1 # Copyright (C) 2015 all contributors <meta@public-inbox.org>
2 # License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
3 package PublicInbox::NNTP;
6 use base qw(Danga::Socket);
7 use fields qw(nntpd article ng long_res);
8 use PublicInbox::Msgmap;
9 use PublicInbox::GitCatFile;
10 use PublicInbox::MID qw(mid2path);
12 use Data::Dumper qw(Dumper);
13 use POSIX qw(strftime);
14 use Time::HiRes qw(clock_gettime ualarm CLOCK_MONOTONIC);
16 r501 => '501 command syntax error',
17 long_response_limit => 0xffffffff,
20 sub now () { clock_gettime(CLOCK_MONOTONIC) };
22 my @OVERVIEW = qw(Subject From Date Message-ID References Bytes Lines);
23 my %OVERVIEW = map { $_ => 1 } @OVERVIEW;
25 # disable commands with easy DoS potential:
26 # LISTGROUP could get pretty bad, too...
27 my %DISABLED; # = map { $_ => 1 } qw(xover list_overview_fmt newnews xhdr);
30 my ($class, $sock, $nntpd) = @_;
31 my $self = fields::new($class);
32 $self->SUPER::new($sock);
33 $self->{nntpd} = $nntpd;
34 res($self, '201 server ready - post via email');
41 my $tot = prototype $cb;
42 my ($nreq, undef) = split(';', $tot);
43 $nreq = ($nreq =~ tr/$//) - 1;
44 $tot = ($tot =~ tr/$//) - 1;
45 ($argc <= $tot && $argc >= $nreq);
48 # returns 1 if we can continue, 0 if not due to buffered writes or disconnect
49 sub process_line ($$) {
51 my ($req, @args) = split(/\s+/, $l);
55 $req = $DISABLED{$req} ? undef : *{'cmd_'.$req}{CODE};
57 return res($self, '500 command not recognized') unless $req;
58 return res($self, r501) unless args_ok($req, scalar @args);
60 my $res = eval { $req->($self, @args) };
62 if ($err && !$self->{closed}) {
63 chomp($l = Dumper(\$l));
64 err($self, "error from: $l $err");
65 $res = '503 program fault - command not performed';
67 return 0 unless defined $res;
72 my ($self, $arg) = @_;
74 return r501 unless $arg eq 'READER';
75 '200 reader status acknowledged';
78 sub cmd_slave ($) { '202 slave status noted' }
80 sub cmd_xgtitle ($;$) {
81 my ($self, $wildmat) = @_;
82 more($self, '282 list of groups and descriptions follows');
83 list_newsgroups($self, $wildmat);
87 sub list_overview_fmt ($$) {
89 more($self, $_ . ':') foreach @OVERVIEW;
92 sub list_active ($;$) {
93 my ($self, $wildmat) = @_;
95 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
96 $ng->{name} =~ $wildmat or next;
97 group_line($self, $ng);
101 sub list_active_times ($;$) {
102 my ($self, $wildmat) = @_;
103 wildmat2re($wildmat);
104 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
105 $ng->{name} =~ $wildmat or next;
106 my $c = eval { $ng->mm->created_at } || time;
107 more($self, "$ng->{name} $c $ng->{address}");
111 sub list_newsgroups ($;$) {
112 my ($self, $wildmat) = @_;
113 wildmat2re($wildmat);
114 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
115 $ng->{name} =~ $wildmat or next;
116 my $d = $ng->description;
117 more($self, "$ng->{name} $d");
121 # LIST SUBSCRIPTIONS not supported
122 sub cmd_list ($;$$) {
123 my ($self, @args) = @_;
125 my $arg = shift @args;
126 $arg =~ tr/A-Z./a-z_/;
128 return '503 function not performed' if $DISABLED{$arg};
134 return r501 unless $arg && args_ok($arg, scalar @args);
135 more($self, '215 information follows');
136 $arg->($self, @args);
138 more($self, '215 list of newsgroups follows');
139 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
140 group_line($self, $ng);
146 sub cmd_listgroup ($;$) {
147 my ($self, $group) = @_;
148 if (defined $group) {
149 my $res = cmd_group($self, $group);
150 return $res if ($res !~ /\A211 /);
154 $self->{ng} or return '412 no newsgroup selected';
155 $self->long_response(0, long_response_limit, sub {
