1 # Copyright (C) 2016-2020 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;
10 use PublicInbox::InboxWritable;
11 use PublicInbox::Filter::Base qw(REJECT);
12 use PublicInbox::Spamcheck;
13 use PublicInbox::Sigfd;
14 use PublicInbox::DS qw(now);
15 use PublicInbox::MID qw(mids);
16 use PublicInbox::ContentHash qw(content_hash);
18 *mime_from_path = \&PublicInbox::InboxWritable::mime_from_path;
20 sub compile_watchheaders ($) {
23 if (my $whs = $ibx->{watchheader}) {
25 my ($k, $v) = split(/:/, $_, 2);
26 # XXX should this be case-insensitive?
27 # Or, mutt-style, case-sensitive iff
28 # a capital letter exists?
29 push @$watch_hdrs, [ $k, qr/\Q$v\E/ ];
32 if (my $list_ids = $ibx->{listid}) {
34 # RFC2919 section 6 stipulates
35 # "case insensitive equality"
36 my $re = qr/<[ \t]*\Q$_\E[ \t]*>/i;
37 push @$watch_hdrs, ['List-Id', $re ];
40 $ibx->{-watchheaders} = $watch_hdrs if scalar @$watch_hdrs;
44 my ($class, $config) = @_;
46 my (%imap, %nntp); # url => [inbox objects] or 'watchspam'
48 # "publicinboxwatch" is the documented namespace
49 # "publicinboxlearn" is legacy but may be supported
51 foreach my $pfx (qw(publicinboxwatch publicinboxlearn)) {
52 my $k = "$pfx.watchspam";
53 defined(my $dirs = $config->{$k}) or next;
54 $dirs = PublicInbox::Config::_array($dirs);
55 for my $dir (@$dirs) {
57 if (is_maildir($dir)) {
58 # skip "new", no MUA has seen it, yet.
59 $mdmap{"$dir/cur"} = 'watchspam';
60 } elsif ($url = imap_url($dir)) {
61 $imap{$url} = 'watchspam';
62 } elsif ($url = nntp_url($dir)) {
63 $nntp{$url} = 'watchspam';
65 warn "unsupported $k=$dir\n";
70 my $k = 'publicinboxwatch.spamcheck';
72 my $spamcheck = PublicInbox::Spamcheck::get($config, $k, $default);
73 $spamcheck = _spamcheck_cb($spamcheck) if $spamcheck;
75 $config->each_inbox(sub {
76 # need to make all inboxes writable for spam removal:
77 my $ibx = $_[0] = PublicInbox::InboxWritable->new($_[0]);
79 my $watches = $ibx->{watch} or return;
80 $watches = PublicInbox::Config::_array($watches);
81 for my $watch (@$watches) {
83 if (is_maildir($watch)) {
84 compile_watchheaders($ibx);
85 my ($new, $cur) = ("$watch/new", "$watch/cur");
86 my $cur_dst = $mdmap{$cur} //= [];
87 return if is_watchspam($cur, $cur_dst, $ibx);
88 push @{$mdmap{$new} //= []}, $ibx;
90 } elsif ($url = imap_url($watch)) {
91 return if is_watchspam($url, $imap{$url}, $ibx);
92 compile_watchheaders($ibx);
93 push @{$imap{$url} ||= []}, $ibx;
94 } elsif ($url = nntp_url($watch)) {
95 return if is_watchspam($url, $nntp{$url}, $ibx);
96 compile_watchheaders($ibx);
97 push @{$nntp{$url} ||= []}, $ibx;
99 warn "watch unsupported: $k=$watch\n";
105 if (scalar keys %mdmap) {
106 $mdre = join('|', map { quotemeta($_) } keys %mdmap);
107 $mdre = qr!\A($mdre)/!;
109 return unless $mdre || scalar(keys %imap) || scalar(keys %nntp);
112 spamcheck => $spamcheck,
116 imap => scalar keys %imap ? \%imap : undef,
117 nntp => scalar keys %nntp? \%nntp : undef,
119 opendirs => {}, # dirname => dirhandle (in progress scans)
120 ops => [], # 'quit', 'full'
126 local $PublicInbox::DS::in_loop = 0; # waitpid() synchronously
127 for my $im (values %{$self->{importers}}) {
128 next if !$im; # $im may be undef during cleanup
130 warn "$im->{ibx}->{name} ->done: $@\n" if $@;
134 sub remove_eml_i { # each_inbox callback
135 my ($ibx, $arg) = @_;
136 my ($self, $eml, $loc) = @$arg;
138 my $im = _importer_for($self, $ibx);
139 $im->remove($eml, 'spam');
140 if (my $scrub = $ibx->filter($im)) {
141 my $scrubbed = $scrub->scrub($eml, 1);
142 if ($scrubbed && $scrubbed != REJECT) {
143 $im->remove($scrubbed, 'spam');
148 warn "error removing spam at: $loc from $ibx->{name}: $@\n";
