1 # Copyright (C) 2020-2021 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # Combine any combination of PublicInbox::Search,
5 # PublicInbox::ExtSearch, and PublicInbox::LeiSearch objects
7 package PublicInbox::LeiXSearch;
10 use parent qw(PublicInbox::LeiSearch PublicInbox::IPC);
11 use PublicInbox::DS qw(now);
12 use File::Temp 0.19 (); # 0.19 for ->newdir
14 use PublicInbox::Search qw(xap_terms);
15 use PublicInbox::Spawn qw(popen_rd spawn which);
16 use PublicInbox::MID qw(mids);
17 use PublicInbox::Smsg;
19 use Fcntl qw(SEEK_SET F_SETFL O_APPEND O_RDWR);
20 use PublicInbox::ContentHash qw(git_sha);
24 PublicInbox::Search::load_xapian();
26 qp_flags => $PublicInbox::Search::QP_FLAGS |
27 PublicInbox::Search::FLAG_PURE_NOT(),
32 my ($self, $ibxish) = @_; # ibxish = ExtSearch or Inbox
33 my $desc = $ibxish->{inboxdir} // $ibxish->{topdir};
34 my $srch = $ibxish->search or
35 return warn("$desc not indexed for Xapian\n");
36 my @shards = $srch->xdb_shards_flat or
37 return warn("$desc has no Xapian shards\n");
39 if (delete $self->{xdb}) { # XXX: do we need this?
40 # clobber existing {xdb} if amending
41 my $expect = delete $self->{nshard};
42 my $shards = delete $self->{shards_flat};
43 scalar(@$shards) == $expect or die
44 "BUG: {nshard}$expect != shards=".scalar(@$shards);
47 for my $old_ibxish (@{$self->{shard2ibx}}) {
48 next if $prev == $old_ibxish;
50 my @shards = $old_ibxish->search->xdb_shards_flat;
51 push @{$self->{shards_flat}}, @shards;
53 my $nr = scalar(@{$self->{shards_flat}});
55 "BUG: reloaded $nr shards, expected $expect"
57 push @{$self->{shards_flat}}, @shards;
58 push(@{$self->{shard2ibx}}, $ibxish) for (@shards);
61 # returns a list of local inboxes (or count in scalar context)
62 sub locals { @{$_[0]->{locals} // []} }
64 sub remotes { @{$_[0]->{remotes} // []} }
66 # called by PublicInbox::Search::xdb (usually via ->mset)
67 sub xdb_shards_flat { @{$_[0]->{shards_flat} // []} }
69 sub _mitem_kw { # retry_reopen callback
70 my ($srch, $smsg, $mitem, $flagged) = @_;
71 my $doc = $mitem->get_document;
72 my $kw = xap_terms('K', $doc);
73 $kw->{flagged} = 1 if $flagged;
74 my $L = xap_terms('L', $doc);
75 # we keep the empty {kw} array here to prevent expensive work in
76 # ->xsmsg_vmd, _unbless_smsg will clobber it iff it's empty
77 $smsg->{kw} = [ sort keys %$kw ];
78 $smsg->{L} = [ sort keys %$L ] if scalar(keys %$L);
81 sub mitem_kw ($$$;$) {
82 my ($srch, $smsg, $mitem, $flagged) = @_;
83 $srch->retry_reopen(\&_mitem_kw, $smsg, $mitem, $flagged);
88 my ($self, $mitem) = @_;
89 # cf. https://trac.xapian.org/wiki/FAQ/MultiDatabaseDocumentID
90 my $nshard = $self->{nshard};
91 my $docid = $mitem->get_docid;
92 my $shard = ($docid - 1) % $nshard;
93 my $num = int(($docid - 1) / $nshard) + 1;
94 my $ibx = $self->{shard2ibx}->[$shard];
95 my $smsg = $ibx->over->get_art($num);
96 return if $smsg->{bytes} == 0; # external message
97 if ($ibx->can('msg_keywords')) {
98 mitem_kw($self, $smsg, $mitem);
104 my ($self, $qstr, $opt) = @_;
106 $opt->{relevance} //= -2;
107 $self->mset($qstr //= 'z:1..', $opt);
112 sub _mset_more ($$) {
113 my ($mset, $mo) = @_;
114 my $size = $mset->size;
115 $size >= $mo->{limit} && (($mo->{offset} += $size) < $mo->{limit});
118 # $startq will EOF when do_augment is done augmenting and allow
119 # query_combined_mset and query_thread_mset to proceed.
