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 ($$) {
51 my ($req, @args) = split(/\s+/, $l);
55 $req = $DISABLED{$req} ? undef : *{'cmd_'.$req}{CODE};
57 return res($self, '500 command not recognized') unless $req;
58 return res($self, r501) unless args_ok($req, scalar @args);
60 my $res = eval { $req->($self, @args) };
62 if ($err && !$self->{closed}) {
63 chomp($l = Dumper(\$l));
64 warning('error from: ', $l, ' ', $err);
65 $res = '503 program fault - command not performed';
67 return 0 unless defined $res;
72 my ($self, $arg) = @_;
74 return r501 unless $arg eq 'READER';
75 '200 reader status acknowledged';
78 sub cmd_slave ($) { '202 slave status noted' }
80 sub cmd_xgtitle ($;$) {
81 my ($self, $wildmat) = @_;
82 more($self, '282 list of groups and descriptions follows');
83 list_newsgroups($self, $wildmat);
87 sub list_overview_fmt ($$) {
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 ($$$) {
332 my ($self, $art, $set_headers) = @_;
333 my $ng = $self->{ng} or return '412 no newsgroup has been selected';
337 if ($art =~ /\A\d+\z/o) {
338 $err = '423 no such article number in this group';
341 } elsif ($art =~ /\A<([^>]+)>\z/) {
342 $err = '430 no such article found';
344 $n = $ng->mm->num_for($mid);
345 defined $mid or return $err;
350 $err = '420 no current article has been selected';
351 $n = $self->{article};
352 defined $n or return $err;
354 $mid = $ng->mm->mid_for($n);
355 defined $mid or return $err;
358 my $o = 'HEAD:' . mid2path($mid);
359 my $s = eval { Email::Simple->new($ng->gcf->cat_file($o)) };
360 return $err unless $s;
362 $s->header_set('Newsgroups', $ng->{name});
363 $s->header_set('Lines', $s->body =~ tr!\n!\n!);
364 $s->header_set('Xref', "$ng->{domain} $ng->{name}:$n");
367 if ($set_headers == 2) {
368 $s->header_set('Bytes', bytes::length($s->as_string));
375 sub simple_body_write ($$) {
379 $body =~ s/^\./../smg;
380 do_more($self, $body);
386 my $h = $s->header_obj;
387 $h->header_set('Bytes');
392 my ($self, $art) = @_;
393 $self->{article} = $art if defined $art && $art =~ /\A\d+\z/;
396 sub cmd_article ($;$) {
397 my ($self, $art) = @_;
398 my $r = $self->art_lookup($art, 1);
399 return $r unless ref $r;
400 my ($n, $mid, $s) = @$r;
401 set_art($self, $art);
402 more($self, "220 $n <$mid> article retrieved - head and body follow");
403 do_more($self, header_str($s));
404 do_more($self, "\r\n");
405 simple_body_write($self, $s);
409 my ($self, $art) = @_;
410 my $r = $self->art_lookup($art, 2);
411 return $r unless ref $r;
412 my ($n, $mid, $s) = @$r;
413 set_art($self, $art);
414 more($self, "221 $n <$mid> article retrieved - head follows");
415 do_more($self, header_str($s));
420 my ($self, $art) = @_;
421 my $r = $self->art_lookup($art, 0);
422 return $r unless ref $r;
423 my ($n, $mid, $s) = @$r;
424 set_art($self, $art);
425 more($self, "222 $n <$mid> article retrieved - body follows");
426 simple_body_write($self, $s);
430 my ($self, $art) = @_;
431 my $r = $self->art_lookup($art, 0);
432 return $r unless ref $r;
433 my ($n, $mid, undef) = @$r;
434 set_art($self, $art);
435 "223 $n <$mid> article retrieved - request text separately";
438 sub cmd_ihave ($) { '435 article not wanted - do not send it' }
440 sub cmd_date ($) { '111 '.strftime('%Y%m%d%H%M%S', gmtime(time)) }
444 more($self, '100 help text follows');
449 my ($self, $range) = @_;
450 my $ng = $self->{ng} or return '412 no news group has been selected';
451 defined $range or return '420 No article(s) selected';
453 my ($min, $max) = $ng->mm->minmax;
454 if ($range =~ /\A(\d+)\z/) {
456 } elsif ($range =~ /\A(\d+)-\z/) {
457 ($beg, $end) = ($1, $max);
458 } elsif ($range =~ /\A(\d+)-(\d+)\z/) {
