Recursively remove duplicate files

remove_duplicates.py

#!/usr/bin/env python

import os
from optparse import OptionParser

def remove_duplicates(working_path, compare_path, pretend = False,
                      compare_size = False, recurse = False):
	'''Compares contents of two paths, removing
       identical content in the working path.'''

	contents = os.listdir(working_path)

	files = [f for f in contents if
		os.path.isfile(os.path.join(working_path, f)) and
		os.path.exists(os.path.join(compare_path, f)) and
		os.path.isfile(os.path.join(compare_path, f))]

	for file in files:
		working_file = os.path.join(working_path, file)
		compare_file = os.path.join(compare_path, file)
		if compare_size:
			if os.path.getsize(working_file) == os.path.getsize(compare_file):
				print 'Deleting %s' % working_file
				if not pretend: os.unlink(working_file)
		else:
			print 'Deleting %s' % working_file
			if not pretend: os.unlink(working_file)

	if recurse:
		directories = [d for d in contents if
			os.path.isdir(os.path.join(working_path, d)) and
			os.path.exists(os.path.join(compare_path, d)) and
			os.path.isdir(os.path.join(compare_path, d))]
		for dir in directories:
			working_dir = os.path.join(working_path, dir)
			remove_duplicates(
				working_dir, os.path.join(compare_path, dir),
				pretend, compare_size, recurse)
			if not os.listdir(working_dir):
				print 'Removing empty directory: %s' % working_dir
				os.rmdir(working_dir)


if __name__ == '__main__':

	parser = OptionParser(
		usage = '%prog [-p] [-s] [-r] working_path compare_path',
		version = '%prog 0.1')

	parser.add_option('-p', '--pretend',
		action = "store_true", dest = "pretend", default = False,
		help = 'Shows the files which will be deleted without deleting them.')
	parser.add_option('-s', '--size',
		action = "store_true", dest = "compare_size", default = False,
		help = 'Compares the file sizes in addition to names.')
	parser.add_option('-r', '--recursive',
		action = "store_true", dest = "recurse", default = False,
		help = 'Make a recursive comparison through directories.')

	(options, args) = parser.parse_args()

	if len(args) != 2:
		parser.error('Incorrect number of arguments.')

	remove_duplicates(args[0], args[1], options.pretend,
		options.compare_size, options.recurse)