1 # Copyright (C) 2016-2021 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 # httsp://wiki2.dovecot.org/MailboxFormat/Maildir
6 package PublicInbox::Watch;
10 use PublicInbox::InboxWritable qw(eml_from_path);
11 use PublicInbox::MdirReader;
12 use PublicInbox::Filter::Base qw(REJECT);
13 use PublicInbox::Spamcheck;
14 use PublicInbox::Sigfd;
15 use PublicInbox::DS qw(now add_timer);
16 use PublicInbox::MID qw(mids);
17 use PublicInbox::ContentHash qw(content_hash);
18 use PublicInbox::EOFpipe;
19 use POSIX qw(_exit WNOHANG);
21 sub compile_watchheaders ($) {
24 if (my $whs = $ibx->{watchheader}) {
26 my ($k, $v) = split(/:/, $_, 2);
27 # XXX should this be case-insensitive?
28 # Or, mutt-style, case-sensitive iff
29 # a capital letter exists?
30 push @$watch_hdrs, [ $k, qr/\Q$v\E/ ];
33 if (my $list_ids = $ibx->{listid}) {
35 # RFC2919 section 6 stipulates
36 # "case insensitive equality"
37 my $re = qr/<[ \t]*\Q$_\E[ \t]*>/i;
38 push @$watch_hdrs, ['List-Id', $re ];
41 $ibx->{-watchheaders} = $watch_hdrs if scalar @$watch_hdrs;
45 my ($class, $cfg) = @_;
47 my (%imap, %nntp); # url => [inbox objects] or 'watchspam'
49 # "publicinboxwatch" is the documented namespace
50 # "publicinboxlearn" is legacy but may be supported
52 foreach my $pfx (qw(publicinboxwatch publicinboxlearn)) {
53 my $k = "$pfx.watchspam";
54 defined(my $dirs = $cfg->{$k}) or next;
55 $dirs = PublicInbox::Config::_array($dirs);
56 for my $dir (@$dirs) {
58 if (is_maildir($dir)) {
59 # skip "new", no MUA has seen it, yet.
60 $mdmap{"$dir/cur"} = 'watchspam';
61 } elsif ($url = imap_url($dir)) {
62 $imap{$url} = 'watchspam';
63 } elsif ($url = nntp_url($dir)) {
64 $nntp{$url} = 'watchspam';
66 warn "unsupported $k=$dir\n";
71 my $k = 'publicinboxwatch.spamcheck';
73 my $spamcheck = PublicInbox::Spamcheck::get($cfg, $k, $default);
74 $spamcheck = _spamcheck_cb($spamcheck) if $spamcheck;
76 $cfg->each_inbox(sub {
77 # need to make all inboxes writable for spam removal:
78 my $ibx = $_[0] = PublicInbox::InboxWritable->new($_[0]);
80 my $watches = $ibx->{watch} or return;
81 $watches = PublicInbox::Config::_array($watches);
82 for my $watch (@$watches) {
84 if (is_maildir($watch)) {
85 compile_watchheaders($ibx);
86 my ($new, $cur) = ("$watch/new", "$watch/cur");
87 my $cur_dst = $mdmap{$cur} //= [];
88 return if is_watchspam($cur, $cur_dst, $ibx);
89 push @{$mdmap{$new} //= []}, $ibx;
91 } elsif ($url = imap_url($watch)) {
92 return if is_watchspam($url, $imap{$url}, $ibx);
93 compile_watchheaders($ibx);
94 push @{$imap{$url} ||= []}, $ibx;
95 } elsif ($url = nntp_url($watch)) {
96 return if is_watchspam($url, $nntp{$url}, $ibx);
97 compile_watchheaders($ibx);
98 push @{$nntp{$url} ||= []}, $ibx;
100 warn "watch unsupported: $k=$watch\n";
106 if (scalar keys %mdmap) {
107 $mdre = join('|', map { quotemeta($_) } keys %mdmap);
108 $mdre = qr!\A($mdre)/!;
110 return unless $mdre || scalar(keys %imap) || scalar(keys %nntp);
113 max_batch => 10, # avoid hogging locks for too long
114 spamcheck => $spamcheck,
118 imap => scalar keys %imap ? \%imap : undef,
119 nntp => scalar keys %nntp? \%nntp : undef,
121 opendirs => {}, # dirname => dirhandle (in progress scans)
122 ops => [], # 'quit', 'full'
128 local $PublicInbox::DS::in_loop = 0; # waitpid() synchronously
129 for my $im (values %{$self->{importers}}) {
