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 $self->SUPER::new($sock);
33 $self->{nntpd} = $nntpd;
34 res($self, '201 server ready - post via email');
42 my $tot = prototype $cb;
43 my ($nreq, undef) = split(';', $tot);
44 $nreq = ($nreq =~ tr/$//) - 1;
45 $tot = ($tot =~ tr/$//) - 1;
46 ($argc <= $tot && $argc >= $nreq);
49 # returns 1 if we can continue, 0 if not due to buffered writes or disconnect
50 sub process_line ($$) {
52 my ($req, @args) = split(/\s+/, $l);
56 $req = $DISABLED{$req} ? undef : *{'cmd_'.$req}{CODE};
58 return res($self, '500 command not recognized') unless $req;
59 return res($self, r501) unless args_ok($req, scalar @args);
61 my $res = eval { $req->($self, @args) };
63 if ($err && !$self->{closed}) {
64 chomp($l = Dumper(\$l));
65 err($self, "error from: $l $err");
66 $res = '503 program fault - command not performed';
68 return 0 unless defined $res;
73 my ($self, $arg) = @_;
75 return r501 unless $arg eq 'READER';
76 '200 reader status acknowledged';
79 sub cmd_slave ($) { '202 slave status noted' }
81 sub cmd_xgtitle ($;$) {
82 my ($self, $wildmat) = @_;
83 more($self, '282 list of groups and descriptions follows');
84 list_newsgroups($self, $wildmat);
88 sub list_overview_fmt ($$) {
90 more($self, $_ . ':') foreach @OVERVIEW;
93 sub list_active ($;$) {
94 my ($self, $wildmat) = @_;
96 foreach my $ng (@{$self->{nntpd}->{grouplist}}) {
97 $ng->{name} =~ $wildmat or next;
98 group_line($self, $ng);
102 sub list_active_times ($;$) {
103 my ($self, $wildmat) = @_;
104 wildmat2re($wildmat);
105 foreach my $ng (@{$self->{nntpd}->{grouplist}}) {
106 $ng->{name} =~ $wildmat or next;
107 my $c = eval { $ng->mm->created_at } || time;
108 more($self, "$ng->{name} $c $ng->{address}");
112 sub list_newsgroups ($;$) {
113 my ($self, $wildmat) = @_;
114 wildmat2re($wildmat);
115 foreach my $ng (@{$self->{nntpd}->{grouplist}}) {
116 $ng->{name} =~ $wildmat or next;
117 my $d = $ng->description;
118 more($self, "$ng->{name} $d");
122 # LIST SUBSCRIPTIONS not supported
123 sub cmd_list ($;$$) {
124 my ($self, @args) = @_;
126 my $arg = shift @args;
127 $arg =~ tr/A-Z./a-z_/;
129 return '503 function not performed' if $DISABLED{$arg};
135 return r501 unless $arg && args_ok($arg, scalar @args);
136 more($self, '215 information follows');
137 $arg->($self, @args);
139 more($self, '215 list of newsgroups follows');
140 foreach my $ng (@{$self->{nntpd}->{grouplist}}) {
141 group_line($self, $ng);
147 sub cmd_listgroup ($;$) {
148 my ($self, $group) = @_;
149 if (defined $group) {
150 my $res = cmd_group($self, $group);
151 return $res if ($res !~ /\A211 /);
155 $self->{ng} or return '412 no newsgroup selected';
156 $self->long_response(0, long_response_limit, sub {
158 my $nr = $self->{ng}->mm->id_batch($$i, sub {
160 more($self, join("\r\n", @$ary));
163 # -1 to adjust for implicit increment in long_response
164 $$i = $nr ? $$i + $nr - 1 : long_response_limit;
168 sub parse_time ($$;$) {
169 my ($date, $time, $gmt) = @_;
170 use Time::Local qw();
171 my ($YY, $MM, $DD) = unpack('A2A2A2', $date);
172 my ($hh, $mm, $ss) = unpack('A2A2A2', $time);
174 $gmt =~ /\A(?:UTC|GMT)\z/i or die "GM invalid: $gmt\n";
