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(gettimeofday tv_interval ualarm);
16 r501 => '501 command syntax error',
17 long_response_limit => 0xffffffff,
20 my @OVERVIEW = qw(Subject From Date Message-ID References Bytes Lines);
21 my %OVERVIEW = map { $_ => 1 } @OVERVIEW;
23 # disable commands with easy DoS potential:
24 # LISTGROUP could get pretty bad, too...
25 my %DISABLED; # = map { $_ => 1 } qw(xover list_overview_fmt newnews xhdr);
28 my ($class, $sock, $nntpd) = @_;
29 my $self = fields::new($class);
30 $self->SUPER::new($sock);
31 $self->{nntpd} = $nntpd;
32 res($self, '201 server ready - post via email');
37 # returns 1 if we can continue, 0 if not due to buffered writes or disconnect
40 my ($req, @args) = split(/\s+/, $l);
44 $req = $DISABLED{$req} ? undef : *{'cmd_'.$req}{CODE};
46 return res($self, '500 command not recognized') unless $req;
48 my $res = eval { $req->($self, @args) };
50 if ($err && !$self->{closed}) {
51 chomp($l = Dumper(\$l));
52 warning('error from: ', $l, ' ', $err);
53 $res = '503 program fault - command not performed';
55 return 0 unless defined $res;
60 my ($self, $arg) = @_;
61 return r501 unless defined $arg;
63 return r501 unless $arg eq 'READER';
64 '200 reader status acknowledged';
70 '202 slave status noted';
74 my ($self, $wildmat) = @_;
75 more($self, '282 list of groups and descriptions follows');
76 list_newsgroups($self, $wildmat);
80 sub list_overview_fmt {
82 more($self, $_ . ':') foreach @OVERVIEW;
86 my ($self, $wildmat) = @_;
88 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
89 $ng->{name} =~ $wildmat or next;
90 group_line($self, $ng);
94 sub list_active_times {
95 my ($self, $wildmat) = @_;
97 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
98 $ng->{name} =~ $wildmat or next;
99 my $c = eval { $ng->mm->created_at } || time;
100 more($self, "$ng->{name} $c $ng->{address}");
104 sub list_newsgroups {
105 my ($self, $wildmat) = @_;
106 wildmat2re($wildmat);
107 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
108 $ng->{name} =~ $wildmat or next;
109 my $d = $ng->description;
110 more($self, "$ng->{name} $d");
114 # LIST SUBSCRIPTIONS not supported
116 my ($self, $arg, $wildmat, @x) = @_;
121 return '503 function not performed' if $DISABLED{$arg};
126 return r501 unless $arg;
127 more($self, '215 information follows');
128 $arg->($self, $wildmat, @x);
130 more($self, '215 list of newsgroups follows');
131 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
132 group_line($self, $ng);
139 my ($self, $group) = @_;
140 if (defined $group) {
141 my $res = cmd_group($self, $group);
142 return $res if ($res !~ /\A211 /);
146 $self->{ng} or return '412 no newsgroup selected';
147 $self->long_response(0, long_response_limit, sub {
149 my $nr = $self->{ng}->mm->id_batch($$i, sub {
151 more($self, join("\r\n", @$ary));
154 # -1 to adjust for implicit increment in long_response
155 $$i = $nr ? $$i + $nr - 1 : long_response_limit;
160 my ($date, $time, $gmt) = @_;
161 use Time::Local qw();
162 my ($YY, $MM, $DD) = unpack('A2A2A2', $date);
163 my ($hh, $mm, $ss) = unpack('A2A2A2', $time);
