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);
21 use POSIX qw(strftime);
25 PublicInbox::Search::load_xapian();
27 qp_flags => $PublicInbox::Search::QP_FLAGS |
28 PublicInbox::Search::FLAG_PURE_NOT(),
33 my ($self, $ibxish) = @_; # ibxish = ExtSearch or Inbox
34 my $desc = $ibxish->{inboxdir} // $ibxish->{topdir};
35 my $srch = $ibxish->search or
36 return warn("$desc not indexed for Xapian ($@ $!)\n");
37 my @shards = $srch->xdb_shards_flat or
38 return warn("$desc has no Xapian shards\n");
40 if (delete $self->{xdb}) { # XXX: do we need this?
41 # clobber existing {xdb} if amending
42 my $expect = delete $self->{nshard};
43 my $shards = delete $self->{shards_flat};
44 scalar(@$shards) == $expect or die
45 "BUG: {nshard}$expect != shards=".scalar(@$shards);
48 for my $old_ibxish (@{$self->{shard2ibx}}) {
49 next if $prev == $old_ibxish;
51 my @shards = $old_ibxish->search->xdb_shards_flat;
52 push @{$self->{shards_flat}}, @shards;
54 my $nr = scalar(@{$self->{shards_flat}});
56 "BUG: reloaded $nr shards, expected $expect"
58 push @{$self->{shards_flat}}, @shards;
59 push(@{$self->{shard2ibx}}, $ibxish) for (@shards);
62 # returns a list of local inboxes (or count in scalar context)
63 sub locals { @{$_[0]->{locals} // []} }
65 sub remotes { @{$_[0]->{remotes} // []} }
67 # called by PublicInbox::Search::xdb (usually via ->mset)
68 sub xdb_shards_flat { @{$_[0]->{shards_flat} // []} }
70 sub _mitem_kw { # retry_reopen callback
71 my ($srch, $smsg, $mitem, $flagged) = @_;
72 my $doc = $mitem->get_document;
73 my $kw = xap_terms('K', $doc);
74 $kw->{flagged} = 1 if $flagged;
75 my @L = xap_terms('L', $doc);
76 # we keep the empty {kw} array here to prevent expensive work in
77 # ->xsmsg_vmd, _unbless_smsg will clobber it iff it's empty
78 $smsg->{kw} = [ sort keys %$kw ];
79 $smsg->{L} = \@L if scalar(@L);
82 sub mitem_kw ($$$;$) {
83 my ($srch, $smsg, $mitem, $flagged) = @_;
84 $srch->retry_reopen(\&_mitem_kw, $smsg, $mitem, $flagged);
89 my ($self, $mitem) = @_;
90 # cf. https://trac.xapian.org/wiki/FAQ/MultiDatabaseDocumentID
91 my $nshard = $self->{nshard};
92 my $docid = $mitem->get_docid;
93 my $shard = ($docid - 1) % $nshard;
94 my $num = int(($docid - 1) / $nshard) + 1;
95 my $ibx = $self->{shard2ibx}->[$shard];
96 my $smsg = $ibx->over->get_art($num);
97 return if $smsg->{bytes} == 0; # external message
98 if ($ibx->can('msg_keywords')) {
99 mitem_kw($self, $smsg, $mitem);
105 my ($self, $qstr, $opt) = @_;
107 $opt->{relevance} //= -2;
108 $self->mset($qstr //= 'z:1..', $opt);
113 sub _check_mset_limit ($$$) {
114 my ($lei, $desc, $mset) = @_;
115 return if defined($lei->{opt}->{limit}); # user requested limit
116 my $est = $mset->get_matches_estimated;
117 my $tot = $lei->{mset_opt}->{total};
118 $est > $tot and $lei->qerr(<<"");
119 # $desc estimated matches ($est) exceeds default --limit=$tot
123 sub _mset_more ($$) {
124 my ($mset, $mo) = @_;
125 my $size = $mset->size;
126 $size >= $mo->{limit} && (($mo->{offset} += $size) < $mo->{total});
129 # $startq will EOF when do_augment is done augmenting and allow
130 # query_combined_mset and query_thread_mset to proceed.
