#!/usr/bin/python

import string
import math
import bisect

#wordlist = map(lambda x: x[:-1], open('/usr/share/dict/web2').xreadlines())
wordlist = [ 'Carlin', 'shit', 'piss', 'fuck', 'cunt', 'cocksucker', 'motherfucker', 'tits' ]
wordlist.sort()
wlcnt = len(wordlist)
ngrps = 8
bitspgrp = math.log(wlcnt) / math.log(2)
bits = int(bitspgrp * ngrps)

def bin2word(bin):
	ret = []
	g = 0l
	nbits = 0
	#print len(bin)
	#print len(bin) * 8
	#print (len(bin) * 8) % bits
	#print bits - (len(bin) * 8) % bits
	#print (bits - (len(bin) * 8) % bits + 7) / 8
	nz = ((bits - ((len(bin) * 8) % bits) + 7) / 8)
	bin += '\0' * nz
	for i in bin:
		g = g * 256 + ord(i)
		nbits = nbits + 8
		#print hex(ord(i)), g, nbits
		if nbits >= bits:
			g, r = divmod(g, 2**(nbits - bits))
			t = []
			for j in range(ngrps):
				g, indx = divmod(g, wlcnt)
				t.append(wordlist[indx])
			t.reverse()
			ret.extend(t)
			g = r
			nbits -= bits

	ret.append(wordlist[nz + nbits * (((bits + 7) / 8) + 1)])
	return ' '.join(ret)

def word2bin(words):
	words = words.split(' ')
	cnt = len(words)

	ret = []
	g = 0
	nbits = 0
	for i in range((cnt - 1) / ngrps):
		r = g
		g = 0
		for j in range(ngrps):
			g *= wlcnt
			g += bisect.bisect(wordlist, words[i * ngrps + j]) - 1
		g = r * 2**bits + g
		nbits += bits
		#print nbits, g
		while nbits >= 8:
			r, g = divmod(g, 2**(nbits - 8))
			ret.append(chr(r))
			nbits -= 8
			#print nbits, hex(r), g

	z = bisect.bisect(wordlist, words[-1]) - 1
	eb, nz = divmod(z, (bits + 7) / 8 + 1)
	if nbits != eb:
		raise ValueError, ("input isn't valid", nbits, eb, nz)

	return ''.join(ret)[:-nz]

if __name__ == '__main__':
	import sys

	inp = open('/dev/urandom')
	for l in xrange(100):
		i = inp.read(l)
		j = bin2word(i)
		a = 0
		try:
			k = word2bin(j)
			a = 1
			if k != i:
				raise ValueError
		except ValueError, x:
			print 'failed at length:', l
			print repr(i)
			print j
			if a:
				print repr(k)
			raise ValueError, x
	print 'testing completed...'