Filtering

By default, It is defined as

use_article_filter_reject_notice	=	yes
article_filter_reject_notice_recipient	=	maintainer sender

To change the recipient to the sender (From: address) only, set

article_filter_reject_notice_recipient	=	sender

To notify the rejection to both ML maintainer and the sender, set

article_filter_reject_notice_recipient	=	maintainer sender

To disable notification of rejection, set

use_article_filter_reject_notice	=	no

Case 1: use internal filter.

use_article_spam_filter         =       yes
article_spam_filter_drivers     =       spamassassin

Case 2: In this case, fml8 use not spamassassin internal filter but use a hook.

$distribute_verify_request_end_hook = q{
        my $spamassassin = '/usr/pkg/bin/spamc -c';

        use FileHandle;
        my $wh  = new FileHandle "| $spamassassin";

        if (defined $wh) {
                $wh->autoflush(1);
                my $msg = $curproc->incoming_message();
                $msg->print($wh);
                $wh->close();
                if ($?) {
                        $curproc->log("spam: (code = $?)");
                        $curproc->stop_this_process();  
                }
        }
};

$distribute_verify_request_end_hook = q{
	my $spamassassin = '/usr/pkg/bin/spamc -c';

	use FileHandle;
	my $wh  = new FileHandle "| $spamassassin";

	if (defined $wh) {
		$wh->autoflush(1);
		my $msg = $curproc->incoming_message();
		$msg->print($wh);
		$wh->close();
		if ($?) {
			$curproc->log("spam: (code = $?)");
			my $hdr = $curproc->incoming_message_header();
			$hdr->add('X-Spam-Status', 'Yes');
		}
	}
};

This is a little tricky but it works well.

fml8 analyzed the incoming message firstly and creates a chain of Mail::Message objects on memory. It is easy for fml8 to analyze the chain to check the message content.

The following examples uses hooks. In all cases, if matched, call stop_this_process() to stop further processing. Pay attension that these examples do not try to return error messsages.

If you need to return error messages, use reply_message(). It is better not to return it since this message must be a virus or a spam.

Here is an example to check attachment keywords e.g. .exe in mesages. Before fml8 2004/12/08 current, it follows:

$distribute_verify_request_start_hook = q{
    my $msg = $curproc->incoming_message() || undef;
    for (my $m = $msg; $m ; $m = $m->{ next } ) {
	my $hs = $m->message_fields() || '';
	if ($hs =~ /filename=.*\.(com|vbs|vbe|wsh|wse|js|exe|doc|rtf)/o) {
	    $curproc->log("attachment \.$1 found");
	    $curproc->stop_this_process();
	}
    }
};

$distribute_verify_request_start_hook = q{
    my $msg  = $curproc->incoming_message() || undef;
    my $list = $msg->message_chain_as_array_ref();
    for my $m (@$list) {
	my $hs = $m->message_fields() || '';
	if ($hs =~ /filename=.*\.(com|vbs|vbe|wsh|wse|js|exe|doc|rtf)/o) {
	    $curproc->log("[new] attachment \.$1 found");
	    $curproc->stop_this_process();
	}
    }
};

Another solution is to trap

Content-Disposition: attachment;

$distribute_verify_request_start_hook = q{
    my $msg = $curproc->incoming_message() || undef;
    for (my $m = $msg; $m ; $m = $m->{ next } ) {
	my $hs = $m->message_fields() || '';
	if ($hs =~ /Content-Disposition:.*attachment;/o) {
	    $curproc->log("attachment \.$1 found");
	    $curproc->stop_this_process();
	}
    }
};

$distribute_verify_request_start_hook = q{
    my $msg  = $curproc->incoming_message() || undef;
    my $list = $msg->message_chain_as_array_ref();
    for my $m (@$list) {
	my $hs = $m->message_fields() || '';
	if ($hs =~ /Content-Disposition:.*attachment;/o) {
	    $curproc->log("[new] attachment \.$1 found");
	    $curproc->stop_this_process();
	}
    }
};