131 sub wait_startq ($) {
133 my $startq = delete $lei->{startq} or return;
135 my $n = sysread($startq, my $do_augment_done, 1);
137 return if $n == 0; # no MUA
138 if ($do_augment_done eq 'q') {
139 $lei->{opt}->{quiet} = 1;
140 delete $lei->{opt}->{verbose};
141 delete $lei->{-progress};
143 $lei->fail("$$ WTF `$do_augment_done'");
147 return $lei->fail("$$ wait_startq: $!") unless $!{EINTR};
153 return if $lei->{early_mua} || !$lei->{-progress};
154 if ($lei->{pkt_op_p}) {
155 $lei->{pkt_op_p}->pkt_do('mset_progress', @_);
156 } else { # single lei-daemon consumer
157 my ($desc, $mset_size, $mset_total_est) = @_;
158 $lei->{-mset_total} += $mset_size if $mset_total_est ne '?';
159 $lei->qerr("# $desc $mset_size/$mset_total_est");
165 $lei->{-nr_write} += $nr;
168 sub query_one_mset { # for --threads and l2m w/o sort
169 my ($self, $ibxish) = @_;
170 local $0 = "$0 query_one_mset";
171 my $lei = $self->{lei};
172 my ($srch, $over) = ($ibxish->search, $ibxish->over);
173 my $dir = $ibxish->{inboxdir} // $ibxish->{topdir};
174 return warn("$dir not indexed by Xapian\n") unless ($srch && $over);
175 bless $srch, 'PublicInbox::LeiSearch'; # for ->qparse_new
176 my $mo = { %{$lei->{mset_opt}} }; # copy
178 my $each_smsg = $lei->{ovv}->ovv_each_smsg_cb($lei);
179 my $can_kw = !!$ibxish->can('msg_keywords');
180 my $threads = $lei->{opt}->{threads} // 0;
181 my $fl = $threads > 1 ? 1 : undef;
182 my $lss = $lei->{lss};
183 my $maxk = "external.$dir.maxuid";
184 my $stop_at = $lss ? $lss->{-cfg}->{$maxk} : undef;
185 if (defined $stop_at) {
186 die "$maxk=$stop_at has multiple values" if ref $stop_at;
188 local $SIG{__WARN__} = sub { push @e, @_ };
190 return warn("$maxk=$stop_at: @e") if @e;
194 $mset = $srch->mset($mo->{qstr}, $mo);
195 mset_progress($lei, $dir, $mo->{offset} + $mset->size,
196 $mset->get_matches_estimated);
197 wait_startq($lei); # wait for keyword updates
198 my $ids = $srch->mset_to_artnums($mset, $mo);
199 @$ids = grep { $_ > $stop_at } @$ids if defined($stop_at);
202 # copy $ids if $lss since over->expand_thread
203 # shifts @{$ctx->{ids}}
204 $first_ids = [ @$ids ] if $lss;
205 my $ctx = { ids => $ids };
206 my %n2item = map { ($ids->[$i++], $_) } $mset->items;
207 while ($over->expand_thread($ctx)) {
208 for my $n (@{$ctx->{xids}}) {
209 my $smsg = $over->get_art($n) or next;
210 my $mitem = delete $n2item{$n};
211 next if $smsg->{bytes} == 0;
212 if ($mitem && $can_kw) {
213 mitem_kw($srch, $smsg, $mitem,
215 } elsif ($mitem && $fl) {
216 # call ->xsmsg_vmd, later
217 $smsg->{lei_q_tt_flagged} = 1;
219 $each_smsg->($smsg, $mitem);
221 @{$ctx->{xids}} = ();
225 my @items = $mset->items;
227 my $mitem = $items[$i++];
228 my $smsg = $over->get_art($n) or next;
229 next if $smsg->{bytes} == 0;
230 mitem_kw($srch, $smsg, $mitem, $fl) if $can_kw;
231 $each_smsg->($smsg, $mitem);
234 } while (_mset_more($mset, $mo));
