X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bin%2Fdebbugs-spam;h=1c44290543420946f53db6942b0c4de3a50961a6;hb=b10a87938eb116b8871081643058052564e39603;hp=4e5d3538ba6ae8b76cc991a1c98bace45e000187;hpb=bbc81477dea7d070a1240a11a355bb0f4cd6c59c;p=debbugs.git

diff --git a/bin/debbugs-spam b/bin/debbugs-spam
index 4e5d353..1c44290 100755
--- a/bin/debbugs-spam
+++ b/bin/debbugs-spam
@@ -62,6 +62,12 @@ ham.
 
 Output the score of all of the messages in a bug
 
+=over
+
+=item B<--skip-seen> Skip messages which have previously been classified
+
+=back
+
 =item B<mark-spam>
 
 Mark messages as spam if there is a regex match to subject or message
@@ -72,11 +78,25 @@ id
 Mark messages as ham if there is a regex match to subject or message
 id
 
+=item B<learn>
+
+Learn from messages which are ham/spam
+
 =back
 
 
 =head1 EXAMPLES
 
+Start spamd:
+
+  /usr/sbin/spamd --socketpath=/home/debbugs/spamd_socket \
+      --nouser-config --cf='include /home/debbugs/.spamassassin/user_prefs' \
+      --cf='allow_user_rules 1' --allow-tell;
+
+Then score bugs:
+
+  debbugs-spam --spamc-opts '-U' --spamc-opts '/home/debbugs/spamd_socket' \
+      score 859123;
 
 =cut
 
@@ -112,15 +132,20 @@ handle_main_arguments(\%options,
 
 my %subcommands =
     ('auto-scan' => {function => \&auto_spamscan,
-                     arguments => {'ham_threshold=s' => -5,
+                     arguments => {'ham_threshold|ham-threshold=s' => 0,
                                   },
+                     defaults => {ham_threshold => -5},
                     },
      'score' => {function => \&score_bug,
+                 arguments => {'skip_seen|skip-seen!' => 0
+                              },
                 },
      'mark-spam' => {function => \&mark_spam,
                     },
      'mark-ham' => {function => \&mark_ham,
                    },
+     'learn' => {function => \&learn,
+                },
      'help' => {function => sub {pod2usage({verbose => 2});}}
     );
 
@@ -151,7 +176,10 @@ if ($subcommand ne 'help') {
     chdir($config{spool_dir}) or die "chdir $config{spool_dir} failed: $!";
 }
 my $opts =
-    handle_subcommand_arguments(\@ARGV,$subcommands{$subcommand}{arguments});
+    handle_subcommand_arguments(\@ARGV,
+                                $subcommands{$subcommand}{arguments},
+                                $subcommands{$subcommand}{defaults},
+                               );
 $subcommands{$subcommand}{function}->(\%options,$opts,\%config,\@ARGV);
 
 
@@ -174,17 +202,55 @@ sub mark_it {
                        my $body = $rec->{text};
                        my ($subject) = $body =~ /^Subject: *(.+)$/mi;
                        my $is_match = 0;
-                       if ($subject =~ /\Q$regex\E/) {
+                       if ($subject =~ /$regex/) {
                            $is_match = 1;
                        }
-                       if ($mid =~ /\Q$regex\E/) {
+                       if ($mid =~ /$regex/) {
                            $is_match = 1;
                        }
-                       if ($spam_ham eq 'spam') {
-                           $spam->add_spam($mid);
+                       if ($is_match) {
+                           print STDERR "it's a match" if $DEBUG;
+                           if ($spam_ham eq 'spam') {
+                               $spam->add_spam($mid);
+                           } else {
+                               $spam->add_ham($mid);
+                           }
+                       }
+                   },
+                   $bug_num
+                  );
+        $spam->save();
+    }
+}
+
+sub learn {
+    my ($options,$opts,$config,$argv) = @_;
+    for my $bug_num (@{$argv}) {
+        my $spam = Debbugs::Log::Spam->new(bug_num => $bug_num) or
+            die "Unable to open bug log spam for $bug_num";
+        foreachmsg(sub {
+                       my ($bn,$rec,$mid) = @_;
+                       my $score;
+                       if ($spam->is_spam($mid)) {
+                           $score //=
+                               spam_score($rec,$options->{spamc},
+                                          [@{$options->{spamc_opts}},
+                                           '-L','spam'
+                                          ]
+                                         );
+                           print STDERR "learning spam" if $DEBUG;
+                       } elsif ($spam->is_ham($mid)) {
+                           $score //=
+                               spam_score($rec,$options->{spamc},
+                                          [@{$options->{spamc_opts}},
+                                           '-L','ham'
+                                          ]
+                                         );
+                           print STDERR "learning ham" if $DEBUG;
                        } else {
-                           $spam->add_ham($mid);
+                           print STDERR "not learning" if $DEBUG;
                        }
+                       print STDERR " from $mid" if $DEBUG;
                    },
                    $bug_num
                   );
@@ -198,7 +264,9 @@ sub score_bug {
         my @bug_score =
             spam_score_bug($bug_num,
                            $options->{spamc},
-                           $options->{spamc_opts});
+                           $options->{spamc_opts},
+                           $opts->{skip_seen},
+                          );
         print "$_->{score} $_->{message_id} $_->{subject}\n"
             foreach @bug_score;
     }
@@ -228,7 +296,7 @@ sub auto_spamscan {
                        if ($is_spam) {
                            print STDERR "it's spam ($score)\n" if $DEBUG;
                            $spam->add_spam($mid);
-                       } elsif ($score < $options->{ham_threshold}) {
+                       } elsif ($score < $opts->{ham_threshold}) {
                            print STDERR "it's really ham ($score)\n" if $DEBUG;
                            $spam->add_ham($mid);
                        }
@@ -243,12 +311,25 @@ sub auto_spamscan {
 }
 
