Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- #!/usr/bin/python
- import os
- from sys import argv
- from glob import glob
- from hashlib import md5 as MD5
- targetDirectory = "." if len(argv) < 2 else argv[1]
- foundFiles = {}
- for f in glob(os.path.join(targetDirectory, "*.swf")):
- if not os.path.isfile(f): continue
- fsize = os.path.getsize(f)
- if fsize in foundFiles:
- foundFiles[fsize].append(f)
- else:
- foundFiles[fsize] = [f]
- duplicateFiles = {}
- for size, files in foundFiles.items():
- if len(files) > 1:
- for f in files:
- data = open(f, "r").read()
- md5sum = MD5(data).hexdigest()
- if md5sum in duplicateFiles:
- duplicateFiles[md5sum].append(f)
- else:
- duplicateFiles[md5sum] = [f]
- for md5sum, files in duplicateFiles.items():
- if len(files) > 1:
- print "Duplicates for MD5:", md5sum
- for f in files:
- print " ", f
Advertisement
Add Comment
Please, Sign In to add comment