235 _check_mset_limit($lei, $dir, $mset);
236 if ($lss && scalar(@$first_ids)) {
238 my $max = $first_ids->[0];
239 $lss->cfg_set($maxk, $max);
242 undef $each_smsg; # may commit
243 $lei->{ovv}->ovv_atexit_child($lei);
246 sub query_combined_mset { # non-parallel for non-"--threads" users
248 local $0 = "$0 query_combined_mset";
249 my $lei = $self->{lei};
250 my $mo = { %{$lei->{mset_opt}} };
252 for my $loc (locals($self)) {
253 attach_external($self, $loc);
255 my $each_smsg = $lei->{ovv}->ovv_each_smsg_cb($lei);
257 $mset = $self->mset($mo->{qstr}, $mo);
258 mset_progress($lei, 'xsearch', $mo->{offset} + $mset->size,
259 $mset->get_matches_estimated);
260 wait_startq($lei); # wait for keyword updates
261 for my $mitem ($mset->items) {
262 my $smsg = smsg_for($self, $mitem) or next;
263 $each_smsg->($smsg, $mitem);
265 } while (_mset_more($mset, $mo));
266 _check_mset_limit($lei, 'xsearch', $mset);
267 undef $each_smsg; # may commit
268 $lei->{ovv}->ovv_atexit_child($lei);
271 sub _smsg_fill ($$) {
272 my ($smsg, $eml) = @_;
273 $smsg->populate($eml);
274 $smsg->parse_references($eml, mids($eml));
275 $smsg->{$_} //= '' for qw(from to cc ds subject references mid);
276 delete @$smsg{qw(From Subject -ds -ts)};
279 sub each_remote_eml { # callback for MboxReader->mboxrd
280 my ($eml, $self, $lei, $each_smsg) = @_;
281 my $xoids = $lei->{ale}->xoids_for($eml, 1);
282 my $smsg = bless {}, 'PublicInbox::Smsg';
283 if ($self->{import_sto} && !$xoids) {
284 my $res = $self->{import_sto}->wq_do('add_eml', $eml);
285 if (ref($res) eq ref($smsg)) { # totally new message
287 $smsg->{kw} = []; # short-circuit xsmsg_vmd
290 $smsg->{blob} //= $xoids ? (keys(%$xoids))[0]
291 : $lei->git_oid($eml)->hexdigest;
292 _smsg_fill($smsg, $eml);
294 if ($lei->{-progress}) {
295 ++$lei->{-nr_remote_eml};
297 my $next = $lei->{-next_progress} //= ($now + 1);
299 $lei->{-next_progress} = $now + 1;
300 my $nr = $lei->{-nr_remote_eml};
301 mset_progress($lei, $lei->{-current_url}, $nr, '?');
304 $each_smsg->($smsg, undef, $eml);
307 sub fudge_qstr_time ($$$) {
308 my ($lei, $uri, $qstr) = @_;
309 return ($qstr, undef) unless $lei->{lss};
310 my $cfg = $lei->{lss}->{-cfg} // die 'BUG: no lss->{-cfg}';
311 my $cfg_key = "external.$uri.lastresult";
312 my $lr = $cfg->{$cfg_key} or return ($qstr, $cfg_key);
313 if ($lr !~ /\A\-?[0-9]+\z/) {
315 "$cfg->{-f}: $cfg_key=$lr not an integer, ignoring");
316 return ($qstr, $cfg_key);
318 my $rft = $lei->{opt}->{'remote-fudge-time'};
319 if ($rft && $rft !~ /\A-?[0-9]+\z/) {
320 my @t = $lei->{lss}->git->date_parse($rft);
321 my $diff = time - $t[0];
322 $lei->qerr("# $rft => $diff seconds");
325 $lr -= ($rft || (48 * 60 * 60));
326 $lei->qerr("# $uri limiting to ".
327 strftime('%Y-%m-%d %k:%M %z', localtime($lr)). ' and newer');
328 # this should really be rt: (received-time), but no stable
329 # public-inbox releases support it, yet.
