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
use Debbugs::Log::Spam;
use Debbugs::Config qw(:config);
use Debbugs::Command qw(:all);
+use Debbugs::Common qw(getbuglocation);
use IPC::Open3 qw(open3);
use Carp;
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 =
('auto-scan' => {function => \&auto_spamscan,
- arguments => {'ham_threshold=s' => -5,
+ arguments => {'ham_threshold|ham-threshold=s' => 0,
+ 'skip_missing|skip-missing!' => 0,
},
+ defaults => {ham_threshold => -5,
+ skip_missing => 0,
+ },
},
'score' => {function => \&score_bug,
arguments => {'skip_seen|skip-seen!' => 0
},
'mark-ham' => {function => \&mark_ham,
},
+ 'learn' => {function => \&learn,
+ arguments => {'skip_missing|skip-missing!' => 0,
+ },
+ defaults => {skip_missing => 0,
+ },
+ },
'help' => {function => sub {pod2usage({verbose => 2});}}
);
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,$opts,\%config,\@ARGV);
if ($mid =~ /\Q$regex\E/) {
$is_match = 1;
}
- if ($spam_ham eq 'spam') {
- $spam->add_spam($mid);
- } else {
- $spam->add_ham($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}) {
+ 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) = @_;
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 $bug_num";
foreachmsg(sub {
if ($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);
}
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"};
};
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"} = $old_sig;
};
carp "processing of message failed [$@]\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 {
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;
+ 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;
}
}
projects / debbugs.git / blobdiff