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 rbuf 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 binmode $sock, ':utf8'; # RFC 3977
33 $self->SUPER::new($sock);
34 $self->{nntpd} = $nntpd;
35 res($self, '201 server ready - post via email');
43 my $tot = prototype $cb;
44 my ($nreq, undef) = split(';', $tot);
45 $nreq = ($nreq =~ tr/$//) - 1;
46 $tot = ($tot =~ tr/$//) - 1;
47 ($argc <= $tot && $argc >= $nreq);
50 # returns 1 if we can continue, 0 if not due to buffered writes or disconnect
51 sub process_line ($$) {
53 my ($req, @args) = split(/\s+/, $l);
57 $req = $DISABLED{$req} ? undef : *{'cmd_'.$req}{CODE};
59 return res($self, '500 command not recognized') unless $req;
60 return res($self, r501) unless args_ok($req, scalar @args);
62 my $res = eval { $req->($self, @args) };
64 if ($err && !$self->{closed}) {
65 chomp($l = Dumper(\$l));
66 err($self, "error from: $l $err");
67 $res = '503 program fault - command not performed';
69 return 0 unless defined $res;
74 my ($self, $arg) = @_;
76 return r501 unless $arg eq 'READER';
77 '200 reader status acknowledged';
80 sub cmd_slave ($) { '202 slave status noted' }
82 sub cmd_xgtitle ($;$) {
83 my ($self, $wildmat) = @_;
84 more($self, '282 list of groups and descriptions follows');
85 list_newsgroups($self, $wildmat);
89 sub list_overview_fmt ($$) {
91 more($self, $_ . ':') foreach @OVERVIEW;
94 sub list_active ($;$) {
95 my ($self, $wildmat) = @_;
97 foreach my $ng (@{$self->{nntpd}->{grouplist}}) {
98 $ng->{name} =~ $wildmat or next;
99 group_line($self, $ng);
103 sub list_active_times ($;$) {
104 my ($self, $wildmat) = @_;
105 wildmat2re($wildmat);
106 foreach my $ng (@{$self->{nntpd}->{grouplist}}) {
107 $ng->{name} =~ $wildmat or next;
108 my $c = eval { $ng->mm->created_at } || time;
109 more($self, "$ng->{name} $c $ng->{address}");
113 sub list_newsgroups ($;$) {
114 my ($self, $wildmat) = @_;
115 wildmat2re($wildmat);
116 foreach my $ng (@{$self->{nntpd}->{grouplist}}) {
117 $ng->{name} =~ $wildmat or next;
118 my $d = $ng->description;
119 more($self, "$ng->{name} $d");
123 # LIST SUBSCRIPTIONS not supported
124 sub cmd_list ($;$$) {
125 my ($self, @args) = @_;
127 my $arg = shift @args;
128 $arg =~ tr/A-Z./a-z_/;
130 return '503 function not performed' if $DISABLED{$arg};
136 return r501 unless $arg && args_ok($arg, scalar @args);
137 more($self, '215 information follows');
138 $arg->($self, @args);
140 more($self, '215 list of newsgroups follows');
141 foreach my $ng (@{$self->{nntpd}->{grouplist}}) {
142 group_line($self, $ng);
148 sub cmd_listgroup ($;$) {
149 my ($self, $group) = @_;
150 if (defined $group) {
151 my $res = cmd_group($self, $group);
152 return $res if ($res !~ /\A211 /);
156 $self->{ng} or return '412 no newsgroup selected';
157 $self->long_response(0, long_response_limit, sub {
159 my $nr = $self->{ng}->mm->id_batch($$i, sub {
161 more($self, join("\r\n", @$ary));
164 # -1 to adjust for implicit increment in long_response
165 $$i = $nr ? $$i + $nr - 1 : long_response_limit;
169 sub parse_time ($$;$) {
170 my ($date, $time, $gmt) = @_;
171 use Time::Local qw();
172 my ($hh, $mm, $ss) = unpack('A2A2A2', $time);
174 $gmt =~ /\A(?:UTC|GMT)\z/i or die "GM invalid: $gmt\n";
177 my @now = $gmt ? gmtime : localtime;
178 my ($YYYY, $MM, $DD);
179 if (length($date) == 8) { # RFC 3977 allows YYYYMMDD