330 my $dt = 'dt:'.strftime('%Y%m%d%H%M%S', gmtime($lr)).'..';
332 substr($qstr, 0, 0, '(');
335 ($qstr .= $dt, $cfg_key);
338 sub query_remote_mboxrd {
339 my ($self, $uris) = @_;
340 local $0 = "$0 query_remote_mboxrd";
341 local $SIG{TERM} = sub { exit(0) }; # for DESTROY (File::Temp, $reap)
342 my $lei = $self->{lei};
343 my $opt = $lei->{opt};
344 my $qstr = $lei->{mset_opt}->{qstr};
345 $qstr =~ s/[ \n\t]+/ /sg; # make URLs less ugly
346 my @qform = (x => 'm');
347 push(@qform, t => 1) if $opt->{threads};
348 my $verbose = $opt->{verbose};
349 my ($reap_tail, $reap_curl);
350 my $cerr = File::Temp->new(TEMPLATE => 'curl.err-XXXX', TMPDIR => 1);
351 fcntl($cerr, F_SETFL, O_APPEND|O_RDWR) or warn "set O_APPEND: $!";
352 my $rdr = { 2 => $cerr, pgid => 0 };
353 my $sigint_reap = $lei->can('sigint_reap');
355 # spawn a process to force line-buffering, otherwise curl
356 # will write 1 character at-a-time and parallel outputs
357 # mmmaaayyy llloookkk llliiikkkeee ttthhhiiisss
358 my $o = { 1 => $lei->{2}, 2 => $lei->{2}, pgid => 0 };
359 my $pid = spawn(['tail', '-f', $cerr->filename], undef, $o);
360 $reap_tail = PublicInbox::OnDestroy->new($sigint_reap, $pid);
362 my $curl = PublicInbox::LeiCurl->new($lei, $self->{curl}) or return;
363 push @$curl, '-s', '-d', '';
364 my $each_smsg = $lei->{ovv}->ovv_each_smsg_cb($lei);
365 $self->{import_sto} = $lei->{sto} if $lei->{opt}->{'import-remote'};
366 for my $uri (@$uris) {
367 $lei->{-current_url} = $uri->as_string;
368 $lei->{-nr_remote_eml} = 0;
370 my ($q, $key) = fudge_qstr_time($lei, $uri, $qstr);
371 $uri->query_form(@qform, q => $q);
372 my $cmd = $curl->for_uri($lei, $uri);
373 $lei->qerr("# $cmd");
374 my ($fh, $pid) = popen_rd($cmd, undef, $rdr);
375 $reap_curl = PublicInbox::OnDestroy->new($sigint_reap, $pid);
376 $fh = IO::Uncompress::Gunzip->new($fh, MultiStream => 1);
377 PublicInbox::MboxReader->mboxrd($fh, \&each_remote_eml, $self,
379 my $err = waitpid($pid, 0) == $pid ? undef
380 : "BUG: waitpid($cmd): $!";
381 @$reap_curl = (); # cancel OnDestroy
383 my $nr = $lei->{-nr_remote_eml};
384 my $wait = $lei->{sto}->wq_do('done') if $nr && $lei->{sto};
386 # don't update if no results, maybe MTA is down
388 $lei->{lss}->cfg_set($key, $start);
389 mset_progress($lei, $lei->{-current_url}, $nr, $nr);
394 seek($cerr, 0, SEEK_SET) or
395 $lei->err("seek($cmd stderr): $!");
396 $err = do { local $/; <$cerr> } //
397 "read($cmd stderr): $!";
398 truncate($cerr, 0) or
399 $lei->err("truncate($cmd stderr): $!");
401 next if (($? >> 8) == 22 && $err =~ /\b404\b/);
402 $uri->query_form(q => $qstr);
403 $lei->child_error($?, "E: <$uri> $err");
406 $lei->{ovv}->ovv_atexit_child($lei);
409 sub git { $_[0]->{git} // die 'BUG: git uninitialized' }
411 sub xsearch_done_wait { # dwaitpid callback
412 my ($arg, $pid) = @_;
413 my ($wq, $lei) = @$arg;
414 $lei->child_error($?, 'non-fatal error from '.ref($wq)) if $?;
417 sub query_done { # EOF callback for main daemon
419 my $l2m = delete $lei->{l2m};
420 $l2m->wq_wait_old(\&xsearch_done_wait, $lei) if $l2m;
421 if (my $lxs = delete $lei->{lxs}) {
422 $lxs->wq_wait_old(\&xsearch_done_wait, $lei);
424 if ($lei->{opt}->{'mail-sync'} && !$lei->{sto}) {
425 warn "BUG: {sto} missing with --mail-sync";
427 $lei->sto_done_request if $lei->{sto};
428 my $wait = $lei->{v2w} ? $lei->{v2w}->wq_do('done') : undef;
429 $lei->{ovv}->ovv_end($lei);
431 if ($l2m) { # close() calls LeiToMail reap_compress
432 if (my $out = delete $lei->{old_1}) {
433 if (my $mbout = $lei->{1}) {
434 close($mbout) or return $lei->fail(<<"");
435 Error closing $lei->{ovv}->{dst}: $!
440 if ($l2m->lock_free) {
443 } else { # mbox users
444 delete $l2m->{mbl}; # drop dotlock
448 if ($lei->{-progress}) {
449 my $tot = $lei->{-mset_total} // 0;
450 my $nr = $lei->{-nr_write} // 0;
452 my $m = "# $nr written to " .
