[project @ 2005-02-01 07:54:01 by blarson]

Update spamscan to current version.  Uses crossassin (my perl version),
avoids spamassassin on duplicate message IDs.

diff --git a/scripts/spamscan.in b/scripts/spamscan.in
index 29d46bbfca86986751c0f5b87ec4d9e11fc2620a..7fe3e8f51be406e4d39f160a340f681e331dd11c 100755 (executable)
--- a/scripts/spamscan.in
+++ b/scripts/spamscan.in
@@ -1,5 +1,5 @@
 #! /usr/bin/perl -T
-# $Id: spamscan.in,v 1.7 2004/01/14 19:41:18 cjwatson Exp $
+# $Id: spamscan.in,v 1.8 2005/02/01 07:54:01 blarson Exp $
 #
 # Usage: spamscan
 #
@@ -26,6 +26,9 @@ push @INC, $lib_path;
 use Mail::SpamAssassin;
 use Mail::SpamAssassin::NoMailAudit;
 
+use lib '/usr/lib/debbugs';
+use Mail::CrossAssassin;
+
 umask 002;
 
 eval {
@@ -33,6 +36,10 @@ eval {
 };
 exit if $@;
 
+ca_init('\b\d{3,8}(?:-(?:close|done|forwarded|maintonly|submitter|quiet))?\@bugs\.debian\.org', '/org/bugs.debian.org/CrossAssassinDb');
+
+my %spamseen = ();
+
 my $user_prefs = "$ENV{HOME}/.spamassassin/user_prefs";
 my $user_prefs_time;
 if (-e $user_prefs) {
@@ -45,6 +52,7 @@ my $spam = Mail::SpamAssassin->new({
     userprefs_filename => $user_prefs,
     local_tests_only => ($gSpamLocalTestsOnly || 0),
     debug => ($ENV{DEBBUGS_SPAM_DEBUG} || 0),
+    check_mx_delay => 2, # bit of a hack until we have parallelization
 });
 $spam->compile_now(1); # use all user preferences
 
@@ -116,38 +124,59 @@ for (;;) {
     my $mail = Mail::SpamAssassin::NoMailAudit->new(data => \@textarray);
     $mail->{noexit} = 1;
 
+    my $messageid = header_or_empty($mail, 'Message-Id');
     print "  From: ", header_or_empty($mail, 'From'), "\n";
     print "  Subject: ", header_or_empty($mail, 'Subject'), "\n";
     print "  Date: ", header_or_empty($mail, 'Date'), "\n";
-    print "  Message-Id: ", header_or_empty($mail, 'Message-Id'), "\n";
-    my $status = $spam->check($mail);
-    $status->rewrite_mail();
-
-    if ($status->is_spam()) {
+    print "  Message-Id: $messageid\n";
+    my $ca_score = ca_set(ca_keys($mail->get_body));
+    if (exists $spamseen{$messageid}) {
        $mail->accept($gSpamMailbox);
        unlink "incoming/R$id" or warn "unlink incoming/R$id: $!";
-       printf "  spam %.1f/%.1f\n",
-              $status->get_hits(), $status->get_required_hits()
+       print "  spam $spamseen{$messageid} duplicate\n"
            or die "printf log: $!";
     } else {
-       open OUT, "> incoming/I$id" or die "open incoming/I$id: $!";
-       my @headers = $mail->get_all_headers();
-       if ($headers[0] =~ /^From /) {
-           my $from = $headers[0];
-           $headers[0] = $headers[1];
-           $headers[1] = $from;
+       my $status = $spam->check($mail);
+       $status->rewrite_mail();
+
+       if ($status->is_spam()) {
+           $mail->accept($gSpamMailbox);
+           unlink "incoming/R$id" or warn "unlink incoming/R$id: $!";
+           my $score = sprintf "%.1f/%.1f %d",
+               $status->get_hits(), $status->get_required_hits(), $ca_score;
+           print "  spam $score\n" or die "print log: $!";
+           $spamseen{$messageid} = $score;
+       } elsif ($status->get_hits() > 0 && $ca_score >= 4) {
+           $mail->accept($gCrossMailbox);
+           unlink "incoming/R$id" or warn "unlink incoming/R$id: $!";
+           my $score = sprintf "%.1f/%.1f %d",
+               $status->get_hits(), $status->get_required_hits(), $ca_score;
+           printf "  spam $score\n" or die "printf log: $!";
+           $spamseen{$messageid} = $score;
+       } else {
+           open OUT, "> incoming/I$id" or die "open incoming/I$id: $!";
+           my @headers = $mail->get_all_headers();
+           if ($headers[0] =~ /^From /) {
+               my $from = $headers[0];
+               $headers[0] = $headers[1];
+               $headers[1] = $from;
+           }
+           print OUT join '', @headers or die "print incoming/I$id: $!";
+           if ($ca_score > 1) {
+               print OUT "X-CrossAssassin-Score: $ca_score\n"
+                   or die "print incoming/I$id: $!";
+           }
+           print OUT "\n" or die "print incoming/I$id: $!";
+           print OUT @{$mail->get_body()} or die "print incoming/I$id: $!";
+           close OUT or die "close incoming/I$id: $!";
+           unlink "incoming/R$id" or warn "unlink incoming/R$id: $!";
+           printf "  ok %.1f/%.1f %d\n",
+               $status->get_hits(), $status->get_required_hits(), $ca_score
+               or die "printf log: $!";
        }
-       print OUT join '', @headers or die "print incoming/I$id: $!";
-       print OUT "\n" or die "print incoming/I$id: $!";
-       print OUT @{$mail->get_body()} or die "print incoming/I$id: $!";
-       close OUT or die "close incoming/I$id: $!";
-       unlink "incoming/R$id" or warn "unlink incoming/R$id: $!";
-       printf "  ok %.1f/%.1f\n",
-              $status->get_hits(), $status->get_required_hits()
-           or die "printf log: $!";
-    }
 
-    $status->finish();
+       $status->finish();
+    }
 }
 &unfilelock;