http://stackoverflow.com/a/3956975/309483

utf-8-binencode.py

import operator
import functools
import itertools
import os
import json
import struct
#import math

import bitarray

def encode(minbytes):

  mbinarr = bitarray.bitarray()
  mbinarr.frombytes(minbytes)

  orglength = len(mbinarr)
  for i in itertools.count(0,8):
    pos = orglength - 7*(i//8) - 7
    if pos < 0: break
    mbinarr.insert(pos,0)

  padlen = (8 - (len(mbinarr) % 8)) % 8
  #padlen2 = math.ceil(len(mbinarr) / 8) * 8 - len(mbinarr)
  #if padlen != padlen2: raise Exception("uoverenstemmelse!" + str((len(mbinarr), padlen, padlen2, mbinarr)))

  for i in range(padlen): mbinarr.insert(0,0)

  mstring = mbinarr.tobytes().decode("utf-8")

  return (orglength, mstring)

#tochars1 = lambda mbinarr: '' if len(mbinarr) == 0 else bin(int.from_bytes(mbinarr.tobytes(), 'big'))[2:].zfill(len(mbinarr))
#tochars2 = lambda mbinarr: ''.join(list(map(lambda x: '1' if x else '0', mbinarr.tolist()))) # .zfill(mbinarr.length())
#tochars3 = lambda mbinarr: mbinarr.to01()

def decode(pair):
  (rawlen, mstring) = pair

  mbinarr = bitarray.bitarray()
  mbinarr.frombytes(mstring.encode())

  orglength = len(mbinarr)

  for i in itertools.count(0,8):
    pos = orglength - i - 8
    if pos < 0: break
    mbinarr.pop(pos)

  for i in range(len(mbinarr)-rawlen):
    mbinarr.pop(0)

  return mbinarr.tobytes()
    
#data = (0x12345678abcdef).to_bytes(7,'big')
#data = (0xffffffffffffff).to_bytes(7,'big')
#data = functools.reduce(operator.add, (map(lambda x: x.to_bytes(4,'big'), [0xaaaaaaaa, 0xbbbbbbbb, 0xcccccccc, 0xdddddddd, 0xeeeeeeee, 0xffffffff])))

def test():
  def iteration(data):
    data2 = decode(encode(data))

    if data2 != data:
      raise Exception("error!\n{}\n{}".format(data, data2))

  iteration(b"")

  n = 100
  for i in range(1,n+1):
    l = int.from_bytes(os.urandom(2),'big') >> 7
    data = os.urandom(l)
    iteration(data)
    if (i) % (n // 10) == 0: print(i // (n // 10) * 10 , '% complete')

if __name__ == "__main__": test()