120 sub wait_startq ($) {
122 my $startq = delete $lei->{startq} or return;
124 my $n = sysread($startq, my $do_augment_done, 1);
126 return if $n == 0; # no MUA
127 if ($do_augment_done eq 'q') {
128 $lei->{opt}->{quiet} = 1;
129 delete $lei->{opt}->{verbose};
130 delete $lei->{-progress};
132 $lei->fail("$$ WTF `$do_augment_done'");
136 return $lei->fail("$$ wait_startq: $!") unless $!{EINTR};
142 return if $lei->{early_mua} || !$lei->{-progress};
143 if ($lei->{pkt_op_p}) {
144 $lei->{pkt_op_p}->pkt_do('mset_progress', @_);
145 } else { # single lei-daemon consumer
146 my ($desc, $mset_size, $mset_total_est) = @_;
147 $lei->{-mset_total} += $mset_size if $mset_total_est ne '?';
148 $lei->qerr("# $desc $mset_size/$mset_total_est");
154 $lei->{-nr_write} += $nr;
157 sub query_one_mset { # for --threads and l2m w/o sort
158 my ($self, $ibxish) = @_;
159 local $0 = "$0 query_one_mset";
160 my $lei = $self->{lei};
161 my ($srch, $over) = ($ibxish->search, $ibxish->over);
162 my $dir = $ibxish->{inboxdir} // $ibxish->{topdir};
163 return warn("$dir not indexed by Xapian\n") unless ($srch && $over);
164 my $mo = { %{$lei->{mset_opt}} }; # copy
166 my $each_smsg = $lei->{ovv}->ovv_each_smsg_cb($lei);
167 my $can_kw = !!$ibxish->can('msg_keywords');
168 my $threads = $lei->{opt}->{threads} // 0;
169 my $fl = $threads > 1 ? 1 : undef;
170 my $lss = $lei->{dedupe};
171 $lss = undef unless $lss && $lss->can('cfg_set'); # saved search
172 my $maxk = "external.$dir.maxuid";
173 my $stop_at = $lss ? $lss->{-cfg}->{$maxk} : undef;
174 if (defined $stop_at) {
175 die "$maxk=$stop_at has multiple values" if ref $stop_at;
177 local $SIG{__WARN__} = sub { push @e, @_ };
179 return warn("$maxk=$stop_at: @e") if @e;
183 $mset = $srch->mset($mo->{qstr}, $mo);
184 mset_progress($lei, $dir, $mset->size,
185 $mset->get_matches_estimated);
186 wait_startq($lei); # wait for keyword updates
187 my $ids = $srch->mset_to_artnums($mset, $mo);
188 @$ids = grep { $_ > $stop_at } @$ids if defined($stop_at);
191 # copy $ids if $lss since over->expand_thread
192 # shifts @{$ctx->{ids}}
193 $first_ids = [ @$ids ] if $lss;
194 my $ctx = { ids => $ids };
195 my %n2item = map { ($ids->[$i++], $_) } $mset->items;
196 while ($over->expand_thread($ctx)) {
197 for my $n (@{$ctx->{xids}}) {
198 my $smsg = $over->get_art($n) or next;
199 my $mitem = delete $n2item{$n};
200 next if $smsg->{bytes} == 0;
201 if ($mitem && $can_kw) {
202 mitem_kw($srch, $smsg, $mitem,
204 } elsif ($mitem && $fl) {
205 # call ->xsmsg_vmd, later
206 $smsg->{lei_q_tt_flagged} = 1;
208 $each_smsg->($smsg, $mitem);
210 @{$ctx->{xids}} = ();
214 my @items = $mset->items;
