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);
17 my ($class, $config) = @_;
18 my (%mdmap, @mdir, $spamc, $spamdir);
20 # "publicinboxwatch" is the documented namespace
21 # "publicinboxlearn" is legacy but may be supported
23 foreach my $pfx (qw(publicinboxwatch publicinboxlearn)) {
24 my $k = "$pfx.watchspam";
25 if (my $dir = $config->{$k}) {
26 if ($dir =~ s/\Amaildir://) {
28 # skip "new", no MUA has seen it, yet.
32 $mdmap{$cur} = 'watchspam';
34 warn "unsupported $k=$dir\n";
39 my $k = 'publicinboxwatch.spamcheck';
40 my $spamcheck = $config->{$k};
42 if ($spamcheck eq 'spamc') {
43 $spamcheck = 'PublicInbox::Spamcheck::Spamc';
45 if ($spamcheck =~ /::/) {
46 eval "require $spamcheck";
47 $spamcheck = _spamcheck_cb($spamcheck->new);
49 warn "unsupported $k=$spamcheck\n";
53 foreach $k (keys %$config) {
54 $k =~ /\Apublicinbox\.([^\.]+)\.watch\z/ or next;
56 my $watch = $config->{$k};
57 if ($watch =~ s/\Amaildir://) {
59 my $inbox = $config->lookup_name($name);
60 if (my $wm = $inbox->{watchheader}) {
61 my ($k, $v) = split(/:/, $wm, 2);
62 $inbox->{-watchheader} = [ $k, qr/\Q$v\E/ ];
64 my $new = "$watch/new";
65 my $cur = "$watch/cur";
66 push @mdir, $new, $cur;
67 die "$new already in use\n" if $mdmap{$new};
68 die "$cur already in use\n" if $mdmap{$cur};
69 $mdmap{$new} = $mdmap{$cur} = $inbox;
71 warn "watch unsupported: $k=$watch\n";
76 my $mdre = join('|', map { quotemeta($_) } @mdir);
77 $mdre = qr!\A($mdre)/!;
79 spamcheck => $spamcheck,
86 opendirs => {}, # dirname => dirhandle (in progress scans)
92 my $importers = $self->{importers};
93 foreach my $im (values %$importers) {
99 my ($self, $scan_re, $paths) = @_;
101 my $path = $_->{path};
102 if ($path =~ $scan_re) {
105 _try_path($self, $path);
108 _done_for_now($self);
112 my ($self, $path) = @_;
113 # path must be marked as (S)een
114 $path =~ /:2,[A-R]*S[T-Za-z]*\z/ or return;
115 my $mime = _path_to_mime($path) or return;
116 $self->{config}->each_inbox(sub {
119 my $im = _importer_for($self, $ibx);
120 $im->remove($mime, 'spam');
121 if (my $scrub = _scrubber_for($ibx)) {
122 my $scrubbed = $scrub->scrub($mime) or return;
123 $scrubbed == 100 and return;
124 $im->remove($scrubbed, 'spam');
128 warn "error removing spam at: ", $path,
129 " from ", $ibx->{name}, ': ', $@, "\n";
135 my ($self, $path) = @_;
136 my @p = split(m!/+!, $path);
137 return if $p[-1] !~ /\A[a-zA-Z0-9][\-\w:,=\.]+\z/;
138 if ($p[-1] =~ /:2,([A-Z]+)\z/i) {
140 return if $flags =~ /[DT]/; # no [D]rafts or [T]rashed mail
142 return unless -f $path;
143 if ($path !~ $self->{mdre}) {
144 warn "unrecognized path: $path\n";
147 my $inbox = $self->{mdmap}->{$1};
149 warn "unmappable dir: $1\n";
152 if (!ref($inbox) && $inbox eq 'watchspam') {
153 return _remove_spam($self, $path);
155 my $im = _importer_for($self, $inbox);
156 my $mime = _path_to_mime($path) or return;
157 $mime->header_set($_) foreach @PublicInbox::MDA::BAD_HEADERS;
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 $scrub = _scrubber_for($inbox)) {
164 my $ret = $scrub->scrub($mime) or return;
165 $ret == 100 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, $inbox) = @_;
257 my $im = $inbox->{-import} ||= eval {
258 my $v = $inbox->{version} || 1;
260 eval { require PublicInbox::V2Writable };
261 die "v2 not supported: $@\n" if $@;
262 my $v2w = PublicInbox::V2Writable->new($inbox);
263 $v2w->{parallel} = 0;
266 my $git = $inbox->git;
267 my $name = $inbox->{name};
268 my $addr = $inbox->{-primary_address};
269 PublicInbox::Import->new($git, $name, $addr, $inbox);
271 die "unsupported inbox version: $v\n";
275 my $importers = $self->{importers};
276 if (scalar(keys(%$importers)) > 2) {
277 delete $importers->{"$im"};
278 _done_for_now($self);
281 $importers->{"$im"} = $im;
286 my $f = $inbox->{filter};
287 if ($f && $f =~ /::/) {
288 my @args = (-inbox => $inbox);
289 # basic line splitting, only
290 # Perhaps we can have proper quote splitting one day...
291 ($f, @args) = split(/\s+/, $f) if $f =~ /\s+/;
297 # e.g: PublicInbox::Filter::Vger->new(@args)
298 return $f->new(@args);
309 if ($sc->spamcheck($mime, \$tmp)) {
310 return PublicInbox::MIME->new(\$tmp);
312 warn $mime->header('Message-ID')." failed spam check\n";