130 next if !$im; # $im may be undef during cleanup
132 warn "$im->{ibx}->{name} ->done: $@\n" if $@;
136 sub remove_eml_i { # each_inbox callback
137 my ($ibx, $self, $eml, $loc) = @_;
140 # try to avoid taking a lock or unnecessary spawning
141 my $im = $self->{importers}->{"$ibx"};
143 if ((!$im || !$im->active) && $ibx->over) {
144 if (content_exists($ibx, $eml)) {
146 } elsif (my $scrub = $ibx->filter($im)) {
147 $scrubbed = $scrub->scrub($eml, 1);
148 if ($scrubbed && $scrubbed != REJECT &&
149 !content_exists($ibx, $scrubbed)) {
157 $im //= _importer_for($self, $ibx); # may spawn fast-import
158 $im->remove($eml, 'spam');
160 my $scrub = $ibx->filter($im);
161 $scrub ? $scrub->scrub($eml, 1) : undef;
163 if ($scrubbed && $scrubbed != REJECT) {
164 $im->remove($scrubbed, 'spam');
168 warn "error removing spam at: $loc from $ibx->{name}: $@\n";
169 _done_for_now($self);
174 my ($self, $path) = @_;
175 # path must be marked as (S)een
176 $path =~ /:2,[A-R]*S[T-Za-z]*\z/ or return;
177 my $eml = eml_from_path($path) or return;
178 local $SIG{__WARN__} = PublicInbox::Eml::warn_ignore_cb();
179 $self->{pi_cfg}->each_inbox(\&remove_eml_i, $self, $eml, $path);
182 sub import_eml ($$$) {
183 my ($self, $ibx, $eml) = @_;
185 # any header match means it's eligible for the inbox:
186 if (my $watch_hdrs = $ibx->{-watchheaders}) {
188 for my $wh (@$watch_hdrs) {
189 my @v = $eml->header_raw($wh->[0]);
190 $ok = grep(/$wh->[1]/, @v) and last;
195 my $im = _importer_for($self, $ibx);
196 if (my $scrub = $ibx->filter($im)) {
197 my $scrubbed = $scrub->scrub($eml) or return;
198 $scrubbed == REJECT and return;
201 $im->add($eml, $self->{spamcheck});
204 warn "$ibx->{name} add failed: $@\n";
205 _done_for_now($self);
210 my ($self, $path) = @_;
211 my $fl = PublicInbox::MdirReader::maildir_path_flags($path) // return;
212 return if $fl =~ /[DT]/; # no Drafts or Trash
213 if ($path !~ $self->{mdre}) {
214 warn "unrecognized path: $path\n";
217 my $inboxes = $self->{mdmap}->{$1};
219 warn "unmappable dir: $1\n";
222 my $warn_cb = $SIG{__WARN__} || \&CORE::warn;
223 local $SIG{__WARN__} = sub {
224 my $pfx = ($_[0] // '') =~ /^([A-Z]: )/g ? $1 : '';
225 $warn_cb->($pfx, "path: $path\n", @_);
227 if (!ref($inboxes) && $inboxes eq 'watchspam') {
228 return _remove_spam($self, $path);
230 foreach my $ibx (@$inboxes) {
231 my $eml = eml_from_path($path) or next;
232 import_eml($self, $ibx, $eml);
238 return unless $self->{quit};
240 # don't have reliable wakeups, keep signalling
242 for (qw(idle_pids poll_pids)) {
243 my $pids = $self->{$_} or next;
245 $done = undef if kill('QUIT', $_);
254 %{$self->{opendirs}} = ();
255 _done_for_now($self);
257 if (my $idle_mic = $self->{idle_mic}) {
258 eval { $idle_mic->done };
260 warn "IDLE DONE error: $@\n";
261 eval { $idle_mic->disconnect };
262 warn "IDLE LOGOUT error: $@\n" if $@;
267 sub watch_fs_init ($) {
270 delete $self->{done_timer};
271 _done_for_now($self);
273 my $cb = sub { # called by PublicInbox::DirIdle::event_step
274 _try_path($self, $_[0]->fullname);
275 $self->{done_timer} //= PublicInbox::DS::requeue($done);
277 require PublicInbox::DirIdle;
278 # inotify_create + EPOLL_CTL_ADD
279 PublicInbox::DirIdle->new([keys %{$self->{mdmap}}], $cb);
282 # avoid exposing deprecated "snews" to users.
