hashing.py

   1 #! /usr/bin/env python
   2
   3 import md5
   4 import sys
   5 import os
   6 import platform
   7 import base64
   8
   9 # Fast hashing of files.
  10 # Initial implementation isn't intended to be fast, just work.
  11 #
  12 # TODO:
  13 #  - Use mmap() in cases where that would work.  Not sure this would
  14 #    really be any faster.
  15 #  - Optionally bind to faster MD5 library.
  16
  17 # When running on Linux, and root, avoid updating the atime of the
  18 # file.
  19 extra_flags = 0
  20 if platform.system() == 'Linux' and os.getuid() == 0:
  21     extra_flags = 01000000
  22
  23 def hashof(path):
  24     ufd = os.open(path, os.O_RDONLY | extra_flags)
  25     fd = os.fdopen(ufd, 'rb', 0)
  26     # fd = open(path, 'rb')
  27     hash = md5.new()
  28     while True:
  29         buf = fd.read(32768)
  30         if buf == '':
  31             break
  32         hash.update(buf)
  33     fd.close()
  34     return hash.digest()
  35
  36 if __name__ == '__main__':
  37     for name in sys.argv[1:]:
  38         print ("%s  %s" % (base64.b16encode(hashof(name)), name))