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');
39 my $tot = prototype $cb;
40 my ($nreq, undef) = split(';', $tot);
41 $nreq = ($nreq =~ tr/$//) - 1;
42 $tot = ($tot =~ tr/$//) - 1;
43 ($argc <= $tot && $argc >= $nreq);
46 # returns 1 if we can continue, 0 if not due to buffered writes or disconnect
47 sub process_line ($$) {
49 my ($req, @args) = split(/\s+/, $l);
53 $req = $DISABLED{$req} ? undef : *{'cmd_'.$req}{CODE};
55 return res($self, '500 command not recognized') unless $req;
56 return res($self, r501) unless args_ok($req, scalar @args);
58 my $res = eval { $req->($self, @args) };
60 if ($err && !$self->{closed}) {
61 chomp($l = Dumper(\$l));
62 warning('error from: ', $l, ' ', $err);
63 $res = '503 program fault - command not performed';
65 return 0 unless defined $res;
70 my ($self, $arg) = @_;
72 return r501 unless $arg eq 'READER';
73 '200 reader status acknowledged';
76 sub cmd_slave ($) { '202 slave status noted' }
78 sub cmd_xgtitle ($;$) {
79 my ($self, $wildmat) = @_;
80 more($self, '282 list of groups and descriptions follows');
81 list_newsgroups($self, $wildmat);
85 sub list_overview_fmt ($$) {
87 more($self, $_ . ':') foreach @OVERVIEW;
90 sub list_active ($;$) {
91 my ($self, $wildmat) = @_;
93 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
94 $ng->{name} =~ $wildmat or next;
95 group_line($self, $ng);
99 sub list_active_times ($;$) {
100 my ($self, $wildmat) = @_;
101 wildmat2re($wildmat);
102 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
103 $ng->{name} =~ $wildmat or next;
104 my $c = eval { $ng->mm->created_at } || time;
105 more($self, "$ng->{name} $c $ng->{address}");
109 sub list_newsgroups ($;$) {
110 my ($self, $wildmat) = @_;
111 wildmat2re($wildmat);
112 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
113 $ng->{name} =~ $wildmat or next;
114 my $d = $ng->description;
115 more($self, "$ng->{name} $d");
119 # LIST SUBSCRIPTIONS not supported
120 sub cmd_list ($;$$) {
121 my ($self, @args) = @_;
123 my $arg = shift @args;
124 $arg =~ tr/A-Z./a-z_/;
126 return '503 function not performed' if $DISABLED{$arg};
132 return r501 unless $arg && args_ok($arg, scalar @args);
133 more($self, '215 information follows');
134 $arg->($self, @args);
136 more($self, '215 list of newsgroups follows');
137 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
138 group_line($self, $ng);
144 sub cmd_listgroup ($;$) {
145 my ($self, $group) = @_;
146 if (defined $group) {
147 my $res = cmd_group($self, $group);
148 return $res if ($res !~ /\A211 /);
152 $self->{ng} or return '412 no newsgroup selected';
153 $self->long_response(0, long_response_limit, sub {
155 my $nr = $self->{ng}->mm->id_batch($$i, sub {
157 more($self, join("\r\n", @$ary));
160 # -1 to adjust for implicit increment in long_response
161 $$i = $nr ? $$i + $nr - 1 : long_response_limit;
165 sub parse_time ($$;$) {
166 my ($date, $time, $gmt) = @_;
167 use Time::Local qw();
168 my ($YY, $MM, $DD) = unpack('A2A2A2', $date);
169 my ($hh, $mm, $ss) = unpack('A2A2A2', $time);
171 $gmt =~ /\A(?:UTC|GMT)\z/i or die "GM invalid: $gmt\n";
174 my @now = $gmt ? gmtime : localtime;
175 if ($YY > strftime('%y', @now)) {