453 "$lei->{ovv}->{dst} ($tot matches)";
454 $nr ? $lei->qfin($m) : $lei->qerr($m);
456 $lei->qerr("# $tot matches");
459 $lei->start_mua if $start_mua;
463 sub do_post_augment {
465 my $l2m = $lei->{l2m} or return; # client disconnected
466 $lei->fchdir or return;
468 eval { $l2m->post_augment($lei) };
471 if (my $lxs = delete $lei->{lxs}) {
473 $lxs->wq_close(0, undef, $lei);
477 if (!$err && delete $lei->{early_mua}) { # non-augment case
480 close(delete $lei->{au_done}); # triggers wait_startq in lei_xsearch
483 sub incr_post_augment { # called whenever an l2m shard finishes augment
485 my $l2m = $lei->{l2m} or return; # client disconnected
486 return if ++$lei->{nr_post_augment} != $l2m->{-wq_nr_workers};
487 do_post_augment($lei);
490 my $MAX_PER_HOST = 4;
493 my ($self, $opt) = @_;
494 my $nl = $opt->{threads} ? locals($self) : 1;
495 my $nr = remotes($self);
496 $nr = $MAX_PER_HOST if $nr > $MAX_PER_HOST;
500 sub start_query ($;$) { # always runs in main (lei-daemon) process
501 my ($self, $l2m) = @_;
502 if ($self->{opt_threads} || ($l2m && !$self->{opt_sort})) {
503 for my $ibxish (locals($self)) {
504 $self->wq_io_do('query_one_mset', [], $ibxish);
506 } elsif (locals($self)) {
507 $self->wq_io_do('query_combined_mset', []);
511 for my $uri (remotes($self)) {
512 push @{$q->[$i++ % $MAX_PER_HOST]}, $uri;
515 $self->wq_io_do('query_remote_mboxrd', [], $uris);
517 if ($self->{-do_lcat}) {
518 $self->wq_io_do('lcat_dump', []);
520 $self->wq_close(1); # lei_xsearch workers stop when done
523 sub incr_start_query { # called whenever an l2m shard starts do_post_auth
524 my ($self, $l2m) = @_;
525 return if ++$self->{nr_start_query} != $l2m->{-wq_nr_workers};
526 start_query($self, $l2m);
529 sub ipc_atfork_child {
531 $self->{lei}->_lei_atfork_child;
532 $self->SUPER::ipc_atfork_child;
536 my ($self, $lei) = @_;
537 my $l2m = $lei->{l2m};
539 'sigpipe_handler' => [ $lei ],
540 'fail_handler' => [ $lei ],
541 'do_post_augment' => [ \&do_post_augment, $lei ],
542 'incr_post_augment' => [ \&incr_post_augment, $lei ],
543 '' => [ \&query_done, $lei ],
544 'mset_progress' => [ \&mset_progress, $lei ],
545 'l2m_progress' => [ \&l2m_progress, $lei ],
547 'child_error' => [ $lei ],
548 'incr_start_query' => [ $self, $l2m ],
550 $lei->{auth}->op_merge($ops, $l2m) if $l2m && $lei->{auth};
551 my $end = $lei->pkt_op_pair;
552 $lei->{1}->autoflush(1);
553 $lei->start_pager if delete $lei->{need_pager};
554 $lei->{ovv}->ovv_begin($lei);
555 die 'BUG: xdb|over open' if $lei->{lse}->{xdb} || $lei->{lse}->{over};
557 $l2m->pre_augment($lei);
558 if ($lei->{opt}->{augment} && delete $lei->{early_mua}) {
561 my $F_SETPIPE_SZ = $^O eq 'linux' ? 1031 : undef;
562 if ($l2m->{-wq_nr_workers} > 1 &&
563 $l2m->{base_type} =~ /\A(?:maildir|mbox)\z/) {
564 # setup two barriers to coordinate ->has_entries
565 # between l2m workers
566 pipe(my ($a_r, $a_w)) or die "pipe: $!";
567 fcntl($a_r, $F_SETPIPE_SZ, 4096) if $F_SETPIPE_SZ;
568 pipe(my ($b_r, $b_w)) or die "pipe: $!";