157 my $nr = $self->{ng}->mm->id_batch($$i, sub {
159 more($self, join("\r\n", @$ary));
162 # -1 to adjust for implicit increment in long_response
163 $$i = $nr ? $$i + $nr - 1 : long_response_limit;
167 sub parse_time ($$;$) {
168 my ($date, $time, $gmt) = @_;
169 use Time::Local qw();
170 my ($YY, $MM, $DD) = unpack('A2A2A2', $date);
171 my ($hh, $mm, $ss) = unpack('A2A2A2', $time);
173 $gmt =~ /\A(?:UTC|GMT)\z/i or die "GM invalid: $gmt\n";
176 my @now = $gmt ? gmtime : localtime;
177 if ($YY > strftime('%y', @now)) {
178 my $cur_year = $now[5] + 1900;
179 $YY += int($cur_year / 1000) * 1000 - 100;
183 Time::Local::timegm($ss, $mm, $hh, $DD, $MM - 1, $YY);
185 Time::Local::timelocal($ss, $mm, $hh, $DD, $MM - 1, $YY);
189 sub group_line ($$) {
190 my ($self, $ng) = @_;
191 my ($min, $max) = $ng->mm->minmax;
192 more($self, "$ng->{name} $max $min n") if defined $min && defined $max;
195 sub cmd_newgroups ($$$;$$) {
196 my ($self, $date, $time, $gmt, $dists) = @_;
197 my $ts = eval { parse_time($date, $time, $gmt) };
201 more($self, '231 list of new newsgroups follows');
202 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
203 my $c = eval { $ng->mm->created_at } || 0;
204 next unless $c > $ts;
205 group_line($self, $ng);
210 sub wildmat2re (;$) {
211 return $_[0] = qr/.*/ if (!defined $_[0] || $_[0] eq '*');
214 use Digest::SHA qw(sha1_hex);
217 $tmp =~ s#(?<!\\)\[(.+)(?<!\\)\]#
219 my $key = sha1_hex($orig . $salt);
220 $orig =~ s/([^\w\-])+/\Q$1/g;
224 my %map = ('*' => '.*', '?' => '.' );
225 $tmp =~ s#(?<!\\)([^\w\\])#$map{$1} || "\Q$1"#ge;
227 $tmp =~ s#([a-f0-9]{40})#
228 my $orig = $keep{$1};
229 defined $orig ? $orig : $1;
232 $_[0] = qr/\A$tmp\z/;
236 return $_[0] = qr/\A\z/ unless defined $_[0];
237 my %map = ('*' => '.*', ',' => '|');
238 $_[0] =~ s!(.)!$map{$1} || "\Q$1"!ge;
239 $_[0] = qr/\A(?:$_[0])\z/;
242 sub cmd_newnews ($$$$;$$) {
243 my ($self, $newsgroups, $date, $time, $gmt, $dists) = @_;
244 my $ts = eval { parse_time($date, $time, $gmt) };
246 more($self, '230 list of new articles by message-id follows');
247 my ($keep, $skip) = split('!', $newsgroups, 2);
251 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
252 $ng->{name} =~ $keep or next;
253 $ng->{name} =~ $skip and next;
254 my $srch = $ng->search or next;
257 return '.' unless @srch;
260 my $opts = { asc => 1, limit => 1000, offset => 0 };
261 $self->long_response(0, long_response_limit, sub {
264 my $res = $srch->query($ts, $opts);
265 my $msgs = $res->{msgs};
266 if (my $nr = scalar @$msgs) {
268 join(">\r\n<", map { $_->mid } @$msgs ).
270 $opts->{offset} += $nr;
273 if (@srch) { # continue onto next newsgroup
275 } else { # break out of the long response.
276 $$i = long_response_limit;
283 my ($self, $group) = @_;
284 my $no_such = '411 no such news group';
285 my $ng = $self->{nntpd}->{groups}->{$group} or return $no_such;
288 my ($min, $max) = $ng->mm->minmax;
291 $self->{article} = $min;
292 my $est_size = $max - $min;
293 "211 $est_size $min $max $group";
296 sub article_adj ($$) {
297 my ($self, $off) = @_;
298 my $ng = $self->{ng} or return '412 no newsgroup selected';
300 my $n = $self->{article};
301 defined $n or return '420 no current article has been selected';
304 my $mid = $ng->mm->mid_for($n);
306 $n = $off > 0 ? 'next' : 'previous';
307 return "421 no $n article in this group";
309 $self->{article} = $n;
310 "223 $n <$mid> article retrieved - request text separately";
313 sub cmd_next ($) { article_adj($_[0], 1) }
314 sub cmd_last ($) { article_adj($_[0], -1) }
316 # We want to encourage using email and CC-ing everybody involved to avoid
317 # the single-point-of-failure a single server provides.