149 _done_for_now($self);
154 my ($self, $path) = @_;
155 # path must be marked as (S)een
156 $path =~ /:2,[A-R]*S[T-Za-z]*\z/ or return;
157 my $eml = mime_from_path($path) or return;
158 $self->{config}->each_inbox(\&remove_eml_i, [ $self, $eml, $path ]);
161 sub import_eml ($$$) {
162 my ($self, $ibx, $eml) = @_;
164 # any header match means it's eligible for the inbox:
165 if (my $watch_hdrs = $ibx->{-watchheaders}) {
167 my $hdr = $eml->header_obj;
168 for my $wh (@$watch_hdrs) {
169 my @v = $hdr->header_raw($wh->[0]);
170 $ok = grep(/$wh->[1]/, @v) and last;
175 my $im = _importer_for($self, $ibx);
176 if (my $scrub = $ibx->filter($im)) {
177 my $scrubbed = $scrub->scrub($eml) or return;
178 $scrubbed == REJECT and return;
181 $im->add($eml, $self->{spamcheck});
184 warn "$ibx->{name} add failed: $@\n";
185 _done_for_now($self);
190 my ($self, $path) = @_;
191 return unless PublicInbox::InboxWritable::is_maildir_path($path);
192 if ($path !~ $self->{mdre}) {
193 warn "unrecognized path: $path\n";
196 my $inboxes = $self->{mdmap}->{$1};
198 warn "unmappable dir: $1\n";
201 my $warn_cb = $SIG{__WARN__} || sub { print STDERR @_ };
202 local $SIG{__WARN__} = sub {
203 $warn_cb->("path: $path\n");
206 if (!ref($inboxes) && $inboxes eq 'watchspam') {
207 return _remove_spam($self, $path);
209 foreach my $ibx (@$inboxes) {
210 my $eml = mime_from_path($path) or next;
211 import_eml($self, $ibx, $eml);
217 return unless $self->{quit};
219 # don't have reliable wakeups, keep signalling
221 for (qw(idle_pids poll_pids)) {
222 my $pids = $self->{$_} or next;
224 $done = undef if kill('QUIT', $_);
233 %{$self->{opendirs}} = ();
234 _done_for_now($self);
236 if (my $idle_mic = $self->{idle_mic}) {
237 eval { $idle_mic->done };
239 warn "IDLE DONE error: $@\n";
240 eval { $idle_mic->disconnect };
241 warn "IDLE LOGOUT error: $@\n" if $@;
246 sub watch_fs_init ($) {
249 delete $self->{done_timer};
250 _done_for_now($self);
253 _try_path($self, $_[0]->fullname);
254 $self->{done_timer} //= PublicInbox::DS::requeue($done);
256 require PublicInbox::DirIdle;
257 # inotify_create + EPOLL_CTL_ADD
258 PublicInbox::DirIdle->new([keys %{$self->{mdmap}}], $cb);
261 # avoid exposing deprecated "snews" to users.
262 my %SCHEME_MAP = ('snews' => 'nntps');
266 my $scheme = $uri->scheme;
267 $SCHEME_MAP{$scheme} // $scheme;
270 # returns the git config section name, e.g [imap "imaps://user@example.com"]
271 # without the mailbox, so we can share connections between different inboxes
272 sub uri_section ($) {
274 uri_scheme($uri) . '://' . $uri->authority;
277 sub cfg_intvl ($$$) {
278 my ($cfg, $key, $url) = @_;
279 my $v = $cfg->urlmatch($key, $url) // return;
280 $v =~ /\A[0-9]+(?:\.[0-9]+)?\z/s and return $v + 0;
281 if (ref($v) eq 'ARRAY') {
282 $v = join(', ', @$v);
283 warn "W: $key has multiple values: $v\nW: $key ignored\n";
285 warn "W: $key=$v is not a numeric value in seconds\n";
290 my ($cfg, $key, $url) = @_;
291 my $orig = $cfg->urlmatch($key, $url) // return;
292 my $bool = PublicInbox::Config::_git_config_bool($orig);
293 warn "W: $key=$orig for $url is not boolean\n" unless defined($bool);
297 # flesh out common IMAP-specific data structures
298 sub imap_common_init ($) {
300 my $cfg = $self->{config};
301 my $mic_args = {}; # scheme://authority => Mail:IMAPClient arg
302 for my $url (sort keys %{$self->{imap}}) {
303 my $uri = PublicInbox::URIimap->new($url);
304 my $sec = uri_section($uri);
305 for my $k (qw(Starttls Debug Compress)) {
306 my $bool = cfg_bool($cfg, "imap.$k", $url) // next;
307 $mic_args->{$sec}->{$k} = $bool;
309 my $to = cfg_intvl($cfg, 'imap.timeout', $url);
