z85.py

"""Python implementation of Z85 85-bit encoding

Z85 encoding is a plaintext encoding for a bytestring interpreted as 32bit integers.
Since the chunks are 32bit, a bytestring must be a multiple of 4 bytes.
See ZMQ RFC 32 for details.


"""

#-----------------------------------------------------------------------------
#  Copyright (c) 2013 Brian Granger, Min Ragan-Kelley
#
#  This file is part of pyzmq
#
#  Distributed under the terms of the New BSD License.  The full license is in
#  the file COPYING.BSD, distributed as part of this software.
#-----------------------------------------------------------------------------

import struct
# Z85CHARS is the base 85 symbol table
Z85CHARS = b"0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ.-:+=^!/*?&<>()[]{}@%$#"
# Z85MAP maps integers in [0,84] to the appropriate character in Z85CHARS
Z85MAP = dict([(c, idx) for idx, c in enumerate(Z85CHARS)])

_85s = [ 85**i for i in range(5) ][::-1]

def encode(rawbytes):
    """encode raw bytes into Z85"""
    # Accepts only byte arrays bounded to 4 bytes
    if len(rawbytes) % 4:
        raise ValueError("length must be multiple of 4, not %i" % len(rawbytes))
    
    nvalues = len(rawbytes) / 4
    
    values = struct.unpack('>%dI' % nvalues, rawbytes)
    encoded = []
    for v in values:
        for offset in _85s:
            encoded.append(Z85CHARS[(v // offset) % 85])
    
    return b''.join(encoded)

def decode(z85bytes):
    """decode Z85 bytes to raw bytes"""
    if len(z85bytes) % 5:
        raise ValueError("Z85 length must be multiple of 5, not %i" % len(z85bytes))
    
    nvalues = len(z85bytes) / 5
    values = []
    for i in range(0, len(z85bytes), 5):
        value = 0
        for j, offset in enumerate(_85s):
            value += Z85MAP[z85bytes[i+j]] * offset
        values.append(value)
    return struct.pack('>%dI' % nvalues, *values)

@remdragon: Your version is much appreciated, I'm going to use it only for myself (to encode zipped data as a PNG image comment) so it's not a problem that it is incompatible.

The + padding in this line is not necessary, as the rawbytes array is already a multiple of 4 bytes long:
nvalues = (len(rawbytes) + padding) // 4