codecs backslashreplace for decode()

backslashreplace.py

# codecs backslashreplace for decode()
# encoding: utf-8

from __future__ import unicode_literals

import sys
import codecs


_backslashreplace_errors = codecs.backslashreplace_errors


def backslashreplace_errors(exc):
    if isinstance(exc, UnicodeDecodeError):
        if sys.version_info[0] >= 3:
            tohex = lambda c: "\\x{0:02x}".format(c)
        else:
            tohex = lambda c: "\\x{0:02x}".format(ord(c))
        u = "".join(tohex(c) for c in exc.object[exc.start:exc.end])
        return (u, exc.end)
    return _backslashreplace_errors(exc)


def test():
    codecs.register_error("backslashreplace", backslashreplace_errors)
    u = "あいうえお"
    s = u.encode("utf-8")
    x = s.decode("ascii", "backslashreplace")
    print(x)


if __name__ == "__main__":
    test()

This code seems to create inconsistent results with python 3.4.1 for byte strings with embedded nulls.
Out of:

• b"nm\xf6p\x00p"
• b"nm\xf6p\x00pj"
• b"nm\xf6p\x00pjk"

only the longest gets the expected result for decode('utf-8', 'backslashreplace')