165 $gmt =~ /\A(?:UTC|GMT)\z/i or die "GM invalid: $gmt\n";
168 my @now = $gmt ? gmtime : localtime;
169 if ($YY > strftime('%y', @now)) {
170 my $cur_year = $now[5] + 1900;
171 $YY += int($cur_year / 1000) * 1000 - 100;
175 Time::Local::timegm($ss, $mm, $hh, $DD, $MM - 1, $YY);
177 Time::Local::timelocal($ss, $mm, $hh, $DD, $MM - 1, $YY);
182 my ($self, $ng) = @_;
183 my ($min, $max) = $ng->mm->minmax;
184 more($self, "$ng->{name} $max $min n") if defined $min && defined $max;
188 my ($self, $date, $time, $gmt, $dists) = @_;
189 my $ts = eval { parse_time($date, $time, $gmt) };
193 more($self, '231 list of new newsgroups follows');
194 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
195 my $c = eval { $ng->mm->created_at } || 0;
196 next unless $c > $ts;
197 group_line($self, $ng);
203 return $_[0] = qr/.*/ if (!defined $_[0] || $_[0] eq '*');
206 use Digest::SHA qw(sha1_hex);
209 $tmp =~ s#(?<!\\)\[(.+)(?<!\\)\]#
211 my $key = sha1_hex($orig . $salt);
212 $orig =~ s/([^\w\-])+/\Q$1/g;
216 my %map = ('*' => '.*', '?' => '.' );
217 $tmp =~ s#(?<!\\)([^\w\\])#$map{$1} || "\Q$1"#ge;
219 $tmp =~ s#([a-f0-9]{40})#
220 my $orig = $keep{$1};
221 defined $orig ? $orig : $1;
224 $_[0] = qr/\A$tmp\z/;
228 return $_[0] = qr/\A\z/ unless defined $_[0];
229 my %map = ('*' => '.*', ',' => '|');
230 $_[0] =~ s!(.)!$map{$1} || "\Q$1"!ge;
231 $_[0] = qr/\A(?:$_[0])\z/;
235 my ($self, $newsgroups, $date, $time, $gmt, $dists) = @_;
236 my $ts = eval { parse_time($date, $time, $gmt) };
238 more($self, '230 list of new articles by message-id follows');
239 my ($keep, $skip) = split('!', $newsgroups, 2);
243 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
244 $ng->{name} =~ $keep or next;
245 $ng->{name} =~ $skip and next;
246 my $srch = $ng->search or next;
249 return '.' unless @srch;
252 my $opts = { asc => 1, limit => 1000, offset => 0 };
253 $self->long_response(0, long_response_limit, sub {
256 my $res = $srch->query($ts, $opts);
257 my $msgs = $res->{msgs};
258 if (my $nr = scalar @$msgs) {
260 join(">\r\n<", map { $_->mid } @$msgs ).
262 $opts->{offset} += $nr;
265 if (@srch) { # continue onto next newsgroup
267 } else { # break out of the long response.
268 $$i = long_response_limit;
275 my ($self, $group) = @_;
276 my $no_such = '411 no such news group';
277 my $ng = $self->{nntpd}->{groups}->{$group} or return $no_such;
280 my ($min, $max) = $ng->mm->minmax;
283 $self->{article} = $min;
284 my $est_size = $max - $min;
285 "211 $est_size $min $max $group";
289 my ($self, $off) = @_;
290 my $ng = $self->{ng} or return '412 no newsgroup selected';
292 my $n = $self->{article};
293 defined $n or return '420 no current article has been selected';
296 my $mid = $ng->mm->mid_for($n);
298 $n = $off > 0 ? 'next' : 'previous';
299 return "421 no $n article in this group";
301 $self->{article} = $n;
302 "223 $n <$mid> article retrieved - request text separately";
305 sub cmd_next { article_adj($_[0], 1) }
306 sub cmd_last { article_adj($_[0], -1) }
308 # We want to encourage using email and CC-ing everybody involved to avoid
309 # the single-point-of-failure a single server provides.