177 my @now = $gmt ? gmtime : localtime;
178 if ($YY > strftime('%y', @now)) {
179 my $cur_year = $now[5] + 1900;
180 $YY += int($cur_year / 1000) * 1000 - 100;
184 Time::Local::timegm($ss, $mm, $hh, $DD, $MM - 1, $YY);
186 Time::Local::timelocal($ss, $mm, $hh, $DD, $MM - 1, $YY);
190 sub group_line ($$) {
191 my ($self, $ng) = @_;
192 my ($min, $max) = $ng->mm->minmax;
193 more($self, "$ng->{name} $max $min n") if defined $min && defined $max;
196 sub cmd_newgroups ($$$;$$) {
197 my ($self, $date, $time, $gmt, $dists) = @_;
198 my $ts = eval { parse_time($date, $time, $gmt) };
202 more($self, '231 list of new newsgroups follows');
203 foreach my $ng (@{$self->{nntpd}->{grouplist}}) {
204 my $c = eval { $ng->mm->created_at } || 0;
205 next unless $c > $ts;
206 group_line($self, $ng);
211 sub wildmat2re (;$) {
212 return $_[0] = qr/.*/ if (!defined $_[0] || $_[0] eq '*');
215 use Digest::SHA qw(sha1_hex);
218 $tmp =~ s#(?<!\\)\[(.+)(?<!\\)\]#
220 my $key = sha1_hex($orig . $salt);
221 $orig =~ s/([^\w\-])+/\Q$1/g;
225 my %map = ('*' => '.*', '?' => '.' );
226 $tmp =~ s#(?<!\\)([^\w\\])#$map{$1} || "\Q$1"#ge;
228 $tmp =~ s#([a-f0-9]{40})#
229 my $orig = $keep{$1};
230 defined $orig ? $orig : $1;
233 $_[0] = qr/\A$tmp\z/;
237 return $_[0] = qr/\A\z/ unless defined $_[0];
238 my %map = ('*' => '.*', ',' => '|');
239 $_[0] =~ s!(.)!$map{$1} || "\Q$1"!ge;
240 $_[0] = qr/\A(?:$_[0])\z/;
243 sub cmd_newnews ($$$$;$$) {
244 my ($self, $newsgroups, $date, $time, $gmt, $dists) = @_;
245 my $ts = eval { parse_time($date, $time, $gmt) };
247 more($self, '230 list of new articles by message-id follows');
248 my ($keep, $skip) = split('!', $newsgroups, 2);
252 foreach my $ng (@{$self->{nntpd}->{grouplist}}) {
253 $ng->{name} =~ $keep or next;
254 $ng->{name} =~ $skip and next;
255 my $srch = $ng->search or next;
258 return '.' unless @srch;
261 my $opts = { asc => 1, limit => 1000, offset => 0 };
262 $self->long_response(0, long_response_limit, sub {
265 my $res = $srch->query($ts, $opts);
266 my $msgs = $res->{msgs};
267 if (my $nr = scalar @$msgs) {
269 join(">\r\n<", map { $_->mid } @$msgs ).
271 $opts->{offset} += $nr;
274 if (@srch) { # continue onto next newsgroup
276 } else { # break out of the long response.
277 $$i = long_response_limit;
284 my ($self, $group) = @_;
285 my $no_such = '411 no such news group';
286 my $ng = $self->{nntpd}->{groups}->{$group} or return $no_such;
289 my ($min, $max) = $ng->mm->minmax;
292 $self->{article} = $min;
293 my $est_size = $max - $min;
294 "211 $est_size $min $max $group";
297 sub article_adj ($$) {
298 my ($self, $off) = @_;
299 my $ng = $self->{ng} or return '412 no newsgroup selected';
301 my $n = $self->{article};
302 defined $n or return '420 no current article has been selected';
305 my $mid = $ng->mm->mid_for($n);
307 $n = $off > 0 ? 'next' : 'previous';
308 return "421 no $n article in this group";
310 $self->{article} = $n;
311 "223 $n <$mid> article retrieved - request text separately";
314 sub cmd_next ($) { article_adj($_[0], 1) }
315 sub cmd_last ($) { article_adj($_[0], -1) }
317 # We want to encourage using email and CC-ing everybody involved to avoid
318 # the single-point-of-failure a single server provides.
