copykill 10-Sep-2003 17:12 7275

copykill.py

#!/usr/bin/python

import os, sys, string, getopt, filecmp
from path import path
from types import *
from stat import *

optlist = []

"""Global default parameters"""

doclean = 0
doquar = 0
interactive = 0
quiet = 0
recurse = 0
mode = 0

src = ''
dest = ''
quarantine = ''

quarantine = os.path.join(os.environ['HOME'], 'copyquarantine')

def dispver():
	"""Displays version information"""
	print "copykill copyright 2003 Markus T�rnqvist, [email protected]"
	print "http://mjt.nysv.org/projects/copykill/"
	print "distributed under the gpl"
	print ""
	
def disphelp():
	dispver()
	"""Displays help switches"""
	print "Usage: copykill [OPTION]... --src=SOURCE --dest=DESTINATION"
	print "   or: copykill [OPTION]... -d DIRECTORY"
	print "Looks for identical files in DIRECTORY or SOURCE and DESTINATION"
	print ""

	helpdict = ({'-d, --dest=DESTINATION': 'destination to compare',
		'-c, --clean': 'clean duplicates',
		'-h, --help': 'this help here',
		'-i': 'interactive, overrides -q',
		'-q, --quiet': 'no output',
		'-Q, --quarantine=QUARANTINE': 'quarantine dir, def: [' + quarantine + ']',
		'-r': 'recurse subdirectories',
		'-s, --src=SOURCE': 'source to compare'})
		
	alflist = []
	
	for switch in helpdict.keys():
		alflist.append(switch)
		alflist.sort()
		
	for switch in alflist:
		print '  %-27s %s' % (switch, helpdict[switch])

	print ""
	print "When quaranteening, only one of identically named identical files"
	print "survives the process, that's the one in the source directory."
	print "Files are paired in numerically named subdirectories."
	print "You may give deeper hierarchies (quarantine/music/mp3 for example)"
	print "as the quarantine directory."
	
	sys.exit(1)
	
def parsargs(args):
	"""Parses the arguments"""

	opts, args = getopt.getopt(args,
		"cd:hiqQ:rs:",
		["dest",
		"clean",
		"help",
		"quiet",
		"quarantine=",
		"src="])

	for opt, arg in opts:
		global optlist

		optlist.append(opt)
	
		"All of the Above"
	
		if opt in ('-q', '--quiet'):
			global quiet
			quiet = 1
		elif opt == '-r':
			global recurse
			recurse = 1
		elif opt in ('-c', '--clean'):
			global doclean
			doclean = 1
	
		elif opt in ('-h', '--help'):
			disphelp()
			break
		
		elif opt == '-i':
			global interactive
			interactive = 1

		elif opt in ('-d', '--dest'):
			global dest
			dest = path(arg)
		elif opt in ('-Q', '--quarantine'):
			global quarantine
			global doquar
			if not arg:
				arg = quarantine
			
			quarantine = arg
			quarantine = path(quarantine)
			doquar = 1
			doclean = 1
		elif opt in ('-s', '--src'):
			global src
			src = path(arg)

	if not opts:
		os.system(sys.argv[0] + " --help")
		sys.exit(1)

	if src == dest:
		if not '--help' in optlist:
			os.system(sys.argv[0] + " --help")
			sys.exit(1)
	
	if interactive and quiet:
		quiet = 0

# End of parsargs(args)

# Do all sorts of funky stuff with the data
parsargs(sys.argv[1:])

# So, we're done here as well.

# We need these.
if not '--src' in optlist:
	if not '-s' in optlist:
		if '-d' in optlist:
			mode = 1
			src = dest
		else:
			if not '--help' in optlist:
				os.system(sys.argv[0] + " --help")

if not '--dest' in optlist:
	if not '-d' in optlist:
		if not '--help' in optlist:
			os.system(sys.argv[0] + " --help")

if not quiet:
	dispver()
	if mode == 1:
		print "Directory:", dest
	elif mode == 0:
		print "Source:", src
		print "Destination:", dest