 sub spam_score_bug {
-    my ($bug,$spamc,$spamc_opts) = @_;
+    my ($bug,$spamc,$spamc_opts,$skip_seen) = @_;
 
+    my $spam;
+    if ($skip_seen) {
+        $spam = Debbugs::Log::Spam->new(bug_num => $bug) or
+            die "Unable to open bug log spam for $bug";
+    }
     my @records;
     foreachmsg(sub {
                    my ($bn,$rec,$mid) = @_;
-                   my $score =
+                   my $score;
+                   if ($skip_seen) {
+                       if ($spam->is_spam($mid)) {
+                           $score = 999;
+                       } elsif ($spam->is_ham($mid)) {
+                           $score = -999;
+                       }
+                   }
+                   $score //=
                        spam_score($rec,$spamc,$spamc_opts);
                    my ($subject) = $rec->{text} =~ /^Subject: *(.+)$/mi;
                    push @records,
@@ -267,6 +348,9 @@ sub spam_score {
     my ($score,$threshold,$report);
     my $is_spam = 0;
     eval {
+        $report = '';
+        $score = 0;
+        $threshold = 5;
         my ($spamc_in,$spamc_out);
         my $old_sig = $SIG{"PIPE"};
         $SIG{"PIPE"} = sub {
@@ -274,7 +358,7 @@ sub spam_score {
         };
         my $childpid =
             open3($spamc_in,$spamc_out,0,
-                  $spamc,'-E',@{$spamc_opts}) or
+                  $spamc,'-E','--headers',@{$spamc_opts}) or
                       die "Unable to fork spamc: $!";
         if (not $childpid) {
             die "Unable to fork spamc";
@@ -282,19 +366,28 @@ sub spam_score {
         print {$spamc_in} $record->{text};
         close($spamc_in) or die "Unable to close spamc_in: $!";
         waitpid($childpid,0);
-        if ($? >> 8) {
+        my $exit_code = $? >> 8;
+        if ($exit_code) {
             $is_spam = 1;
         }
-        my ($first_line,@report) = <$spamc_out>;
-        if ($DEBUG) {
-            print STDERR "[$?;".($? >> 8)."] ";
-            print STDERR $first_line,@report;
-            print STDERR " ";
+        my $in_spam_header = 0;
+        while (<$spamc_out>) {
+            if (/^X-Spam/ or (/^\s+/ and $in_spam_header)) {
+                $in_spam_header = 1;
+                $report .= $_;
+                if (/^X-Spam-Status: (Yes|No), score=(-?[\d\.]+) required=(-?[\d\.]+)/) {
+                    $threshold = $3;
+                    $score = $2;
+                }
+            } else {
+               $in_spam_header = 0;
+            }
+            if (/^\s*$/) {
+                last;
+            }
         }
-        if (defined $first_line) {
-            chomp $first_line;
-            ($score,$threshold) = $first_line =~ m{^(-?[\d\.]+)/(-?[\d\.]+)$};
-            $report = join('',@report);
+        if ($DEBUG) {
+            print STDERR "[$exit_code] [$score/$threshold]\n$report\n";
         }
         close($spamc_out);
         $SIG{"PIPE"} = $old_sig;
@@ -317,17 +410,18 @@ sub foreachmsg {
         my ($msg_id) = record_regex($record,
                                     qr/^Message-Id:\s+<(.+)>/mi);
         next unless defined $msg_id;
+        print STDERR "examining $msg_id: " if $DEBUG;
         if ($msg_id =~ /$config{email_domain}$/) {
-            print STDERR "skipping $msg_id\n" if $DEBUG;
+            print STDERR "skipping\n" if $DEBUG;
             next;
         }
-        print STDERR "examining $msg_id: " if $DEBUG;
         if ($seen_msgids{$msg_id}) {
             print STDERR "already seen\n" if $DEBUG;
             next;
         }
         $seen_msgids{$msg_id}=1;
         $sub->($bug_num,$record,$msg_id);
+        print STDERR "\n" if $DEBUG;
     }
 }