1 # Copyright (C) 2015-2020 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
3 # based on notmuch, but with no concept of folders, files or flags
5 # Indexes mail with Xapian and our (SQLite-based) ::Msgmap for use
6 # with the web and NNTP interfaces. This index maintains thread
7 # relationships for use by PublicInbox::SearchThread.
8 # This writes to the search index.
9 package PublicInbox::SearchIdx;
12 use base qw(PublicInbox::Search PublicInbox::Lock);
14 use PublicInbox::InboxWritable;
15 use PublicInbox::MID qw/mid_clean mid_mime mids_for_index/;
16 use PublicInbox::MsgIter;
18 use POSIX qw(strftime);
19 use PublicInbox::OverIdx;
20 use PublicInbox::Spawn qw(spawn);
21 use PublicInbox::Git qw(git_unquote);
22 use PublicInbox::MsgTime qw(msg_timestamp msg_datestamp);
23 my $X = \%PublicInbox::Search::X;
24 my ($DB_CREATE_OR_OPEN, $DB_OPEN);
26 BATCH_BYTES => defined($ENV{XAPIAN_FLUSH_THRESHOLD}) ?
27 0x7fffffff : 1_000_000,
28 DEBUG => !!$ENV{DEBUG},
31 my $xapianlevels = qr/\A(?:full|medium)\z/;
34 my ($class, $ibx, $creat, $shard) = @_;
35 ref $ibx or die "BUG: expected PublicInbox::Inbox object: $ibx";
36 my $levels = qr/\A(?:full|medium|basic)\z/;
37 my $inboxdir = $ibx->{inboxdir};
38 my $version = $ibx->version;
39 my $indexlevel = 'full';
40 my $altid = $ibx->{altid};
42 require PublicInbox::AltId;
43 $altid = [ map { PublicInbox::AltId->new($ibx, $_); } @$altid ];
45 if ($ibx->{indexlevel}) {
46 if ($ibx->{indexlevel} =~ $levels) {
47 $indexlevel = $ibx->{indexlevel};
49 die("Invalid indexlevel $ibx->{indexlevel}\n");
52 $ibx = PublicInbox::InboxWritable->new($ibx);
54 inboxdir => $inboxdir,
59 indexlevel => $indexlevel,
61 $self->{-set_indexlevel_once} = 1 if $indexlevel eq 'medium';
64 $self->{lock_path} = "$inboxdir/ssoma.lock";
65 my $dir = $self->xdir;
66 $self->{over} = PublicInbox::OverIdx->new("$dir/over.sqlite3");
67 $self->{index_max_size} = $ibx->{index_max_size};
68 } elsif ($version == 2) {
69 defined $shard or die "shard is required for v2\n";
71 $self->{shard} = $shard;
72 $self->{lock_path} = undef;
74 die "unsupported inbox version=$version\n";
76 $self->{creat} = ($creat || 0) == 1;
80 sub need_xapian ($) { $_[0]->{indexlevel} =~ $xapianlevels }
84 if (need_xapian($self)) {
85 my $xdb = delete $self->{xdb} or croak 'not acquired';
88 $self->lock_release if $self->{creat};
92 sub load_xapian_writable () {
93 return 1 if $X->{WritableDatabase};
94 PublicInbox::Search::load_xapian() or return;
95 my $xap = $PublicInbox::Search::Xap;
96 for (qw(Document TermGenerator WritableDatabase)) {
97 $X->{$_} = $xap.'::'.$_;
99 eval 'require '.$X->{WritableDatabase} or die;
100 *sortable_serialise = $xap.'::sortable_serialise';
101 $DB_CREATE_OR_OPEN = eval($xap.'::DB_CREATE_OR_OPEN()');
102 $DB_OPEN = eval($xap.'::DB_OPEN()');
109 my $dir = $self->xdir;
110 if (need_xapian($self)) {
111 croak 'already acquired' if $self->{xdb};
112 load_xapian_writable();
113 $flag = $self->{creat} ? $DB_CREATE_OR_OPEN : $DB_OPEN;
115 if ($self->{creat}) {
