1 # Copyright (C) 2016 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
3 package PublicInbox::WatchMaildir;
7 use Email::MIME::ContentType;
8 $Email::MIME::ContentType::STRICT_PARAMS = 0; # user input is imperfect
10 use PublicInbox::Import;
12 use PublicInbox::Spawn qw(spawn);
15 my ($class, $config) = @_;
16 my (%mdmap, @mdir, $spamc);
18 # XXX is "publicinboxlearn" really a good namespace for this?
19 my $k = 'publicinboxlearn.watchspam';
20 if (my $spamdir = $config->{$k}) {
21 if ($spamdir =~ s/\Amaildir://) {
23 # skip "new", no MUA has seen it, yet.
24 my $cur = "$spamdir/cur";
26 $mdmap{$cur} = 'watchspam';
28 warn "unsupported $k=$spamdir\n";
32 $k = 'publicinboxwatch.spamcheck';
33 my $spamcheck = $config->{$k};
35 if ($spamcheck eq 'spamc') {
36 $spamcheck = 'PublicInbox::Spamcheck::Spamc';
38 if ($spamcheck =~ /::/) {
39 eval "require $spamcheck";
40 $spamcheck = _spamcheck_cb($spamcheck->new);
42 warn "unsupported $k=$spamcheck\n";
46 foreach $k (keys %$config) {
47 $k =~ /\Apublicinbox\.([^\.]+)\.watch\z/ or next;
49 my $watch = $config->{$k};
50 if ($watch =~ s/\Amaildir://) {
52 my $inbox = $config->lookup_name($name);
53 if (my $wm = $inbox->{watchheader}) {
54 my ($k, $v) = split(/:/, $wm, 2);
55 $inbox->{-watchheader} = [ $k, qr/\Q$v\E/ ];
57 my $new = "$watch/new";
58 my $cur = "$watch/cur";
59 push @mdir, $new, $cur;
60 die "$new already in use\n" if $mdmap{$new};
61 die "$cur already in use\n" if $mdmap{$cur};
62 $mdmap{$new} = $mdmap{$cur} = $inbox;
64 warn "watch unsupported: $k=$watch\n";
69 my $mdre = join('|', map { quotemeta($_) } @mdir);
70 $mdre = qr!\A($mdre)/!;
72 spamcheck => $spamcheck,
81 $_->done foreach values %{$_[0]->{importers}};
85 my ($self, $paths) = @_;
86 _try_path($self, $_->{path}) foreach @$paths;
91 my ($self, $path) = @_;
92 $path =~ /:2,[A-R]*S[T-Z]*\z/ or return;
93 my $mime = _path_to_mime($path) or return;
95 foreach my $inbox (values %{$self->{mdmap}}) {
96 next unless ref $inbox;
97 my $im = _importer_for($self, $inbox);
99 if (my $scrub = _scrubber_for($inbox)) {
100 my $scrubbed = $scrub->scrub($mime) or next;
101 $im->remove($scrubbed);
106 # used to hash the relevant portions of a message when there are conflicts
110 my $dig = Digest::SHA->new('SHA-1');
111 $dig->add($mime->header_obj->header_raw('Subject'));
112 $dig->add($mime->body_raw);
118 # probably a bad idea, but we inject a Message-Id if
119 # one is missing, here..
120 my $mid = $mime->header_obj->header_raw('Message-Id');
121 if (!defined $mid || $mid =~ /\A\s*\z/) {
122 $mid = '<' . _hash_mime2($mime) . '@generated>';
123 $mime->header_set('Message-Id', $mid);
128 my ($self, $path) = @_;
129 my @p = split(m!/+!, $path);
130 return unless $p[-1] =~ /\A[a-zA-Z0-9][\w:,=\.]+\z/;
131 return unless -f $path;
132 if ($path !~ $self->{mdre}) {
133 warn "unrecognized path: $path\n";
136 my $inbox = $self->{mdmap}->{$1};
138 warn "unmappable dir: $1\n";
141 if (!ref($inbox) && $inbox eq 'watchspam') {
142 return _remove_spam($self, $path);
144 my $im = _importer_for($self, $inbox);
145 my $mime = _path_to_mime($path) or return;
146 $mime->header_set($_) foreach @PublicInbox::MDA::BAD_HEADERS;
147 my $wm = $inbox->{-watchheader};
149 my $v = $mime->header_obj->header_raw($wm->[0]);
150 return unless ($v && $v =~ $wm->[1]);
152 if (my $scrub = _scrubber_for($inbox)) {
153 $mime = $scrub->scrub($mime) or return;
157 $im->add($mime, $self->{spamcheck});
162 my $cb = sub { _try_fsn_paths($self, \@_) };
163 my $mdir = $self->{mdir};
165 # lazy load here, we may support watching via IMAP IDLE
167 require Filesys::Notify::Simple;
168 my $watcher = Filesys::Notify::Simple->new($mdir);
169 $watcher->wait($cb) while (1);
174 my $mdir = $self->{mdir};
175 foreach my $dir (@$mdir) {
176 my $ok = opendir(my $dh, $dir);
178 warn "failed to open $dir: $!\n";
181 while (my $fn = readdir($dh)) {
182 _try_path($self, "$dir/$fn");
186 _done_for_now($self);
191 if (open my $fh, '<', $path) {
195 return Email::MIME->new(\$str);
196 } elsif ($!{ENOENT}) {
199 warn "failed to open $path: $!\n";
205 my ($self, $inbox) = @_;
206 my $im = $inbox->{-import} ||= eval {
207 my $git = $inbox->git;
208 my $name = $inbox->{name};
209 my $addr = $inbox->{-primary_address};
210 PublicInbox::Import->new($git, $name, $addr);
212 $self->{importers}->{"$im"} = $im;
217 my $f = $inbox->{filter};
218 if ($f && $f =~ /::/) {
234 if ($sc->spamcheck($mime, \$tmp)) {
235 return Email::MIME->new(\$tmp);