321 my $ng = $self->{ng};
322 $ng ? "440 mailto:$ng->{address} to post" : '440 posting not allowed'
327 res($self, '205 closing connection - goodbye!');
332 sub art_lookup ($$$) {
333 my ($self, $art, $set_headers) = @_;
334 my $ng = $self->{ng};
338 if ($art =~ /\A\d+\z/o) {
339 $err = '423 no such article number in this group';
342 } elsif ($art =~ /\A<([^>]+)>\z/) {
344 $err = '430 no such article found';
345 $n = $ng->mm->num_for($mid) if $ng;
346 goto found if defined $n;
347 foreach my $g (values %{$self->{nntpd}->{groups}}) {
348 $n = $g->mm->num_for($mid);
359 $err = '420 no current article has been selected';
360 $n = $self->{article};
361 defined $n or return $err;
363 $ng or return '412 no newsgroup has been selected';
364 $mid = $ng->mm->mid_for($n);
365 defined $mid or return $err;
368 my $o = 'HEAD:' . mid2path($mid);
369 my $s = eval { Email::Simple->new($ng->gcf->cat_file($o)) };
370 return $err unless $s;
372 $s->header_set('Newsgroups', $ng->{name});
373 $s->header_set('Lines', $s->body =~ tr!\n!\n!);
374 $s->header_set('Xref', "$ng->{domain} $ng->{name}:$n");
377 if ($set_headers == 2) {
378 $s->header_set('Bytes', bytes::length($s->as_string));
385 sub simple_body_write ($$) {
389 $body =~ s/^\./../smg;
390 do_more($self, $body);
396 my $h = $s->header_obj;
397 $h->header_set('Bytes');
402 my ($self, $art) = @_;
403 $self->{article} = $art if defined $art && $art =~ /\A\d+\z/;
406 sub cmd_article ($;$) {
407 my ($self, $art) = @_;
408 my $r = $self->art_lookup($art, 1);
409 return $r unless ref $r;
410 my ($n, $mid, $s) = @$r;
411 set_art($self, $art);
412 more($self, "220 $n <$mid> article retrieved - head and body follow");
413 do_more($self, header_str($s));
414 do_more($self, "\r\n");
415 simple_body_write($self, $s);
419 my ($self, $art) = @_;
420 my $r = $self->art_lookup($art, 2);
421 return $r unless ref $r;
422 my ($n, $mid, $s) = @$r;
423 set_art($self, $art);
424 more($self, "221 $n <$mid> article retrieved - head follows");
425 do_more($self, header_str($s));
430 my ($self, $art) = @_;
431 my $r = $self->art_lookup($art, 0);
432 return $r unless ref $r;
433 my ($n, $mid, $s) = @$r;
434 set_art($self, $art);
435 more($self, "222 $n <$mid> article retrieved - body follows");
436 simple_body_write($self, $s);
440 my ($self, $art) = @_;
441 my $r = $self->art_lookup($art, 0);
442 return $r unless ref $r;
443 my ($n, $mid, undef) = @$r;
444 set_art($self, $art);
445 "223 $n <$mid> article retrieved - request text separately";
448 sub cmd_ihave ($) { '435 article not wanted - do not send it' }
450 sub cmd_date ($) { '111 '.strftime('%Y%m%d%H%M%S', gmtime(time)) }
454 more($self, '100 help text follows');
459 my ($self, $range) = @_;
460 my $ng = $self->{ng} or return '412 no news group has been selected';
461 defined $range or return '420 No article(s) selected';
463 my ($min, $max) = $ng->mm->minmax;
464 if ($range =~ /\A(\d+)\z/) {
466 } elsif ($range =~ /\A(\d+)-\z/) {
467 ($beg, $end) = ($1, $max);
468 } elsif ($range =~ /\A(\d+)-(\d+)\z/) {
469 ($beg, $end) = ($1, $2);
473 $beg = $min if ($beg < $min);
474 $end = $max if ($end > $max);
475 return '420 No article(s) selected' if ($beg > $end);
480 my ($r, $header) = @_;
481 $r = $r->[2]->header_obj->header($header);
482 defined $r or return;
483 $r =~ s/[\r\n\t]+/ /sg;
487 sub long_response ($$$$) {
488 my ($self, $beg, $end, $cb) = @_;
489 die "BUG: nested long response" if $self->{long_res};
491 # make sure we disable reading during a long response,
492 # clients should not be sending us stuff and making us do more
493 # work while we are stream a response to them
494 $self->watch_read(0);
495 my $fd = fileno $self->{sock};
497 $self->{long_res} = sub {
498 # limit our own running time for fairness with other
499 # clients and to avoid buffering too much:
501 local $SIG{ALRM} = sub { $yield = 1 };
506 eval { $cb->(\$beg) };
507 } until (($err = $@) || $self->{closed} || $yield ||
508 $self->{write_buf_size} || ++$beg > $end);
511 if ($err || $self->{closed}) {
512 $self->{long_res} = undef;
516 "$err during long response[$fd] - %0.6f",
519 if ($self->{closed}) {
520 out($self, " deferred[$fd] aborted - %0.6f",
523 $self->watch_read(1);
525 } elsif ($yield || $self->{write_buf_size}) {
526 # no recursion, schedule another call ASAP
527 # but only after all pending writes are done
528 Danga::Socket->AddTimer(0, sub {
529 $self->write($self->{long_res});
532 $self->{long_res} = undef;
533 $self->watch_read(1);
535 out($self, " deferred[$fd] done - %0.6f", now() - $t0);
538 $self->{long_res}->(); # kick off!
