]> Sergey Matveev's repositories - public-inbox.git/blobdiff - lib/PublicInbox/WatchMaildir.pm
mda, watch: wire up List-ID header support
[public-inbox.git] / lib / PublicInbox / WatchMaildir.pm
index 10dc618416a377387652f1ebab198be4bb91f60b..08b1aab43ad1e8a4f6b06e6def310a4ecfb29eb0 100644 (file)
@@ -1,4 +1,4 @@
-# Copyright (C) 2016-2018 all contributors <meta@public-inbox.org>
+# Copyright (C) 2016-2019 all contributors <meta@public-inbox.org>
 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
 #
 # ref: https://cr.yp.to/proto/maildir.html
@@ -7,31 +7,40 @@ package PublicInbox::WatchMaildir;
 use strict;
 use warnings;
 use PublicInbox::MIME;
-use PublicInbox::Git;
-use PublicInbox::Import;
-use PublicInbox::MDA;
 use PublicInbox::Spawn qw(spawn);
 use PublicInbox::InboxWritable;
 use File::Temp qw//;
 use PublicInbox::Filter::Base;
+use PublicInbox::Spamcheck;
 *REJECT = *PublicInbox::Filter::Base::REJECT;
 
 sub new {
        my ($class, $config) = @_;
-       my (%mdmap, @mdir, $spamc, $spamdir);
+       my (%mdmap, @mdir, $spamc);
+       my %uniq;
 
        # "publicinboxwatch" is the documented namespace
        # "publicinboxlearn" is legacy but may be supported
        # indefinitely...
        foreach my $pfx (qw(publicinboxwatch publicinboxlearn)) {
                my $k = "$pfx.watchspam";
-               if (my $dir = $config->{$k}) {
-                       if ($dir =~ s/\Amaildir://) {
-                               $dir =~ s!/+\z!!;
+               defined(my $dirs = $config->{$k}) or next;
+               $dirs = [ $dirs ] if !ref($dirs);
+               for my $dir (@$dirs) {
+                       if (is_maildir($dir)) {
                                # skip "new", no MUA has seen it, yet.
                                my $cur = "$dir/cur";
-                               $spamdir = $cur;
+                               my $old = $mdmap{$cur};
+                               if (ref($old)) {
+                                       foreach my $ibx (@$old) {
+                                               warn <<"";
+"$cur already watched for `$ibx->{name}'
+
+                                       }
+                                       die;
+                               }
                                push @mdir, $cur;
+                               $uniq{$cur}++;
                                $mdmap{$cur} = 'watchspam';
                        } else {
                                warn "unsupported $k=$dir\n";
@@ -40,51 +49,47 @@ sub new {
        }
 
        my $k = 'publicinboxwatch.spamcheck';
-       my $spamcheck = $config->{$k};
-       if ($spamcheck) {
-               if ($spamcheck eq 'spamc') {
-                       $spamcheck = 'PublicInbox::Spamcheck::Spamc';
-               }
-               if ($spamcheck =~ /::/) {
-                       eval "require $spamcheck";
-                       $spamcheck = _spamcheck_cb($spamcheck->new);
-               } else {
-                       warn "unsupported $k=$spamcheck\n";
-                       $spamcheck = undef;
-               }
-       }
+       my $default = undef;
+       my $spamcheck = PublicInbox::Spamcheck::get($config, $k, $default);
+       $spamcheck = _spamcheck_cb($spamcheck) if $spamcheck;
 
-       # need to make all inboxes writable for spam removal:
-       $config->each_inbox(sub { PublicInbox::InboxWritable->new($_[0]) });
+       $config->each_inbox(sub {
+               # need to make all inboxes writable for spam removal:
+               my $ibx = $_[0] = PublicInbox::InboxWritable->new($_[0]);
 
