disas.py

#coding=utf-8
import sys,os
import re
from collections import namedtuple
from itertools import *

class Asm(object):

	def __init__(self,address,binary,opecode,operand):
		self.address = address
		self.binary = binary
		self.opecode = opecode
		self.operand = operand

	def next_address(self):
		return self.address + len(self.binary)

	def __str__(self):
		address = "%x" % self.address
		binary = ' '.join(['%02x' % i for i in self.binary])
		opecode = self.opecode if self.opecode is not None else ''
		operand = self.operand if self.operand is not None else ''
		return " %s:\t%-22s\t%s\t%s" % (address,binary,opecode,operand)

def load_assembler(file_name):
	part = re.compile(r"\s(?P<address>\w+):\s+(?P<binary>(\w\w\s)+)\s+((?P<opecode>\w+)(\s+(?P<operand>.*))?)?$")
	data = [i.strip('\n') for i in open(file_name)]
	ret = []
	i = 0
	text_section = False
	while i < len(data):
		line = data[i]
		if line == "Disassembly of section .text:":
			print line
			text_section = True
			i += 3
			continue
		if text_section:
			if line == "":
				break
			m = part.match(line)
			address = int(m.group("address"),16)
			binary = [int(j,16) for j in m.group("binary").split()]
			opecode = m.group("opecode")
			operand = m.group("operand")
			asm = Asm(address,binary,opecode,operand)
			ret.append(asm)
		i += 1

	return ret

def analyze(asms):
	next_instruction_address = asms[0].address
	current_graph = []
	for asm in asms:
		if next_instruction_address < asm.address:
			yield current_graph
			current_graph = []
			next_instruction_address = asm.address
		current_graph.append(asm)
		if asm.opecode == 'ret':
			continue
		elif asm.opecode.startswith('j'):
			jump_to = int(asm.operand.split()[0],16)
			to = max(jump_to,asm.next_address())
			next_instruction_address = max(next_instruction_address,to)
		else:
			next_instruction_address = max(next_instruction_address,asm.next_address())

def main():
	asms = load_assembler(sys.argv[1])
	for graph in analyze(asms):
		print "-" * 30
		for i in graph:
			print i


if __name__ == '__main__':
	main()