localetest.py

#!/usr/bin/env python
# encoding: utf-8

"""
localetest.py - tests collation / sort order for various Latin-script locales

Correct output:

The order for German is:
LATIN SMALL LETTER A
LATIN SMALL LETTER A WITH DIAERESIS
LATIN SMALL LETTER Z

The order for British English is:
LATIN SMALL LETTER A
LATIN SMALL LETTER E
LATIN SMALL LETTER Z

The order for Polish is:
LATIN SMALL LETTER A
LATIN SMALL LETTER A WITH OGONEK
LATIN SMALL LETTER Z

Tests show Polish and German characters with diacritics are sorted incorrectly 
after "z" on OS X and maybe other systems.
See also http://stackoverflow.com/questions/3412933/python-not-sorting-unicode-properly-strcoll-doesnt-help
"""

import locale  
import unicodedata

testdata = {
    'en': {'chars': [u'a', u'z', u'e'], 'localestring': 'en_GB.UTF-8', 'lang': 'British English' },
    'de': {'chars': [u'a', u'z', u'ä'], 'localestring': 'de_DE.UTF-8', 'lang': 'German' },
    'pl': {'chars': [u'a', u'z', u'ą'], 'localestring': 'pl_PL.UTF-8', 'lang': 'Polish' }
}

for l in testdata:
    try:
        locale.setlocale(locale.LC_ALL, testdata[l]['localestring'])
    except locale.Error as e:
        print "Error for %s and locale %s: %s\n" % (l, testdata[l]['localestring'], e)
        continue

    print "The order for %s is:" % testdata[l]['lang']
    for item in sorted(testdata[l]['chars'], cmp=locale.strcoll):
        print unicodedata.name(item)
    print "The LC_COLLATE culture and encoding settings were %s." % ', '.join(locale.getlocale(locale.LC_COLLATE))
    print

OS X 10.6.4, gcc 4.2.1, python 2.6.5:
wrong for all, but English.

Same for me. Works fine on Ubuntu Linux 10.4, Python 2.6.5 built with gcc 4.4.3. Also works fine, with changed locale strings, on Windows XP.