-       foreach $k (keys %$config) {
-               $k =~ /\Apublicinbox\.([^\.]+)\.watch\z/ or next;
-               my $name = $1;
-               my $watch = $config->{$k};
-               if ($watch =~ s/\Amaildir://) {
-                       $watch =~ s!/+\z!!;
-                       my $inbox = $config->lookup_name($name);
-                       if (my $wm = $inbox->{watchheader}) {
-                               my ($k, $v) = split(/:/, $wm, 2);
-                               $inbox->{-watchheader} = [ $k, qr/\Q$v\E/ ];
+               my $watch = $ibx->{watch} or return;
+               if (is_maildir($watch)) {
+                       my $watch_hdrs = [];
+                       if (my $wh = $ibx->{watchheader}) {
+                               my ($k, $v) = split(/:/, $wh, 2);
+                               push @$watch_hdrs, [ $k, qr/\Q$v\E/ ];
+                       }
+                       if (my $list_ids = $ibx->{listid}) {
+                               for (@$list_ids) {
+                                       my $re = qr/<[ \t]*\Q$_\E[ \t]*>/;
+                                       push @$watch_hdrs, ['List-Id', $re ];
+                               }
+                       }
+                       if (scalar @$watch_hdrs) {
+                               $ibx->{-watchheaders} = $watch_hdrs;
                        }
                        my $new = "$watch/new";
                        my $cur = "$watch/cur";
-                       push @mdir, $new, $cur;
-                       die "$new already in use\n" if $mdmap{$new};
-                       die "$cur already in use\n" if $mdmap{$cur};
-                       $mdmap{$new} = $mdmap{$cur} = $inbox;
+                       push @mdir, $new unless $uniq{$new}++;
+                       push @mdir, $cur unless $uniq{$cur}++;
+
+                       push @{$mdmap{$new} ||= []}, $ibx;
+                       push @{$mdmap{$cur} ||= []}, $ibx;
                } else {
                        warn "watch unsupported: $k=$watch\n";
                }
-       }
+       });
        return unless @mdir;
 
        my $mdre = join('|', map { quotemeta($_) } @mdir);
        $mdre = qr!\A($mdre)/!;
        bless {
                spamcheck => $spamcheck,
-               spamdir => $spamdir,
                mdmap => \%mdmap,
                mdir => \@mdir,
                mdre => $mdre,
@@ -125,7 +130,7 @@ sub _remove_spam {
                eval {
                        my $im = _importer_for($self, $ibx);
                        $im->remove($mime, 'spam');
-                       if (my $scrub = $ibx->filter) {
+                       if (my $scrub = $ibx->filter($im)) {
                                my $scrubbed = $scrub->scrub($mime, 1);
                                $scrubbed or return;
                                $scrubbed == REJECT() and return;
@@ -146,28 +151,44 @@ sub _try_path {
                warn "unrecognized path: $path\n";
                return;
        }
-       my $inbox = $self->{mdmap}->{$1};
-       unless ($inbox) {
+       my $inboxes = $self->{mdmap}->{$1};
+       unless ($inboxes) {
                warn "unmappable dir: $1\n";
                return;
        }
-       if (!ref($inbox) && $inbox eq 'watchspam') {
+       if (!ref($inboxes) && $inboxes eq 'watchspam') {
                return _remove_spam($self, $path);
        }
-       my $im = _importer_for($self, $inbox);
-       my $mime = _path_to_mime($path) or return;
-       my $wm = $inbox->{-watchheader};
-       if ($wm) {
-               my $v = $mime->header_obj->header_raw($wm->[0]);
-               return unless ($v && $v =~ $wm->[1]);
-       }
-       if (my $scrub = $inbox->filter) {
-               my $ret = $scrub->scrub($mime) or return;
-               $ret == REJECT() and return;
-               $mime = $ret;
-       }
 
-       $im->add($mime, $self->{spamcheck});
+       my $warn_cb = $SIG{__WARN__} || sub { print STDERR @_ };
+       local $SIG{__WARN__} = sub {
+               $warn_cb->("path: $path\n");
+               $warn_cb->(@_);
+       };
+       foreach my $ibx (@$inboxes) {
+               my $mime = _path_to_mime($path) or next;
+               my $im = _importer_for($self, $ibx);
+
+               # any header match means it's eligible for the inbox:
+               if (my $watch_hdrs = $ibx->{-watchheaders}) {
+                       my $ok;
+                       my $hdr = $mime->header_obj;
+                       for my $wh (@$watch_hdrs) {
+                               my $v = $hdr->header_raw($wh->[0]);
+                               next unless defined($v) && $v =~ $wh->[1];
+                               $ok = 1;
+                               last;
+                       }
+                       next unless $ok;
+               }
+
+               if (my $scrub = $ibx->filter($im)) {
+                       my $ret = $scrub->scrub($mime) or next;
+                       $ret == REJECT() and next;
+                       $mime = $ret;
+               }
+               $im->add($mime, $self->{spamcheck});
+       }
 }
 
 sub quit { trigger_scan($_[0], 'quit') }
@@ -278,4 +299,11 @@ sub _spamcheck_cb {
        }
 }
 
+sub is_maildir {
+       $_[0] =~ s!\Amaildir:!! or return;
+       $_[0] =~ tr!/!/!s;
+       $_[0] =~ s!/\z!!;
+       $_[0];
+}
+
 1;