Converts binary data to an arbitrary alphabet

alphabet-codec.py

"""
Converts binary data to an arbitrary alphabet
"""

class AlphabetCodec(object):

    def __init__(self, alphabet):
        self._alphabet = alphabet
        self._base = len(self._alphabet)

    def encode(self, data):
        # convert the bytes string into a bigint
        bigint = int(data.encode('hex'), 16)

        if bigint == 0:
            return self._alphabet[0]

        result = []
        while bigint > 0:
            mod = bigint % self._base
            bigint = bigint // self._base
            result.append(self._alphabet[mod])

        # Order and convert to a bytes string
        result.reverse()
        return ''.join(result)

    def decode(self, encoded):
        num = 0
        l = len(encoded) - 1
        for i, ch in enumerate(encoded):
            factor = self._base ** (l - i)
            num += self._alphabet.index(ch) * factor

        # convert the bigint into a bytes string
        h = '%x' % num
        return ('0'*(len(h) % 2) + h).decode('hex')



if __name__ == "__main__":
    import sys
    import base64

    original = 'aE1lTm9KMWw1eVZEd3hVU0tiNTRiSWVpV0Y0ZlZVUFdGeTFVdw=='
    if len(sys.argv) > 1:
        original = sys.argv[1]

    codec = AlphabetCodec('23456789abcdefghijkmnopqrstuvwxyzABCDEFGHJKLMNPQRSTUVWXYZ')

    data = base64.b64decode(original)
    print 'ORIGINAL: %s' % data

    encoded = codec.encode(data)
    print ' ENCODED: %s' % encoded

    decoded = codec.decode(encoded)
    print ' DECODED: %s' % decoded

    b64 = base64.b64encode(decoded)
    print '     B64: %s' % b64