#!/usr/bin/env python

#
# mpfextract.py:	A script to extract clipart files from Microsoft Office Clipart Manager
# 			clipart packages.
#			Written by DarkWyrm <darkwyrm@gmail.com>
#			Licensed under the GPL
import base64, sys, os
import xml.dom.minidom

def ExtractMPF(filename, path):
	fileHandle = open(filename)
	doc = xml.dom.minidom.parse(fileHandle)
	
	if path != '.':
		os.chdir(path)
	
	filenameList = doc.getElementsByTagName('C:filepath')
	filedataList = doc.getElementsByTagName('C:contents')
	for i in range(0,filenameList.length):
		name = filenameList[i].firstChild.data		
		encodedData = filedataList[i].firstChild.data
		decodedData = base64.b64decode(encodedData)
		outfileHandle = open(name,'w')
		outfileHandle.write(decodedData)
		outfileHandle.close()

def main():
	args = sys.argv[1:]
	argcount = len(args)
	infile = 'ClipArt.mpf'
	if  argcount < 1 or argcount > 2:
		print 'usage: mpfextract.py infile.mpf [outfolder]\n'
		print 'defaulting to extracting ClipArt.mpf to the current folder.'
	else:
		infile = args[0]
	outfolder = '.'
	if argcount == 2:
		outfolder = args[1]
	ExtractMPF(infile,outfolder)


if __name__ == '__main__':
	main()