1 # Copyright (C) 2016-2018 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # ref: https://cr.yp.to/proto/maildir.html
5 # http://wiki2.dovecot.org/MailboxFormat/Maildir
6 package PublicInbox::WatchMaildir;
11 use PublicInbox::Import;
13 use PublicInbox::Spawn qw(spawn);
14 use PublicInbox::InboxWritable;
16 use PublicInbox::Filter::Base;
17 *REJECT = *PublicInbox::Filter::Base::REJECT;
20 my ($class, $config) = @_;
21 my (%mdmap, @mdir, $spamc, $spamdir);
23 # "publicinboxwatch" is the documented namespace
24 # "publicinboxlearn" is legacy but may be supported
26 foreach my $pfx (qw(publicinboxwatch publicinboxlearn)) {
27 my $k = "$pfx.watchspam";
28 if (my $dir = $config->{$k}) {
29 if ($dir =~ s/\Amaildir://) {
31 # skip "new", no MUA has seen it, yet.
35 $mdmap{$cur} = 'watchspam';
37 warn "unsupported $k=$dir\n";
42 my $k = 'publicinboxwatch.spamcheck';
43 my $spamcheck = $config->{$k};
45 if ($spamcheck eq 'spamc') {
46 $spamcheck = 'PublicInbox::Spamcheck::Spamc';
48 if ($spamcheck =~ /::/) {
49 eval "require $spamcheck";
50 $spamcheck = _spamcheck_cb($spamcheck->new);
52 warn "unsupported $k=$spamcheck\n";
57 # need to make all inboxes writable for spam removal:
58 $config->each_inbox(sub { PublicInbox::InboxWritable->new($_[0]) });
60 foreach $k (keys %$config) {
61 $k =~ /\Apublicinbox\.([^\.]+)\.watch\z/ or next;
63 my $watch = $config->{$k};
64 if ($watch =~ s/\Amaildir://) {
66 my $inbox = $config->lookup_name($name);
67 if (my $wm = $inbox->{watchheader}) {
68 my ($k, $v) = split(/:/, $wm, 2);
69 $inbox->{-watchheader} = [ $k, qr/\Q$v\E/ ];
71 my $new = "$watch/new";
72 my $cur = "$watch/cur";
73 push @mdir, $new, $cur;
74 die "$new already in use\n" if $mdmap{$new};
75 die "$cur already in use\n" if $mdmap{$cur};
76 $mdmap{$new} = $mdmap{$cur} = $inbox;
78 warn "watch unsupported: $k=$watch\n";
83 my $mdre = join('|', map { quotemeta($_) } @mdir);
84 $mdre = qr!\A($mdre)/!;
86 spamcheck => $spamcheck,
93 opendirs => {}, # dirname => dirhandle (in progress scans)
99 my $importers = $self->{importers};
100 foreach my $im (values %$importers) {
106 my ($self, $scan_re, $paths) = @_;
108 my $path = $_->{path};
109 if ($path =~ $scan_re) {
112 _try_path($self, $path);
115 _done_for_now($self);
119 my ($self, $path) = @_;
120 # path must be marked as (S)een
121 $path =~ /:2,[A-R]*S[T-Za-z]*\z/ or return;
122 my $mime = _path_to_mime($path) or return;
123 $self->{config}->each_inbox(sub {
126 my $im = _importer_for($self, $ibx);
127 $im->remove($mime, 'spam');
128 if (my $scrub = $ibx->filter) {
129 my $scrubbed = $scrub->scrub($mime) or return;
130 $scrubbed == REJECT() and return;
131 $im->remove($scrubbed, 'spam');
135 warn "error removing spam at: ", $path,
136 " from ", $ibx->{name}, ': ', $@, "\n";
142 my ($self, $path) = @_;
143 return unless PublicInbox::InboxWritable::is_maildir_path($path);
144 if ($path !~ $self->{mdre}) {
145 warn "unrecognized path: $path\n";
148 my $inbox = $self->{mdmap}->{$1};
150 warn "unmappable dir: $1\n";
153 if (!ref($inbox) && $inbox eq 'watchspam') {
154 return _remove_spam($self, $path);
156 my $im = _importer_for($self, $inbox);
157 my $mime = _path_to_mime($path) or return;
158 my $wm = $inbox->{-watchheader};
160 my $v = $mime->header_obj->header_raw($wm->[0]);
161 return unless ($v && $v =~ $wm->[1]);
163 if (my $scrub = $inbox->filter) {
164 my $ret = $scrub->scrub($mime) or return;
165 $ret == REJECT() and return;
169 $im->add($mime, $self->{spamcheck});
172 sub quit { trigger_scan($_[0], 'quit') }
176 my $scan = File::Temp->newdir("public-inbox-watch.$$.scan.XXXXXX",
178 my $scandir = $self->{scandir} = $scan->dirname;
179 my $re = qr!\A$scandir/!;
180 my $cb = sub { _try_fsn_paths($self, $re, \@_) };
182 # lazy load here, we may support watching via IMAP IDLE
184 require Filesys::Notify::Simple;
185 my $fsn = Filesys::Notify::Simple->new([@{$self->{mdir}}, $scandir]);
186 $fsn->wait($cb) until $self->{quit};
190 my ($self, $base) = @_;
191 my $dir = $self->{scandir} or return;
192 open my $fh, '>', "$dir/$base" or die "open $dir/$base failed: $!\n";
193 close $fh or die "close $dir/$base failed: $!\n";
197 my ($self, $path) = @_;
198 if ($path =~ /quit\z/) {
199 %{$self->{opendirs}} = ();
200 _done_for_now($self);
201 delete $self->{scandir};
205 # else: $path =~ /(cont|full)\z/
206 return if $self->{quit};
208 my $opendirs = $self->{opendirs};
209 my @dirnames = keys %$opendirs;
210 foreach my $dir (@dirnames) {
211 my $dh = delete $opendirs->{$dir};
213 while (my $fn = readdir($dh)) {
214 _try_path($self, "$dir/$fn");
217 $opendirs->{$dir} = $dh if $n < 0;
219 if ($path =~ /full\z/) {
220 foreach my $dir (@{$self->{mdir}}) {
221 next if $opendirs->{$dir}; # already in progress
222 my $ok = opendir(my $dh, $dir);
224 warn "failed to open $dir: $!\n";
228 while (my $fn = readdir($dh)) {
229 _try_path($self, "$dir/$fn");
232 $opendirs->{$dir} = $dh if $n < 0;
235 _done_for_now($self);
236 # do we have more work to do?
237 trigger_scan($self, 'cont') if keys %$opendirs;
242 if (open my $fh, '<', $path) {
246 return PublicInbox::MIME->new(\$str);
247 } elsif ($!{ENOENT}) {
250 warn "failed to open $path: $!\n";
256 my ($self, $ibx) = @_;
257 my $importers = $self->{importers};
258 my $im = $importers->{"$ibx"} ||= $ibx->importer(0);
259 if (scalar(keys(%$importers)) > 2) {
260 delete $importers->{"$ibx"};
261 _done_for_now($self);
264 $importers->{"$ibx"} = $im;
272 if ($sc->spamcheck($mime, \$tmp)) {
273 return PublicInbox::MIME->new(\$tmp);
275 warn $mime->header('Message-ID')." failed spam check\n";