176 my $cur_year = $now[5] + 1900;
177 $YY += int($cur_year / 1000) * 1000 - 100;
181 Time::Local::timegm($ss, $mm, $hh, $DD, $MM - 1, $YY);
183 Time::Local::timelocal($ss, $mm, $hh, $DD, $MM - 1, $YY);
187 sub group_line ($$) {
188 my ($self, $ng) = @_;
189 my ($min, $max) = $ng->mm->minmax;
190 more($self, "$ng->{name} $max $min n") if defined $min && defined $max;
193 sub cmd_newgroups ($$$;$$) {
194 my ($self, $date, $time, $gmt, $dists) = @_;
195 my $ts = eval { parse_time($date, $time, $gmt) };
199 more($self, '231 list of new newsgroups follows');
200 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
201 my $c = eval { $ng->mm->created_at } || 0;
202 next unless $c > $ts;
203 group_line($self, $ng);
208 sub wildmat2re (;$) {
209 return $_[0] = qr/.*/ if (!defined $_[0] || $_[0] eq '*');
212 use Digest::SHA qw(sha1_hex);
215 $tmp =~ s#(?<!\\)\[(.+)(?<!\\)\]#
217 my $key = sha1_hex($orig . $salt);
218 $orig =~ s/([^\w\-])+/\Q$1/g;
222 my %map = ('*' => '.*', '?' => '.' );
223 $tmp =~ s#(?<!\\)([^\w\\])#$map{$1} || "\Q$1"#ge;
225 $tmp =~ s#([a-f0-9]{40})#
226 my $orig = $keep{$1};
227 defined $orig ? $orig : $1;
230 $_[0] = qr/\A$tmp\z/;
234 return $_[0] = qr/\A\z/ unless defined $_[0];
235 my %map = ('*' => '.*', ',' => '|');
236 $_[0] =~ s!(.)!$map{$1} || "\Q$1"!ge;
237 $_[0] = qr/\A(?:$_[0])\z/;
240 sub cmd_newnews ($$$$;$$) {
241 my ($self, $newsgroups, $date, $time, $gmt, $dists) = @_;
242 my $ts = eval { parse_time($date, $time, $gmt) };
244 more($self, '230 list of new articles by message-id follows');
245 my ($keep, $skip) = split('!', $newsgroups, 2);
249 foreach my $ng (values %{$self->{nntpd}->{groups}}) {
250 $ng->{name} =~ $keep or next;
251 $ng->{name} =~ $skip and next;
252 my $srch = $ng->search or next;
255 return '.' unless @srch;
258 my $opts = { asc => 1, limit => 1000, offset => 0 };
259 $self->long_response(0, long_response_limit, sub {
262 my $res = $srch->query($ts, $opts);
263 my $msgs = $res->{msgs};
264 if (my $nr = scalar @$msgs) {
266 join(">\r\n<", map { $_->mid } @$msgs ).
268 $opts->{offset} += $nr;
271 if (@srch) { # continue onto next newsgroup
273 } else { # break out of the long response.
274 $$i = long_response_limit;
281 my ($self, $group) = @_;
282 my $no_such = '411 no such news group';
283 my $ng = $self->{nntpd}->{groups}->{$group} or return $no_such;
286 my ($min, $max) = $ng->mm->minmax;
289 $self->{article} = $min;
290 my $est_size = $max - $min;
291 "211 $est_size $min $max $group";
294 sub article_adj ($$) {
295 my ($self, $off) = @_;
296 my $ng = $self->{ng} or return '412 no newsgroup selected';
298 my $n = $self->{article};
299 defined $n or return '420 no current article has been selected';
302 my $mid = $ng->mm->mid_for($n);
304 $n = $off > 0 ? 'next' : 'previous';
305 return "421 no $n article in this group";
307 $self->{article} = $n;
308 "223 $n <$mid> article retrieved - request text separately";
311 sub cmd_next ($) { article_adj($_[0], 1) }
312 sub cmd_last ($) { article_adj($_[0], -1) }
314 # We want to encourage using email and CC-ing everybody involved to avoid
315 # the single-point-of-failure a single server provides.
