brandl <georg@python.org>
date Tue, 14 Jun 2011 21:09:55 +0200
parents fcfe324663a6
children 98f64cbed2ac f39b26ca7f3d
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
59522
fcfe324663a6 convert shebang lines: python -> python3
Benjamin Peterson <benjamin@python.org>
parents: 47103
diff changeset
1 #!/usr/bin/env python3
35761
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
2
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
3 """ Compare the output of two codecs.
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
4
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
5 (c) Copyright 2005, Marc-Andre Lemburg (mal@lemburg.com).
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
6
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
7 Licensed to PSF under a Contributor Agreement.
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
8
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
9 """
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
10 import sys
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
11
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
12 def compare_codecs(encoding1, encoding2):
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
13
42427
46625e6122e9 Convert print statements to function calls in Tools/.
Collin Winter <collinw@gmail.com>
parents: 40587
diff changeset
14 print('Comparing encoding/decoding of %r and %r' % (encoding1, encoding2))
35761
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
15 mismatch = 0
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
16 # Check encoding
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
17 for i in range(sys.maxunicode):
47103
63741c7521aa More 2to3 fixes in the Tools directory. Fixes #2893.
Georg Brandl <georg@python.org>
parents: 42427
diff changeset
18 u = chr(i)
35761
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
19 try:
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
20 c1 = u.encode(encoding1)
40587
6b18a095c2e7 SF patch 1631942 by Collin Winter:
Guido van Rossum <guido@python.org>
parents: 35761
diff changeset
21 except UnicodeError as reason:
35761
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
22 c1 = '<undefined>'
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
23 try:
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
24 c2 = u.encode(encoding2)
40587
6b18a095c2e7 SF patch 1631942 by Collin Winter:
Guido van Rossum <guido@python.org>
parents: 35761
diff changeset
25 except UnicodeError as reason:
35761
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
26 c2 = '<undefined>'
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
27 if c1 != c2:
42427
46625e6122e9 Convert print statements to function calls in Tools/.
Collin Winter <collinw@gmail.com>
parents: 40587
diff changeset
28 print(' * encoding mismatch for 0x%04X: %-14r != %r' % \
46625e6122e9 Convert print statements to function calls in Tools/.
Collin Winter <collinw@gmail.com>
parents: 40587
diff changeset
29 (i, c1, c2))
35761
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
30 mismatch += 1
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
31 # Check decoding
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
32 for i in range(256):
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
33 c = chr(i)
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
34 try:
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
35 u1 = c.decode(encoding1)
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
36 except UnicodeError:
47103
63741c7521aa More 2to3 fixes in the Tools directory. Fixes #2893.
Georg Brandl <georg@python.org>
parents: 42427
diff changeset
37 u1 = '<undefined>'
35761
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
38 try:
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
39 u2 = c.decode(encoding2)
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
40 except UnicodeError:
47103
63741c7521aa More 2to3 fixes in the Tools directory. Fixes #2893.
Georg Brandl <georg@python.org>
parents: 42427
diff changeset
41 u2 = '<undefined>'
35761
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
42 if u1 != u2:
42427
46625e6122e9 Convert print statements to function calls in Tools/.
Collin Winter <collinw@gmail.com>
parents: 40587
diff changeset
43 print(' * decoding mismatch for 0x%04X: %-14r != %r' % \
46625e6122e9 Convert print statements to function calls in Tools/.
Collin Winter <collinw@gmail.com>
parents: 40587
diff changeset
44 (i, u1, u2))
35761
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
45 mismatch += 1
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
46 if mismatch:
42427
46625e6122e9 Convert print statements to function calls in Tools/.
Collin Winter <collinw@gmail.com>
parents: 40587
diff changeset
47 print()
46625e6122e9 Convert print statements to function calls in Tools/.
Collin Winter <collinw@gmail.com>
parents: 40587
diff changeset
48 print('Found %i mismatches' % mismatch)
35761
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
49 else:
42427
46625e6122e9 Convert print statements to function calls in Tools/.
Collin Winter <collinw@gmail.com>
parents: 40587
diff changeset
50 print('-> Codecs are identical.')
35761
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
51
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
52 if __name__ == '__main__':
7a5ac203154a Add two new tools to compare codecs and show differences and to
Marc-André Lemburg <mal@egenix.com>
parents:
diff changeset
53 compare_codecs(sys.argv[1], sys.argv[2])