312 my $ng = $self->{ng};
313 $ng ? "440 mailto:$ng->{address} to post" : '440 posting not allowed'
318 res($self, '205 closing connection - goodbye!');
324 my ($self, $art, $set_headers) = @_;
325 my $ng = $self->{ng} or return '412 no newsgroup has been selected';
329 if ($art =~ /\A\d+\z/o) {
330 $err = '423 no such article number in this group';
333 } elsif ($art =~ /\A<([^>]+)>\z/) {
334 $err = '430 no such article found';
336 $n = $ng->mm->num_for($mid);
337 defined $mid or return $err;
342 $err = '420 no current article has been selected';
343 $n = $self->{article};
344 defined $n or return $err;
346 $mid = $ng->mm->mid_for($n);
347 defined $mid or return $err;
350 my $o = 'HEAD:' . mid2path($mid);
351 my $s = eval { Email::Simple->new($ng->gcf->cat_file($o)) };
352 return $err unless $s;
354 $s->header_set('Newsgroups', $ng->{name});
355 $s->header_set('Lines', $s->body =~ tr!\n!\n!);
356 $s->header_set('Xref', "$ng->{domain} $ng->{name}:$n");
359 if ($set_headers == 2) {
360 $s->header_set('Bytes', bytes::length($s->as_string));
367 sub simple_body_write {
371 $body =~ s/^\./../smg;
372 do_more($self, $body);
378 my $h = $s->header_obj;
379 $h->header_set('Bytes');
384 my ($self, $art) = @_;
385 my $r = $self->art_lookup($art, 1);
386 return $r unless ref $r;
387 my ($n, $mid, $s) = @$r;
388 more($self, "220 $n <$mid> article retrieved - head and body follow");
389 do_more($self, header_str($s));
390 do_more($self, "\r\n");
391 simple_body_write($self, $s);
395 my ($self, $art) = @_;
396 my $r = $self->art_lookup($art, 2);
397 return $r unless ref $r;
398 my ($n, $mid, $s) = @$r;
399 more($self, "221 $n <$mid> article retrieved - head follows");
400 do_more($self, header_str($s));
405 my ($self, $art) = @_;
406 my $r = $self->art_lookup($art, 0);
407 return $r unless ref $r;
408 my ($n, $mid, $s) = @$r;
409 more($self, "222 $n <$mid> article retrieved - body follows");
410 simple_body_write($self, $s);
414 my ($self, $art) = @_;
415 my $r = $self->art_lookup($art, 0);
416 return $r unless ref $r;
417 my ($n, $mid, undef) = @$r;
418 "223 $n <$mid> article retrieved - request text separately";
421 sub cmd_ihave { '435 article not wanted - do not send it' }
423 sub cmd_date { '111 '.strftime('%Y%m%d%H%M%S', gmtime(time)) }
427 more($self, '100 help text follows');
432 my ($self, $range) = @_;
433 my $ng = $self->{ng} or return '412 no news group has been selected';
434 defined $range or return '420 No article(s) selected';
436 my ($min, $max) = $ng->mm->minmax;
437 if ($range =~ /\A(\d+)\z/) {
439 } elsif ($range =~ /\A(\d+)-\z/) {
440 ($beg, $end) = ($1, $max);
441 } elsif ($range =~ /\A(\d+)-(\d+)\z/) {
442 ($beg, $end) = ($1, $2);
446 $beg = $min if ($beg < $min);
447 $end = $max if ($end > $max);
448 return '420 No article(s) selected' if ($beg > $end);
453 my ($r, $header) = @_;
454 $r = $r->[2]->header_obj->header($header);
455 defined $r or return;
456 $r =~ s/[\r\n\t]+/ /sg;
461 my ($self, $beg, $end, $cb) = @_;
462 die "BUG: nested long response" if $self->{long_res};
464 # make sure we disable reading during a long response,
465 # clients should not be sending us stuff and making us do more
466 # work while we are stream a response to them
467 $self->watch_read(0);
468 $self->{long_res} = sub {
469 # limit our own running time for fairness with other
470 # clients and to avoid buffering too much:
472 local $SIG{ALRM} = sub { $yield = 1 };
477 eval { $cb->(\$beg) };
478 } until (($err = $@) || $self->{closed} || $yield ||
479 $self->{write_buf_size} || ++$beg > $end);
482 if ($err || $self->{closed}) {
483 $self->{long_res} = undef;
484 warning("$err during long response") if $err;
485 $self->watch_read(1) unless $self->{closed};
486 } elsif ($yield || $self->{write_buf_size}) {
487 # no recursion, schedule another call ASAP
488 # but only after all pending writes are done
489 Danga::Socket->AddTimer(0, sub {
490 $self->write($self->{long_res});
493 $self->{long_res} = undef;
494 $self->watch_read(1);
498 $self->{long_res}->(); # kick off!
