# find and unify duplicate files
# -- drt@un.bewaff.net - http://c0re.jp/

import os
from stat import *

doreplace = 0
doprint = 1

dupecheck = {}
dupelist = {}

find = os.popen('find -x /usr/local/xitami/webpages/aldi/ -type f -exec gmd5sum {} \;' 'r')
for x in find.readlines():
	md5 = long("0x%s" % x[:32], 16)
	name = x[32:].strip()
	if md5 not in dupecheck:
		dupecheck[md5] = name
	else:
		if md5 not in dupelist:
			dupelist[md5] = [dupecheck[md5]]
		dupelist[md5].append(name)

if doprint:
	for l in dupelist.values():
		print l
		
if doreplace:
	wastesize = 0
	for l in dupelist.values():
		modes = []
		for x in l:
			modes.append((x, os.stat(x)))
			if len(modes) > 1:
				wastesize += modes[-1][-1][ST_SIZE]
		base = modes[0][0]
		basestat = modes [0][1]
		for (name, modes) in modes[1:]:
			if basestat[ST_UID] != modes[ST_UID] or \
			   basestat[ST_UID] != modes[ST_UID] or \
			   basestat[ST_MODE] != modes[ST_MODE]:
				print "permissions differ: %r" % name
			else:
				os.unlink(name)
				os.link(base, name)
				#os.utime(name, (modes[ST_ATIME], modes[ST_MTIME]))
	print wastesize