119 # don't create empty Xapian directories if we don't need Xapian
120 my $is_shard = defined($self->{shard});
121 if (!$is_shard || ($is_shard && need_xapian($self))) {
122 File::Path::mkpath($dir);
125 return unless defined $flag;
126 my $xdb = eval { ($X->{WritableDatabase})->new($dir, $flag) };
128 die "Failed opening $dir: ", $@;
134 my ($doc, $col, $num) = @_;
135 $num = sortable_serialise($num);
136 $doc->add_value($col, $num);
139 sub term_generator ($) { # write-only
142 $self->{term_generator} //= do {
143 my $tg = $X->{TermGenerator}->new;
144 $tg->set_stemmer($self->stemmer);
149 sub index_text ($$$$) {
150 my ($self, $text, $wdf_inc, $prefix) = @_;
151 my $tg = term_generator($self); # man Search::Xapian::TermGenerator
153 if ($self->{indexlevel} eq 'full') {
154 $tg->index_text($text, $wdf_inc, $prefix);
155 $tg->increase_termpos;
157 $tg->index_text_without_positions($text, $wdf_inc, $prefix);
161 sub index_users ($$) {
162 my ($self, $smsg) = @_;
164 my $from = $smsg->from;
168 index_text($self, $from, 1, 'A'); # A - author
169 index_text($self, $to, 1, 'XTO') if $to ne '';
170 index_text($self, $cc, 1, 'XCC') if $cc ne '';
173 sub index_diff_inc ($$$$) {
174 my ($self, $text, $pfx, $xnq) = @_;
176 index_text($self, join("\n", @$xnq), 1, 'XNQ');
179 index_text($self, $text, 1, $pfx);
182 sub index_old_diff_fn {
183 my ($self, $seen, $fa, $fb, $xnq) = @_;
185 # no renames or space support for traditional diffs,
186 # find the number of leading common paths to strip:
187 my @fa = split('/', $fa);
188 my @fb = split('/', $fb);
189 while (scalar(@fa) && scalar(@fb)) {
190 $fa = join('/', @fa);
191 $fb = join('/', @fb);
193 unless ($seen->{$fa}++) {
194 index_diff_inc($self, $fa, 'XDFN', $xnq);
204 sub index_diff ($$$) {
205 my ($self, $txt, $doc) = @_;
210 foreach (split(/\n/, $txt)) {
211 if ($in_diff && s/^ //) { # diff context
212 index_diff_inc($self, $_, 'XDFCTX', $xnq);
213 } elsif (/^-- $/) { # email signature begins
215 } elsif (m!^diff --git "?[^/]+/.+ "?[^/]+/.+\z!) {
216 # wait until "---" and "+++" to capture filenames
219 } elsif (m/^diff -(.+) (\S+) (\S+)$/) {
220 my ($opt, $fa, $fb) = ($1, $2, $3);
222 # only support unified:
223 next unless $opt =~ /[uU]/;
224 $in_diff = index_old_diff_fn($self, \%seen, $fa, $fb,
226 } elsif (m!^--- ("?[^/]+/.+)!) {
228 $fn = (split('/', git_unquote($fn), 2))[1];
229 $seen{$fn}++ or index_diff_inc($self, $fn, 'XDFN', $xnq);
231 } elsif (m!^\+\+\+ ("?[^/]+/.+)!) {
233 $fn = (split('/', git_unquote($fn), 2))[1];
234 $seen{$fn}++ or index_diff_inc($self, $fn, 'XDFN', $xnq);
236 } elsif (/^--- (\S+)/) {
239 } elsif (defined $in_diff && /^\+\+\+ (\S+)/) {
240 $in_diff = index_old_diff_fn($self, \%seen, $in_diff,
242 } elsif ($in_diff && s/^\+//) { # diff added
243 index_diff_inc($self, $_, 'XDFB', $xnq);
244 } elsif ($in_diff && s/^-//) { # diff removed
245 index_diff_inc($self, $_, 'XDFA', $xnq);
246 } elsif (m!^index ([a-f0-9]+)\.\.([a-f0-9]+)!) {
247 my ($ba, $bb) = ($1, $2);
248 index_git_blob_id($doc, 'XDFPRE', $ba);