310 $mic_args->{$sec}->{Timeout} = $to if $to;
311 for my $k (qw(pollInterval idleInterval)) {
312 $to = cfg_intvl($cfg, "imap.$k", $url) // next;
313 $self->{imap_opt}->{$sec}->{$k} = $to;
315 my $k = 'imap.fetchBatchSize';
316 my $bs = $cfg->urlmatch($k, $url) // next;
317 if ($bs =~ /\A([0-9]+)\z/) {
318 $self->{imap_opt}->{$sec}->{batch_size} = $bs;
320 warn "$k=$bs is not an integer\n";
326 sub auth_anon_cb { '' }; # for Mail::IMAPClient::Authcallback
328 sub mic_for ($$$) { # mic = Mail::IMAPClient
329 my ($self, $url, $mic_args) = @_;
330 my $uri = PublicInbox::URIimap->new($url);
331 require PublicInbox::GitCredential;
334 protocol => $uri->scheme,
336 username => $uri->user,
337 password => $uri->password,
338 }, 'PublicInbox::GitCredential';
339 my $common = $mic_args->{uri_section($uri)} // {};
340 # IMAPClient and Net::Netrc both mishandles `0', so we pass `127.0.0.1'
341 my $host = $cred->{host};
342 $host = '127.0.0.1' if $host eq '0';
346 Ssl => $uri->scheme eq 'imaps',
347 Keepalive => 1, # SO_KEEPALIVE
348 %$common, # may set Starttls, Compress, Debug ....
350 my $mic = PublicInbox::IMAPClient->new(%$mic_arg) or
351 die "E: <$url> new: $@\n";
353 # default to using STARTTLS if it's available, but allow
354 # it to be disabled since I usually connect to localhost
355 if (!$mic_arg->{Ssl} && !defined($mic_arg->{Starttls}) &&
356 $mic->has_capability('STARTTLS') &&
357 $mic->can('starttls')) {
358 $mic->starttls or die "E: <$url> STARTTLS: $@\n";
361 # do we even need credentials?
362 if (!defined($cred->{username}) &&
363 $mic->has_capability('AUTH=ANONYMOUS')) {
367 $cred->check_netrc unless defined $cred->{password};
368 $cred->fill; # may prompt user here
369 $mic->User($mic_arg->{User} = $cred->{username});
370 $mic->Password($mic_arg->{Password} = $cred->{password});
371 } else { # AUTH=ANONYMOUS
372 $mic->Authmechanism($mic_arg->{Authmechanism} = 'ANONYMOUS');
373 $mic->Authcallback($mic_arg->{Authcallback} = \&auth_anon_cb);
375 if ($mic->login && $mic->IsAuthenticated) {
376 # success! keep IMAPClient->new arg in case we get disconnected
377 $self->{mic_arg}->{uri_section($uri)} = $mic_arg;
379 warn "E: <$url> LOGIN: $@\n";
382 $cred->run($mic ? 'approve' : 'reject') if $cred;
386 sub imap_import_msg ($$$$) {
387 my ($self, $url, $uid, $raw) = @_;
388 # our target audience expects LF-only, save storage
389 $$raw =~ s/\r\n/\n/sg;
391 my $inboxes = $self->{imap}->{$url};
393 for my $ibx (@$inboxes) {
394 my $eml = PublicInbox::Eml->new($$raw);
395 my $x = import_eml($self, $ibx, $eml);
397 } elsif ($inboxes eq 'watchspam') {
398 my $eml = PublicInbox::Eml->new($raw);
399 my $arg = [ $self, $eml, "$url UID:$uid" ];
400 $self->{config}->each_inbox(\&remove_eml_i, $arg);
402 die "BUG: destination unknown $inboxes";
406 sub imap_fetch_all ($$$) {
407 my ($self, $mic, $url) = @_;
408 my $uri = PublicInbox::URIimap->new($url);
409 my $sec = uri_section($uri);
410 my $mbx = $uri->mailbox;
411 $mic->Clear(1); # trim results history
412 $mic->examine($mbx) or return "E: EXAMINE $mbx ($sec) failed: $!";
413 my ($r_uidval, $r_uidnext);
414 for ($mic->Results) {
415 /^\* OK \[UIDVALIDITY ([0-9]+)\].*/ and $r_uidval = $1;
416 /^\* OK \[UIDNEXT ([0-9]+)\].*/ and $r_uidnext = $1;
417 last if $r_uidval && $r_uidnext;
419 $r_uidval //= $mic->uidvalidity($mbx) //
420 return "E: $url cannot get UIDVALIDITY";
421 $r_uidnext //= $mic->uidnext($mbx) //
422 return "E: $url cannot get UIDNEXT";
423 my $itrk = PublicInbox::IMAPTracker->new($url);
424 my ($l_uidval, $l_uid) = $itrk->get_last;
425 $l_uidval //= $r_uidval; # first time
427 if ($l_uidval != $r_uidval) {
428 return "E: $url UIDVALIDITY mismatch\n".
