Copies files into a directory by SHA hash. This allows flat archival of files without worrying about filename clashes.

hashcopy.py

#!/usr/bin/env python


"""Utility to archive files by hash.

Gets the shasum of each file and copies the file to the specified dest_dir
with the hash inserted into the name.  For example, IMG_011.JPG would become
IMG_011.ab98feb.JPG, etc This is useful for image files that might share the
same filename but have differing contents (such as after editing).
"""

__author__ = '(Nathan Naze)'


import logging
import argparse
import os
import sys
import subprocess
import shutil


def _GetArgsParser():
  """Get the options parser."""

  parser = argparse.ArgumentParser(description='Copy into directory by path.')
  parser.add_argument('--dest_dir')
  parser.add_argument('paths', metavar='path', type=str, nargs='+',
                   help='Source paths to be copied.')
  return parser


def _GetShaHash(path):
  output = subprocess.check_output(['shasum', path])
  return output.split()[0]

def _CopyToHashFile(src_path, dir):
  sha_hash = _GetShaHash(src_path)
  logging.info('File %s shasum: %s', src_path, sha_hash)

  basename = os.path.basename(src_path)
  root, ext = os.path.splitext(basename)

  
  hash_filename = '%s.%s%s' % (root, sha_hash, ext)
  dest = os.path.join(dir, hash_filename)


  if os.path.exists(dest):
    logging.info('File exists, skipping. %s', dest)
    return

  logging.info('Copying file. %s to %s', src_path, dest)
  shutil.copy2(src_path, dest)
  

def main():
  logging.basicConfig(format=(sys.argv[0] + ': %(message)s'),
                      level=logging.INFO)
  
  parser = _GetArgsParser()
  args = parser.parse_args()

  dest_dir = args.dest_dir
  if not dest_dir:
    parser.print_help()
    sys.exit(1)
    return

  dest_dir = os.path.expanduser(dest_dir)
  if not os.path.isdir(dest_dir):
    logging.error('Did not find destination directory %s', dest_dir)
    parser.print_help()
    sys.exit(1)
    return
  
  for path in args.paths:
    path = os.path.expanduser(path)
    _CopyToHashFile(path, dest_dir)



if __name__ == '__main__':
  main()