should be Return-path, not Return-Path

[debbugs.git] / bin / debbugs-spam

diff --git a/bin/debbugs-spam b/bin/debbugs-spam
index 164013bce1da73970de2a9ce52b29255b87c9a9b..5dca08edf01de69653388521c21f3c4aa6e6d785 100755 (executable)
--- a/bin/debbugs-spam
+++ b/bin/debbugs-spam
@@ -78,6 +78,10 @@ id
 Mark messages as ham if there is a regex match to subject or message
 id
 
+=item B<learn>
+
+Learn from messages which have been marked as spam/ham
+
 =back
 
 
@@ -103,18 +107,18 @@ my %options =
      man     => 0,
      verbose => 0,
      quiet   => 0,
-     quick   => 0,
      spamc   => 'spamc',
      spamc_opts => [],
     );
 
 handle_main_arguments(\%options,
-                      'quick|q',
                       'service|s',
                       'sysconfdir|c',
                       'spamc=s' => 0,
                       'spamc_opts|spamc-opts=s@' => 0,
                       'spool_dir|spool-dir=s',
+                      'quiet|q:+',
+                      'verbose|v:+',
                       'debug|d+','help|h|?','man|m');
 
 my %subcommands =
@@ -134,6 +138,12 @@ my %subcommands =
                     },
      'mark-ham' => {function => \&mark_ham,
                    },
+     'learn' => {function => \&learn,
+                 arguments => {'skip_missing|skip-missing!' => 0,
+                              },
+                 defaults => {skip_missing => 0,
+                             },
+                },
      'help' => {function => sub {pod2usage({verbose => 2});}}
     );
 
@@ -226,6 +236,34 @@ sub score_bug {
     }
 }
 
+sub learn {
+    my ($options,$opts,$config,$argv) = @_;
+
+    for my $bug_num (@{$argv}) {
+        if ($opts->{skip_missing} and
+            not defined getbuglocation($bug_num,'log')) {
+            print STDERR "bug $bug_num does not exist\n" if $options->{verbose} > -1;
+            next;
+        }
+        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) = @_;
+                       if ($spam->is_spam($mid)) {
+                           spamc_learn_spam($rec,$options->{spamc},$options->{spamc_opts});
+                           return;
+                       }
+                       if ($spam->is_ham($mid)) {
+                           spamc_learn_ham($rec,$options->{spamc},$options->{spamc_opts});
+                           return;
+                       }
+                   },
+                   $bug_num,
+                  );
+        $spam->save();
+    }
+}
+
 sub auto_spamscan {
     my ($options,$opts,$config,$argv) = @_;
 
@@ -302,10 +340,22 @@ sub spam_score_bug {
     return @records;
 }
 
-sub spam_score {
+sub add_return_path {
+    my ($message) = @_;
+    ## debbugs is kind of odd, and puts "Received:" first, them "From
+    ## ", and doesn't have a Return-Path. Fix that up so spamassassin
+    ## is happy.
+    $message =~
+        s{^(Received: \(at \S+\) by \S+;[^\n]+\n)(From (\S+) [^\n]+\n)}
+        {${2}Return-path: $3\n$1};
+    return $message;
+}
+
+sub spamc_bug {
     my ($record,$spamc,$spamc_opts) = @_;
-    my ($score,$threshold,$report);
-    my $is_spam = 0;
+    my $first_line = '';
+    my $report = '';
+    my $exit_code = 0;
     eval {
         my ($spamc_in,$spamc_out);
         my $old_sig = $SIG{"PIPE"};
@@ -314,28 +364,20 @@ sub spam_score {
         };
         my $childpid =
             open3($spamc_in,$spamc_out,0,
-                  $spamc,'-E',@{$spamc_opts}) or
+                  $spamc,@{$spamc_opts}) or
                       die "Unable to fork spamc: $!";
         if (not $childpid) {
             die "Unable to fork spamc";
         }
-        print {$spamc_in} $record->{text};
+        print STDERR add_return_path($record->{text}) if $DEBUG > 1;
+        print {$spamc_in} add_return_path($record->{text});
         close($spamc_in) or die "Unable to close spamc_in: $!";
         waitpid($childpid,0);
         if ($? >> 8) {
-            $is_spam = 1;
-        }
-        my ($first_line,@report) = <$spamc_out>;
-        if ($DEBUG) {
-            print STDERR "[$?;".($? >> 8)."] ";
-            print STDERR $first_line,@report;
-            print STDERR " ";
-        }
-        if (defined $first_line) {
-            chomp $first_line;
-            ($score,$threshold) = $first_line =~ m{^(-?[\d\.]+)/(-?[\d\.]+)$};
-            $report = join('',@report);
+            $exit_code = $? >> 8;
         }
+        local $/;
+        $report = <$spamc_out>;
         close($spamc_out);
         $SIG{"PIPE"} = $old_sig;
     };
@@ -343,7 +385,35 @@ sub spam_score {
         carp "processing of message failed [$@]\n";
         return undef;
     }
-    return wantarray?($score,$is_spam,$report):$score;
+    return ($exit_code,$report);
+}
+
+sub spam_score {
+    my ($record,$spamc,$spamc_opts) = @_;
+    my ($score,$threshold,$report,$exit_code);
+    ($exit_code,$report) =
+        spamc_bug($record,$spamc,[@{$spamc_opts},'-c']);
+    if (defined $report) {
+        ($score,$threshold) = $report =~ s{^(-?[\d\.]+)/(-?[\d\.]+)\n?}{};
+    }
+    return wantarray?($score,$exit_code,$report):$score;
+}
+
+sub spamc_learn_ham {
+    spamc_learn('ham',@_);
+}
+
+sub spamc_learn_forget {
+    spamc_learn('forget',@_);
+}
+
+sub spamc_learn_spam {
+    spamc_learn('spam',@_);
+}
+
+sub spamc_learn {
+    my ($type,$record,$spamc,$spamc_opts) = @_;
+    spamc_bug($record,$spamc,[@{$spamc_opts},'-L',$type])
 }
 
 sub foreachmsg {
@@ -352,8 +422,7 @@ sub foreachmsg {
         die "Unable to open bug log for $bug_num";
     my %seen_msgids;
     while (my $record = $log->read_record()) {
-        next if $record->{type} eq 'html';
-        next if $record->{type} eq 'autocheck';
+        next unless $record->{type} eq 'incoming-recv';
         my ($msg_id) = record_regex($record,
                                     qr/^Message-Id:\s+<(.+)>/mi);
         next unless defined $msg_id;