""" Compare .NET code pages to Python

"""
import sys, codecs
import System
import System.Text

_debug = 0

CHECK_CODECS = [
    # (.NET Code page, Python codec name)
    (65000, 'utf-7'),
    (65001, 'utf-8'),
    #(65005, 'utf-32-le'), # not supported by .NET 2.0
    #(65006, 'utf-32-be'), # not supported by .NET 2.0
    (1200, 'utf-16-le'),
    (1201, 'utf-16-be'),
    (28591, 'iso-8859-1'),
    #(28592, 'iso-8859-2'), # not supported by IronPython 2.6
    #(28593, 'iso-8859-3'), # not supported by IronPython 2.6
    #(28594, 'iso-8859-4'), # not supported by IronPython 2.6
    #(28595, 'iso-8859-5'), # not supported by IronPython 2.6
    ]
UNICODE_RANGE = sys.maxunicode
    
def compare_codec(code_page, encoding):
    net_codec = System.Text.Encoding.GetEncoding(code_page)
    try:
        python_encoder = codecs.getencoder(encoding)
    except LookupError:
        print 'Unknown encoding in Python: %r' % encoding
        return -1
    errors = 0
    for i in range(UNICODE_RANGE):
        u = unichr(i)
        try:
            python_bytes, count = python_encoder(u)
        except ValueError:
            if _debug: print '%i: skipped (Python encode error)' % i
            continue
        net_bytes = ''.join([chr(int(c)) for c in net_codec.GetBytes(u)])
        if python_bytes == net_bytes:
            if _debug: print '%i: OK' % i
        else:
            print '%i: .NET=%r, Python=%r' % (
                i, net_bytes, python_bytes)
            errors += 1
    return errors

for code_page, encoding in CHECK_CODECS:
    print 'Code Page %i vs. encoding %r' % (code_page, encoding)
    errors = compare_codec(code_page, encoding)
    print
    print '%i errors' % errors
    print