249 index_git_blob_id($doc, 'XDFPOST', $bb);
251 } elsif (/^@@ (?:\S+) (?:\S+) @@\s*$/) {
252 # traditional diff w/o -p
253 } elsif (/^@@ (?:\S+) (?:\S+) @@\s*(\S+.*)$/) {
254 # hunk header context
255 index_diff_inc($self, $1, 'XDFHH', $xnq);
256 # ignore the following lines:
257 } elsif (/^(?:dis)similarity index/ ||
258 /^(?:old|new) mode/ ||
259 /^(?:deleted|new) file mode/ ||
260 /^(?:copy|rename) (?:from|to) / ||
261 /^(?:dis)?similarity index / ||
262 /^\\ No newline at end of file/ ||
263 /^Binary files .* differ/) {
266 # possible to be in diff context, some mail may be
267 # stripped by MUA or even GNU diff(1). "git apply"
268 # treats a bare "\n" as diff context, too
271 warn "non-diff line: $_\n" if DEBUG && $_ ne '';
276 index_text($self, join("\n", @xnq), 1, 'XNQ');
279 sub index_xapian { # msg_iter callback
280 my $part = $_[0]->[0]; # ignore $depth and $idx
281 my ($self, $doc) = @{$_[1]};
282 my $ct = $part->content_type || 'text/plain';
283 my $fn = $part->filename;
284 if (defined $fn && $fn ne '') {
285 index_text($self, $fn, 1, 'XFN');
287 if ($part->{is_submsg}) {
288 my $mids = mids_for_index($part);
289 index_ids($self, $doc, $part, $mids);
290 my $smsg = PublicInbox::Smsg->new($part);
291 index_users($self, $smsg);
292 index_text($self, $smsg->subject, 1, 'S') if $smsg->subject;
295 my ($s, undef) = msg_part_text($part, $ct);
296 defined $s or return;
297 $_[0]->[0] = $part = undef; # free memory
299 # split off quoted and unquoted blocks:
300 my @sections = PublicInbox::MsgIter::split_quotes($s);
301 undef $s; # free memory
302 for my $txt (@sections) {
304 index_text($self, $txt, 0, 'XQUOT');
306 # does it look like a diff?
307 if ($txt =~ /^(?:diff|---|\+\+\+) /ms) {
308 index_diff($self, $txt, $doc);
310 index_text($self, $txt, 1, 'XNQ');
313 undef $txt; # free memory
317 sub index_ids ($$$$) {
318 my ($self, $doc, $hdr, $mids) = @_;
319 for my $mid (@$mids) {
320 index_text($self, $mid, 1, 'XM');
322 # because too many Message-IDs are prefixed with
324 if ($mid =~ /\w{12,}/) {
325 my @long = ($mid =~ /(\w{3,}+)/g);
326 index_text($self, join(' ', @long), 1, 'XM');
329 $doc->add_boolean_term('Q' . $_) for @$mids;
330 for my $l ($hdr->header_raw('List-Id')) {
331 $l =~ /<([^>]+)>/ or next;
333 $doc->add_boolean_term('G' . $lid);
334 index_text($self, $lid, 1, 'XL'); # probabilistic
338 sub add_xapian ($$$$) {
339 my ($self, $mime, $smsg, $mids) = @_;
340 $smsg->{mime} = $mime; # XXX dangerous
341 my $hdr = $mime->header_obj;
342 my $doc = $X->{Document}->new;
343 my $subj = $smsg->subject;
344 add_val($doc, PublicInbox::Search::TS(), $smsg->{ts});
345 my @ds = gmtime($smsg->{ds});
346 my $yyyymmdd = strftime('%Y%m%d', @ds);
347 add_val($doc, PublicInbox::Search::YYYYMMDD(), $yyyymmdd);
348 my $dt = strftime('%Y%m%d%H%M%S', @ds);
349 add_val($doc, PublicInbox::Search::DT(), $dt);
351 my $tg = term_generator($self);
352 $tg->set_document($doc);
353 index_text($self, $subj, 1, 'S') if $subj;
354 index_users($self, $smsg);
356 msg_iter($mime, \&index_xapian, [ $self, $doc ]);
