#!/usr/bin/env python import io import os import re import sys #path = os.getcwd() #print path startpattern = """mod/public_WC/""" imgFormats = ['jpg','jpeg','gif','png','mp4'] images = [] ifile = io.open( "mod_public_WC.txt", mode='rt', encoding='utf-16' ) text = ifile.read() for match in re.finditer(r"mod/public_WC/.+?[.](gif|jpg|jpeg|png|mp4)", text, flags=re.U): imgfile = match.group().encode("utf-8") randmatch = re.search(r"'\s*[+]\s*rand\s*[(]\s*(\d+)\s*[,]\s*(\d+)\s*[)]\s*[+]\s*'", imgfile) if randmatch != None: for i in range(int(randmatch.group(1)), 1+int(randmatch.group(2))): images.append(re.sub(r"'\s*[+]\s*rand\s*[(].*?[)]\s*[+]\s*'", str(i), imgfile)) else: images.append(imgfile) ifile.close() for image in images: if not re.search(r"[<$]", image) and not os.path.isfile(image): print "Image not found:", image