429 "E: local=$l_uidval != remote=$r_uidval";
431 my $r_uid = $r_uidnext - 1;
432 if ($l_uid != 1 && $l_uid > $r_uid) {
433 return "E: $url local UID exceeds remote ($l_uid > $r_uid)\n".
434 "E: $url strangely, UIDVALIDLITY matches ($l_uidval)\n";
436 return if $l_uid >= $r_uid; # nothing to do
438 warn "I: $url fetching UID $l_uid:$r_uid\n";
439 $mic->Uid(1); # the default, we hope
440 my $bs = $self->{imap_opt}->{$sec}->{batch_size} // 1;
441 my $req = $mic->imap4rev1 ? 'BODY.PEEK[]' : 'RFC822.PEEK';
443 # TODO: FLAGS may be useful for personal use
447 my $warn_cb = $SIG{__WARN__} || sub { print STDERR @_ };
448 local $SIG{__WARN__} = sub {
450 $warn_cb->("$url UID:$batch\n");
455 # I wish "UID FETCH $START:*" could work, but:
456 # 1) servers do not need to return results in any order
457 # 2) Mail::IMAPClient doesn't offer a streaming API
458 $uids = $mic->search("UID $l_uid:*") or
459 return "E: $url UID SEARCH $l_uid:* error: $!";
460 return if scalar(@$uids) == 0;
462 # RFC 3501 doesn't seem to indicate order of UID SEARCH
463 # responses, so sort it ourselves. Order matters so
464 # IMAPTracker can store the newest UID.
465 @$uids = sort { $a <=> $b } @$uids;
467 # Did we actually get new messages?
468 return if $uids->[0] < $l_uid;
470 $l_uid = $uids->[-1] + 1; # for next search
473 while (scalar @$uids) {
474 my @batch = splice(@$uids, 0, $bs);
475 $batch = join(',', @batch);
476 local $0 = "UID:$batch $mbx $sec";
477 my $r = $mic->fetch_hash($batch, $req);
478 unless ($r) { # network error?
479 $err = "E: $url UID FETCH $batch error: $!";
482 for my $uid (@batch) {
483 # messages get deleted, so holes appear
484 my $per_uid = delete $r->{$uid} // next;
485 my $raw = delete($per_uid->{$key}) // next;
486 imap_import_msg($self, $url, $uid, \$raw);
488 last if $self->{quit};
490 last if $self->{quit};
492 _done_for_now($self);
493 $itrk->update_last($r_uidval, $last_uid) if defined $last_uid;
494 } until ($err || $self->{quit});
498 sub imap_idle_once ($$$$) {
499 my ($self, $mic, $intvl, $url) = @_;
500 my $i = $intvl //= (29 * 60);
501 my $end = now() + $intvl;
502 warn "I: $url idling for ${intvl}s\n";
503 local $0 = "IDLE $0";
504 unless ($mic->idle) {
505 return if $self->{quit};
506 return "E: IDLE failed on $url: $!";
508 $self->{idle_mic} = $mic; # for ->quit
510 until ($self->{quit} || !$mic->IsConnected ||
511 grep(/^\* [0-9]+ EXISTS/, @res) || $i <= 0) {
512 @res = $mic->idle_data($i);
515 delete $self->{idle_mic};
516 unless ($self->{quit}) {
517 $mic->IsConnected or return "E: IDLE disconnected on $url";
518 $mic->done or return "E: IDLE DONE failed on $url: $!";
523 # idles on a single URI
524 sub watch_imap_idle_1 ($$$) {
525 my ($self, $url, $intvl) = @_;
526 my $uri = PublicInbox::URIimap->new($url);
527 my $sec = uri_section($uri);
528 my $mic_arg = $self->{mic_arg}->{$sec} or
529 die "BUG: no Mail::IMAPClient->new arg for $sec";
531 local $0 = $uri->mailbox." $sec";
532 until ($self->{quit}) {
533 $mic //= PublicInbox::IMAPClient->new(%$mic_arg);
535 if ($mic && $mic->IsConnected) {
536 $err = imap_fetch_all($self, $mic, $url);
537 $err //= imap_idle_once($self, $mic, $intvl, $url);
539 $err = "not connected: $!";
541 if ($err && !$self->{quit}) {
544 sleep 60 unless $self->{quit};
549 sub watch_atfork_child ($) {
551 delete $self->{idle_pids};
552 delete $self->{poll_pids};
553 delete $self->{opendirs};
554 PublicInbox::DS->Reset;
555 %SIG = (%SIG, %{$self->{sig}}, CHLD => 'DEFAULT');
556 PublicInbox::Sigfd::sig_setmask($self->{oldset});
