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 use PublicInbox::Spamcheck;
18 *REJECT = *PublicInbox::Filter::Base::REJECT;
21 my ($class, $config) = @_;
22 my (%mdmap, @mdir, $spamc, $spamdir);
24 # "publicinboxwatch" is the documented namespace
25 # "publicinboxlearn" is legacy but may be supported
27 foreach my $pfx (qw(publicinboxwatch publicinboxlearn)) {
28 my $k = "$pfx.watchspam";
29 if (my $dir = $config->{$k}) {
30 if ($dir =~ s/\Amaildir://) {
32 # skip "new", no MUA has seen it, yet.
36 $mdmap{$cur} = 'watchspam';
38 warn "unsupported $k=$dir\n";
43 my $k = 'publicinboxwatch.spamcheck';
45 my $spamcheck = PublicInbox::Spamcheck::get($config, $k, $default);
46 $spamcheck = _spamcheck_cb($spamcheck) if $spamcheck;
48 # need to make all inboxes writable for spam removal:
49 $config->each_inbox(sub { PublicInbox::InboxWritable->new($_[0]) });
51 foreach $k (keys %$config) {
52 $k =~ /\Apublicinbox\.([^\.]+)\.watch\z/ or next;
54 my $watch = $config->{$k};
55 if ($watch =~ s/\Amaildir://) {
57 my $inbox = $config->lookup_name($name);
58 if (my $wm = $inbox->{watchheader}) {
59 my ($k, $v) = split(/:/, $wm, 2);
60 $inbox->{-watchheader} = [ $k, qr/\Q$v\E/ ];
62 my $new = "$watch/new";
63 my $cur = "$watch/cur";
64 push @mdir, $new, $cur;
65 die "$new already in use\n" if $mdmap{$new};
66 die "$cur already in use\n" if $mdmap{$cur};
67 $mdmap{$new} = $mdmap{$cur} = $inbox;
69 warn "watch unsupported: $k=$watch\n";
74 my $mdre = join('|', map { quotemeta($_) } @mdir);
75 $mdre = qr!\A($mdre)/!;
77 spamcheck => $spamcheck,
84 opendirs => {}, # dirname => dirhandle (in progress scans)
90 my $importers = $self->{importers};
91 foreach my $im (values %$importers) {
97 my ($self, $scan_re, $paths) = @_;
99 my $path = $_->{path};
100 if ($path =~ $scan_re) {
103 _try_path($self, $path);
106 _done_for_now($self);
110 my ($self, $path) = @_;
111 # path must be marked as (S)een
112 $path =~ /:2,[A-R]*S[T-Za-z]*\z/ or return;
113 my $mime = _path_to_mime($path) or return;
114 $self->{config}->each_inbox(sub {
117 my $im = _importer_for($self, $ibx);
118 $im->remove($mime, 'spam');
119 if (my $scrub = $ibx->filter) {
120 my $scrubbed = $scrub->scrub($mime, 1);
122 $scrubbed == REJECT() and return;
123 $im->remove($scrubbed, 'spam');
127 warn "error removing spam at: ", $path,
128 " from ", $ibx->{name}, ': ', $@, "\n";
134 my ($self, $path) = @_;
135 return unless PublicInbox::InboxWritable::is_maildir_path($path);
136 if ($path !~ $self->{mdre}) {
137 warn "unrecognized path: $path\n";
140 my $inbox = $self->{mdmap}->{$1};
142 warn "unmappable dir: $1\n";
145 if (!ref($inbox) && $inbox eq 'watchspam') {
146 return _remove_spam($self, $path);
148 my $im = _importer_for($self, $inbox);
149 my $mime = _path_to_mime($path) or return;
150 my $wm = $inbox->{-watchheader};
152 my $v = $mime->header_obj->header_raw($wm->[0]);
153 return unless ($v && $v =~ $wm->[1]);
155 if (my $scrub = $inbox->filter) {
156 my $ret = $scrub->scrub($mime) or return;
157 $ret == REJECT() and return;
161 $im->add($mime, $self->{spamcheck});
164 sub quit { trigger_scan($_[0], 'quit') }
168 my $scan = File::Temp->newdir("public-inbox-watch.$$.scan.XXXXXX",
170 my $scandir = $self->{scandir} = $scan->dirname;
171 my $re = qr!\A$scandir/!;
172 my $cb = sub { _try_fsn_paths($self, $re, \@_) };
174 # lazy load here, we may support watching via IMAP IDLE
176 require Filesys::Notify::Simple;
177 my $fsn = Filesys::Notify::Simple->new([@{$self->{mdir}}, $scandir]);
178 $fsn->wait($cb) until $self->{quit};
182 my ($self, $base) = @_;
183 my $dir = $self->{scandir} or return;
184 open my $fh, '>', "$dir/$base" or die "open $dir/$base failed: $!\n";
185 close $fh or die "close $dir/$base failed: $!\n";
189 my ($self, $path) = @_;
190 if ($path =~ /quit\z/) {
191 %{$self->{opendirs}} = ();
192 _done_for_now($self);
193 delete $self->{scandir};
197 # else: $path =~ /(cont|full)\z/
198 return if $self->{quit};
200 my $opendirs = $self->{opendirs};
201 my @dirnames = keys %$opendirs;
202 foreach my $dir (@dirnames) {
203 my $dh = delete $opendirs->{$dir};
205 while (my $fn = readdir($dh)) {
206 _try_path($self, "$dir/$fn");
209 $opendirs->{$dir} = $dh if $n < 0;
211 if ($path =~ /full\z/) {
212 foreach my $dir (@{$self->{mdir}}) {
213 next if $opendirs->{$dir}; # already in progress
214 my $ok = opendir(my $dh, $dir);
216 warn "failed to open $dir: $!\n";
220 while (my $fn = readdir($dh)) {
221 _try_path($self, "$dir/$fn");
224 $opendirs->{$dir} = $dh if $n < 0;
227 _done_for_now($self);
228 # do we have more work to do?
229 trigger_scan($self, 'cont') if keys %$opendirs;
234 if (open my $fh, '<', $path) {
238 return PublicInbox::MIME->new(\$str);
239 } elsif ($!{ENOENT}) {
242 warn "failed to open $path: $!\n";
248 my ($self, $ibx) = @_;
249 my $importers = $self->{importers};
250 my $im = $importers->{"$ibx"} ||= $ibx->importer(0);
251 if (scalar(keys(%$importers)) > 2) {
252 delete $importers->{"$ibx"};
253 _done_for_now($self);
256 $importers->{"$ibx"} = $im;
264 if ($sc->spamcheck($mime, \$tmp)) {
265 return PublicInbox::MIME->new(\$tmp);
267 warn $mime->header('Message-ID')." failed spam check\n";