try harder to parse headers

gistfile1.py

def utf8decode(thing):
    """Try to convert from ASCII to UTF-8."""

    try:
        return thing.decode('utf-8')
    except UnicodeDecodeError, _exc:
        # many badly created messages will be like this
        try:
            return thing.decode('iso-8859-15')
        except UnicodeDecodeError, exc:
            # haven't run into this case yet.  why not?
            # do i need to try and parse more bad spam?
            logging.error('failed to decode %s: %s', thing, str(exc))
            sys.exit(1)


def rfc2047decode(name):
    """A few fields can be encoded with RFC 2047.

    Use as

        subject = rfc2047decode(message.get('Subject', ''))

    Details at http://tools.ietf.org/html/rfc2047.html

    Note: does not preserve whitespace.  For example,

    Re: [Loungers] =?utf-8?q?Korea?=

    and

    Re: [Loungers]     =?utf-8?q?Korea?=

    both decode to

    Re: [Loungers] Korea

    TODO: check if this behavior is appropriate.
    """

    if name:
        decodelist = decode_header(name)
        retls = []
        for pair in decodelist:
            if pair[1]:
                try:
                    dname = pair[0].decode(pair[1])
                except UnicodeDecodeError, exc:
                    logging.warning('unable to decode %s as %s: %s',
                                    pair[0], pair[1], str(exc))
                    retls.append(utf8decode(pair[0]))
                else:
                    retls.append(dname)
            else:
                retls.append(utf8decode(pair[0]))
        return u' '.join(retls)

    return name