559 sub watch_atfork_parent ($) {
561 _done_for_now($self);
564 sub imap_idle_requeue ($) { # DS::add_timer callback
565 my ($self, $url_intvl) = @{$_[0]};
566 return if $self->{quit};
567 push @{$self->{idle_todo}}, $url_intvl;
571 sub imap_idle_reap { # PublicInbox::DS::dwaitpid callback
572 my ($self, $pid) = @_;
573 my $url_intvl = delete $self->{idle_pids}->{$pid} or
574 die "BUG: PID=$pid (unknown) reaped: \$?=$?\n";
576 my ($url, $intvl) = @$url_intvl;
577 return if $self->{quit};
578 warn "W: PID=$pid on $url died: \$?=$?\n" if $?;
579 PublicInbox::DS::add_timer(60,
580 \&imap_idle_requeue, [ $self, $url_intvl ]);
583 sub imap_idle_fork ($$) {
584 my ($self, $url_intvl) = @_;
585 my ($url, $intvl) = @$url_intvl;
586 defined(my $pid = fork) or die "fork: $!";
588 watch_atfork_child($self);
589 watch_imap_idle_1($self, $url, $intvl);
592 $self->{idle_pids}->{$pid} = $url_intvl;
593 PublicInbox::DS::dwaitpid($pid, \&imap_idle_reap, $self);
598 return if $self->{quit};
599 my $idle_todo = $self->{idle_todo};
600 if ($idle_todo && @$idle_todo) {
601 watch_atfork_parent($self);
602 while (my $url_intvl = shift(@$idle_todo)) {
603 imap_idle_fork($self, $url_intvl);
606 goto(&fs_scan_step) if $self->{mdre};
609 sub watch_imap_fetch_all ($$) {
610 my ($self, $urls) = @_;
611 for my $url (@$urls) {
612 my $uri = PublicInbox::URIimap->new($url);
613 my $sec = uri_section($uri);
614 my $mic_arg = $self->{mic_arg}->{$sec} or
615 die "BUG: no Mail::IMAPClient->new arg for $sec";
616 my $mic = PublicInbox::IMAPClient->new(%$mic_arg) or next;
617 my $err = imap_fetch_all($self, $mic, $url);
618 last if $self->{quit};
619 warn $err, "\n" if $err;
623 sub watch_nntp_fetch_all ($$) {
624 my ($self, $urls) = @_;
625 for my $url (@$urls) {
626 my $uri = uri_new($url);
627 my $sec = uri_section($uri);
628 my $nn_arg = $self->{nn_arg}->{$sec} or
629 die "BUG: no Net::NNTP->new arg for $sec";
630 my $nntp_opt = $self->{nntp_opt}->{$sec};
631 my $nn = nn_new($nn_arg, $nntp_opt, $url);
633 warn "E: $url: \$!=$!\n";
636 last if $self->{quit};
637 if (my $postconn = $nntp_opt->{-postconn}) {
638 for my $m_arg (@$postconn) {
639 my ($method, @args) = @$m_arg;
640 $nn->$method(@args) and next;
641 warn "E: <$url> $method failed\n";
646 last if $self->{quit};
648 my $err = nntp_fetch_all($self, $nn, $url);
649 warn $err, "\n" if $err;
654 sub poll_fetch_fork ($) { # DS::add_timer callback
655 my ($self, $intvl, $urls) = @{$_[0]};
656 return if $self->{quit};
657 watch_atfork_parent($self);
658 defined(my $pid = fork) or die "fork: $!";
660 watch_atfork_child($self);
661 if ($urls->[0] =~ m!\Aimaps?://!i) {
662 watch_imap_fetch_all($self, $urls);
664 watch_nntp_fetch_all($self, $urls);
668 $self->{poll_pids}->{$pid} = [ $intvl, $urls ];
669 PublicInbox::DS::dwaitpid($pid, \&poll_fetch_reap, $self);
672 sub poll_fetch_reap { # PublicInbox::DS::dwaitpid callback
673 my ($self, $pid) = @_;
674 my $intvl_urls = delete $self->{poll_pids}->{$pid} or
675 die "BUG: PID=$pid (unknown) reaped: \$?=$?\n";
676 return if $self->{quit};
677 my ($intvl, $urls) = @$intvl_urls;
679 warn "W: PID=$pid died: \$?=$?\n", map { "$_\n" } @$urls;
681 warn("I: will check $_ in ${intvl}s\n") for @$urls;
682 PublicInbox::DS::add_timer($intvl, \&poll_fetch_fork,
683 [$self, $intvl, $urls]);
686 sub watch_imap_init ($$) {
687 my ($self, $poll) = @_;
688 eval { require PublicInbox::IMAPClient } or
689 die "Mail::IMAPClient is required for IMAP:\n$@\n";
690 eval { require PublicInbox::IMAPTracker } or
691 die "DBD::SQLite is required for IMAP\n:$@\n";
