No-character-mapping-nonsense Python2/3 String class

string.py

import six
from six.moves import xrange


class String(six.text_type):

    @staticmethod
    def as_raw_unicode(value=u""):
        if isinstance(value, six.text_type):
            return value
        elif isinstance(value, six.binary_type):
            return u"".join(six.unichr(c) for c in six.iterbytes(value))
        else:
            return six.text_type(value)

    def raw_encode(self):
        return b''.join(b''.join(six.int2byte((ord(uc) & 0xFF00) >> 8),
                                 six.int2byte(ord(uc) & 0xFF)) for uc in self)

    @classmethod
    def raw_decode(cls, encoded):
        return cls(u"".join(six.unichr((six.byte2int(encoded[i]) << 8) | six.byte2int(encoded[i+1]))
                             for i in xrange(0, len(encoded), 2)))

    @staticmethod
    def __new__(cls, *more):
        return six.text_type.__new__(cls, cls.as_raw_unicode(more[0]) if more else u"")

This does look cool. It would be great to have some test strings with expected output for each.

def as_raw_unicode(value=u""):
    if isinstance(value, bytes):
        return codecs.decode(value, 'raw_unicode_escape')
    else:
        return six.text_type(value)

Also, on raw_decode() couldn't [ and [ and ] and ] all be removed to remain in generator mode and avoid creating lists of list?

Also, on the last line unicode should be replaced with six.text_type.

@cclauss nice catches, thanks! Updated the code.