283 my %SCHEME_MAP = ('snews' => 'nntps');
287 my $scheme = $uri->scheme;
288 $SCHEME_MAP{$scheme} // $scheme;
291 # returns the git config section name, e.g [imap "imaps://user@example.com"]
292 # without the mailbox, so we can share connections between different inboxes
293 sub uri_section ($) {
295 uri_scheme($uri) . '://' . $uri->authority;
298 sub cfg_intvl ($$$) {
299 my ($cfg, $key, $url) = @_;
300 my $v = $cfg->urlmatch($key, $url) // return;
301 $v =~ /\A[0-9]+(?:\.[0-9]+)?\z/s and return $v + 0;
302 if (ref($v) eq 'ARRAY') {
303 $v = join(', ', @$v);
304 warn "W: $key has multiple values: $v\nW: $key ignored\n";
306 warn "W: $key=$v is not a numeric value in seconds\n";
311 my ($cfg, $key, $url) = @_;
312 my $orig = $cfg->urlmatch($key, $url) // return;
313 my $bool = $cfg->git_bool($orig);
314 warn "W: $key=$orig for $url is not boolean\n" unless defined($bool);
318 # flesh out common IMAP-specific data structures
319 sub imap_common_init ($) {
321 my $cfg = $self->{pi_cfg};
322 my $mic_args = {}; # scheme://authority => Mail:IMAPClient arg
323 for my $url (sort keys %{$self->{imap}}) {
324 my $uri = PublicInbox::URIimap->new($url);
325 my $sec = uri_section($uri);
326 for my $k (qw(Starttls Debug Compress)) {
327 my $bool = cfg_bool($cfg, "imap.$k", $url) // next;
328 $mic_args->{$sec}->{$k} = $bool;
330 my $to = cfg_intvl($cfg, 'imap.timeout', $url);
331 $mic_args->{$sec}->{Timeout} = $to if $to;
332 for my $k (qw(pollInterval idleInterval)) {
333 $to = cfg_intvl($cfg, "imap.$k", $url) // next;
334 $self->{imap_opt}->{$sec}->{$k} = $to;
336 my $k = 'imap.fetchBatchSize';
337 my $bs = $cfg->urlmatch($k, $url) // next;
338 if ($bs =~ /\A([0-9]+)\z/) {
339 $self->{imap_opt}->{$sec}->{batch_size} = $bs;
341 warn "$k=$bs is not an integer\n";
347 sub auth_anon_cb { '' }; # for Mail::IMAPClient::Authcallback
349 sub mic_for ($$$) { # mic = Mail::IMAPClient
350 my ($self, $url, $mic_args) = @_;
351 my $uri = PublicInbox::URIimap->new($url);
352 require PublicInbox::GitCredential;
355 protocol => $uri->scheme,
357 username => $uri->user,
358 password => $uri->password,
359 }, 'PublicInbox::GitCredential';
360 my $common = $mic_args->{uri_section($uri)} // {};
361 # IMAPClient and Net::Netrc both mishandles `0', so we pass `127.0.0.1'
362 my $host = $cred->{host};
363 $host = '127.0.0.1' if $host eq '0';
367 Ssl => $uri->scheme eq 'imaps',
368 Keepalive => 1, # SO_KEEPALIVE
369 %$common, # may set Starttls, Compress, Debug ....
371 my $mic = PublicInbox::IMAPClient->new(%$mic_arg) or
372 die "E: <$url> new: $@\n";
374 # default to using STARTTLS if it's available, but allow
375 # it to be disabled since I usually connect to localhost
376 if (!$mic_arg->{Ssl} && !defined($mic_arg->{Starttls}) &&
377 $mic->has_capability('STARTTLS') &&
378 $mic->can('starttls')) {
379 $mic->starttls or die "E: <$url> STARTTLS: $@\n";
382 # do we even need credentials?
383 if (!defined($cred->{username}) &&
384 $mic->has_capability('AUTH=ANONYMOUS')) {
388 $cred->check_netrc unless defined $cred->{password};
389 $cred->fill; # may prompt user here
390 $mic->User($mic_arg->{User} = $cred->{username});
391 $mic->Password($mic_arg->{Password} = $cred->{password});
392 } else { # AUTH=ANONYMOUS
393 $mic->Authmechanism($mic_arg->{Authmechanism} = 'ANONYMOUS');
394 $mic->Authcallback($mic_arg->{Authcallback} = \&auth_anon_cb);
396 if ($mic->login && $mic->IsAuthenticated) {
397 # success! keep IMAPClient->new arg in case we get disconnected
398 $self->{mic_arg}->{uri_section($uri)} = $mic_arg;
400 warn "E: <$url> LOGIN: $@\n";
403 $cred->run($mic ? 'approve' : 'reject') if $cred;
407 sub imap_import_msg ($$$$$) {
408 my ($self, $url, $uid, $raw, $flags) = @_;
409 # our target audience expects LF-only, save storage
410 $$raw =~ s/\r\n/\n/sg;
412 my $inboxes = $self->{imap}->{$url};
414 for my $ibx (@$inboxes) {
415 my $eml = PublicInbox::Eml->new($$raw);
416 import_eml($self, $ibx, $eml);
418 } elsif ($inboxes eq 'watchspam') {
419 return if $flags !~ /\\Seen\b/; # don't remove unseen messages
420 local $SIG{__WARN__} = PublicInbox::Eml::warn_ignore_cb();
421 my $eml = PublicInbox::Eml->new($raw);
422 $self->{pi_cfg}->each_inbox(\&remove_eml_i,
423 $self, $eml, "$url UID:$uid");
425 die "BUG: destination unknown $inboxes";
429 sub imap_fetch_all ($$$) {
430 my ($self, $mic, $url) = @_;
431 my $uri = PublicInbox::URIimap->new($url);
432 my $sec = uri_section($uri);
433 my $mbx = $uri->mailbox;
434 $mic->Clear(1); # trim results history
435 $mic->examine($mbx) or return "E: EXAMINE $mbx ($sec) failed: $!";
436 my ($r_uidval, $r_uidnext);
437 for ($mic->Results) {
438 /^\* OK \[UIDVALIDITY ([0-9]+)\].*/ and $r_uidval = $1;
439 /^\* OK \[UIDNEXT ([0-9]+)\].*/ and $r_uidnext = $1;
440 last if $r_uidval && $r_uidnext;
442 $r_uidval //= $mic->uidvalidity($mbx) //
443 return "E: $url cannot get UIDVALIDITY";
444 $r_uidnext //= $mic->uidnext($mbx) //
445 return "E: $url cannot get UIDNEXT";
446 my $itrk = PublicInbox::IMAPTracker->new($url);
447 my ($l_uidval, $l_uid) = $itrk->get_last;
448 $l_uidval //= $r_uidval; # first time
450 if ($l_uidval != $r_uidval) {
451 return "E: $url UIDVALIDITY mismatch\n".
