Parses intel's microcode dat file into the actual inc's

parse_microcode_dat.py

# Author : Matan M. Mates
# Purpose: Parse Intel's microcode files into binaries

import re
import os
MICROCODE_FILE = "./microcode.dat"
LICENSE_MARKER = '/---'
OUTPUT_DIR = '.' + os.sep + "parsed_files"
COMMENT_CLOSE_MARKER_RE, COMMENT_START_MARKER_RE = "\*/", "/\*"
COMMENT_CLOSE_MARKER, COMMENT_START_MARKER = "*/", "/*"

def clean_data(raw_data):
	# Clear License
	data = raw_data[raw_data.index(LICENSE_MARKER)+len(LICENSE_MARKER):]
	# Clear date, first comment is the date
	return data[re.search(COMMENT_CLOSE_MARKER_RE, data).end():]

def binarize_file_data(data):
	""" File data is in comma seperated yucky hex..., fix it """
	return data.replace('0x','').replace(',','').decode('hex')


def parse_data(data):
	"""
	Return list of (filename,bindata)
	"""
	stop = False
	# Spaces and newlines are for losers
	data = data.replace('\n','').replace(' ','').replace('\t','')
	files = []
	while COMMENT_START_MARKER in data:
		file_name = data[data.index(COMMENT_START_MARKER) + len(COMMENT_START_MARKER):data.index(COMMENT_CLOSE_MARKER)]
		data = data[data.index(COMMENT_CLOSE_MARKER) + len(COMMENT_CLOSE_MARKER):]
		# Handle the last file with try
		try:
			file_data = data[:data.index(COMMENT_START_MARKER)]
			data = data[data.index(COMMENT_START_MARKER):]
		except:
			file_data = data
			pass
		files.append((file_name, binarize_file_data(file_data)))
	return files

def main():
	data = open(MICROCODE_FILE, 'rb').read()
	actual_data = clean_data(data)
	files = parse_data(actual_data)
	print "[*] There are {filec} files".format(filec=len(files))
	print "[*] Attempting to craete output directory: " + OUTPUT_DIR
	try:
		os.stat(OUTPUT_DIR)
		print "[!] Directory exists, Removing it."
		os.rmdir(OUTPUT_DIR)
	except:
		pass
	os.mkdir(OUTPUT_DIR)
	print "[*] Saving Files"
	for f in files:
		open(OUTPUT_DIR + os.sep + f[0], 'wb').write(f[1])
	print "[*] Finished"


if __name__ == '__main__':
	main()