enable body based loop check

3 files changed, 295 insertions, 2 deletions

diff --git a/fml/lib/FML/Body.pm b/fml/lib/FML/Body.pm
new file mode 100644
index 00000000..97b05e3a
--- /dev/null
+++ b/fml/lib/FML/Body.pm
@@ -0,0 +1,200 @@
+#-*- perl -*-
+#
+#  Copyright (C) 2006 Ken'ichi Fukamachi
+#   All rights reserved. This program is free software; you can
+#   redistribute it and/or modify it under the same terms as Perl itself.
+#
+# $FML: @template.pm,v 1.10 2006/01/07 13:16:41 fukachan Exp $
+#
+
+package FML::Body;
+use strict;
+use vars qw(@ISA @EXPORT @EXPORT_OK $AUTOLOAD);
+use Carp;
+
+=head1 NAME
+
+FML::Body - operations for mail body.
+
+=head1 SYNOPSIS
+
+=head1 DESCRIPTION
+
+=head1 METHODS
+
+=head2 new()
+
+constructor.
+
+=cut
+
+
+# Descriptions: constructor
+#    Arguments: OBJ($self) OBJ($curproc)
+# Side Effects: none
+# Return Value: OBJ
+sub new
+{
+    my ($self, $curproc) = @_;
+    my ($type) = ref($self) || $self;
+    my $me     = { _curproc => $curproc };
+    return bless $me, $type;
+}
+
+
+# Descriptions: checksum of mail body part.
+#               return 1 if the same checksum is found in the database.
+#    Arguments: OBJ($self) OBJ($config)
+# Side Effects: update database.
+# Return Value: NUM
+sub check_body_checksum
+{
+    my ($self, $config) = @_;
+    my $curproc   = $self->{ _curproc };
+    my $body_file = $curproc->incoming_message_print_body_as_file();
+
+    # calculate checksum of $body_file.
+    use Mail::Message::Checksum;
+    my $cksum = new Mail::Message::Checksum;
+    my $md5   = $cksum->md5_file($body_file);
+
+    # compare md5 value with the checksum database.
+    my $retval = 0;
+    my $db_dir = $config->{ incoming_mail_body_checksum_cache_dir };
+    my $db     = $self->db_open( { directory => $db_dir } );
+    if (defined $db) {
+	if ($db->{ $md5 }) {
+	    $retval = 1;
+	}
+	else {
+	    $db->{ $md5 } = time;
+	    $retval = 0;
+	}
+	$self->db_close();
+    }
+    return $retval;
+}
+
+
+=head1 DATABASE
+
+=head2 db_open($db_args)
+
+open database (journalized db).
+
+=head2 db_close($db_args)
+
+dummy.
+
+=cut
+
+
+# Descriptions: open database.
+#    Arguments: OBJ($self) HASH_REF($db_args)
+# Side Effects: open database.
+# Return Value: HASH_ERF
+sub db_open
+{
+    my ($self, $db_args) = @_;
+    my $dir  = $db_args->{ 'directory' } || '';
+    my $mode = 'temporal';
+    my $days = 14;
+
+    if ($dir) {
+	unless (-d $dir) {
+	    # XXX-TODO: dir_mode is hard-coded ?
+	    my $dir_mode = $self->{ _dir_mode } || 0700;
+
+	    use File::Path;
+	    mkpath( [ $dir ], 0, $dir_mode );
+	}
+
+	my %db = ();
+	use Tie::JournaledDir;
+	tie %db, 'Tie::JournaledDir', { dir => $dir };
+
+	$self->{ _db } = \%db;
+	return \%db;
+    }
+
+    return undef;
+}
+
+
+sub db_close
+{
+    ;
+}
+
+
+=head1 ACCESS METHODS
+
+=head2 set_checksum_type($type)
+
+set checksum method type.
+
+=head2 get_checksum_type()
+
+get checksum method type. 
+return 'md5' by default.
+
+=cut
+
+
+my $global_default_checksum_type = 'md5';
+
+
+# Descriptions: set checksum method type.
+#    Arguments: OBJ($self) STR($type)
+# Side Effects: update $self
+# Return Value: none
+sub set_checksum_type
+{
+    my ($self, $type) = @_;
+    my $curproc = $self->{ _curproc };
+
+    if ($type eq 'md5') {
+	$self->{ _type } = $type;
+    }
+    else {
+	$curproc->logerror("unsupported checksum: $type");
+    }
+}
+
+
+# Descriptions: return checksum method type.
+#    Arguments: OBJ($self)
+# Side Effects: none
+# Return Value: STR
+sub get_checksum_type
+{
+    my ($self) = @_;
+
+    return( $self->{ _type } || $global_default_checksum_type );
+}
+
+
+=head1 CODING STYLE
+
+See C<http://www.fml.org/software/FNF/> on fml coding style guide.
+
+=head1 AUTHOR
+
+Ken'ichi Fukamachi
+
+=head1 COPYRIGHT
+
+Copyright (C) 2006 Ken'ichi Fukamachi
+
+All rights reserved. This program is free software; you can
+redistribute it and/or modify it under the same terms as Perl itself.
+
+=head1 HISTORY
+
+FML::Body appeared in fml8 mailing list driver package.
+See C<http://www.fml.org/> for more details.
+
+=cut
+
+
+1;
diff --git a/fml/lib/FML/Process/Kernel.pm b/fml/lib/FML/Process/Kernel.pm
index 45d2f7d9..56553981 100644
--- a/fml/lib/FML/Process/Kernel.pm
+++ b/fml/lib/FML/Process/Kernel.pm
@@ -4,7 +4,7 @@
 #   All rights reserved. This program is free software; you can
 #   redistribute it and/or modify it under the same terms as Perl itself.
 #
-# $FML: Kernel.pm,v 1.282 2006/05/04 05:21:29 fukachan Exp $
+# $FML: Kernel.pm,v 1.283 2006/05/12 13:50:11 fukachan Exp $
 #
 