452 "E: local=$l_uidval != remote=$r_uidval";
454 my $r_uid = $r_uidnext - 1;
455 if ($l_uid != 1 && $l_uid > $r_uid) {
456 return "E: $url local UID exceeds remote ($l_uid > $r_uid)\n".
457 "E: $url strangely, UIDVALIDLITY matches ($l_uidval)\n";
459 return if $l_uid >= $r_uid; # nothing to do
461 warn "I: $url fetching UID $l_uid:$r_uid\n";
462 $mic->Uid(1); # the default, we hope
463 my $bs = $self->{imap_opt}->{$sec}->{batch_size} // 1;
464 my $req = $mic->imap4rev1 ? 'BODY.PEEK[]' : 'RFC822.PEEK';
466 # TODO: FLAGS may be useful for personal use
470 my $warn_cb = $SIG{__WARN__} || \&CORE::warn;
471 local $SIG{__WARN__} = sub {
472 my $pfx = ($_[0] // '') =~ /^([A-Z]: )/g ? $1 : '';
474 $warn_cb->("$pfx$url UID:$batch\n", @_);
478 # I wish "UID FETCH $START:*" could work, but:
479 # 1) servers do not need to return results in any order
480 # 2) Mail::IMAPClient doesn't offer a streaming API
481 $uids = $mic->search("UID $l_uid:*") or
482 return "E: $url UID SEARCH $l_uid:* error: $!";
483 return if scalar(@$uids) == 0;
485 # RFC 3501 doesn't seem to indicate order of UID SEARCH
486 # responses, so sort it ourselves. Order matters so
487 # IMAPTracker can store the newest UID.
488 @$uids = sort { $a <=> $b } @$uids;
490 # Did we actually get new messages?
491 return if $uids->[0] < $l_uid;
493 $l_uid = $uids->[-1] + 1; # for next search
495 my $n = $self->{max_batch};
497 while (scalar @$uids) {
499 _done_for_now($self);
500 $itrk->update_last($r_uidval, $last_uid);
501 $n = $self->{max_batch};
503 my @batch = splice(@$uids, 0, $bs);
504 $batch = join(',', @batch);
505 local $0 = "UID:$batch $mbx $sec";
506 my $r = $mic->fetch_hash($batch, $req, 'FLAGS');
507 unless ($r) { # network error?
508 $err = "E: $url UID FETCH $batch error: $!";
511 for my $uid (@batch) {
512 # messages get deleted, so holes appear
513 my $per_uid = delete $r->{$uid} // next;
514 my $raw = delete($per_uid->{$key}) // next;
515 my $fl = $per_uid->{FLAGS} // '';
516 imap_import_msg($self, $url, $uid, \$raw, $fl);
518 last if $self->{quit};
520 last if $self->{quit};
522 _done_for_now($self);
523 $itrk->update_last($r_uidval, $last_uid);
524 } until ($err || $self->{quit});
528 sub imap_idle_once ($$$$) {
529 my ($self, $mic, $intvl, $url) = @_;
530 my $i = $intvl //= (29 * 60);
531 my $end = now() + $intvl;
532 warn "I: $url idling for ${intvl}s\n";
533 local $0 = "IDLE $0";
534 unless ($mic->idle) {
535 return if $self->{quit};
536 return "E: IDLE failed on $url: $!";
538 $self->{idle_mic} = $mic; # for ->quit
540 until ($self->{quit} || !$mic->IsConnected ||
541 grep(/^\* [0-9]+ EXISTS/, @res) || $i <= 0) {
542 @res = $mic->idle_data($i);
545 delete $self->{idle_mic};
546 unless ($self->{quit}) {
547 $mic->IsConnected or return "E: IDLE disconnected on $url";
548 $mic->done or return "E: IDLE DONE failed on $url: $!";
553 # idles on a single URI
554 sub watch_imap_idle_1 ($$$) {
555 my ($self, $url, $intvl) = @_;
556 my $uri = PublicInbox::URIimap->new($url);
557 my $sec = uri_section($uri);
558 my $mic_arg = $self->{mic_arg}->{$sec} or
559 die "BUG: no Mail::IMAPClient->new arg for $sec";
561 local $0 = $uri->mailbox." $sec";
562 until ($self->{quit}) {
563 $mic //= PublicInbox::IMAPClient->new(%$mic_arg);
565 if ($mic && $mic->IsConnected) {
566 $err = imap_fetch_all($self, $mic, $url);
567 $err //= imap_idle_once($self, $mic, $intvl, $url);
569 $err = "E: not connected: $!";
571 if ($err && !$self->{quit}) {
574 sleep 60 unless $self->{quit};
579 sub watch_atfork_child ($) {
581 delete $self->{idle_pids};
582 delete $self->{poll_pids};
583 delete $self->{opendirs};
584 PublicInbox::DS->Reset;
585 %SIG = (%SIG, %{$self->{sig}}, CHLD => 'DEFAULT');
586 PublicInbox::DS::sig_setmask($self->{oldset});
589 sub watch_atfork_parent ($) {
591 _done_for_now($self);