357 index_ids($self, $doc, $hdr, $mids);
358 $smsg->{to} = $smsg->{cc} = ''; # WWW doesn't need these, only NNTP
359 PublicInbox::OverIdx::parse_references($smsg, $hdr, $mids);
360 my $data = $smsg->to_doc_data;
361 $doc->set_data($data);
362 if (my $altid = $self->{-altid}) {
363 foreach my $alt (@$altid) {
364 my $pfx = $alt->{xprefix};
365 foreach my $mid (@$mids) {
366 my $id = $alt->mid2alt($mid);
367 next unless defined $id;
368 $doc->add_boolean_term($pfx . $id);
372 $self->{xdb}->replace_document($smsg->{num}, $doc);
375 sub _msgmap_init ($) {
377 die "BUG: _msgmap_init is only for v1\n" if $self->{ibx_ver} != 1;
378 $self->{mm} //= eval {
379 require PublicInbox::Msgmap;
380 PublicInbox::Msgmap->new($self->{inboxdir}, 1);
385 # mime = PublicInbox::Eml or Email::MIME object
386 my ($self, $mime, $smsg) = @_;
387 my $hdr = $mime->header_obj;
388 my $mids = mids_for_index($hdr);
389 $smsg //= bless { blob => '' }, 'PublicInbox::Smsg'; # test-only compat
390 $smsg->{mid} //= $mids->[0]; # v1 compatibility
391 $smsg->{num} //= do { # v1
393 index_mm($self, $mime);
397 $smsg->{ds} //= msg_datestamp($hdr, $self->{autime});
398 $smsg->{ts} //= msg_timestamp($hdr, $self->{cotime});
401 # order matters, overview stores every possible piece of
402 # data in doc_data (deflated). Xapian only stores a subset
403 # of the fields which exist in over.sqlite3. We may stop
404 # storing doc_data in Xapian sometime after we get multi-inbox
406 if (my $over = $self->{over}) { # v1 only
407 $over->add_overview($mime, $smsg);
409 if (need_xapian($self)) {
410 add_xapian($self, $mime, $smsg, $mids);
415 warn "failed to index message <".join('> <',@$mids).">: $@\n";
421 # returns begin and end PostingIterator
423 my ($self, $termval) = @_;
424 my $db = $self->{xdb};
426 ($db->postlist_begin($termval), $db->postlist_end($termval));
431 my ($self, $termval, $cb) = @_;
432 my $batch_size = 1000; # don't let @ids grow too large to avoid OOM
434 my ($head, $tail) = $self->find_doc_ids($termval);
435 return if $head == $tail;
437 for (; $head != $tail && @ids < $batch_size; $head++) {
438 push @ids, $head->get_docid;
444 # v1 only, where $mid is unique
446 my ($self, $mid) = @_;
447 $mid = mid_clean($mid);
449 if (my $over = $self->{over}) {
450 my $nr = eval { $over->remove_oid(undef, $mid) };
452 warn "failed to remove <$mid> from overview: $@\n";
454 warn "<$mid> missing for removal from overview\n";
457 return unless need_xapian($self);
458 my $db = $self->{xdb};
461 batch_do($self, 'Q' . $mid, sub {
463 $db->delete_document($_) for @$ids;
468 warn "failed to remove <$mid> from Xapian: $@\n";
470 warn "<$mid> missing for removal from Xapian\n";
474 # MID is a hint in V2
476 my ($self, $oid, $mid) = @_;
478 $self->{over}->remove_oid($oid, $mid) if $self->{over};
480 return unless need_xapian($self);
481 my $db = $self->{xdb};
483 # XXX careful, we cannot use batch_do here since we conditionally
484 # delete documents based on other factors, so we cannot call
485 # find_doc_ids twice.