459 ($beg, $end) = ($1, $2);
463 $beg = $min if ($beg < $min);
464 $end = $max if ($end > $max);
465 return '420 No article(s) selected' if ($beg > $end);
470 my ($r, $header) = @_;
471 $r = $r->[2]->header_obj->header($header);
472 defined $r or return;
473 $r =~ s/[\r\n\t]+/ /sg;
477 sub long_response ($$$$) {
478 my ($self, $beg, $end, $cb) = @_;
479 die "BUG: nested long response" if $self->{long_res};
481 # make sure we disable reading during a long response,
482 # clients should not be sending us stuff and making us do more
483 # work while we are stream a response to them
484 $self->watch_read(0);
485 my $fd = fileno $self->{sock};
487 $self->{long_res} = sub {
488 # limit our own running time for fairness with other
489 # clients and to avoid buffering too much:
491 local $SIG{ALRM} = sub { $yield = 1 };
496 eval { $cb->(\$beg) };
497 } until (($err = $@) || $self->{closed} || $yield ||
498 $self->{write_buf_size} || ++$beg > $end);
501 if ($err || $self->{closed}) {
502 $self->{long_res} = undef;
505 warning("$err during long response[$fd] - ".
506 sprintf('%0.6', now() - $t0));
508 if ($self->{closed}) {
510 " deferred[$fd] aborted - %0.6f\n",
513 $self->watch_read(1);
515 } elsif ($yield || $self->{write_buf_size}) {
516 # no recursion, schedule another call ASAP
517 # but only after all pending writes are done
518 Danga::Socket->AddTimer(0, sub {
519 $self->write($self->{long_res});
522 $self->{long_res} = undef;
523 $self->watch_read(1);
525 printf(STDERR " deferred[$fd] done - %0.6f\n",
529 $self->{long_res}->(); # kick off!
533 sub xhdr_message_id ($$) { # optimize XHDR Message-ID [range] for slrnpull.
534 my ($self, $range) = @_;
536 my $mm = $self->{ng}->mm;
537 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
538 my $n = $mm->num_for($range);
539 more($self, '221 Header follows');
540 more($self, "<$range> <$range>") if defined $n;
542 } else { # numeric range
543 $range = $self->{article} unless defined $range;
544 my $r = get_range($self, $range);
545 return $r unless ref $r;
546 my ($beg, $end) = @$r;
547 more($self, '221 Header follows');
548 $self->long_response($beg, $end, sub {
550 my $mid = $mm->mid_for($$i);
551 more($self, "$$i <$mid>") if defined $mid;
557 my ($srch, $mid) = @_;
559 my $smsg = $srch->lookup_message($mid);
560 $smsg = PublicInbox::SearchMsg->load_doc($smsg->{doc});
561 $smsg->mini_mime->header_obj;
565 sub xhdr_searchmsg ($$$) {
566 my ($self, $sub, $range) = @_;
567 my $srch = $self->{ng}->search;
568 my $emit = ($sub eq 'date') ? sub {
570 my @t = gmtime($m->header('X-PI-TS'));
571 more($self, "$pfx ". strftime('%a, %d %b %Y %T %z', @t));
574 my $h = $m->header($sub);
575 more($self, "$pfx $h") if defined $h;
578 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
579 more($self, '221 Header follows');
580 my $m = header_obj_for($srch, $1);
581 $emit->($range, $m) if defined $m;
583 } else { # numeric range
584 $range = $self->{article} unless defined $range;
585 my $mm = $self->{ng}->mm;
586 my $r = get_range($self, $range);
587 return $r unless ref $r;
588 my ($beg, $end) = @$r;
589 more($self, '221 Header follows');
590 $self->long_response($beg, $end, sub {
592 my $mid = $mm->mid_for($$i) or return;
593 my $m = header_obj_for($srch, $mid) or return;
599 sub cmd_xhdr ($$;$) {
600 my ($self, $header, $range) = @_;
601 my $ng = $self->{ng};
602 defined $ng or return '412 no news group currently selected';
603 my $sub = lc $header;
604 if ($sub eq 'message-id') {
605 xhdr_message_id($self, $range);
606 } elsif ($sub =~ /\A(subject|references|date)\z/ && $ng->search) {
607 xhdr_searchmsg($self, $sub, $range);
609 xhdr_slow($self, $header, $range);