592 PublicInbox::DS::block_signals();
595 sub imap_idle_requeue { # DS::add_timer callback
596 my ($self, $url_intvl) = @_;
597 return if $self->{quit};
598 push @{$self->{idle_todo}}, $url_intvl;
602 sub imap_idle_reap { # PublicInbox::DS::dwaitpid callback
603 my ($self, $pid) = @_;
604 my $url_intvl = delete $self->{idle_pids}->{$pid} or
605 die "BUG: PID=$pid (unknown) reaped: \$?=$?\n";
607 my ($url, $intvl) = @$url_intvl;
608 return if $self->{quit};
609 warn "W: PID=$pid on $url died: \$?=$?\n" if $?;
610 add_timer(60, \&imap_idle_requeue, $self, $url_intvl);
613 sub reap { # callback for EOFpipe
614 my ($pid, $cb, $self) = @{$_[0]};
615 my $ret = waitpid($pid, 0);
617 $cb->($self, $pid); # poll_fetch_reap || imap_idle_reap
619 warn "W: waitpid($pid) => ", $ret // "($!)", "\n";
623 sub imap_idle_fork ($$) {
624 my ($self, $url_intvl) = @_;
625 my ($url, $intvl) = @$url_intvl;
626 pipe(my ($r, $w)) or die "pipe: $!";
627 my $seed = rand(0xffffffff);
628 my $pid = fork // die "fork: $!";
631 eval { Net::SSLeay::randomize() };
633 watch_atfork_child($self);
634 watch_imap_idle_1($self, $url, $intvl);
638 $self->{idle_pids}->{$pid} = $url_intvl;
639 PublicInbox::EOFpipe->new($r, \&reap, [$pid, \&imap_idle_reap, $self]);
644 return if $self->{quit};
645 my $idle_todo = $self->{idle_todo};
646 if ($idle_todo && @$idle_todo) {
647 my $oldset = watch_atfork_parent($self);
649 while (my $url_intvl = shift(@$idle_todo)) {
650 imap_idle_fork($self, $url_intvl);
653 PublicInbox::DS::sig_setmask($oldset);
656 fs_scan_step($self) if $self->{mdre};
659 sub watch_imap_fetch_all ($$) {
660 my ($self, $urls) = @_;
661 for my $url (@$urls) {
662 my $uri = PublicInbox::URIimap->new($url);
663 my $sec = uri_section($uri);
664 my $mic_arg = $self->{mic_arg}->{$sec} or
665 die "BUG: no Mail::IMAPClient->new arg for $sec";
666 my $mic = PublicInbox::IMAPClient->new(%$mic_arg) or next;
667 my $err = imap_fetch_all($self, $mic, $url);
668 last if $self->{quit};
669 warn $err, "\n" if $err;
673 sub watch_nntp_fetch_all ($$) {
674 my ($self, $urls) = @_;
675 for my $url (@$urls) {
676 my $uri = uri_new($url);
677 my $sec = uri_section($uri);
678 my $nn_arg = $self->{nn_arg}->{$sec} or
679 die "BUG: no Net::NNTP->new arg for $sec";
680 my $nntp_opt = $self->{nntp_opt}->{$sec};
681 my $nn = nn_new($nn_arg, $nntp_opt, $url);
683 warn "E: $url: \$!=$!\n";
686 last if $self->{quit};
687 if (my $postconn = $nntp_opt->{-postconn}) {
688 for my $m_arg (@$postconn) {
689 my ($method, @args) = @$m_arg;
690 $nn->$method(@args) and next;
691 warn "E: <$url> $method failed\n";
696 last if $self->{quit};
698 my $err = nntp_fetch_all($self, $nn, $url);
699 warn $err, "\n" if $err;
704 sub poll_fetch_fork { # DS::add_timer callback
705 my ($self, $intvl, $urls) = @_;
706 return if $self->{quit};
707 pipe(my ($r, $w)) or die "pipe: $!";
708 my $oldset = watch_atfork_parent($self);
709 my $seed = rand(0xffffffff);
711 if (defined($pid) && $pid == 0) {
713 eval { Net::SSLeay::randomize() };
715 watch_atfork_child($self);
716 if ($urls->[0] =~ m!\Aimaps?://!i) {
717 watch_imap_fetch_all($self, $urls);
719 watch_nntp_fetch_all($self, $urls);
724 PublicInbox::DS::sig_setmask($oldset);
725 die "fork: $!" unless defined $pid;
726 $self->{poll_pids}->{$pid} = [ $intvl, $urls ];
727 PublicInbox::EOFpipe->new($r, \&reap, [$pid, \&poll_fetch_reap, $self]);
730 sub poll_fetch_reap {
731 my ($self, $pid) = @_;
732 my $intvl_urls = delete $self->{poll_pids}->{$pid} or
733 die "BUG: PID=$pid (unknown) reaped: \$?=$?\n";
734 return if $self->{quit};
735 my ($intvl, $urls) = @$intvl_urls;
737 warn "W: PID=$pid died: \$?=$?\n", map { "$_\n" } @$urls;
739 warn("I: will check $_ in ${intvl}s\n") for @$urls;
740 add_timer($intvl, \&poll_fetch_fork, $self, $intvl, $urls);