320 my $ng = $self->{ng};
321 $ng ? "440 mailto:$ng->{address} to post" : '440 posting not allowed'
326 res($self, '205 closing connection - goodbye!');
331 sub art_lookup ($$$) {
332 my ($self, $art, $set_headers) = @_;
333 my $ng = $self->{ng};
337 if ($art =~ /\A\d+\z/o) {
338 $err = '423 no such article number in this group';
341 } elsif ($art =~ /\A<([^>]+)>\z/) {
343 $err = '430 no such article found';
344 $n = $ng->mm->num_for($mid) if $ng;
345 goto found if defined $n;
346 foreach my $g (values %{$self->{nntpd}->{groups}}) {
347 $n = $g->mm->num_for($mid);
358 $err = '420 no current article has been selected';
359 $n = $self->{article};
360 defined $n or return $err;
362 $ng or return '412 no newsgroup has been selected';
363 $mid = $ng->mm->mid_for($n);
364 defined $mid or return $err;
367 my $o = 'HEAD:' . mid2path($mid);
368 my $s = eval { Email::Simple->new($ng->gcf->cat_file($o)) };
369 return $err unless $s;
371 $s->header_set('Newsgroups', $ng->{name});
372 $s->header_set('Lines', $s->body =~ tr!\n!\n!);
373 $s->header_set('Xref', "$ng->{domain} $ng->{name}:$n");
376 if ($set_headers == 2) {
377 $s->header_set('Bytes', bytes::length($s->as_string));
384 sub simple_body_write ($$) {
388 $body =~ s/^\./../smg;
389 do_more($self, $body);
395 my $h = $s->header_obj;
396 $h->header_set('Bytes');
401 my ($self, $art) = @_;
402 $self->{article} = $art if defined $art && $art =~ /\A\d+\z/;
405 sub cmd_article ($;$) {
406 my ($self, $art) = @_;
407 my $r = $self->art_lookup($art, 1);
408 return $r unless ref $r;
409 my ($n, $mid, $s) = @$r;
410 set_art($self, $art);
411 more($self, "220 $n <$mid> article retrieved - head and body follow");
412 do_more($self, header_str($s));
413 do_more($self, "\r\n");
414 simple_body_write($self, $s);
418 my ($self, $art) = @_;
419 my $r = $self->art_lookup($art, 2);
420 return $r unless ref $r;
421 my ($n, $mid, $s) = @$r;
422 set_art($self, $art);
423 more($self, "221 $n <$mid> article retrieved - head follows");
424 do_more($self, header_str($s));
429 my ($self, $art) = @_;
430 my $r = $self->art_lookup($art, 0);
431 return $r unless ref $r;
432 my ($n, $mid, $s) = @$r;
433 set_art($self, $art);
434 more($self, "222 $n <$mid> article retrieved - body follows");
435 simple_body_write($self, $s);
439 my ($self, $art) = @_;
440 my $r = $self->art_lookup($art, 0);
441 return $r unless ref $r;
442 my ($n, $mid, undef) = @$r;
443 set_art($self, $art);
444 "223 $n <$mid> article retrieved - request text separately";
447 sub cmd_ihave ($) { '435 article not wanted - do not send it' }
449 sub cmd_date ($) { '111 '.strftime('%Y%m%d%H%M%S', gmtime(time)) }
453 more($self, '100 help text follows');
458 my ($self, $range) = @_;
459 my $ng = $self->{ng} or return '412 no news group has been selected';
460 defined $range or return '420 No article(s) selected';
462 my ($min, $max) = $ng->mm->minmax;
463 if ($range =~ /\A(\d+)\z/) {
465 } elsif ($range =~ /\A(\d+)-\z/) {
466 ($beg, $end) = ($1, $max);
467 } elsif ($range =~ /\A(\d+)-(\d+)\z/) {
468 ($beg, $end) = ($1, $2);
472 $beg = $min if ($beg < $min);
473 $end = $max if ($end > $max);
474 return '420 No article(s) selected' if ($beg > $end);
479 my ($r, $header) = @_;
480 $r = $r->[2]->header_obj->header($header);
481 defined $r or return;
482 $r =~ s/[\r\n\t]+/ /sg;
486 sub long_response ($$$$) {
487 my ($self, $beg, $end, $cb) = @_;
488 die "BUG: nested long response" if $self->{long_res};
490 # make sure we disable reading during a long response,