216 my $mitem = $items[$i++];
217 my $smsg = $over->get_art($n) or next;
218 next if $smsg->{bytes} == 0;
219 mitem_kw($srch, $smsg, $mitem, $fl) if $can_kw;
220 $each_smsg->($smsg, $mitem);
223 } while (_mset_more($mset, $mo));
224 if ($lss && scalar(@$first_ids)) {
226 my $max = $first_ids->[0];
227 $lss->cfg_set($maxk, $max);
230 undef $each_smsg; # may commit
231 $lei->{ovv}->ovv_atexit_child($lei);
234 sub query_combined_mset { # non-parallel for non-"--threads" users
236 local $0 = "$0 query_combined_mset";
237 my $lei = $self->{lei};
238 my $mo = { %{$lei->{mset_opt}} };
240 for my $loc (locals($self)) {
241 attach_external($self, $loc);
243 my $each_smsg = $lei->{ovv}->ovv_each_smsg_cb($lei);
245 $mset = $self->mset($mo->{qstr}, $mo);
246 mset_progress($lei, 'xsearch', $mset->size,
247 $mset->get_matches_estimated);
248 wait_startq($lei); # wait for keyword updates
249 for my $mitem ($mset->items) {
250 my $smsg = smsg_for($self, $mitem) or next;
251 $each_smsg->($smsg, $mitem);
253 } while (_mset_more($mset, $mo));
254 undef $each_smsg; # may commit
255 $lei->{ovv}->ovv_atexit_child($lei);
258 sub _smsg_fill ($$) {
259 my ($smsg, $eml) = @_;
260 $smsg->populate($eml);
261 $smsg->parse_references($eml, mids($eml));
262 $smsg->{$_} //= '' for qw(from to cc ds subject references mid);
263 delete @$smsg{qw(From Subject -ds -ts)};
266 sub each_remote_eml { # callback for MboxReader->mboxrd
267 my ($eml, $self, $lei, $each_smsg) = @_;
268 my $xoids = $lei->{ale}->xoids_for($eml, 1);
269 if ($self->{import_sto} && !$xoids) {
270 $self->{import_sto}->ipc_do('add_eml', $eml);
272 my $smsg = bless {}, 'PublicInbox::Smsg';
273 $smsg->{blob} = $xoids ? (keys(%$xoids))[0]
274 : git_sha(1, $eml)->hexdigest;
275 _smsg_fill($smsg, $eml);
277 if ($lei->{-progress}) {
278 ++$lei->{-nr_remote_eml};
280 my $next = $lei->{-next_progress} //= ($now + 1);
282 $lei->{-next_progress} = $now + 1;
283 my $nr = $lei->{-nr_remote_eml};
284 mset_progress($lei, $lei->{-current_url}, $nr, '?');
287 $each_smsg->($smsg, undef, $eml);
290 sub query_remote_mboxrd {
291 my ($self, $uris) = @_;
292 local $0 = "$0 query_remote_mboxrd";
293 local $SIG{TERM} = sub { exit(0) }; # for DESTROY (File::Temp, $reap)
294 my $lei = $self->{lei};
295 my $opt = $lei->{opt};
296 my @qform = (q => $lei->{mset_opt}->{qstr}, x => 'm');
297 push(@qform, t => 1) if $opt->{threads};
298 my $verbose = $opt->{verbose};
299 my ($reap_tail, $reap_curl);
300 my $cerr = File::Temp->new(TEMPLATE => 'curl.err-XXXX', TMPDIR => 1);
301 fcntl($cerr, F_SETFL, O_APPEND|O_RDWR) or warn "set O_APPEND: $!";
302 my $rdr = { 2 => $cerr, pgid => 0 };
303 my $sigint_reap = $lei->can('sigint_reap');