318 my $ng = $self->{ng};
319 $ng ? "440 mailto:$ng->{address} to post" : '440 posting not allowed'
324 res($self, '205 closing connection - goodbye!');
329 sub art_lookup ($$$) {
330 my ($self, $art, $set_headers) = @_;
331 my $ng = $self->{ng} or return '412 no newsgroup has been selected';
335 if ($art =~ /\A\d+\z/o) {
336 $err = '423 no such article number in this group';
339 } elsif ($art =~ /\A<([^>]+)>\z/) {
340 $err = '430 no such article found';
342 $n = $ng->mm->num_for($mid);
343 defined $mid or return $err;
348 $err = '420 no current article has been selected';
349 $n = $self->{article};
350 defined $n or return $err;
352 $mid = $ng->mm->mid_for($n);
353 defined $mid or return $err;
356 my $o = 'HEAD:' . mid2path($mid);
357 my $s = eval { Email::Simple->new($ng->gcf->cat_file($o)) };
358 return $err unless $s;
360 $s->header_set('Newsgroups', $ng->{name});
361 $s->header_set('Lines', $s->body =~ tr!\n!\n!);
362 $s->header_set('Xref', "$ng->{domain} $ng->{name}:$n");
365 if ($set_headers == 2) {
366 $s->header_set('Bytes', bytes::length($s->as_string));
373 sub simple_body_write ($$) {
377 $body =~ s/^\./../smg;
378 do_more($self, $body);
384 my $h = $s->header_obj;
385 $h->header_set('Bytes');
389 sub cmd_article ($$) {
390 my ($self, $art) = @_;
391 my $r = $self->art_lookup($art, 1);
392 return $r unless ref $r;
393 my ($n, $mid, $s) = @$r;
394 more($self, "220 $n <$mid> article retrieved - head and body follow");
395 do_more($self, header_str($s));
396 do_more($self, "\r\n");
397 simple_body_write($self, $s);
401 my ($self, $art) = @_;
402 my $r = $self->art_lookup($art, 2);
403 return $r unless ref $r;
404 my ($n, $mid, $s) = @$r;
405 more($self, "221 $n <$mid> article retrieved - head follows");
406 do_more($self, header_str($s));
411 my ($self, $art) = @_;
412 my $r = $self->art_lookup($art, 0);
413 return $r unless ref $r;
414 my ($n, $mid, $s) = @$r;
415 more($self, "222 $n <$mid> article retrieved - body follows");
416 simple_body_write($self, $s);
420 my ($self, $art) = @_;
421 my $r = $self->art_lookup($art, 0);
422 return $r unless ref $r;
423 my ($n, $mid, undef) = @$r;
424 "223 $n <$mid> article retrieved - request text separately";
427 sub cmd_ihave ($) { '435 article not wanted - do not send it' }
429 sub cmd_date ($) { '111 '.strftime('%Y%m%d%H%M%S', gmtime(time)) }
433 more($self, '100 help text follows');
438 my ($self, $range) = @_;
439 my $ng = $self->{ng} or return '412 no news group has been selected';
440 defined $range or return '420 No article(s) selected';
442 my ($min, $max) = $ng->mm->minmax;
443 if ($range =~ /\A(\d+)\z/) {
445 } elsif ($range =~ /\A(\d+)-\z/) {
446 ($beg, $end) = ($1, $max);
447 } elsif ($range =~ /\A(\d+)-(\d+)\z/) {
448 ($beg, $end) = ($1, $2);
452 $beg = $min if ($beg < $min);
453 $end = $max if ($end > $max);
454 return '420 No article(s) selected' if ($beg > $end);
459 my ($r, $header) = @_;
460 $r = $r->[2]->header_obj->header($header);
461 defined $r or return;
462 $r =~ s/[\r\n\t]+/ /sg;
466 sub long_response ($$$$) {
467 my ($self, $beg, $end, $cb) = @_;
468 die "BUG: nested long response" if $self->{long_res};
470 # make sure we disable reading during a long response,
471 # clients should not be sending us stuff and making us do more
472 # work while we are stream a response to them
473 $self->watch_read(0);
474 $self->{long_res} = sub {
475 # limit our own running time for fairness with other
476 # clients and to avoid buffering too much:
478 local $SIG{ALRM} = sub { $yield = 1 };
483 eval { $cb->(\$beg) };
484 } until (($err = $@) || $self->{closed} || $yield ||
485 $self->{write_buf_size} || ++$beg > $end);
488 if ($err || $self->{closed}) {
489 $self->{long_res} = undef;
490 warning("$err during long response") if $err;
491 $self->watch_read(1) unless $self->{closed};
492 } elsif ($yield || $self->{write_buf_size}) {
493 # no recursion, schedule another call ASAP
494 # but only after all pending writes are done
495 Danga::Socket->AddTimer(0, sub {
496 $self->write($self->{long_res});
499 $self->{long_res} = undef;
500 $self->watch_read(1);
504 $self->{long_res}->(); # kick off!