491 # clients should not be sending us stuff and making us do more
492 # work while we are stream a response to them
493 $self->watch_read(0);
494 my $fd = fileno $self->{sock};
496 $self->{long_res} = sub {
497 # limit our own running time for fairness with other
498 # clients and to avoid buffering too much:
500 local $SIG{ALRM} = sub { $yield = 1 };
505 eval { $cb->(\$beg) };
506 } until (($err = $@) || $self->{closed} || $yield ||
507 $self->{write_buf_size} || ++$beg > $end);
510 if ($err || $self->{closed}) {
511 $self->{long_res} = undef;
515 "$err during long response[$fd] - %0.6f",
518 if ($self->{closed}) {
519 out($self, " deferred[$fd] aborted - %0.6f",
522 $self->watch_read(1);
524 } elsif ($yield || $self->{write_buf_size}) {
525 # no recursion, schedule another call ASAP
526 # but only after all pending writes are done
527 Danga::Socket->AddTimer(0, sub {
528 $self->write($self->{long_res});
531 $self->{long_res} = undef;
532 $self->watch_read(1);
534 out($self, " deferred[$fd] done - %0.6f", now() - $t0);
537 $self->{long_res}->(); # kick off!
541 sub xhdr_message_id ($$) { # optimize XHDR Message-ID [range] for slrnpull.
542 my ($self, $range) = @_;
544 my $mm = $self->{ng}->mm;
545 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
546 my $n = $mm->num_for($range);
547 more($self, '221 Header follows');
548 more($self, "<$range> <$range>") if defined $n;
550 } else { # numeric range
551 $range = $self->{article} unless defined $range;
552 my $r = get_range($self, $range);
553 return $r unless ref $r;
554 my ($beg, $end) = @$r;
555 more($self, '221 Header follows');
556 $self->long_response($beg, $end, sub {
558 my $mid = $mm->mid_for($$i);
559 more($self, "$$i <$mid>") if defined $mid;
565 my ($srch, $mid) = @_;
567 my $smsg = $srch->lookup_message($mid);
568 $smsg = PublicInbox::SearchMsg->load_doc($smsg->{doc});
569 $smsg->mini_mime->header_obj;
573 sub xhdr_searchmsg ($$$) {
574 my ($self, $sub, $range) = @_;
575 my $srch = $self->{ng}->search;
576 my $emit = ($sub eq 'date') ? sub {
578 my @t = gmtime($m->header('X-PI-TS'));
579 more($self, "$pfx ". strftime('%a, %d %b %Y %T %z', @t));
582 my $h = $m->header($sub);
583 more($self, "$pfx $h") if defined $h;
586 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
587 more($self, '221 Header follows');
588 my $m = header_obj_for($srch, $1);
589 $emit->($range, $m) if defined $m;
591 } else { # numeric range
592 $range = $self->{article} unless defined $range;
593 my $mm = $self->{ng}->mm;
594 my $r = get_range($self, $range);
595 return $r unless ref $r;
596 my ($beg, $end) = @$r;
597 more($self, '221 Header follows');
598 $self->long_response($beg, $end, sub {
600 my $mid = $mm->mid_for($$i) or return;
601 my $m = header_obj_for($srch, $mid) or return;
607 sub cmd_xhdr ($$;$) {
608 my ($self, $header, $range) = @_;
609 my $ng = $self->{ng};
610 defined $ng or return '412 no news group currently selected';
611 my $sub = lc $header;
612 if ($sub eq 'message-id') {
613 xhdr_message_id($self, $range);
614 } elsif ($sub =~ /\A(subject|references|date)\z/ && $ng->search) {
615 xhdr_searchmsg($self, $sub, $range);
617 xhdr_slow($self, $header, $range);
621 sub xhdr_slow ($$$) {
622 my ($self, $header, $range) = @_;
624 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
625 my $r = $self->art_lookup($range, 2);
626 return $r unless ref $r;
627 more($self, '221 Header follows');
628 if (defined($r = xhdr($r, $header))) {
629 more($self, "<$range> $r");
632 } else { # numeric range