305 # spawn a process to force line-buffering, otherwise curl
306 # will write 1 character at-a-time and parallel outputs
307 # mmmaaayyy llloookkk llliiikkkeee ttthhhiiisss
308 my $o = { 1 => $lei->{2}, 2 => $lei->{2}, pgid => 0 };
309 my $pid = spawn(['tail', '-f', $cerr->filename], undef, $o);
310 $reap_tail = PublicInbox::OnDestroy->new($sigint_reap, $pid);
312 my $curl = PublicInbox::LeiCurl->new($lei, $self->{curl}) or return;
313 push @$curl, '-s', '-d', '';
314 my $each_smsg = $lei->{ovv}->ovv_each_smsg_cb($lei);
315 $self->{import_sto} = $lei->{sto} if $lei->{opt}->{'import-remote'};
316 for my $uri (@$uris) {
317 $lei->{-current_url} = $uri->as_string;
318 $lei->{-nr_remote_eml} = 0;
319 $uri->query_form(@qform);
320 my $cmd = $curl->for_uri($lei, $uri);
321 $lei->qerr("# $cmd");
322 my ($fh, $pid) = popen_rd($cmd, undef, $rdr);
323 $reap_curl = PublicInbox::OnDestroy->new($sigint_reap, $pid);
324 $fh = IO::Uncompress::Gunzip->new($fh, MultiStream => 1);
325 PublicInbox::MboxReader->mboxrd($fh, \&each_remote_eml, $self,
327 my $err = waitpid($pid, 0) == $pid ? undef
328 : "BUG: waitpid($cmd): $!";
329 @$reap_curl = (); # cancel OnDestroy
331 my $nr = $lei->{-nr_remote_eml};
332 if ($nr && $lei->{sto}) {
333 my $wait = $lei->{sto}->ipc_do('done');
336 mset_progress($lei, $lei->{-current_url}, $nr, $nr);
341 seek($cerr, 0, SEEK_SET) or
342 $lei->err("seek($cmd stderr): $!");
343 $err = do { local $/; <$cerr> } //
344 "read($cmd stderr): $!";
345 truncate($cerr, 0) or
346 $lei->err("truncate($cmd stderr): $!");
348 next if (($? >> 8) == 22 && $err =~ /\b404\b/);
349 $uri->query_form(q => $lei->{mset_opt}->{qstr});
350 $lei->child_error($?, "E: <$uri> $err");
353 $lei->{ovv}->ovv_atexit_child($lei);
356 sub git { $_[0]->{git} // die 'BUG: git uninitialized' }
358 sub xsearch_done_wait { # dwaitpid callback
359 my ($arg, $pid) = @_;
360 my ($wq, $lei) = @$arg;
361 $lei->child_error($?, 'non-fatal error from '.ref($wq)) if $?;
364 sub query_done { # EOF callback for main daemon
366 my $l2m = delete $lei->{l2m};
367 $l2m->wq_wait_old(\&xsearch_done_wait, $lei) if $l2m;
368 if (my $lxs = delete $lei->{lxs}) {
369 $lxs->wq_wait_old(\&xsearch_done_wait, $lei);
371 if ($lei->{opt}->{'mail-sync'} && !$lei->{sto}) {
372 warn "BUG: {sto} missing with --mail-sync";
374 my $wait = $lei->{sto} ? $lei->{sto}->ipc_do('done') : undef;
375 $wait = $lei->{v2w} ? $lei->{v2w}->ipc_do('done') : undef;
376 $lei->{ovv}->ovv_end($lei);
378 if ($l2m) { # close() calls LeiToMail reap_compress
379 if (my $out = delete $lei->{old_1}) {
380 if (my $mbout = $lei->{1}) {
381 close($mbout) or return $lei->fail(<<"");
