#!/usr/bin/env python

# md5coll parser
# by Javantea aka. Joel R. Voss

"""
md5coll finds an md5 collision and outputs a C formatted file.
Luckily for us, it is easy to parse this file with python.
This gives us a pair of binaries with the same md5.
To test, we use Python's built-in md5 hash. It's easier than having to run 
md5sum on the two resulting files, but why not do that also.
Hash collisions are cool because you can append anything to them and they still 
collide.
cat md5coll31.bin md5coll3.py | md5sum
d5dc7227898bdc128fd3fd9255d6d457
cat md5coll32.bin md5coll3.py | md5sum
d5dc7227898bdc128fd3fd9255d6d457
"""

import md5
import base64
import struct

def parse_md5coll(filename='md5coll2a.txt', outfile='md5coll3a'):
	input_file = open(filename, 'r')
	data = input_file.read()
	input_file.close()
	lines = data.split("\n")
	Start1 = False
	Start2 = False
	Data1  = ''
	Data2  = ''
	for line in lines:
		if Start1:
			if line == '};':
				Start1 = False
				continue
			#end if
			Data1 += line
		elif Start2:
			if line == '};':
				Start2 = False
				continue
			#end if
			Data2 += line
		else:
			if line == 'unsigned int m0[32] = {':
				Start1 = True
			elif line == 'unsigned int m1[32] = {':
				Start2 = True
			#else:
			#	print line
			#end if
		#end if
	#next line
	
	DataList1 = eval('[' + Data1 + ']')
	Binary1 = ''
	for i in DataList1: Binary1 += struct.pack("L", i)
	#print "Data1:", Data1, base64.b16encode(Binary1)
	print "md5 1:", base64.b16encode(md5.new(Binary1).digest())
	
	d = open(outfile + '1.bin', 'wb')
	d.write(Binary1)
	d.close()
	
	DataList2 = eval('[' + Data2 + ']')
	Binary2 = ''
	for i in DataList2: Binary2 += struct.pack("L", i)
	#print "Data2:", Data2, base64.b16encode(Binary2)
	print "md5 2:", base64.b16encode(md5.new(Binary2).digest())
	
	d = open(outfile + '2.bin', 'wb')
	d.write(Binary2)
	d.close()
#end def parse_md5coll(filename, outfile)

def main():
	from sys import argv
	if len(argv) > 1:
		infile = argv[1]
		if infile == '-h' or infile == '--help':
			print 'Usage: python md5coll3.py infile outfile'
			print 'It will read infile and create two files:'
			print '<outfile>1.bin and <outfile>2.bin'
		#end if
	else:
		infile = 'md5coll2a.txt'    # The input file
	#end if
	if len(argv) > 2:
		outfile = argv[2]
	else:
		outfile = 'md5coll3'        # The output file
	#end if
	try:
		parse_md5coll(infile, outfile)
	except Exception, ex:
		print ex
	#end try
#end def main()

if __name__ == '__main__':
	main()
#end if