613 sub xhdr_slow ($$$) {
614 my ($self, $header, $range) = @_;
616 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
617 my $r = $self->art_lookup($range, 2);
618 return $r unless ref $r;
619 more($self, '221 Header follows');
620 if (defined($r = xhdr($r, $header))) {
621 more($self, "<$range> $r");
624 } else { # numeric range
625 $range = $self->{article} unless defined $range;
626 my $r = get_range($self, $range);
627 return $r unless ref $r;
628 my ($beg, $end) = @$r;
629 more($self, '221 Header follows');
630 $self->long_response($beg, $end, sub {
632 $r = $self->art_lookup($$i, 2);
633 return unless ref $r;
634 defined($r = xhdr($r, $header)) or return;
635 more($self, "$$i $r");
640 sub cmd_xrover ($;$) {
641 my ($self, $range) = @_;
642 my $ng = $self->{ng} or return '412 no newsgroup selected';
643 (defined $range && $range =~ /[<>]/) and
644 return '420 No article(s) selected'; # no message IDs
646 $range = $self->{article} unless defined $range;
647 my $r = get_range($self, $range);
648 return $r unless ref $r;
649 my ($beg, $end) = @$r;
651 my $srch = $ng->search;
652 more($self, '224 Overview information follows');
653 $self->long_response($beg, $end, sub {
655 my $mid = $mm->mid_for($$i) or return;
656 my $m = header_obj_for($srch, $mid) or return;
657 my $h = $m->header('references');
658 more($self, "$$i $h") if defined $h;
662 sub cmd_xover ($;$) {
663 my ($self, $range) = @_;
664 $range = $self->{article} unless defined $range;
665 my $r = get_range($self, $range);
666 return $r unless ref $r;
667 my ($beg, $end) = @$r;
668 more($self, "224 Overview information follows for $beg to $end");
669 $self->long_response($beg, $end, sub {
671 my $r = $self->art_lookup($$i, 2);
672 return unless ref $r;
673 more($self, join("\t", $r->[0],
675 my $h = xhdr($r, $_);
676 defined $h ? $h : '';
682 my ($self, $mid) = @_;
683 return r501 unless $mid =~ /\A<(.+)>\z/;
686 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
687 my $n = $ng->mm->num_for($mid);
688 push @paths, "$ng->{name}/$n" if defined $n;
690 return '430 no such article on server' unless @paths;
691 '223 '.join(' ', @paths);
695 my ($self, $line) = @_;
696 do_write($self, $line . "\r\n");
700 my ($self, $line) = @_;
701 do_more($self, $line . "\r\n");
705 my ($self, $data) = @_;
706 my $done = $self->write($data);
707 die if $self->{closed};
709 # Do not watch for readability if we have data in the queue,
710 # instead re-enable watching for readability when we can
711 $self->watch_read(0) if (!$done || $self->{long_res});
716 use constant MSG_MORE => ($^O eq 'linux') ? 0x8000 : 0;
719 my ($self, $data) = @_;
720 if (MSG_MORE && !$self->{write_buf_size}) {
721 my $n = send($self->{sock}, $data, MSG_MORE);
723 my $dlen = bytes::length($data);
724 return 1 if $n == $dlen; # all done!
725 $data = bytes::substr($data, $n, $dlen - $n);
728 $self->do_write($data);
731 # callbacks for by Danga::Socket
733 sub event_hup { $_[0]->close }
734 sub event_err { $_[0]->close }
738 # only continue watching for readability when we are done writing:
739 $self->write(undef) == 1 and $self->watch_read(1);
744 use constant LINE_MAX => 512; # RFC 977 section 2.3
746 my $buf = $self->read(LINE_MAX) or return $self->close;
747 while ($r > 0 && $$buf =~ s/\A\s*([^\r\n]+)\r?\n//) {
750 $r = eval { $self->process_line($line) };
751 my $d = $self->{long_res} ?
752 ' deferred['.fileno($self->{sock}).']' : '';
753 printf(STDERR "$line - %0.6f$d\n", now() - $t0);
755 return $self->close if $r < 0;
756 my $len = bytes::length($$buf);
757 return $self->close if ($len >= LINE_MAX);
758 $self->push_back_read($buf) if ($len);
761 sub warning { print STDERR 'W: ', @_, "\n" }