 package FML::Process::Kernel;
@@ -668,6 +668,9 @@ sub simple_loop_check
     unless ($match) {
 	$match = $curproc->_header_based_loop_check();
     }
+    unless ($match) {
+	$match = $curproc->_body_based_loop_check();
+    }
 
     # $match contains the first matched rule name (== reason).
     if ($match) {
@@ -742,6 +745,38 @@ sub _header_based_loop_check
 }
 
 
+# Descriptions: body based loop check.
+#    Arguments: OBJ($curproc)
+# Side Effects: none
+# Return Value: STR
+sub _body_based_loop_check
+{
+    my ($curproc) = @_;
+    my $config    = $curproc->config();
+    my $match     = undef;
+    my $var_rules = 'incoming_mail_body_loop_check_rules';
+
+    if ($config->yes('use_incoming_mail_body_loop_check')) {
+	use FML::Body;
+	my $body = new FML::Body $curproc;
+
+	my $rules = $config->get_as_array_ref($var_rules);
+      RULE:
+	for my $rule (@$rules) {
+	    if ($body->can($rule)) {
+		$match = $body->$rule($config) ? $rule : undef;
+	    }
+	    else {
+		$curproc->logwarn("body->${rule}() is undefined");
+ 	    }
+	    last RULE if $match;
+	}
+    }
+
+    return $match;
+}
+
+
 # Descriptions: commit message-id cache update transaction.
 #    Arguments: OBJ($curproc)
 # Side Effects: update cache.
diff --git a/fml/lib/FML/Process/Utils.pm b/fml/lib/FML/Process/Utils.pm
index 898a8382..f74c6f90 100644
--- a/fml/lib/FML/Process/Utils.pm
+++ b/fml/lib/FML/Process/Utils.pm
@@ -4,7 +4,7 @@
 #   All rights reserved. This program is free software; you can
 #   redistribute it and/or modify it under the same terms as Perl itself.
 #
-# $FML: Utils.pm,v 1.148 2006/02/15 13:44:04 fukachan Exp $
+# $FML: Utils.pm,v 1.149 2006/04/22 08:29:23 fukachan Exp $
 #
 
 package FML::Process::Utils;
@@ -329,6 +329,64 @@ sub incoming_message_remove_queue
 }
 
 
+=head2 incoming_message_print_body_as_file()
+
+dupliate the body part into a temporary file.
+return a newly created temporary file path.
+
+=cut
+
+
+# Descriptions: dupliate the body part into a temporary file.
+#    Arguments: OBJ($curproc) OBJ($queue)
+# Side Effects: create a temporary file.
+# Return Value: STR
+sub incoming_message_print_body_as_file
+{
+    my ($curproc, $queue) = @_;
+    my $cache_file = $curproc->incoming_message_get_cache_file_path();
+    my $tmp_file   = $curproc->tmp_file_path();
+
+    if (-f $cache_file) {
+	my $rh = new FileHandle $cache_file;
+	my $wh = new FileHandle "> $tmp_file";
+	if (defined $rh && defined $wh) {
+	    my $found = 0;
+	    my $buf;
+	  LINE:
+	    while ($buf = <$rh>) {
+		unless ($found) {
+		    if ($buf =~ /^$/o) {
+			$found = 1;
+		    }
+		    next LINE;
+		}
+		print $wh $buf;
+	    }
+	    $wh->close();
+	    $rh->close();
+
+	    return $tmp_file;
+	}
+	else {
+	    my $myname = "incoming_message_dup_body";
+	    unless (defined $rh) {
+		$curproc->logerror("$myname: cannot open $cache_file");
+	    }
+	    unless (defined $wh) {
+		$curproc->logerror("$myname: cannot open $tmp_file");
+	    }
+
+	    return undef;
+	}
+    }
+    else {
+	$curproc->logerror("incoming_message_dup_body: no cache");
+	return undef;
+    }
+}
+
+
 =head1 access METHODS to handle article
 
 available only in C<libexec/distribute> process.