New version of spamscan that uses internal locking for

[debbugs.git] / scripts / spamscan-sa

diff --git a/scripts/spamscan-sa b/scripts/spamscan-sa
new file mode 100755 (executable)
index 0000000..3ee7682
--- /dev/null
+++ b/scripts/spamscan-sa
@@ -0,0 +1,155 @@
+#! /usr/bin/perl
+# spamassassin handling split from spamscan
+#
+
+# unfortunatly we can't use strict;
+
+use lib qw(/usr/lib/debbugs);
+use Mail::CrossAssassin;
+
+use Mail::SpamAssassin;
+use Mail::SpamAssassin::NoMailAudit;
+
+my $config_path = '/etc/debbugs';
+require "$config_path/config";
+# New versions of debbugs will not allow use in /etc/debbugs/config
+use POSIX qw(strftime);
+$gSpamMailbox = strftime($gSpamMailbox,gmtime);
+$gCrossMailbox = strftime($gCrossMailbox,gmtime);
+
+umask 002;
+$| = 1;
+STDOUT->autoflush(1);
+
+sub header_or_empty ($$) {
+    my ($mail, $hdr) = @_;
+    my $value = $mail->get_header($hdr);
+    if (defined $value) {
+       chomp $value;
+       $value =~ tr/\n/\\n/;
+       return $value;
+    }
+    return '';
+}
+
+my $user_prefs = "$ENV{HOME}/.spamassassin/user_prefs";
+
+my $spam = Mail::SpamAssassin->new({
+        dont_copy_prefs => 1,
+        site_rules_filename => $gSpamRulesDir,
+        userprefs_filename => $user_prefs,
+               local_tests_only => ($gSpamLocalTestsOnly || 0),
+               debug => ($ENV{DEBBUGS_SPAM_DEBUG} || 0),
+});
+$spam->compile_now(1); # use all user preferences
+
+       while (my $id = <STDIN>) {
+           chomp $id;
+           my $nf = <STDIN> or die "Could not read nf: $!";
+           chomp $nf;
+           unless (rename "incoming/S$id", "incoming/R$id") {
+               die "Could not rename incoming/S$id: $!";
+           }
+           my $out = "[$nf] $id scanning ...\n";
+           open MESSAGE, "< incoming/R$id" or die "open incoming/R$id: $!";
+           my @textarray;
+           # Kludge to work around Received: then From_ weirdness in receive;
+           # remove when receive is fixed? We may continue to need it for
+           # reprocessing old messages.
+           $textarray[0] = <MESSAGE>;
+           if ($textarray[0] =~ /^Received:/) {
+               my $maybefrom = <MESSAGE>;
+               if ($maybefrom =~ /^From /) {
+                   $textarray[1] = $textarray[0];
+                   $textarray[0] = $maybefrom;
+               } else {
+                   $textarray[1] = $maybefrom;
+               }
+           }
+           push @textarray, <MESSAGE>;
+           close MESSAGE;
+           my $mail = Mail::SpamAssassin::NoMailAudit->new(data => \@textarray);
+           $mail->{noexit} = 1;
+
+           my $messageid = header_or_empty($mail, 'Message-Id');
+           $out .= "  From: " . header_or_empty($mail, 'From') . "\n";
+           $out .= "  Subject: ". header_or_empty($mail, 'Subject') . "\n";
+           $out .= "  Date: " . header_or_empty($mail, 'Date') . "\n";
+           $out .= "  Message-Id: $messageid\n";
+           my $keys = ca_keys($mail->get_body);
+           print  "$keys\n$messageid\n"
+               or die "Could not send keys: $!";
+           my $ca_score = <STDIN> or die "Could not read ca_score: $!";
+           chomp $ca_score;
+           my $todo = 0;
+           my $seen = <STDIN> or die "Child could not read seen: $!";
+           chomp $seen;
+           my $status;
+           my $nseen = $seen;
+           if ($seen) {
+               $todo = 1;
+               $out .= "  spam $seen duplicate\n";
+           } else {
+               $status = $spam->check($mail);
+               $status->rewrite_mail();
+
+               if ($status->is_spam()) {
+#                  $mail->accept($gSpamMailbox);
+#                  unlink "incoming/R$id" or warn "unlink incoming/R$id: $!";
+                   $todo = 1;
+                   my $score = sprintf "%.1f/%.1f %d",
+                           $status->get_hits(), $status->get_required_hits(),
+                           $ca_score;
+                   $out .= "  spam $score\n";
+                   $nseen = $score;
+               } elsif ($status->get_hits() > 0 && $ca_score >= $gMaxCross) {
+#                  $mail->accept($gCrossMailbox);
+#                  unlink "incoming/R$id" or warn "unlink incoming/R$id: $!";
+                   $todo = 2;
+                   my $score = sprintf "%.1f/%.1f %d",
+                   $status->get_hits(), $status->get_required_hits(), $ca_score;
+                   $out .= "  spam $score\n";
+                   $nseen = $score;
+               } else {
+                   open OUT, "> incoming/I$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: $!";
+                   $out .= sprintf "  ok %.1f/%.1f %d\n",
+                           $status->get_hits(), $status->get_required_hits(),
+                           $ca_score;
+               }
+           }
+           print "$todo\n";
+           my $x = <STDIN>;
+           if ($todo) {
+               open OUT, '>>', ($todo == 1) ? $gSpamMailbox : $gCrossMailbox
+                   or die "Could not open assassinated: $!";
+               my @headers = $mail->get_all_headers();
+               print OUT @headers
+                   or die "print assassinated: $!";
+               print OUT "\n"
+                   or die "print assassinated: $!";
+               foreach (@{$mail->get_body()}) {
+                   s/^From />From /;
+                   print OUT $_
+                       or die "print assassinated: $!";
+               }
+               close OUT or die "Close assassinated: $!";
+           }
+           $out =~ tr/\n/\r/;
+           print "$nseen\n$out\n";
+           $status->finish() unless($seen);
+       }