743 sub watch_imap_init ($$) {
744 my ($self, $poll) = @_;
745 eval { require PublicInbox::IMAPClient } or
746 die "Mail::IMAPClient is required for IMAP:\n$@\n";
747 eval { require PublicInbox::IMAPTracker } or
748 die "DBD::SQLite is required for IMAP\n:$@\n";
750 my $mic_args = imap_common_init($self); # read args from config
752 # make sure we can connect and cache the credentials in memory
753 $self->{mic_arg} = {}; # schema://authority => IMAPClient->new args
754 my $mics = {}; # schema://authority => IMAPClient obj
755 for my $url (sort keys %{$self->{imap}}) {
756 my $uri = PublicInbox::URIimap->new($url);
757 $mics->{uri_section($uri)} //= mic_for($self, $url, $mic_args);
760 my $idle = []; # [ [ url1, intvl1 ], [url2, intvl2] ]
761 for my $url (keys %{$self->{imap}}) {
762 my $uri = PublicInbox::URIimap->new($url);
763 my $sec = uri_section($uri);
764 my $mic = $mics->{$sec};
765 my $intvl = $self->{imap_opt}->{$sec}->{pollInterval};
766 if ($mic->has_capability('IDLE') && !$intvl) {
767 $intvl = $self->{imap_opt}->{$sec}->{idleInterval};
768 push @$idle, [ $url, $intvl // () ];
770 push @{$poll->{$intvl || 120}}, $url;
774 $self->{idle_todo} = $idle;
775 PublicInbox::DS::requeue($self); # ->event_step to fork
779 # flesh out common NNTP-specific data structures
780 sub nntp_common_init ($) {
782 my $cfg = $self->{pi_cfg};
783 my $nn_args = {}; # scheme://authority => Net::NNTP->new arg
784 for my $url (sort keys %{$self->{nntp}}) {
785 my $sec = uri_section(uri_new($url));
787 # Debug and Timeout are passed to Net::NNTP->new
788 my $v = cfg_bool($cfg, 'nntp.Debug', $url);
789 $nn_args->{$sec}->{Debug} = $v if defined $v;
790 my $to = cfg_intvl($cfg, 'nntp.Timeout', $url);
791 $nn_args->{$sec}->{Timeout} = $to if $to;
793 # Net::NNTP post-connect commands
794 for my $k (qw(starttls compress)) {
795 $v = cfg_bool($cfg, "nntp.$k", $url) // next;
796 $self->{nntp_opt}->{$sec}->{$k} = $v;
800 for my $k (qw(pollInterval)) {
801 $to = cfg_intvl($cfg, "nntp.$k", $url) // next;
802 $self->{nntp_opt}->{$sec}->{$k} = $to;
808 # Net::NNTP doesn't support CAPABILITIES, yet
809 sub try_starttls ($) {
811 return if $host =~ /\.onion\z/s;
812 return if $host =~ /\A127\.[0-9]+\.[0-9]+\.[0-9]+\z/s;
813 return if $host eq '::1';
818 my ($nn_arg, $nntp_opt, $url) = @_;
819 my $nn = Net::NNTP->new(%$nn_arg) or die "E: <$url> new: $!\n";
821 # default to using STARTTLS if it's available, but allow
822 # it to be disabled for localhost/VPN users
823 if (!$nn_arg->{SSL} && $nn->can('starttls')) {
824 if (!defined($nntp_opt->{starttls}) &&
825 try_starttls($nn_arg->{Host})) {
826 # soft fail by default
827 $nn->starttls or warn <<"";
828 W: <$url> STARTTLS tried and failed (not requested)
830 } elsif ($nntp_opt->{starttls}) {
831 # hard fail if explicitly configured
832 $nn->starttls or die <<"";
833 E: <$url> STARTTLS requested and failed
836 } elsif ($nntp_opt->{starttls}) {
837 $nn->can('starttls') or
838 die "E: <$url> Net::NNTP too old for STARTTLS\n";
839 $nn->starttls or die <<"";
840 E: <$url> STARTTLS requested and failed
846 sub nn_for ($$$) { # nn = Net::NNTP
847 my ($self, $url, $nn_args) = @_;
848 my $uri = uri_new($url);
849 my $sec = uri_section($uri);
850 my $nntp_opt = $self->{nntp_opt}->{$sec} //= {};
851 my $host = $uri->host;
852 # Net::NNTP and Net::Netrc both mishandle `0', so we pass `127.0.0.1'
853 $host = '127.0.0.1' if $host eq '0';
856 if (defined(my $ui = $uri->userinfo)) {
857 require PublicInbox::GitCredential;
860 protocol => uri_scheme($uri),
862 }, 'PublicInbox::GitCredential';
863 ($u, $p) = split(/:/, $ui, 2);
864 ($cred->{username}, $cred->{password}) = ($u, $p);
865 $cred->check_netrc unless defined $p;
867 my $common = $nn_args->{$sec} // {};
871 SSL => $uri->secure, # snews == nntps
872 %$common, # may Debug ....
