├── README
└── filecomp.py


/README:
--------------------------------------------------------------------------------
 1 | /*********************************\
 2 |            filecomp.py
 3 |      Will compare files in two
 4 |       directories for missing
 5 |       and/or mismatched files
 6 |        (based upon md5sums).
 7 | 
 8 |       Currently a bit sloppy
 9 |        and lacks recursion.
10 | 
11 |      Written by Twitch(Ben) 
12 | \*********************************/
13 | 
14 | This is a very small, as of yet, utility to compare the contents of a pair of directories. This became necessary when managing archived files with their originals and is likely quite useful for managing files/data across multiple systems (without using something overly complex like git/svn/cvs).
15 | 
16 | Recursion is now in place, though not exceedingly well tested.
17 | 
18 | 2010-07-21
19 | 
20 |    Picked up some touch-ups again. This now runs against UNIX files (i.e. '/' vs. '\')  rather than FAT paths. I should try to fix this to do both soon. I can borrow some code from another project I did something similar in.
21 | 
22 | 2010-07-23
23 | 	Fixed the shebang to run in both environments and added os.name checks to use a pathsep so that this will run in *nix and Windows environments.
24 | 


--------------------------------------------------------------------------------
/filecomp.py:
--------------------------------------------------------------------------------
  1 | #!/usr/bin/env python
  2 | import os,sys,hashlib,re
  3 | 
  4 | """
  5 | /*********************************\
  6 |            filecomp.py
  7 |      Will compare files in two
  8 |       directories for missing
  9 |       and/or mismatched files
 10 |        (based on md5sums).
 11 | 
 12 |       Currently a bit sloppy.
 13 | 
 14 |      Written by Twitch(Ben)
 15 | \*********************************/
 16 | """
 17 | 
 18 | """ Confirm input (very) briefly """
 19 | if len(sys.argv) != 3:
 20 |     print "You're doing something wrong.\nUsage: %s dir1 dir2" % sys.argv[0]
 21 |     exit(2)
 22 | 
 23 | (onedir, twodir) = sys.argv[1], sys.argv[2]
 24 | 
 25 | 
 26 | ''' Determine OS type and use appropriate path seperator. '''
 27 | if os.name == "posix":
 28 | 	pathsep = "/"
 29 | elif os.name == "nt":
 30 | 	pathsep = "\\"
 31 | else:
 32 | 	print "I haven't the faintest idea what operating system you have run me on, but I have no intentions of attempting to work under these conditions!"
 33 | 	exit(2)
 34 | 
 35 | 
 36 | def sumfile(fobj):
 37 | 	'''Returns an md5 hash for an object with read() method.'''
 38 | 	m = hashlib.md5()
 39 | 	while True:
 40 | 		d = fobj.read(8096)
 41 | 		if not d:
 42 | 			break
 43 | 		m.update(d)
 44 | 	return m.hexdigest()
 45 | 
 46 | def md5sum(fname):
 47 | 	'''Returns md5 of a file, or stdin if fname is "-".'''
 48 | 	if fname == '-':
 49 | 		ret = sumfile(sys.stdin)
 50 | 	else:
 51 | 		try:
 52 | 			f = file(fname, 'rb')
 53 | 		except:
 54 | 			return 'Failed to open file'
 55 | 		ret = sumfile(f)
 56 | 		f.close()
 57 | 	return ret
 58 | 
 59 | """ Abandon all hope, ye who enter here """
 60 | 
 61 | def walkdirs(dir1, dir2):
 62 | 	dirsa = []
 63 | 	dirsb = []
 64 | 
 65 | 	''' Walk directories and build list of similar structures. '''
 66 | 	''' This creates relative names for the sub-directories and adds them to the list. '''
 67 | 	for directory in os.walk(dir1):
 68 | 		relativedirname = pathsep + re.sub(dir1, "", directory[0])
 69 | 		if relativedirname != pathsep:
 70 | 			dirsa.append(relativedirname)
 71 |     
 72 | 	for directory in os.walk(dir2):
 73 | 		relativedirname = pathsep + re.sub(dir2, "", directory[0])
 74 | 		if relativedirname != pathsep:
 75 | 			dirsb.append(relativedirname)
 76 | 
 77 | 	# Return matches for further inspection.
 78 | 	return set(dirsa).intersection(set(dirsb))
 79 | 
 80 | def complists(dir1, dir2):
 81 | 	''' Compare list of files in the directories. '''
 82 | 	print "#######################################"
 83 | 	print "Comparing \n%s \n%s" % (dir1, dir2)
 84 | 	print "#######################################"
 85 | 	''' List files in target directories. '''
 86 | 	dira = []
 87 | 	dirb = []
 88 | 	
 89 | 	for file in os.listdir(dir1):
 90 | 		dira.append(file)
 91 | 	for file in os.listdir(dir2):
 92 | 		dirb.append(file)
 93 | 
 94 | 	''' Compare files and return disparities. '''
 95 | 	indir1 = set(dira).difference(set(dirb))
 96 | 	indir2 = set(dirb).difference(set(dira))
 97 | 	return (indir1, indir2, set(dira).intersection(set(dirb)))
 98 | 
 99 | def comfiles(files, onedir, twodir):
100 | 	''' Compare files which appear in both directories. '''
101 | 	firstdir = {}
102 | 	secdir = {}
103 | 	# Create an absolute path to them from the relative filename and get the md5.
104 | 	for f in files:
105 | 		of = onedir + pathsep + f
106 | 		sf = twodir + pathsep + f
107 | 		firstdir[f] = md5sum(of)
108 | 		secdir[f] = md5sum(sf)
109 | 	for x in firstdir:
110 | 		if firstdir[x] != secdir[x]:
111 | 			print "File %s in both targets but does not match!" % x
112 | 	print "\n"
113 | 
114 | def outp(datum, fdir=onedir, sdir=twodir):
115 | 	if len(datum[0]) > 0:
116 | 		print "Items in \"%s\" and not in \"%s\":" % (fdir, sdir)
117 | 		for z in datum[0]:
118 | 			print "   " + z
119 | 		print "\n------------------------------"
120 | 	else:
121 | 		print "No unique items in \"%s\"" % (fdir)
122 | 		
123 | 	if len(datum[1]) > 0:
124 | 		print "Items in \"%s\" and not in \"%s\":" %(sdir, fdir)
125 | 		for z in datum[1]:
126 | 			print "   " + z
127 | 		print "\n------------------------------"
128 | 	else:
129 | 		print "No unique items in \"%s\"" % (sdir)
130 | 	
131 | 	if len(datum[2]) > 0:
132 | 		comfiles(datum[2], fdir, sdir)
133 | 
134 | subdirs = walkdirs(onedir, twodir)
135 | setinfo = complists(onedir, twodir)
136 | outp(setinfo)
137 | 
138 | # Perform recursion through directories returned by the walkdirs() function.
139 | 
140 | for subdir in subdirs:
141 |     setinfo = complists(onedir + subdir, twodir + subdir)
142 |     outp(setinfo, onedir + subdir, twodir + subdir)
143 | 


--------------------------------------------------------------------------------