view Tools/unicode/comparecodecs.py @ 58689:514b56efd89a 3.1

Merged revisions 77962 via svnmerge from svn+ssh://pythondev@svn.python.org/python/branches/py3k ................ r77962 | r.david.murray | 2010-02-04 11:41:57 -0500 (Thu, 04 Feb 2010) | 17 lines Merged revisions 77943,77959-77960 via svnmerge from svn+ssh://pythondev@svn.python.org/python/trunk ........ r77943 | r.david.murray | 2010-02-03 08:36:23 -0500 (Wed, 03 Feb 2010) | 2 lines Explicitly mention the default value for formatdate's usegmt parameter. ........ r77959 | nick.coghlan | 2010-02-04 07:43:58 -0500 (Thu, 04 Feb 2010) | 1 line Issue 6760: Clarify args handling for subprocess.Popen. Patch by Chris Rebert ........ r77960 | r.david.murray | 2010-02-04 11:33:31 -0500 (Thu, 04 Feb 2010) | 2 lines Add Chris Rebert to ACKS for issue 6760 Popen doc improvements. ........ ................
author R. David Murray <rdmurray@bitdance.com>
date Thu, 04 Feb 2010 16:44:31 +0000
parents 63741c7521aa
children fcfe324663a6
line wrap: on
line source

#!/usr/bin/env python

""" Compare the output of two codecs.

(c) Copyright 2005, Marc-Andre Lemburg (mal@lemburg.com).

    Licensed to PSF under a Contributor Agreement.

"""
import sys

def compare_codecs(encoding1, encoding2):

    print('Comparing encoding/decoding of   %r and   %r' % (encoding1, encoding2))
    mismatch = 0
    # Check encoding
    for i in range(sys.maxunicode):
        u = chr(i)
        try:
            c1 = u.encode(encoding1)
        except UnicodeError as reason:
            c1 = '<undefined>'
        try:
            c2 = u.encode(encoding2)
        except UnicodeError as reason:
            c2 = '<undefined>'
        if c1 != c2:
            print(' * encoding mismatch for 0x%04X: %-14r != %r' % \
                  (i, c1, c2))
            mismatch += 1
    # Check decoding
    for i in range(256):
        c = chr(i)
        try:
            u1 = c.decode(encoding1)
        except UnicodeError:
            u1 = '<undefined>'
        try:
            u2 = c.decode(encoding2)
        except UnicodeError:
            u2 = '<undefined>'
        if u1 != u2:
            print(' * decoding mismatch for 0x%04X: %-14r != %r' % \
                  (i, u1, u2))
            mismatch += 1
    if mismatch:
        print()
        print('Found %i mismatches' % mismatch)
    else:
        print('-> Codecs are identical.')

if __name__ == '__main__':
    compare_codecs(sys.argv[1], sys.argv[2])