542 sub xhdr_message_id ($$) { # optimize XHDR Message-ID [range] for slrnpull.
543 my ($self, $range) = @_;
545 my $mm = $self->{ng}->mm;
546 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
547 my $n = $mm->num_for($range);
548 more($self, '221 Header follows');
549 more($self, "<$range> <$range>") if defined $n;
551 } else { # numeric range
552 $range = $self->{article} unless defined $range;
553 my $r = get_range($self, $range);
554 return $r unless ref $r;
555 my ($beg, $end) = @$r;
556 more($self, '221 Header follows');
557 $self->long_response($beg, $end, sub {
559 my $mid = $mm->mid_for($$i);
560 more($self, "$$i <$mid>") if defined $mid;
566 my ($srch, $mid) = @_;
568 my $smsg = $srch->lookup_message($mid);
569 $smsg = PublicInbox::SearchMsg->load_doc($smsg->{doc});
570 $smsg->mini_mime->header_obj;
574 sub xhdr_searchmsg ($$$) {
575 my ($self, $sub, $range) = @_;
576 my $srch = $self->{ng}->search;
577 my $emit = ($sub eq 'date') ? sub {
579 my @t = gmtime($m->header('X-PI-TS'));
580 more($self, "$pfx ". strftime('%a, %d %b %Y %T %z', @t));
583 my $h = $m->header($sub);
584 more($self, "$pfx $h") if defined $h;
587 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
588 more($self, '221 Header follows');
589 my $m = header_obj_for($srch, $1);
590 $emit->($range, $m) if defined $m;
592 } else { # numeric range
593 $range = $self->{article} unless defined $range;
594 my $mm = $self->{ng}->mm;
595 my $r = get_range($self, $range);
596 return $r unless ref $r;
597 my ($beg, $end) = @$r;
598 more($self, '221 Header follows');
599 $self->long_response($beg, $end, sub {
601 my $mid = $mm->mid_for($$i) or return;
602 my $m = header_obj_for($srch, $mid) or return;
608 sub cmd_xhdr ($$;$) {
609 my ($self, $header, $range) = @_;
610 my $ng = $self->{ng};
611 defined $ng or return '412 no news group currently selected';
612 my $sub = lc $header;
613 if ($sub eq 'message-id') {
614 xhdr_message_id($self, $range);
615 } elsif ($sub =~ /\A(subject|references|date)\z/ && $ng->search) {
616 xhdr_searchmsg($self, $sub, $range);
618 xhdr_slow($self, $header, $range);
622 sub xhdr_slow ($$$) {
623 my ($self, $header, $range) = @_;
625 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
626 my $r = $self->art_lookup($range, 2);
627 return $r unless ref $r;
628 more($self, '221 Header follows');
629 if (defined($r = xhdr($r, $header))) {
630 more($self, "<$range> $r");
633 } else { # numeric range
634 $range = $self->{article} unless defined $range;
635 my $r = get_range($self, $range);
636 return $r unless ref $r;
637 my ($beg, $end) = @$r;
638 more($self, '221 Header follows');
639 $self->long_response($beg, $end, sub {
641 $r = $self->art_lookup($$i, 2);
642 return unless ref $r;
643 defined($r = xhdr($r, $header)) or return;
644 more($self, "$$i $r");
649 sub cmd_xrover ($;$) {
650 my ($self, $range) = @_;
651 my $ng = $self->{ng} or return '412 no newsgroup selected';
652 (defined $range && $range =~ /[<>]/) and
653 return '420 No article(s) selected'; # no message IDs
655 $range = $self->{article} unless defined $range;
656 my $r = get_range($self, $range);
657 return $r unless ref $r;
658 my ($beg, $end) = @$r;
660 my $srch = $ng->search;