180 ($YYYY, $MM, $DD) = unpack('A4A2A2', $date);
181 } else { # legacy clients send YYMMDD
182 ($YYYY, $MM, $DD) = unpack('A2A2A2', $date);
183 if ($YYYY > strftime('%y', @now)) {
184 my $cur_year = $now[5] + 1900;
185 $YYYY += int($cur_year / 1000) * 1000 - 100;
189 Time::Local::timegm($ss, $mm, $hh, $DD, $MM - 1, $YYYY);
191 Time::Local::timelocal($ss, $mm, $hh, $DD, $MM - 1, $YYYY);
195 sub group_line ($$) {
196 my ($self, $ng) = @_;
197 my ($min, $max) = $ng->mm->minmax;
198 more($self, "$ng->{name} $max $min n") if defined $min && defined $max;
201 sub cmd_newgroups ($$$;$$) {
202 my ($self, $date, $time, $gmt, $dists) = @_;
203 my $ts = eval { parse_time($date, $time, $gmt) };
207 more($self, '231 list of new newsgroups follows');
208 foreach my $ng (@{$self->{nntpd}->{grouplist}}) {
209 my $c = eval { $ng->mm->created_at } || 0;
210 next unless $c > $ts;
211 group_line($self, $ng);
216 sub wildmat2re (;$) {
217 return $_[0] = qr/.*/ if (!defined $_[0] || $_[0] eq '*');
220 use Digest::SHA qw(sha1_hex);
223 $tmp =~ s#(?<!\\)\[(.+)(?<!\\)\]#
225 my $key = sha1_hex($orig . $salt);
226 $orig =~ s/([^\w\-])+/\Q$1/g;
230 my %map = ('*' => '.*', '?' => '.' );
231 $tmp =~ s#(?<!\\)([^\w\\])#$map{$1} || "\Q$1"#ge;
233 $tmp =~ s#([a-f0-9]{40})#
234 my $orig = $keep{$1};
235 defined $orig ? $orig : $1;
238 $_[0] = qr/\A$tmp\z/;
242 return $_[0] = qr/\A\z/ unless defined $_[0];
243 my %map = ('*' => '.*', ',' => '|');
244 $_[0] =~ s!(.)!$map{$1} || "\Q$1"!ge;
245 $_[0] = qr/\A(?:$_[0])\z/;
248 sub cmd_newnews ($$$$;$$) {
249 my ($self, $newsgroups, $date, $time, $gmt, $dists) = @_;
250 my $ts = eval { parse_time($date, $time, $gmt) };
252 more($self, '230 list of new articles by message-id follows');
253 my ($keep, $skip) = split('!', $newsgroups, 2);
257 foreach my $ng (@{$self->{nntpd}->{grouplist}}) {
258 $ng->{name} =~ $keep or next;
259 $ng->{name} =~ $skip and next;
260 my $srch = $ng->search or next;
263 return '.' unless @srch;
266 my $opts = { asc => 1, limit => 1000, offset => 0 };
267 $self->long_response(0, long_response_limit, sub {
270 my $res = $srch->query($ts, $opts);
271 my $msgs = $res->{msgs};
272 if (my $nr = scalar @$msgs) {
274 join(">\r\n<", map { $_->mid } @$msgs ).
276 $opts->{offset} += $nr;
279 if (@srch) { # continue onto next newsgroup
281 } else { # break out of the long response.
282 $$i = long_response_limit;
289 my ($self, $group) = @_;
290 my $no_such = '411 no such news group';
291 my $ng = $self->{nntpd}->{groups}->{$group} or return $no_such;
294 my ($min, $max) = $ng->mm->minmax;
297 $self->{article} = $min;
298 my $est_size = $max - $min;
299 "211 $est_size $min $max $group";
302 sub article_adj ($$) {
303 my ($self, $off) = @_;
304 my $ng = $self->{ng} or return '412 no newsgroup selected';
306 my $n = $self->{article};
307 defined $n or return '420 no current article has been selected';
310 my $mid = $ng->mm->mid_for($n);
312 $n = $off > 0 ? 'next' : 'previous';
313 return "421 no $n article in this group";
315 $self->{article} = $n;
316 "223 $n <$mid> article retrieved - request text separately";
319 sub cmd_next ($) { article_adj($_[0], 1) }
320 sub cmd_last ($) { article_adj($_[0], -1) }
322 # We want to encourage using email and CC-ing everybody involved to avoid
323 # the single-point-of-failure a single server provides.
