[Python-checkins] python/dist/src/Tools/unicode comparecodecs.py, NONE, 1.1 listcodecs.py, NONE, 1.1

lemburg@users.sourceforge.net lemburg at users.sourceforge.net
Fri Oct 21 15:47:06 CEST 2005


Update of /cvsroot/python/python/dist/src/Tools/unicode
In directory sc8-pr-cvs1.sourceforge.net:/tmp/cvs-serv14104

Added Files:
	comparecodecs.py listcodecs.py 
Log Message:
Add two new tools to compare codecs and show differences and to
list all installed codecs.


--- NEW FILE: comparecodecs.py ---
#!/usr/bin/env python

""" Compare the output of two codecs.

(c) Copyright 2005, Marc-Andre Lemburg (mal at 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 = unichr(i)
        try:
            c1 = u.encode(encoding1)
        except UnicodeError, reason:
            c1 = '<undefined>'
        try:
            c2 = u.encode(encoding2)
        except UnicodeError, 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 = u'<undefined>'
        try:
            u2 = c.decode(encoding2)
        except UnicodeError:
            u2 = u'<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])

--- NEW FILE: listcodecs.py ---
""" List all available codec modules.

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

    Licensed to PSF under a Contributor Agreement.

"""

import os, codecs, encodings

_debug = 0

def listcodecs(dir):
    names = []
    for filename in os.listdir(dir):
        if filename[-3:] != '.py':
            continue
        name = filename[:-3]
        # Check whether we've found a true codec
        try:
            codecs.lookup(name)
        except LookupError:
            # Codec not found
            continue
        except Exception, reason:
            # Probably an error from importing the codec; still it's
            # a valid code name
            if _debug:
                print '* problem importing codec %r: %s' % \
                      (name, reason)
        names.append(name)
    return names


if __name__ == '__main__':
    names = listcodecs(encodings.__path__[0])
    names.sort()
    print 'all_codecs = ['
    for name in names:
        print '    %r,' % name
    print ']'



More information about the Python-checkins mailing list