382 Error closing $lei->{ovv}->{dst}: $!
387 if ($l2m->lock_free) {
390 } else { # mbox users
391 delete $l2m->{mbl}; # drop dotlock
395 if ($lei->{-progress}) {
396 my $tot = $lei->{-mset_total} // 0;
397 my $nr = $lei->{-nr_write} // 0;
399 "# $nr written to $lei->{ovv}->{dst} ($tot matches)" :
402 $lei->start_mua if $start_mua;
406 sub do_post_augment {
408 my $l2m = $lei->{l2m} or return; # client disconnected
409 $lei->fchdir or return;
411 eval { $l2m->post_augment($lei) };
414 if (my $lxs = delete $lei->{lxs}) {
416 $lxs->wq_close(0, undef, $lei);
420 if (!$err && delete $lei->{early_mua}) { # non-augment case
423 close(delete $lei->{au_done}); # triggers wait_startq in lei_xsearch
426 sub incr_post_augment { # called whenever an l2m shard finishes augment
428 my $l2m = $lei->{l2m} or return; # client disconnected
429 return if ++$lei->{nr_post_augment} != $l2m->{-wq_nr_workers};
430 do_post_augment($lei);
433 my $MAX_PER_HOST = 4;
436 my ($self, $opt) = @_;
437 my $nl = $opt->{threads} ? locals($self) : 1;
438 my $nr = remotes($self);
439 $nr = $MAX_PER_HOST if $nr > $MAX_PER_HOST;
443 sub start_query ($;$) { # always runs in main (lei-daemon) process
444 my ($self, $l2m) = @_;
445 if ($self->{opt_threads} || ($l2m && !$self->{opt_sort})) {
446 for my $ibxish (locals($self)) {
447 $self->wq_io_do('query_one_mset', [], $ibxish);
449 } elsif (locals($self)) {
450 $self->wq_io_do('query_combined_mset', []);
454 for my $uri (remotes($self)) {
455 push @{$q->[$i++ % $MAX_PER_HOST]}, $uri;
458 $self->wq_io_do('query_remote_mboxrd', [], $uris);
460 if ($self->{-do_lcat}) {
461 $self->wq_io_do('lcat_dump', []);
463 $self->wq_close(1); # lei_xsearch workers stop when done
466 sub incr_start_query { # called whenever an l2m shard starts do_post_auth
467 my ($self, $l2m) = @_;
468 return if ++$self->{nr_start_query} != $l2m->{-wq_nr_workers};
469 start_query($self, $l2m);
472 sub ipc_atfork_child {
474 $self->{lei}->_lei_atfork_child;
475 $self->SUPER::ipc_atfork_child;
479 my ($self, $lei) = @_;
480 my $l2m = $lei->{l2m};
482 '|' => [ $lei->can('sigpipe_handler'), $lei ],
483 '!' => [ $lei->can('fail_handler'), $lei ],
484 '.' => [ \&do_post_augment, $lei ],
485 '+' => [ \&incr_post_augment, $lei ],
486 '' => [ \&query_done, $lei ],
487 'mset_progress' => [ \&mset_progress, $lei ],
488 'l2m_progress' => [ \&l2m_progress, $lei ],
489 'x_it' => [ $lei->can('x_it'), $lei ],
490 'child_error' => [ $lei->can('child_error'), $lei ],
491 'incr_start_query' => [ \&incr_start_query, $self, $l2m ],
493 $lei->{auth}->op_merge($ops, $l2m) if $l2m && $lei->{auth};
494 my $end = $lei->pkt_op_pair;
495 $lei->{1}->autoflush(1);
496 $lei->start_pager if delete $lei->{need_pager};
497 $lei->{ovv}->ovv_begin($lei);
498 die 'BUG: xdb|over open' if $lei->{lse}->{xdb} || $lei->{lse}->{over};
500 $l2m->pre_augment($lei);
501 if ($lei->{opt}->{augment} && delete $lei->{early_mua}) {
504 my $F_SETPIPE_SZ = $^O eq 'linux' ? 1031 : undef;
505 if ($l2m->{-wq_nr_workers} > 1 &&
506 $l2m->{base_type} =~ /\A(?:maildir|mbox)\z/) {
507 # setup two barriers to coordinate dedupe_nr
