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
},
'mark-ham' => {function => \&mark_ham,
},
+ 'learn' => {function => \&learn,
+ },
'help' => {function => sub {pod2usage({verbose => 2});}}
);
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/) {
$is_match = 1;
}
if ($is_match) {
},
$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 {
+ print STDERR "not learning" if $DEBUG;
+ }
+ print STDERR " from $mid" if $DEBUG;
+ },
+ $bug_num
+ );
}
}
my ($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 {
};
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 $childpid) {
die "Unable to fork spamc";
print {$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;