326 my $ng = $self->{ng};
327 $ng ? "440 mailto:$ng->{address} to post" : '440 posting not allowed'
332 res($self, '205 closing connection - goodbye!');
337 sub art_lookup ($$$) {
338 my ($self, $art, $set_headers) = @_;
339 my $ng = $self->{ng};
343 if ($art =~ /\A\d+\z/o) {
344 $err = '423 no such article number in this group';
347 } elsif ($art =~ /\A<([^>]+)>\z/) {
349 $err = '430 no such article found';
350 $n = $ng->mm->num_for($mid) if $ng;
351 goto found if defined $n;
352 foreach my $g (values %{$self->{nntpd}->{groups}}) {
353 $n = $g->mm->num_for($mid);
364 $err = '420 no current article has been selected';
365 $n = $self->{article};
366 defined $n or return $err;
368 $ng or return '412 no newsgroup has been selected';
369 $mid = $ng->mm->mid_for($n);
370 defined $mid or return $err;
373 my $o = 'HEAD:' . mid2path($mid);
375 my $s = eval { Email::MIME->new($ng->gcf->cat_file($o, \$bytes)) };
376 return $err unless $s;
378 $s->header_set('Newsgroups', $ng->{name});
379 $s->header_set('Lines', $s->body =~ tr!\n!\n!);
380 $s->header_set('Xref', xref($ng, $n));
383 if ($set_headers == 2) {
384 $s->header_set('Bytes', $bytes);
391 sub simple_body_write ($$) {
395 $body =~ s/^\./../smg;
396 do_more($self, $body);
402 my $h = $s->header_obj;
403 $h->header_set('Bytes');
408 my ($self, $art) = @_;
409 $self->{article} = $art if defined $art && $art =~ /\A\d+\z/;
412 sub cmd_article ($;$) {
413 my ($self, $art) = @_;
414 my $r = $self->art_lookup($art, 1);
415 return $r unless ref $r;
416 my ($n, $mid, $s) = @$r;
417 set_art($self, $art);
418 more($self, "220 $n <$mid> article retrieved - head and body follow");
419 do_more($self, header_str($s));
420 do_more($self, "\r\n");
421 simple_body_write($self, $s);
425 my ($self, $art) = @_;
426 my $r = $self->art_lookup($art, 2);
427 return $r unless ref $r;
428 my ($n, $mid, $s) = @$r;
429 set_art($self, $art);
430 more($self, "221 $n <$mid> article retrieved - head follows");
431 do_more($self, header_str($s));
436 my ($self, $art) = @_;
437 my $r = $self->art_lookup($art, 0);
438 return $r unless ref $r;
439 my ($n, $mid, $s) = @$r;
440 set_art($self, $art);
441 more($self, "222 $n <$mid> article retrieved - body follows");
442 simple_body_write($self, $s);
446 my ($self, $art) = @_;
447 my $r = $self->art_lookup($art, 0);
448 return $r unless ref $r;
449 my ($n, $mid, undef) = @$r;
450 set_art($self, $art);
451 "223 $n <$mid> article retrieved - request text separately";
454 sub cmd_ihave ($) { '435 article not wanted - do not send it' }
456 sub cmd_date ($) { '111 '.strftime('%Y%m%d%H%M%S', gmtime(time)) }
460 more($self, '100 help text follows');
465 my ($self, $range) = @_;
466 my $ng = $self->{ng} or return '412 no news group has been selected';
467 defined $range or return '420 No article(s) selected';
469 my ($min, $max) = $ng->mm->minmax;
470 if ($range =~ /\A(\d+)\z/) {
472 } elsif ($range =~ /\A(\d+)-\z/) {
473 ($beg, $end) = ($1, $max);
474 } elsif ($range =~ /\A(\d+)-(\d+)\z/) {
475 ($beg, $end) = ($1, $2);
479 $beg = $min if ($beg < $min);
480 $end = $max if ($end > $max);
481 return '420 No article(s) selected' if ($beg > $end);
486 my ($r, $header) = @_;
487 $r = $r->[2]->header_obj->header($header);
488 defined $r or return;
489 $r =~ s/[\r\n\t]+/ /sg;
493 sub long_response ($$$$) {
494 my ($self, $beg, $end, $cb) = @_;
495 die "BUG: nested long response" if $self->{long_res};
497 # make sure we disable reading during a long response,
498 # clients should not be sending us stuff and making us do more
499 # work while we are stream a response to them
500 $self->watch_read(0);
501 my $fd = fileno $self->{sock};
503 $self->{long_res} = sub {
504 # limit our own running time for fairness with other
505 # clients and to avoid buffering too much:
507 local $SIG{ALRM} = sub { $yield = 1 };
512 eval { $cb->(\$beg) };
513 } until (($err = $@) || $self->{closed} || $yield ||
514 $self->{write_buf_size} || ++$beg > $end);
517 if ($err || $self->{closed}) {
518 $self->{long_res} = undef;
522 "$err during long response[$fd] - %0.6f",
525 if ($self->{closed}) {
526 out($self, " deferred[$fd] aborted - %0.6f",
529 $self->watch_read(1);
531 } elsif ($yield || $self->{write_buf_size}) {
532 # no recursion, schedule another call ASAP
533 # but only after all pending writes are done
534 Danga::Socket->AddTimer(0, sub {
535 $self->write($self->{long_res});
538 $self->{long_res} = undef;
539 $self->watch_read(1);
541 out($self, " deferred[$fd] done - %0.6f", now() - $t0);
544 $self->{long_res}->(); # kick off!
