#!/usr/bin/perl -w
# qmail duplicate eliminator.

use strict;
use Email::Fingerprint::App::EliminateDups;

# Make sure group and world permissions are off
umask oct(77);

# Run the app
my $app = Email::Fingerprint::App::EliminateDups->new;
$app->run(@ARGV);

__END__
=head1 NAME

eliminate-dups

=head1 SYNOPSIS

  eliminate-dups [options] [dbname]

If C<dbname> is not specified, it defaults to C<$HOME/.maildups>.

=head1 DESCRIPTION  

Reads an email message on standard input and calculates a fingerprint
based on the mail headers. If the fingerprint already exists in the
hash file, then the message is a duplicate.  If the fingerprint does
not exist, save the fingerprint in the hash file and deliver the
message.

=head1 OPTIONS

The following command-line options are supported:

=over

=item --dump

Dump the fingerprint DB human-readably and exit.  Implies C<--no-check>
and C<--no-purge>.

=item --help

Print a usage message and exit

=item --no-check

Do not compute any message fingerprints (i.e., only purge). The same
exit status will be returned as if a fingerprint check found no
match--i.e., any message on standard input will not be flagged as
a duplicate.

=item --no-purge

Do not purge the fingerprint DB (i.e., only check).  This option
can be used to reduce load on the mail server.  The script should
periodically be run with the C<--no-check> option to keep the fingerprint
database from growing without limit.

=item --strict

Include the message body when calculating fingerprints.

=back

Contradictory options are silently accepted. For example, C<--no-check>
is inconsistent with C<--strict>, but no error is reported if both
are used.  Similarly, C<--no-check> and C<--no-purge> together mean
that the script does nothing (unless C<--dump> or C<--help> is also
specified), but this doesn't result in an error.

=head1 SETUP

Create a ~/.qmail-maildir file

   ./Maildir/

Then add the following lines to your ~/.qmail file

  | eliminate-dups hashfile
  &user-maildir

Forwarding to the user-maildir address is I<very important>.  It
ensures that if delivery to the Maildir is deferred, eliminate-dups
will not be called a second time (which would result in a lost
message).

To reduce server load, the following can be used in the ~/.qmail file
instead:

  | eliminate-dups --no-purge hashfile
  &user-maildir

If the C<--no-purge> option is used, then a cron job must be set up to
purge the fingerprint database periodically. The following can be put in
the user's crontab file:

  23 2 * * *       /path/to/eliminate-dups --no-check $HOME/.maildups

=head1 LICENSE

Copyright (C) 2007-2008 Len Budney

This program is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.