661 more($self, '224 Overview information follows');
662 $self->long_response($beg, $end, sub {
664 my $mid = $mm->mid_for($$i) or return;
665 my $m = header_obj_for($srch, $mid) or return;
666 my $h = $m->header('references');
667 more($self, "$$i $h") if defined $h;
671 sub cmd_xover ($;$) {
672 my ($self, $range) = @_;
673 $range = $self->{article} unless defined $range;
674 my $r = get_range($self, $range);
675 return $r unless ref $r;
676 my ($beg, $end) = @$r;
677 more($self, "224 Overview information follows for $beg to $end");
678 $self->long_response($beg, $end, sub {
680 my $r = $self->art_lookup($$i, 2);
681 return unless ref $r;
682 more($self, join("\t", $r->[0],
684 my $h = xhdr($r, $_);
685 defined $h ? $h : '';
691 my ($self, $mid) = @_;
692 return r501 unless $mid =~ /\A<(.+)>\z/;
695 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
696 my $n = $ng->mm->num_for($mid);
697 push @paths, "$ng->{name}/$n" if defined $n;
699 return '430 no such article on server' unless @paths;
700 '223 '.join(' ', @paths);
704 my ($self, $line) = @_;
705 do_write($self, $line . "\r\n");
709 my ($self, $line) = @_;
710 do_more($self, $line . "\r\n");
714 my ($self, $data) = @_;
715 my $done = $self->write($data);
716 die if $self->{closed};
718 # Do not watch for readability if we have data in the queue,
719 # instead re-enable watching for readability when we can
720 $self->watch_read(0) if (!$done || $self->{long_res});
726 my ($self, $fmt, @args) = @_;
727 printf { $self->{nntpd}->{err} } $fmt."\n", @args;
731 my ($self, $fmt, @args) = @_;
732 printf { $self->{nntpd}->{out} } $fmt."\n", @args;
735 use constant MSG_MORE => ($^O eq 'linux') ? 0x8000 : 0;
738 my ($self, $data) = @_;
739 if (MSG_MORE && !$self->{write_buf_size}) {
740 my $n = send($self->{sock}, $data, MSG_MORE);
742 my $dlen = bytes::length($data);
743 return 1 if $n == $dlen; # all done!
744 $data = bytes::substr($data, $n, $dlen - $n);
747 $self->do_write($data);
750 # callbacks for by Danga::Socket
752 sub event_hup { $_[0]->close }
753 sub event_err { $_[0]->close }
757 # only continue watching for readability when we are done writing:
758 if ($self->write(undef) == 1 && !$self->{long_res}) {
759 $self->watch_read(1);
765 use constant LINE_MAX => 512; # RFC 977 section 2.3
769 while ($r > 0 && $self->{rbuf} =~ s/\A\s*([^\r\n]+)\r?\n//) {
772 $r = eval { $self->process_line($line) };
773 my $d = $self->{long_res} ?
774 ' deferred['.fileno($self->{sock}).']' : '';
775 out($self, "$line - %0.6f$d", now() - $t0);
777 unless (defined $line) {
778 my $buf = $self->read(LINE_MAX) or return $self->close;
779 $self->{rbuf} .= $$buf;
783 return $self->close if $r < 0;
784 my $len = bytes::length($self->{rbuf});
785 return $self->close if ($len >= LINE_MAX);
789 my ($self, $bool) = @_;
790 my $rv = $self->SUPER::watch_read($bool);
791 if ($bool && $self->{rbuf} ne '') {
792 # Force another read if there is a pipelined request.
793 # We don't know if the socket has anything for us to read,
794 # and we must double-check again by the time the timer fires
795 # in case we really did dispatch a read event and started
796 # another long response.
797 Danga::Socket->AddTimer(0, sub {
798 if (&Danga::Socket::POLLIN & $self->{event_watch}) {