548 sub xhdr_message_id ($$) { # optimize XHDR Message-ID [range] for slrnpull.
549 my ($self, $range) = @_;
551 my $mm = $self->{ng}->mm;
552 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
553 my $n = $mm->num_for($range);
554 more($self, '221 Header follows');
555 more($self, "<$range> <$range>") if defined $n;
557 } else { # numeric range
558 $range = $self->{article} unless defined $range;
559 my $r = get_range($self, $range);
560 return $r unless ref $r;
561 my ($beg, $end) = @$r;
562 more($self, '221 Header follows');
563 $self->long_response($beg, $end, sub {
565 my $mid = $mm->mid_for($$i);
566 more($self, "$$i <$mid>") if defined $mid;
573 "$ng->{domain} $ng->{name}:$n"
576 sub xhdr_xref ($$) { # optimize XHDR Xref [range] for rtin
577 my ($self, $range) = @_;
579 my $ng = $self->{ng};
581 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
582 my $n = $mm->num_for($range);
583 more($self, '221 Header follows');
584 more($self, "<$range> ".xref($ng, $n)) if defined $n;
586 } else { # numeric range
587 $range = $self->{article} unless defined $range;
588 my $r = get_range($self, $range);
589 return $r unless ref $r;
590 my ($beg, $end) = @$r;
591 more($self, '221 Header follows');
592 $self->long_response($beg, $end, sub {
594 my $mid = $mm->mid_for($$i);
595 more($self, "$$i ".xref($ng, $$i)) if defined $mid;
601 my ($srch, $mid) = @_;
603 my $smsg = $srch->lookup_message($mid);
604 $smsg = PublicInbox::SearchMsg->load_doc($smsg->{doc});
605 $smsg->mini_mime->header_obj;
609 sub xhdr_searchmsg ($$$) {
610 my ($self, $sub, $range) = @_;
611 my $srch = $self->{ng}->search;
612 my $emit = ($sub eq 'date') ? sub {
614 my @t = gmtime($m->header('X-PI-TS'));
615 more($self, "$pfx ". strftime('%a, %d %b %Y %T %z', @t));
618 my $h = $m->header($sub);
619 more($self, "$pfx $h") if defined $h;
622 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
623 more($self, '221 Header follows');
624 my $m = header_obj_for($srch, $1);
625 $emit->($range, $m) if defined $m;
627 } else { # numeric range
628 $range = $self->{article} unless defined $range;
629 my $mm = $self->{ng}->mm;
630 my $r = get_range($self, $range);
631 return $r unless ref $r;
632 my ($beg, $end) = @$r;
633 more($self, '221 Header follows');
634 $self->long_response($beg, $end, sub {
636 my $mid = $mm->mid_for($$i) or return;
637 my $m = header_obj_for($srch, $mid) or return;
643 sub cmd_xhdr ($$;$) {
644 my ($self, $header, $range) = @_;
645 my $ng = $self->{ng};
646 defined $ng or return '412 no news group currently selected';
647 my $sub = lc $header;
648 if ($sub eq 'message-id') {
649 xhdr_message_id($self, $range);
650 } elsif ($sub eq 'xref') {
651 xhdr_xref($self, $range);
652 } elsif ($sub =~ /\A(subject|references|date)\z/ && $ng->search) {
653 xhdr_searchmsg($self, $sub, $range);
655 xhdr_slow($self, $header, $range);
659 sub xhdr_slow ($$$) {
660 my ($self, $header, $range) = @_;
662 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
663 my $r = $self->art_lookup($range, 2);
664 return $r unless ref $r;
665 more($self, '221 Header follows');
666 if (defined($r = xhdr($r, $header))) {
667 more($self, "<$range> $r");
670 } else { # numeric range
671 $range = $self->{article} unless defined $range;
672 my $r = get_range($self, $range);
673 return $r unless ref $r;
674 my ($beg, $end) = @$r;
675 more($self, '221 Header follows');
676 $self->long_response($beg, $end, sub {
678 $r = $self->art_lookup($$i, 2);
679 return unless ref $r;
680 defined($r = xhdr($r, $header)) or return;
681 more($self, "$$i $r");
686 sub cmd_xrover ($;$) {