633 $range = $self->{article} unless defined $range;
634 my $r = get_range($self, $range);
635 return $r unless ref $r;
636 my ($beg, $end) = @$r;
637 more($self, '221 Header follows');
638 $self->long_response($beg, $end, sub {
640 $r = $self->art_lookup($$i, 2);
641 return unless ref $r;
642 defined($r = xhdr($r, $header)) or return;
643 more($self, "$$i $r");
648 sub cmd_xrover ($;$) {
649 my ($self, $range) = @_;
650 my $ng = $self->{ng} or return '412 no newsgroup selected';
651 (defined $range && $range =~ /[<>]/) and
652 return '420 No article(s) selected'; # no message IDs
654 $range = $self->{article} unless defined $range;
655 my $r = get_range($self, $range);
656 return $r unless ref $r;
657 my ($beg, $end) = @$r;
659 my $srch = $ng->search;
660 more($self, '224 Overview information follows');
661 $self->long_response($beg, $end, sub {
663 my $mid = $mm->mid_for($$i) or return;
664 my $m = header_obj_for($srch, $mid) or return;
665 my $h = $m->header('references');
666 more($self, "$$i $h") if defined $h;
670 sub cmd_xover ($;$) {
671 my ($self, $range) = @_;
672 $range = $self->{article} unless defined $range;
673 my $r = get_range($self, $range);
674 return $r unless ref $r;
675 my ($beg, $end) = @$r;
676 more($self, "224 Overview information follows for $beg to $end");
677 $self->long_response($beg, $end, sub {
679 my $r = $self->art_lookup($$i, 2);
680 return unless ref $r;
681 more($self, join("\t", $r->[0],
683 my $h = xhdr($r, $_);
684 defined $h ? $h : '';
690 my ($self, $mid) = @_;
691 return r501 unless $mid =~ /\A<(.+)>\z/;
694 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
695 my $n = $ng->mm->num_for($mid);
696 push @paths, "$ng->{name}/$n" if defined $n;
698 return '430 no such article on server' unless @paths;
699 '223 '.join(' ', @paths);
703 my ($self, $line) = @_;
704 do_write($self, $line . "\r\n");
708 my ($self, $line) = @_;
709 do_more($self, $line . "\r\n");
713 my ($self, $data) = @_;
714 my $done = $self->write($data);
715 die if $self->{closed};
717 # Do not watch for readability if we have data in the queue,
718 # instead re-enable watching for readability when we can
719 $self->watch_read(0) if (!$done || $self->{long_res});
725 my ($self, $fmt, @args) = @_;
726 printf { $self->{nntpd}->{err} } $fmt."\n", @args;
730 my ($self, $fmt, @args) = @_;
731 printf { $self->{nntpd}->{out} } $fmt."\n", @args;
734 use constant MSG_MORE => ($^O eq 'linux') ? 0x8000 : 0;
737 my ($self, $data) = @_;
738 if (MSG_MORE && !$self->{write_buf_size}) {
739 my $n = send($self->{sock}, $data, MSG_MORE);
741 my $dlen = bytes::length($data);
742 return 1 if $n == $dlen; # all done!
743 $data = bytes::substr($data, $n, $dlen - $n);
746 $self->do_write($data);
749 # callbacks for by Danga::Socket
751 sub event_hup { $_[0]->close }
752 sub event_err { $_[0]->close }
756 # only continue watching for readability when we are done writing:
757 if ($self->write(undef) == 1 && !$self->{long_res}) {
758 $self->watch_read(1);
764 use constant LINE_MAX => 512; # RFC 977 section 2.3
766 my $buf = $self->read(LINE_MAX) or return $self->close;
767 while ($r > 0 && $$buf =~ s/\A\s*([^\r\n]+)\r?\n//) {
770 $r = eval { $self->process_line($line) };
771 my $d = $self->{long_res} ?
772 ' deferred['.fileno($self->{sock}).']' : '';
773 out($self, "$line - %0.6f$d", now() - $t0);
775 return $self->close if $r < 0;
776 my $len = bytes::length($$buf);
777 return $self->close if ($len >= LINE_MAX);
778 $self->push_back_read($buf) if ($len);