693 my $mic_args = imap_common_init($self); # read args from config
695 # make sure we can connect and cache the credentials in memory
696 $self->{mic_arg} = {}; # schema://authority => IMAPClient->new args
697 my $mics = {}; # schema://authority => IMAPClient obj
698 for my $url (sort keys %{$self->{imap}}) {
699 my $uri = PublicInbox::URIimap->new($url);
700 $mics->{uri_section($uri)} //= mic_for($self, $url, $mic_args);
703 my $idle = []; # [ [ url1, intvl1 ], [url2, intvl2] ]
704 for my $url (keys %{$self->{imap}}) {
705 my $uri = PublicInbox::URIimap->new($url);
706 my $sec = uri_section($uri);
707 my $mic = $mics->{$sec};
708 my $intvl = $self->{imap_opt}->{$sec}->{pollInterval};
709 if ($mic->has_capability('IDLE') && !$intvl) {
710 $intvl = $self->{imap_opt}->{$sec}->{idleInterval};
711 push @$idle, [ $url, $intvl // () ];
713 push @{$poll->{$intvl || 120}}, $url;
717 $self->{idle_todo} = $idle;
718 PublicInbox::DS::requeue($self); # ->event_step to fork
722 # flesh out common NNTP-specific data structures
723 sub nntp_common_init ($) {
725 my $cfg = $self->{config};
726 my $nn_args = {}; # scheme://authority => Net::NNTP->new arg
727 for my $url (sort keys %{$self->{nntp}}) {
728 my $sec = uri_section(uri_new($url));
730 # Debug and Timeout are passed to Net::NNTP->new
731 my $v = cfg_bool($cfg, 'nntp.Debug', $url);
732 $nn_args->{$sec}->{Debug} = $v if defined $v;
733 my $to = cfg_intvl($cfg, 'nntp.Timeout', $url);
734 $nn_args->{$sec}->{Timeout} = $to if $to;
736 # Net::NNTP post-connect commands
737 for my $k (qw(starttls compress)) {
738 $v = cfg_bool($cfg, "nntp.$k", $url) // next;
739 $self->{nntp_opt}->{$sec}->{$k} = $v;
743 for my $k (qw(pollInterval)) {
744 $to = cfg_intvl($cfg, "nntp.$k", $url) // next;
745 $self->{nntp_opt}->{$sec}->{$k} = $to;
751 # Net::NNTP doesn't support CAPABILITIES, yet
752 sub try_starttls ($) {
754 return if $host =~ /\.onion\z/s;
755 return if $host =~ /\A127\.[0-9]+\.[0-9]+\.[0-9]+\z/s;
756 return if $host eq '::1';
761 my ($nn_arg, $nntp_opt, $url) = @_;
762 my $nn = Net::NNTP->new(%$nn_arg) or die "E: <$url> new: $!\n";
764 # default to using STARTTLS if it's available, but allow
765 # it to be disabled for localhost/VPN users
766 if (!$nn_arg->{SSL} && $nn->can('starttls')) {
767 if (!defined($nntp_opt->{starttls}) &&
768 try_starttls($nn_arg->{Host})) {
769 # soft fail by default
770 $nn->starttls or warn <<"";
771 W: <$url> STARTTLS tried and failed (not requested)
773 } elsif ($nntp_opt->{starttls}) {
774 # hard fail if explicitly configured
775 $nn->starttls or die <<"";
776 E: <$url> STARTTLS requested and failed
779 } elsif ($nntp_opt->{starttls}) {
780 $nn->can('starttls') or
781 die "E: <$url> Net::NNTP too old for STARTTLS\n";
782 $nn->starttls or die <<"";
783 E: <$url> STARTTLS requested and failed
789 sub nn_for ($$$) { # nn = Net::NNTP
790 my ($self, $url, $nn_args) = @_;
791 my $uri = uri_new($url);
792 my $sec = uri_section($uri);
793 my $nntp_opt = $self->{nntp_opt}->{$sec} //= {};
794 my $host = $uri->host;
795 # Net::NNTP and Net::Netrc both mishandle `0', so we pass `127.0.0.1'
796 $host = '127.0.0.1' if $host eq '0';
799 if (defined(my $ui = $uri->userinfo)) {
800 require PublicInbox::GitCredential;
803 protocol => uri_scheme($uri),
805 }, 'PublicInbox::GitCredential';
806 ($u, $p) = split(/:/, $ui, 2);
807 ($cred->{username}, $cred->{password}) = ($u, $p);
808 $cred->check_netrc unless defined $p;
810 my $common = $nn_args->{$sec} // {};
814 SSL => $uri->secure, # snews == nntps
815 %$common, # may Debug ....