874 my $nn = nn_new($nn_arg, $nntp_opt, $url);
877 $cred->fill; # may prompt user here
878 if ($nn->authinfo($u, $p)) {
879 push @{$nntp_opt->{-postconn}}, [ 'authinfo', $u, $p ];
881 warn "E: <$url> AUTHINFO $u XXXX failed\n";
886 if ($nntp_opt->{compress}) {
887 # https://rt.cpan.org/Ticket/Display.html?id=129967
888 if ($nn->can('compress')) {
890 push @{$nntp_opt->{-postconn}}, [ 'compress' ];
892 warn "W: <$url> COMPRESS failed\n";
895 delete $nntp_opt->{compress};
897 W: <$url> COMPRESS not supported by Net::NNTP
898 W: see https://rt.cpan.org/Ticket/Display.html?id=129967 for updates
903 $self->{nn_arg}->{$sec} = $nn_arg;
904 $cred->run($nn ? 'approve' : 'reject') if $cred;
908 sub nntp_fetch_all ($$$) {
909 my ($self, $nn, $url) = @_;
910 my $uri = uri_new($url);
911 my ($group, $num_a, $num_b) = $uri->group;
912 my $sec = uri_section($uri);
913 my ($nr, $beg, $end) = $nn->group($group);
914 unless (defined($nr)) {
915 chomp(my $msg = $nn->message);
916 return "E: GROUP $group <$sec> $msg";
919 # IMAPTracker is also used for tracking NNTP, UID == article number
920 # LIST.ACTIVE can get the equivalent of UIDVALIDITY, but that's
921 # expensive. So we assume newsgroups don't change:
922 my $itrk = PublicInbox::IMAPTracker->new($url);
923 my (undef, $l_art) = $itrk->get_last;
924 $l_art //= $beg; # initial import
926 # allow users to specify articles to refetch
927 # cf. https://tools.ietf.org/id/draft-gilman-news-url-01.txt
928 # nntp://example.com/inbox.foo/$num_a-$num_b
929 $l_art = $num_a if defined($num_a) && $num_a < $l_art;
930 $end = $num_b if defined($num_b) && $num_b < $end;
932 return if $l_art >= $end; # nothing to do
935 warn "I: $url fetching ARTICLE $beg..$end\n";
936 my $warn_cb = $SIG{__WARN__} || \&CORE::warn;
938 local $SIG{__WARN__} = sub {
939 my $pfx = ($_[0] // '') =~ /^([A-Z]: )/g ? $1 : '';
940 $warn_cb->("$pfx$url ", $art ? ("ARTICLE $art") : (), "\n", @_);
942 my $inboxes = $self->{nntp}->{$url};
944 my $n = $self->{max_batch};
946 last if $self->{quit};
949 _done_for_now($self);
950 $itrk->update_last(0, $last_art);
951 $n = $self->{max_batch};
953 my $raw = $nn->article($art);
954 unless (defined($raw)) {
955 my $msg = $nn->message;
956 if ($nn->code == 421) { # pseudo response from Net::Cmd
959 } else { # probably just a deleted message (spam)
964 s/\r\n/\n/ for @$raw;
965 $raw = join('', @$raw);
967 for my $ibx (@$inboxes) {
968 my $eml = PublicInbox::Eml->new($raw);
969 import_eml($self, $ibx, $eml);
971 } elsif ($inboxes eq 'watchspam') {
972 my $eml = PublicInbox::Eml->new(\$raw);
973 $self->{pi_cfg}->each_inbox(\&remove_eml_i,
974 $self, $eml, "$url ARTICLE $art");
976 die "BUG: destination unknown $inboxes";
980 _done_for_now($self);
981 $itrk->update_last(0, $last_art);
985 sub watch_nntp_init ($$) {
986 my ($self, $poll) = @_;
987 eval { require Net::NNTP } or
988 die "Net::NNTP is required for NNTP:\n$@\n";
989 eval { require PublicInbox::IMAPTracker } or
990 die "DBD::SQLite is required for NNTP\n:$@\n";
992 my $nn_args = nntp_common_init($self); # read args from config
994 # make sure we can connect and cache the credentials in memory
995 $self->{nn_arg} = {}; # schema://authority => Net::NNTP->new args
996 for my $url (sort keys %{$self->{nntp}}) {
997 nn_for($self, $url, $nn_args);
999 for my $url (keys %{$self->{nntp}}) {
1000 my $uri = uri_new($url);
1001 my $sec = uri_section($uri);
1002 my $intvl = $self->{nntp_opt}->{$sec}->{pollInterval};