# End of them

# Define the functions
def do_recurse(dir, list):
	"""Given a directory and a list, do_recurse appends the files to the list"""
	dir = path(dir).relpathto(dir)
	try:
		for file in dir.walkfiles():
			list.append(str(file))
	except OSError, (exnum, exstr):
		if exnum == 13:
			if not quiet:
				print "Permission denied for", file
		else:
			raise
		
def do_getlists(source, destination):
	"""do_getlists gets the file lists of source and destination, it may use
		do_recurse for the dirty work"""
	src_list = []
	dest_list = []

	if not recurse:
		src_list = os.listdir(source)
		dest_list = os.listdir(destination)
	
	elif recurse:
		os.chdir(source)
		do_recurse(source, src_list)
		os.chdir(destination)
		do_recurse(destination, dest_list)

	return src_list, dest_list
	
def do_cmp(src_list, dest_list):
	"""do_cmp runs a filecmp.cmp() over the files in the lists"""
	cmpmatchlist = []
	
	for file in src_list:
		cfile = os.path.join(src, file)
		for dfile in dest_list:
			cdfile = os.path.join(dest, dfile)
			val = filecmp.cmp(cfile, cdfile, 0)
			if val == 1:
				if mode == 0:
					if os.path.getsize(cfile) != 0:
						cmpmatchlist.append((cfile, cdfile))
				if mode == 1:
					if cfile != cdfile:
						if not (cdfile, cfile) in cmpmatchlist:
							if os.path.getsize(cfile) != 0:
								cmpmatchlist.append((cfile, cdfile))
	
	return cmpmatchlist

def do_quarantine(src, dest, quarantine, matchlist):
	"""quaranteens files"""
	numlist = []
	
	if not os.path.exists(quarantine):
		os.makedirs(quarantine)
	
	for num in range(len(matchlist)):
		num = string.zfill(num, 3)
		try:
			os.makedirs(os.path.join(quarantine, num))
		except OSError, (exnum, exstr):
			if exnum == 17:
				if not quiet:
					print "Directory already exists: " + str(num)
				else:	
					pass
			else:
				raise
			
		numlist.append(num)
		
	for match in matchlist:
		file, dfile = match
		num = numlist[0]
		if not quiet:
			print "Quaranteening to " + os.path.join(quarantine, num) + " file " + file
			print "Quaranteening to " + os.path.join(quarantine, num) + " file " + dfile
		sf = open(os.path.join(src, file), 'r')
		df = open(os.path.join(dest, dfile), 'r')

		qsf = open(path(quarantine / num / file.basename()), 'w')
		qdf = open(path(quarantine / num / dfile.basename()), 'w')

		qsf.write(sf.read())
		if dfile.basename() != file.basename():
			qdf.write(df.read())

		qsf.close()
		qdf.close()
		sf.close()
		df.close()

		del numlist[0]

def do_clean(src, dest, matchlist):
	"""cleans up extra files"""
	if interactive == 1:
		dodel = 0
	elif interactive == 0:
		dodel = 1

	def do_interactive(path, file):
		print "Do you want to delete", os.path.join(path, file), "[y/n] ",
		answer = raw_input()
		while answer not in ['y', 'n']:
			print "[y/n] ",
			answer = raw_input()
		if answer == 'y':
			dodel = 1
		elif answer == 'n':
			dodel = 0
		return dodel
	
	for delete in matchlist:
		delfile, deldfile = delete
		
		if interactive == 1:
			dodel = do_interactive(src, delfile)

		if dodel == 1:
			print "[SRC] Cleaning up",
			print os.path.join(src, delfile),
			try:
				os.remove(os.path.join(src, delfile))
				print "_o/"
			except OSError:
				print "_o_"

		if doquar:
			if interactive == 1:
				dodel = do_interactive(dest, deldfile)
			if dodel == 1:
				print "[DST] Cleaning up",
				print os.path.join(dest, deldfile),
				try:
					os.remove(os.path.join(dest, deldfile))
					print "_o/"
				except OSError:
					print "_o_"

# End of function definitions

# Here we get the variables we play with
src_list, dest_list = do_getlists(src, dest)

cmpmatchlist = do_cmp(src_list, dest_list)
# Got them!

# Do something sexy
if not quiet:
	print ""
	for match in cmpmatchlist:
		print "Match: %s <=> %s" % match
	print ""
	if doquar:
		do_quarantine(src, dest, quarantine, cmpmatchlist)
	if doclean:
		do_clean(src, dest, cmpmatchlist)
		
# Stop doing something sexy