817 my $nn = nn_new($nn_arg, $nntp_opt, $url);
820 $cred->fill; # may prompt user here
821 if ($nn->authinfo($u, $p)) {
822 push @{$nntp_opt->{-postconn}}, [ 'authinfo', $u, $p ];
824 warn "E: <$url> AUTHINFO $u XXXX failed\n";
829 if ($nntp_opt->{compress}) {
830 # https://rt.cpan.org/Ticket/Display.html?id=129967
831 if ($nn->can('compress')) {
833 push @{$nntp_opt->{-postconn}}, [ 'compress' ];
835 warn "W: <$url> COMPRESS failed\n";
838 delete $nntp_opt->{compress};
840 W: <$url> COMPRESS not supported by Net::NNTP
841 W: see https://rt.cpan.org/Ticket/Display.html?id=129967 for updates
846 $self->{nn_arg}->{$sec} = $nn_arg;
847 $cred->run($nn ? 'approve' : 'reject') if $cred;
851 sub nntp_fetch_all ($$$) {
852 my ($self, $nn, $url) = @_;
853 my $uri = uri_new($url);
854 my ($group, $num_a, $num_b) = $uri->group;
855 my $sec = uri_section($uri);
856 my ($nr, $beg, $end) = $nn->group($group);
857 unless (defined($nr)) {
858 chomp(my $msg = $nn->message);
859 return "E: GROUP $group <$sec> $msg";
862 # IMAPTracker is also used for tracking NNTP, UID == article number
863 # LIST.ACTIVE can get the equivalent of UIDVALIDITY, but that's
864 # expensive. So we assume newsgroups don't change:
865 my $itrk = PublicInbox::IMAPTracker->new($url);
866 my (undef, $l_art) = $itrk->get_last;
867 $l_art //= $beg; # initial import
869 # allow users to specify articles to refetch
870 # cf. https://tools.ietf.org/id/draft-gilman-news-url-01.txt
871 # nntp://example.com/inbox.foo/$num_a-$num_b
872 $l_art = $num_a if defined($num_a) && $num_a < $l_art;
873 $end = $num_b if defined($num_b) && $num_b < $end;
875 return if $l_art >= $end; # nothing to do
878 warn "I: $url fetching ARTICLE $beg..$end\n";
879 my $warn_cb = $SIG{__WARN__} || sub { print STDERR @_ };
881 local $SIG{__WARN__} = sub {
882 $warn_cb->("$url ", $art ? ("ARTICLE $art") : (), "\n", @_);
884 my $inboxes = $self->{nntp}->{$url};
887 last if $self->{quit};
889 my $raw = $nn->article($art);
890 unless (defined($raw)) {
891 my $msg = $nn->message;
892 if ($nn->code == 421) { # pseudo response from Net::Cmd
895 } else { # probably just a deleted message (spam)
900 s/\r\n/\n/ for @$raw;
901 $raw = join('', @$raw);
903 for my $ibx (@$inboxes) {
904 my $eml = PublicInbox::Eml->new($raw);
905 import_eml($self, $ibx, $eml);
907 } elsif ($inboxes eq 'watchspam') {
908 my $eml = PublicInbox::Eml->new(\$raw);
909 my $arg = [ $self, $eml, "$url ARTICLE $art" ];
910 $self->{config}->each_inbox(\&remove_eml_i, $arg);
912 die "BUG: destination unknown $inboxes";
916 $itrk->update_last(0, $last_art) if defined $last_art;
917 _done_for_now($self);
921 sub watch_nntp_init ($$) {
922 my ($self, $poll) = @_;
923 eval { require Net::NNTP } or
924 die "Net::NNTP is required for NNTP:\n$@\n";
925 eval { require PublicInbox::IMAPTracker } or
926 die "DBD::SQLite is required for NNTP\n:$@\n";
928 my $nn_args = nntp_common_init($self); # read args from config
930 # make sure we can connect and cache the credentials in memory
931 $self->{nn_arg} = {}; # schema://authority => Net::NNTP->new args
932 for my $url (sort keys %{$self->{nntp}}) {
933 nn_for($self, $url, $nn_args);
935 for my $url (keys %{$self->{nntp}}) {
936 my $uri = uri_new($url);
937 my $sec = uri_section($uri);