1003 push @{$poll->{$intvl || 120}}, $url;
1007 sub watch { # main entry point
1008 my ($self, $sig, $oldset) = @_;
1009 $self->{oldset} = $oldset;
1010 $self->{sig} = $sig;
1011 my $poll = {}; # intvl_seconds => [ url1, url2 ]
1012 watch_imap_init($self, $poll) if $self->{imap};
1013 watch_nntp_init($self, $poll) if $self->{nntp};
1014 while (my ($intvl, $urls) = each %$poll) {
1015 # poll all URLs for a given interval sequentially
1016 add_timer(0, \&poll_fetch_fork, $self, $intvl, $urls);
1018 watch_fs_init($self) if $self->{mdre};
1019 PublicInbox::DS->SetPostLoopCallback(sub { !$self->quit_done });
1020 PublicInbox::DS->EventLoop; # calls ->event_step
1021 _done_for_now($self);
1025 my ($self, $op) = @_;
1026 push @{$self->{ops}}, $op;
1027 PublicInbox::DS::requeue($self);
1032 return if $self->{quit};
1033 my $op = shift @{$self->{ops}};
1034 local $PublicInbox::DS::in_loop = 0; # waitpid() synchronously
1036 # continue existing scan
1037 my $opendirs = $self->{opendirs};
1038 my @dirnames = keys %$opendirs;
1039 foreach my $dir (@dirnames) {
1040 my $dh = delete $opendirs->{$dir};
1041 my $n = $self->{max_batch};
1042 while (my $fn = readdir($dh)) {
1043 _try_path($self, "$dir/$fn");
1046 $opendirs->{$dir} = $dh if $n < 0;
1048 if ($op && $op eq 'full') {
1049 foreach my $dir (keys %{$self->{mdmap}}) {
1050 next if $opendirs->{$dir}; # already in progress
1051 my $ok = opendir(my $dh, $dir);
1053 warn "failed to open $dir: $!\n";
1056 my $n = $self->{max_batch};
1057 while (my $fn = readdir($dh)) {
1058 _try_path($self, "$dir/$fn");
1061 $opendirs->{$dir} = $dh if $n < 0;
1064 _done_for_now($self);
1065 # do we have more work to do?
1066 PublicInbox::DS::requeue($self) if keys %$opendirs;
1070 my ($self, $op) = @_;
1071 push @{$self->{ops}}, $op;
1072 fs_scan_step($self);
1076 my ($self, $ibx) = @_;
1077 my $importers = $self->{importers};
1078 my $im = $importers->{"$ibx"} ||= $ibx->importer(0);
1079 if (scalar(keys(%$importers)) > 2) {
1080 delete $importers->{"$ibx"};
1081 _done_for_now($self);
1084 $importers->{"$ibx"} = $im;
1087 # XXX consider sharing with V2Writable, this only requires read-only access
1088 sub content_exists ($$) {
1089 my ($ibx, $eml) = @_;
1090 my $over = $ibx->over or return;
1091 my $mids = mids($eml);
1092 my $chash = content_hash($eml);
1094 for my $mid (@$mids) {
1095 while (my $smsg = $over->next_by_mid($mid, \$id, \$prev)) {
1096 my $cmp = $ibx->smsg_eml($smsg) or return;
1097 return 1 if $chash eq content_hash($cmp);
1105 sub { # this gets called by (V2Writable||Import)->add
1106 my ($mime, $ibx) = @_;
1107 return if content_exists($ibx, $mime);
1109 if ($sc->spamcheck($mime, \$tmp)) {
1110 return PublicInbox::Eml->new(\$tmp);
1112 warn $mime->header('Message-ID')." failed spam check\n";
1118 $_[0] =~ s!\Amaildir:!! or return;
1125 my ($cur, $ws, $ibx) = @_;
1126 if ($ws && !ref($ws) && $ws eq 'watchspam') {
1128 E: $cur is a spam folder and cannot be used for `$ibx->{name}' input
1138 # URI::snews exists, URI::nntps does not, so use URI::snews
1139 $url =~ s!\Anntps://!snews://!i;
1145 require PublicInbox::URIimap;
1146 my $uri = PublicInbox::URIimap->new($url);
1147 $uri ? $uri->canonical->as_string : undef;
1150 my %IS_NNTP = (news => 1, snews => 1, nntp => 1);
1154 my $uri = uri_new($url);
1155 return unless $uri && $IS_NNTP{$uri->scheme} && $uri->group;
1156 $url = $uri->canonical->as_string;
1157 # nntps is IANA registered, snews is deprecated
1158 $url =~ s!\Asnews://!nntps://!;