569 fcntl($b_r, $F_SETPIPE_SZ, 4096) if $F_SETPIPE_SZ;
570 $l2m->{au_peers} = [ $a_r, $a_w, $b_r, $b_w ];
572 $l2m->wq_workers_start('lei2mail', undef,
573 $lei->oldset, { lei => $lei });
574 pipe($lei->{startq}, $lei->{au_done}) or die "pipe: $!";
575 fcntl($lei->{startq}, $F_SETPIPE_SZ, 4096) if $F_SETPIPE_SZ;
576 delete $l2m->{au_peers};
578 $self->wq_workers_start('lei_xsearch', undef,
579 $lei->oldset, { lei => $lei });
580 my $op_c = delete $lei->{pkt_op_c};
581 delete $lei->{pkt_op_p};
583 $self->{opt_threads} = $lei->{opt}->{threads};
584 $self->{opt_sort} = $lei->{opt}->{'sort'};
585 $self->{-do_lcat} = !!(delete $lei->{lcat_todo});
587 $l2m->net_merge_all_done($lei) unless $lei->{auth};
591 $lei->event_step_init; # wait for shutdowns
592 $lei->wait_wq_events($op_c, $ops);
596 my ($self, $uri) = @_;
597 if (my $curl = $self->{curl} //= which('curl') // 0) {
598 require PublicInbox::MboxReader;
599 require IO::Uncompress::Gunzip;
600 require PublicInbox::LeiCurl;
601 push @{$self->{remotes}}, $uri;
603 warn "curl missing, ignoring $uri\n";
607 sub prepare_external {
608 my ($self, $loc, $boost) = @_; # n.b. already ordered by boost
609 if (ref $loc) { # already a URI, or PublicInbox::Inbox-like object
610 return add_uri($self, $loc) if $loc->can('scheme');
611 } elsif ($loc =~ m!\Ahttps?://!) {
613 return add_uri($self, URI->new($loc));
614 } elsif (-f "$loc/ei.lock") {
615 require PublicInbox::ExtSearch;
616 die "`\\n' not allowed in `$loc'\n" if index($loc, "\n") >= 0;
617 $loc = PublicInbox::ExtSearch->new($loc);
618 } elsif (-f "$loc/inbox.lock" || -d "$loc/public-inbox") {
619 die "`\\n' not allowed in `$loc'\n" if index($loc, "\n") >= 0;
620 require PublicInbox::Inbox; # v2, v1
621 $loc = bless { inboxdir => $loc }, 'PublicInbox::Inbox';
623 warn "W: $loc gone, perhaps run: lei forget-external $loc\n";
626 warn "W: $loc ignored, unable to determine external type\n";
629 push @{$self->{locals}}, $loc;
632 sub _lcat_i { # LeiMailSync->each_src iterator callback
633 my ($oidbin, $id, $each_smsg) = @_;
634 $each_smsg->({blob => unpack('H*', $oidbin), pct => 100});
637 sub _lcat2smsg { # git->cat_async callback
638 my ($bref, $oid, $type, $size, $smsg) = @_;
640 my $eml = PublicInbox::Eml->new($bref);
641 my $json_dump = delete $smsg->{-json_dump};
642 bless $smsg, 'PublicInbox::Smsg';
643 _smsg_fill($smsg, $eml);
644 $json_dump->($smsg, undef, $eml);
648 sub lcat_dump { # via wq_io_do
650 my $lei = $self->{lei};
651 my $each_smsg = $lei->{ovv}->ovv_each_smsg_cb($lei);
652 my $git = $lei->{ale}->git;
654 my $json_dump = $each_smsg;
657 $smsg->{-json_dump} = $json_dump;
658 $git->cat_async($smsg->{blob}, \&_lcat2smsg, $smsg);
662 for my $ent (@{$lei->{lcat_todo}}) {
663 if (ref $ent eq 'HASH') { # { fid => $fid ,.. }
664 $lms //= $lei->{lse}->lms;
665 $lms->each_src($ent, \&_lcat_i, $each_smsg);
667 $each_smsg->({ blob => $ent, pct => 100 });
670 $git->async_wait_all;
671 undef $each_smsg; # may commit
672 $lei->{ovv}->ovv_atexit_child($lei);