938 my $intvl = $self->{nntp_opt}->{$sec}->{pollInterval};
939 push @{$poll->{$intvl || 120}}, $url;
944 my ($self, $sig, $oldset) = @_;
945 $self->{oldset} = $oldset;
947 my $poll = {}; # intvl_seconds => [ url1, url2 ]
948 watch_imap_init($self, $poll) if $self->{imap};
949 watch_nntp_init($self, $poll) if $self->{nntp};
950 while (my ($intvl, $urls) = each %$poll) {
951 # poll all URLs for a given interval sequentially
952 PublicInbox::DS::add_timer(0, \&poll_fetch_fork,
953 [$self, $intvl, $urls]);
955 watch_fs_init($self) if $self->{mdre};
956 PublicInbox::DS->SetPostLoopCallback(sub { !$self->quit_done });
957 PublicInbox::DS->EventLoop;
958 _done_for_now($self);
962 my ($self, $op) = @_;
963 push @{$self->{ops}}, $op;
964 PublicInbox::DS::requeue($self);
969 return if $self->{quit};
970 my $op = shift @{$self->{ops}};
971 local $PublicInbox::DS::in_loop = 0; # waitpid() synchronously
973 # continue existing scan
975 my $opendirs = $self->{opendirs};
976 my @dirnames = keys %$opendirs;
977 foreach my $dir (@dirnames) {
978 my $dh = delete $opendirs->{$dir};
980 while (my $fn = readdir($dh)) {
981 _try_path($self, "$dir/$fn");
984 $opendirs->{$dir} = $dh if $n < 0;
986 if ($op && $op eq 'full') {
987 foreach my $dir (keys %{$self->{mdmap}}) {
988 next if $opendirs->{$dir}; # already in progress
989 my $ok = opendir(my $dh, $dir);
991 warn "failed to open $dir: $!\n";
995 while (my $fn = readdir($dh)) {
996 _try_path($self, "$dir/$fn");
999 $opendirs->{$dir} = $dh if $n < 0;
1002 _done_for_now($self);
1003 # do we have more work to do?
1004 PublicInbox::DS::requeue($self) if keys %$opendirs;
1008 my ($self, $op) = @_;
1009 push @{$self->{ops}}, $op;
1014 my ($self, $ibx) = @_;
1015 my $importers = $self->{importers};
1016 my $im = $importers->{"$ibx"} ||= $ibx->importer(0);
1017 if (scalar(keys(%$importers)) > 2) {
1018 delete $importers->{"$ibx"};
1019 _done_for_now($self);
1022 $importers->{"$ibx"} = $im;
1025 # XXX consider sharing with V2Writable, this only requires read-only access
1026 sub content_exists ($$) {
1027 my ($ibx, $eml) = @_;
1028 my $over = $ibx->over or return;
1029 my $mids = mids($eml);
1030 my $chash = content_hash($eml);
1032 for my $mid (@$mids) {
1033 while (my $smsg = $over->next_by_mid($mid, \$id, \$prev)) {
1034 my $cmp = $ibx->smsg_eml($smsg) or return;
1035 return 1 if $chash eq content_hash($cmp);
1044 my ($mime, $ibx) = @_;
1045 return if content_exists($ibx, $mime);
1047 if ($sc->spamcheck($mime, \$tmp)) {
1048 return PublicInbox::Eml->new(\$tmp);
1050 warn $mime->header('Message-ID')." failed spam check\n";
1056 $_[0] =~ s!\Amaildir:!! or return;
1063 my ($cur, $ws, $ibx) = @_;
1064 if ($ws && !ref($ws) && $ws eq 'watchspam') {
1066 E: $cur is a spam folder and cannot be used for `$ibx->{name}' input
1076 # URI::snews exists, URI::nntps does not, so use URI::snews
1077 $url =~ s!\Anntps://!snews://!i;
1083 require PublicInbox::URIimap;
1084 my $uri = PublicInbox::URIimap->new($url);
1085 $uri ? $uri->canonical->as_string : undef;
1088 my %IS_NNTP = (news => 1, snews => 1, nntp => 1);
1092 my $uri = uri_new($url);
1093 return unless $uri && $IS_NNTP{$uri->scheme} && $uri->group;
1094 $url = $uri->canonical->as_string;
1095 # nntps is IANA registered, snews is deprecated
1096 $url =~ s!\Asnews://!nntps://!;