687 my ($self, $range) = @_;
688 my $ng = $self->{ng} or return '412 no newsgroup selected';
689 (defined $range && $range =~ /[<>]/) and
690 return '420 No article(s) selected'; # no message IDs
692 $range = $self->{article} unless defined $range;
693 my $r = get_range($self, $range);
694 return $r unless ref $r;
695 my ($beg, $end) = @$r;
697 my $srch = $ng->search;
698 more($self, '224 Overview information follows');
699 $self->long_response($beg, $end, sub {
701 my $mid = $mm->mid_for($$i) or return;
702 my $m = header_obj_for($srch, $mid) or return;
703 my $h = $m->header('references');
704 more($self, "$$i $h") if defined $h;
708 sub cmd_xover ($;$) {
709 my ($self, $range) = @_;
710 $range = $self->{article} unless defined $range;
711 my $r = get_range($self, $range);
712 return $r unless ref $r;
713 my ($beg, $end) = @$r;
714 more($self, "224 Overview information follows for $beg to $end");
715 $self->long_response($beg, $end, sub {
717 my $r = $self->art_lookup($$i, 2);
718 return unless ref $r;
719 more($self, join("\t", $r->[0],
721 my $h = xhdr($r, $_);
722 defined $h ? $h : '';
728 my ($self, $mid) = @_;
729 return r501 unless $mid =~ /\A<(.+)>\z/;
732 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
733 my $n = $ng->mm->num_for($mid);
734 push @paths, "$ng->{name}/$n" if defined $n;
736 return '430 no such article on server' unless @paths;
737 '223 '.join(' ', @paths);
741 my ($self, $line) = @_;
742 do_write($self, $line . "\r\n");
746 my ($self, $line) = @_;
747 do_more($self, $line . "\r\n");
751 my ($self, $data) = @_;
752 my $done = $self->write($data);
753 die if $self->{closed};
755 # Do not watch for readability if we have data in the queue,
756 # instead re-enable watching for readability when we can
757 $self->watch_read(0) if (!$done || $self->{long_res});
763 my ($self, $fmt, @args) = @_;
764 printf { $self->{nntpd}->{err} } $fmt."\n", @args;
768 my ($self, $fmt, @args) = @_;
769 printf { $self->{nntpd}->{out} } $fmt."\n", @args;
772 use constant MSG_MORE => ($^O eq 'linux') ? 0x8000 : 0;
775 my ($self, $data) = @_;
776 if (MSG_MORE && !$self->{write_buf_size}) {
777 my $n = send($self->{sock}, $data, MSG_MORE);
779 my $dlen = length($data);
780 return 1 if $n == $dlen; # all done!
781 $data = substr($data, $n, $dlen - $n);
784 $self->do_write($data);
787 # callbacks for by Danga::Socket
789 sub event_hup { $_[0]->close }
790 sub event_err { $_[0]->close }
794 # only continue watching for readability when we are done writing:
795 if ($self->write(undef) == 1 && !$self->{long_res}) {
796 $self->watch_read(1);
802 use constant LINE_MAX => 512; # RFC 977 section 2.3
806 while ($r > 0 && $self->{rbuf} =~ s/\A\s*([^\r\n]+)\r?\n//) {
809 $r = eval { $self->process_line($line) };
810 my $d = $self->{long_res} ?
811 ' deferred['.fileno($self->{sock}).']' : '';
812 out($self, "$line - %0.6f$d", now() - $t0);
814 unless (defined $line) {
815 my $buf = $self->read(LINE_MAX) or return $self->close;
816 $self->{rbuf} .= $$buf;
820 return $self->close if $r < 0;
821 my $len = length($self->{rbuf});
822 return $self->close if ($len >= LINE_MAX);
826 my ($self, $bool) = @_;
827 my $rv = $self->SUPER::watch_read($bool);
828 if ($bool && $self->{rbuf} ne '') {
829 # Force another read if there is a pipelined request.
830 # We don't know if the socket has anything for us to read,
831 # and we must double-check again by the time the timer fires
832 # in case we really did dispatch a read event and started
833 # another long response.
834 Danga::Socket->AddTimer(0, sub {
835 if (&Danga::Socket::POLLIN & $self->{event_watch}) {