503 my ($self, $header, $range) = @_;
504 defined $self->{ng} or return '412 no news group currently selected';
505 unless (defined $range) {
506 defined($range = $self->{article}) or
507 return '420 no current article has been selected';
509 if ($range =~ /\A<(.+)>\z/) { # Message-ID
510 my $r = $self->art_lookup($range, 2);
511 return $r unless ref $r;
512 more($self, '221 Header follows');
513 if (defined($r = xhdr($r, $header))) {
514 more($self, "<$range> $r");
517 } else { # numeric range
518 my $r = get_range($self, $range);
519 return $r unless ref $r;
520 my ($beg, $end) = @$r;
521 more($self, '221 Header follows');
522 $self->long_response($beg, $end, sub {
524 $r = $self->art_lookup($$i, 2);
525 return unless ref $r;
526 defined($r = xhdr($r, $header)) or return;
527 more($self, "$$i $r");
533 my ($self, $range) = @_;
534 my $r = get_range($self, $range);
535 return $r unless ref $r;
536 my ($beg, $end) = @$r;
537 more($self, "224 Overview information follows for $beg to $end");
538 $self->long_response($beg, $end, sub {
540 my $r = $self->art_lookup($$i, 2);
541 return unless ref $r;
542 more($self, join("\t", $r->[0],
544 my $h = xhdr($r, $_);
545 defined $h ? $h : '';
551 my ($self, $line) = @_;
552 do_write($self, $line . "\r\n");
556 my ($self, $line) = @_;
557 do_more($self, $line . "\r\n");
561 my ($self, $data) = @_;
562 my $done = $self->write($data);
563 die if $self->{closed};
565 # Do not watch for readability if we have data in the queue,
566 # instead re-enable watching for readability when we can
567 $self->watch_read(0) if (!$done || $self->{long_res});
572 use constant MSG_MORE => ($^O eq 'linux') ? 0x8000 : 0;
575 my ($self, $data) = @_;
576 if (MSG_MORE && !$self->{write_buf_size}) {
577 my $n = send($self->{sock}, $data, MSG_MORE);
579 my $dlen = bytes::length($data);
580 return 1 if $n == $dlen; # all done!
581 $data = bytes::substr($data, $n, $dlen - $n);
584 $self->do_write($data);
587 # callbacks for by Danga::Socket
589 sub event_hup { $_[0]->close }
590 sub event_err { $_[0]->close }
594 # only continue watching for readability when we are done writing:
595 $self->write(undef) == 1 and $self->watch_read(1);
600 use constant LINE_MAX => 512; # RFC 977 section 2.3
602 my $buf = $self->read(LINE_MAX) or return $self->close;
603 while ($r > 0 && $$buf =~ s/\A([^\r\n]+)\r?\n//) {
605 my $t0 = [ gettimeofday ];
606 $r = eval { $self->process_line($line) };
607 printf(STDERR "$line %0.6f\n",
608 tv_interval($t0, [gettimeofday]));
610 return $self->close if $r < 0;
611 my $len = bytes::length($$buf);
612 return $self->close if ($len >= LINE_MAX);
613 $self->push_back_read($buf) if ($len);
616 sub warning { print STDERR @_, "\n" }