508 sub cmd_xhdr ($$;$) {
509 my ($self, $header, $range) = @_;
510 defined $self->{ng} or return '412 no news group currently selected';
511 if (defined $range && $range =~ /\A<(.+)>\z/) { # Message-ID
512 my $r = $self->art_lookup($range, 2);
513 return $r unless ref $r;
514 more($self, '221 Header follows');
515 if (defined($r = xhdr($r, $header))) {
516 more($self, "<$range> $r");
519 } else { # numeric range
520 $range = $self->{article} unless defined $range;
521 my $r = get_range($self, $range);
522 return $r unless ref $r;
523 my ($beg, $end) = @$r;
524 more($self, '221 Header follows');
525 $self->long_response($beg, $end, sub {
527 $r = $self->art_lookup($$i, 2);
528 return unless ref $r;
529 defined($r = xhdr($r, $header)) or return;
530 more($self, "$$i $r");
535 sub cmd_xover ($;$) {
536 my ($self, $range) = @_;
537 $range = $self->{article} unless defined $range;
538 my $r = get_range($self, $range);
539 return $r unless ref $r;
540 my ($beg, $end) = @$r;
541 more($self, "224 Overview information follows for $beg to $end");
542 $self->long_response($beg, $end, sub {
544 my $r = $self->art_lookup($$i, 2);
545 return unless ref $r;
546 more($self, join("\t", $r->[0],
548 my $h = xhdr($r, $_);
549 defined $h ? $h : '';
555 my ($self, $line) = @_;
556 do_write($self, $line . "\r\n");
560 my ($self, $line) = @_;
561 do_more($self, $line . "\r\n");
565 my ($self, $data) = @_;
566 my $done = $self->write($data);
567 die if $self->{closed};
569 # Do not watch for readability if we have data in the queue,
570 # instead re-enable watching for readability when we can
571 $self->watch_read(0) if (!$done || $self->{long_res});
576 use constant MSG_MORE => ($^O eq 'linux') ? 0x8000 : 0;
579 my ($self, $data) = @_;
580 if (MSG_MORE && !$self->{write_buf_size}) {
581 my $n = send($self->{sock}, $data, MSG_MORE);
583 my $dlen = bytes::length($data);
584 return 1 if $n == $dlen; # all done!
585 $data = bytes::substr($data, $n, $dlen - $n);
588 $self->do_write($data);
591 # callbacks for by Danga::Socket
593 sub event_hup { $_[0]->close }
594 sub event_err { $_[0]->close }
598 # only continue watching for readability when we are done writing:
599 $self->write(undef) == 1 and $self->watch_read(1);
604 use constant LINE_MAX => 512; # RFC 977 section 2.3
606 my $buf = $self->read(LINE_MAX) or return $self->close;
607 while ($r > 0 && $$buf =~ s/\A([^\r\n]+)\r?\n//) {
609 my $t0 = [ gettimeofday ];
610 $r = eval { $self->process_line($line) };
611 printf(STDERR "$line %0.6f\n",
612 tv_interval($t0, [gettimeofday]));
614 return $self->close if $r < 0;
615 my $len = bytes::length($$buf);
616 return $self->close if ($len >= LINE_MAX);
617 $self->push_back_read($buf) if ($len);
620 sub warning { print STDERR @_, "\n" }