#!/usr/local/bin/python

"""Usage: finddups [--dupfile <filename>] [--dupcount <count>] <headerlist>

The mail message on standard input is parsed, and an md5 digest of the
headers named in the argument list is computed. This is then checked
against a list of digests from previously seen messages. If a match is
found, finddups exits with a result of 1. If no match is found, the
digest is added to the saved list and finddups exits with a result of
0.

The flags are --dupfile (defaults to $HOME/.finddups-digests),
--dupcount (defaults to 500) and --noisy. dupfile species where the
duplicates are kept. dupcount specifies how many duplicates to keep.
noisy - if present - will cause finddups to announce whether or not
this is a duplicate message.

All other arguments indicate strings to be added to the digest. The
special header names "body" and "headers" indicate the text of the
body, and the text of all the headers (received lines, included.),
respectively. All other argument names are the names mail headers to
be added to the digest."""

import os, sys, getopt, os, md5, mailwrapper, cPickle, string, fcntl, FCNTL


#
# Argument processing happens here
#
count = None
filename = None
verbose = None

options, headers = getopt.getopt(sys.argv[1:], '',
				 ['dupfile=', 'dupcount=', 'noisy'])
for name, value in options:
    if name == '--noisy':
	verbose = 1
    elif name == '--dupfile':
	filename = value
    elif name == '--dupcount':
	try:
	    count = string.atoi(value)
	except ValueError:
	    sys.stderr.write('%s: Invalid value for dupcount: %s\n' %
			     (sys.argv[0], value))
    else:
	sys.stderr.write('%s: Unrecognized flag: %s.\n' % (sys.argv[0], value))
	
count = count or 500		# The Default value is set here.
try:
    filename = filename or os.path.join(os.environ['HOME'], '.finddups-digests')
except KeyError:
    sys.stderr.write("%s: You must either set $HOME or provide a --dupfile option!\n" %  sys.argv[1])
    sys.exit(1)

#
# Now take this messages fingerprint.
#

text = mailwrapper.selector(headers).get_headers()
digest = md5.new()
map(digest.update, text)
fingerprint = digest.digest()

#
# Finally, check (and maybe update) our records.
#
try:
    file =  open(filename, 'r+')
    file.seek(0)
except IOError: file = open(filename, 'w+')
fcntl.flock(file.fileno(), FCNTL.LOCK_EX)
try: digests = cPickle.load(file)
except EOFError: digests = []
		      
# Time for the moment of truth - is it a duplicate or not?
if fingerprint in digests:
    print "Duplcate mail"
    fcntl.flock(file.fileno(), FCNTL.LOCK_UN)
    file.close()
    sys.exit(0)

# Guess not...
digests.append(fingerprint)
overage = len(digests) - count
if overage > 0:
    digests = digests[overage:]
file.seek(0)
cPickle.dump(digests, file, 0)	# 0 for debugging, 1 for production
file.flush()
fcntl.flock(file.fileno(), FCNTL.LOCK_UN)
file.close()
if verbose: print "Original mail."
sys.exit(1)