508 # between l2m workers
509 pipe(my ($a_r, $a_w)) or die "pipe: $!";
510 fcntl($a_r, $F_SETPIPE_SZ, 4096) if $F_SETPIPE_SZ;
511 pipe(my ($b_r, $b_w)) or die "pipe: $!";
512 fcntl($b_r, $F_SETPIPE_SZ, 4096) if $F_SETPIPE_SZ;
513 $l2m->{au_peers} = [ $a_r, $a_w, $b_r, $b_w ];
515 $l2m->wq_workers_start('lei2mail', undef,
516 $lei->oldset, { lei => $lei });
517 pipe($lei->{startq}, $lei->{au_done}) or die "pipe: $!";
518 fcntl($lei->{startq}, $F_SETPIPE_SZ, 4096) if $F_SETPIPE_SZ;
519 delete $l2m->{au_peers};
521 $self->wq_workers_start('lei_xsearch', undef,
522 $lei->oldset, { lei => $lei });
523 my $op_c = delete $lei->{pkt_op_c};
524 delete $lei->{pkt_op_p};
526 $self->{opt_threads} = $lei->{opt}->{threads};
527 $self->{opt_sort} = $lei->{opt}->{'sort'};
528 $self->{-do_lcat} = $lei->{lcat_blob} // $lei->{lcat_fid};
530 $l2m->net_merge_all_done unless $lei->{auth};
534 $lei->event_step_init; # wait for shutdowns
535 $lei->wait_wq_events($op_c, $ops);
539 my ($self, $uri) = @_;
540 if (my $curl = $self->{curl} //= which('curl') // 0) {
541 require PublicInbox::MboxReader;
542 require IO::Uncompress::Gunzip;
543 require PublicInbox::LeiCurl;
544 push @{$self->{remotes}}, $uri;
546 warn "curl missing, ignoring $uri\n";
550 sub prepare_external {
551 my ($self, $loc, $boost) = @_; # n.b. already ordered by boost
552 if (ref $loc) { # already a URI, or PublicInbox::Inbox-like object
553 return add_uri($self, $loc) if $loc->can('scheme');
554 } elsif ($loc =~ m!\Ahttps?://!) {
556 return add_uri($self, URI->new($loc));
557 } elsif (-f "$loc/ei.lock") {
558 require PublicInbox::ExtSearch;
559 die "`\\n' not allowed in `$loc'\n" if index($loc, "\n") >= 0;
560 $loc = PublicInbox::ExtSearch->new($loc);
561 } elsif (-f "$loc/inbox.lock" || -d "$loc/public-inbox") {
562 die "`\\n' not allowed in `$loc'\n" if index($loc, "\n") >= 0;
563 require PublicInbox::Inbox; # v2, v1
564 $loc = bless { inboxdir => $loc }, 'PublicInbox::Inbox';
566 warn "W: ignoring $loc, unable to determine type\n";
569 push @{$self->{locals}}, $loc;
572 sub _lcat_i { # LeiMailSync->each_src iterator callback
573 my ($oidbin, $id, $each_smsg) = @_;
574 $each_smsg->({blob => unpack('H*', $oidbin), pct => 100});
577 sub _lcat2smsg { # git->cat_async callback
578 my ($bref, $oid, $type, $size, $smsg) = @_;
580 my $eml = PublicInbox::Eml->new($bref);
581 my $json_dump = delete $smsg->{-json_dump};
582 bless $smsg, 'PublicInbox::Smsg';
583 _smsg_fill($smsg, $eml);
584 $json_dump->($smsg, undef, $eml);
590 my $lei = $self->{lei};
591 my $each_smsg = $lei->{ovv}->ovv_each_smsg_cb($lei);
592 my $git = $lei->{ale}->git;
594 my $json_dump = $each_smsg;
598 $smsg->{-json_dump} = $json_dump;
599 $git->cat_async($smsg->{blob}, \&_lcat2smsg, $smsg);
602 for my $oid (@{$lei->{lcat_blob} // []}) {
603 $each_smsg->({ blob => $oid, pct => 100 });
605 if (my $fids = delete $lei->{lcat_fid}) {
606 my $lms = $lei->{lse}->lms;
607 for my $fid (@$fids) {
608 $lms->each_src({fid => $fid}, \&_lcat_i, $each_smsg);
611 $git->async_wait_all;
612 undef $each_smsg; # may commit
613 $lei->{ovv}->ovv_atexit_child($lei);