486 my ($head, $tail) = $self->find_doc_ids('Q' . $mid);
487 return if $head == $tail;
489 # there is only ONE element in @delete unless we
490 # have bugs in our v2writable deduplication check
492 for (; $head != $tail; $head++) {
493 my $docid = $head->get_docid;
494 my $doc = $db->get_document($docid);
495 my $smsg = PublicInbox::Smsg->wrap($mid);
496 $smsg->load_expand($doc);
497 if ($smsg->{blob} eq $oid) {
498 push(@delete, $docid);
501 $db->delete_document($_) foreach @delete;
505 sub index_git_blob_id {
506 my ($doc, $pfx, $objid) = @_;
508 my $len = length($objid);
509 for (my $len = length($objid); $len >= 7; ) {
510 $doc->add_term($pfx.$objid);
511 $objid = substr($objid, 0, --$len);
516 my ($self, $mime) = @_;
517 my $mid = eval { mid_mime($mime) };
518 $self->remove_message($mid) if defined $mid;
522 my ($self, $mime) = @_;
523 my $mid = mid_mime($mime);
524 my $mm = $self->{mm};
527 if (defined $self->{regen_down}) {
528 $num = $mm->num_for($mid) and return $num;
530 while (($num = $self->{regen_down}--) > 0) {
531 if ($mm->mid_set($num, $mid) != 0) {
535 } elsif (defined $self->{regen_up}) {
536 $num = $mm->num_for($mid) and return $num;
538 # this is to fixup old bugs due to add-remove-add
539 while (($num = ++$self->{regen_up})) {
540 if ($mm->mid_set($num, $mid) != 0) {
546 $num = $mm->mid_insert($mid) and return $num;
548 # fallback to num_for since filters like RubyLang set the number
553 my ($self, $mime) = @_;
554 $self->{mm}->mid_delete(mid_mime($mime));
558 my ($self, $mime, $smsg) = @_;
559 my $num = index_mm($self, $mime);
561 add_message($self, $mime, $smsg);
565 my ($self, $mime) = @_;
566 unindex_blob($self, $mime);
567 unindex_mm($self, $mime);
571 my ($git, $blob, $sizeref) = @_;
572 my $str = $git->cat_file($blob, $sizeref) or
573 die "BUG: $blob not found in $git->{git_dir}";
574 PublicInbox::Eml->new($str);
577 # called by public-inbox-index
579 my ($self, $opts) = @_;
580 delete $self->{lock_path} if $opts->{-skip_lock};
581 $self->{-inbox}->with_umask(sub { $self->_index_sync($opts) })
584 sub batch_adjust ($$$$$) {
585 my ($max, $bytes, $batch_cb, $latest, $nr) = @_;
589 $batch_cb->($nr, $latest);
594 my ($self, $git, $oid) = @_;
595 my $max_size = $self->{index_max_size} or return;
596 my (undef, undef, $size) = $git->check($oid);
597 die "E: bad $oid in $git->{git_dir}\n" if !defined($size);
598 return if $size <= $max_size;
599 warn "W: skipping $oid ($size > $max_size)\n";
605 my ($self, $log, $add_cb, $del_cb, $batch_cb) = @_;
606 my $hex = '[a-f0-9]';
607 my $h40 = $hex .'{40}';
608 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
609 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
610 my $git = $self->{git};
613 my $max = BATCH_BYTES;
619 while (defined($line = <$log>)) {
620 if ($line =~ /$addmsg/o) {
622 if (delete $D{$blob}) {
623 if (defined $self->{regen_down}) {
624 my $num = $self->{regen_down}--;
625 $self->{mm}->num_highwater($num);
629 next if too_big($self, $git, $blob);
630 my $mime = do_cat_mail($git, $blob, \$bytes);
631 my $smsg = bless {}, 'PublicInbox::Smsg';
632 batch_adjust(\$max, $bytes, $batch_cb, $latest, ++$nr);
633 $smsg->{blob} = $blob;
634 $smsg->{bytes} = $bytes;
635 $add_cb->($self, $mime, $smsg);
636 } elsif ($line =~ /$delmsg/o) {
638 $D{$blob} = 1 unless too_big($self, $git, $blob);
639 } elsif ($line =~ /^commit ($h40)/o) {
642 } elsif ($line =~ /^author .*? ([0-9]+) [\-\+][0-9]+$/) {
643 $self->{autime} = $1;
644 } elsif ($line =~ /^committer .*? ([0-9]+) [\-\+][0-9]+$/) {
645 $self->{cotime} = $1;
648 close($log) or die "git log failed: \$?=$?";
650 foreach my $blob (keys %D) {
651 my $mime = do_cat_mail($git, $blob, \$bytes);
652 $del_cb->($self, $mime);
654 $batch_cb->($nr, $latest, $newest);
658 my ($self, $opts, $range) = @_;
659 my $git = $self->{git};
661 if (index($range, '..') < 0) {
662 # don't show annoying git errors to users who run -index
664 $git->qx(qw(rev-parse -q --verify), "$range^0");
666 open my $fh, '<', '/dev/null' or
667 die "failed to open /dev/null: $!\n";
672 # Count the new files so they can be added newest to oldest
673 # and still have numbers increasing from oldest to newest
675 my $pr = $opts->{-progress};
676 $pr->("counting changes\n\t$range ... ") if $pr;
677 # can't use 'rev-list --count' if we use --diff-filter
678 my $fh = $git->popen(qw(log --pretty=tformat:%h
679 --no-notes --no-color --no-renames
680 --diff-filter=AM), $range);
681 ++$fcount while <$fh>;
682 close $fh or die "git log failed: \$?=$?";
683 my $high = $self->{mm}->num_highwater;
684 $pr->("$fcount\n") if $pr; # continue previous line
685 $self->{ntodo} = $fcount;
687 if (index($range, '..') < 0) {
688 if ($high && $high == $fcount) {
689 # fix up old bugs in full indexes which caused messages to
690 # not appear in Msgmap
691 $self->{regen_up} = $high;
693 # normal regen is for for fresh data
694 $self->{regen_down} = $fcount;
697 # Give oldest messages the smallest numbers
698 $self->{regen_down} = $high + $fcount;
701 $git->popen(qw/log --pretty=raw --no-notes --no-color --no-renames
702 --raw -r --no-abbrev/, $range);
705 # --is-ancestor requires git 1.8.0+
706 sub is_ancestor ($$$) {
707 my ($git, $cur, $tip) = @_;
708 return 0 unless $git->check($cur);
709 my $cmd = [ 'git', "--git-dir=$git->{git_dir}",
710 qw(merge-base --is-ancestor), $cur, $tip ];
711 my $pid = spawn($cmd);
712 waitpid($pid, 0) == $pid or die join(' ', @$cmd) .' did not finish';
716 sub need_update ($$$) {
717 my ($self, $cur, $new) = @_;
718 my $git = $self->{git};
719 return 1 if $cur && !is_ancestor($git, $cur, $new);
720 my $range = $cur eq '' ? $new : "$cur..$new";
721 chomp(my $n = $git->qx(qw(rev-list --count), $range));
722 ($n eq '' || $n > 0);
725 # The last git commit we indexed with Xapian or SQLite (msgmap)
726 # This needs to account for cases where Xapian or SQLite is
727 # out-of-date with respect to the other.
729 my ($self, $mm) = @_;
730 my $lm = $mm->last_commit || '';
732 if (need_xapian($self)) {
733 $lx = $self->{xdb}->get_metadata('last_commit') || '';
737 # Use last_commit from msgmap if it is older or unset
738 if (!$lm || ($lx && $lm && is_ancestor($self->{git}, $lm, $lx))) {
744 sub reindex_from ($$) {
745 my ($reindex, $last_commit) = @_;
746 return $last_commit unless $reindex;
747 ref($reindex) eq 'HASH' ? $reindex->{from} : '';
750 # indexes all unindexed messages (v1 only)
752 my ($self, $opts) = @_;
753 my $tip = $opts->{ref} || 'HEAD';
754 my ($last_commit, $lx, $xlog);
755 my $git = $self->{git};
757 my $pr = $opts->{-progress};
759 my $xdb = $self->begin_txn_lazy;
760 my $mm = _msgmap_init($self);
763 close($xlog) or die "git log failed: \$?=$?";
766 $last_commit = _last_x_commit($self, $mm);
767 $lx = reindex_from($opts->{reindex}, $last_commit);
769 $self->{over}->rollback_lazy;
770 $self->{over}->disconnect;
773 $xdb->cancel_transaction if $xdb;
774 $xdb = _xdb_release($self);
776 # ensure we leak no FDs to "git log" with Xapian <= 1.2
777 my $range = $lx eq '' ? $tip : "$lx..$tip";
778 $xlog = _git_log($self, $opts, $range);
780 $xdb = $self->begin_txn_lazy;
781 } while (_last_x_commit($self, $mm) ne $last_commit);
783 my $dbh = $mm->{dbh} if $mm;
785 my ($nr, $commit, $newest) = @_;
788 my $cur = $mm->last_commit || '';
789 if (need_update($self, $cur, $newest)) {
790 $mm->last_commit($newest);
795 if ($newest && need_xapian($self)) {
796 my $cur = $xdb->get_metadata('last_commit');
797 if (need_update($self, $cur, $newest)) {
798 $xdb->set_metadata('last_commit', $newest);
801 $self->commit_txn_lazy;
803 $xdb = _xdb_release($self);
804 # let another process do some work... <
805 $pr->("indexed $nr/$self->{ntodo}\n") if $pr && $nr;
807 $xdb = $self->begin_txn_lazy;
808 $dbh->begin_work if $dbh;
813 read_log($self, $xlog, *index_both, *unindex_both, $cb);
817 # order matters for unlocking
818 $_[0]->{xdb} = undef;
819 $_[0]->{lockfh} = undef;
822 # remote_* subs are only used by SearchIdxPart
825 if (my $w = $self->{w}) {
826 print $w "commit\n" or die "failed to write commit: $!";
828 $self->commit_txn_lazy;
834 if (my $w = delete $self->{w}) {
835 my $pid = delete $self->{pid} or die "no process to wait on\n";
836 print $w "close\n" or die "failed to write to pid:$pid: $!\n";
837 close $w or die "failed to close pipe for pid:$pid: $!\n";
838 waitpid($pid, 0) == $pid or die "remote process did not finish";
839 $? == 0 or die ref($self)." pid:$pid exited with: $?";
841 die "transaction in progress $self\n" if $self->{txn};
842 $self->_xdb_release if $self->{xdb};
847 my ($self, $oid, $mid) = @_;
848 if (my $w = $self->{w}) {
849 # triggers remove_by_oid in a shard
850 print $w "D $oid $mid\n" or die "failed to write remove $!";
852 $self->begin_txn_lazy;
853 $self->remove_by_oid($oid, $mid);
859 return if $self->{txn};
861 $self->{-inbox}->with_umask(sub {
862 my $xdb = $self->{xdb} || $self->_xdb_acquire;
863 $self->{over}->begin_lazy if $self->{over};
864 $xdb->begin_transaction if $xdb;
870 # store 'indexlevel=medium' in v2 shard=0 and v1 (only one shard)
871 # This metadata is read by Admin::detect_indexlevel:
875 if (!$self->{shard} && # undef or 0, not >0
876 delete($self->{-set_indexlevel_once})) {
877 my $xdb = $self->{xdb};
878 my $level = $xdb->get_metadata('indexlevel');
879 if (!$level || $level ne 'medium') {
880 $xdb->set_metadata('indexlevel', 'medium');
885 sub commit_txn_lazy {
887 delete $self->{txn} or return;
888 $self->{-inbox}->with_umask(sub {
889 if (my $xdb = $self->{xdb}) {
890 set_indexlevel($self);
891 $xdb->commit_transaction;
893 $self->{over}->commit_lazy if $self->{over};
899 if (need_xapian($self)) {
900 die "$$ $0 xdb not released\n" if $self->{xdb};
902 die "$$ $0 still in transaction\n" if $self->{txn};