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);
25 our $BATCH_BYTES = defined($ENV{XAPIAN_FLUSH_THRESHOLD}) ?
26 0x7fffffff : 1_000_000;
27 use constant DEBUG => !!$ENV{DEBUG};
29 my $xapianlevels = qr/\A(?:full|medium)\z/;
32 my ($class, $ibx, $creat, $shard) = @_;
33 ref $ibx or die "BUG: expected PublicInbox::Inbox object: $ibx";
34 my $levels = qr/\A(?:full|medium|basic)\z/;
35 my $inboxdir = $ibx->{inboxdir};
36 my $version = $ibx->version;
37 my $indexlevel = 'full';
38 my $altid = $ibx->{altid};
40 require PublicInbox::AltId;
41 $altid = [ map { PublicInbox::AltId->new($ibx, $_); } @$altid ];
43 if ($ibx->{indexlevel}) {
44 if ($ibx->{indexlevel} =~ $levels) {
45 $indexlevel = $ibx->{indexlevel};
47 die("Invalid indexlevel $ibx->{indexlevel}\n");
50 $ibx = PublicInbox::InboxWritable->new($ibx);
52 inboxdir => $inboxdir,
57 indexlevel => $indexlevel,
59 $self->{-set_indexlevel_once} = 1 if $indexlevel eq 'medium';
62 $self->{lock_path} = "$inboxdir/ssoma.lock";
63 my $dir = $self->xdir;
64 $self->{over} = PublicInbox::OverIdx->new("$dir/over.sqlite3");
65 $self->{index_max_size} = $ibx->{index_max_size};
66 } elsif ($version == 2) {
67 defined $shard or die "shard is required for v2\n";
69 $self->{shard} = $shard;
70 $self->{lock_path} = undef;
72 die "unsupported inbox version=$version\n";
74 $self->{creat} = ($creat || 0) == 1;
78 sub need_xapian ($) { $_[0]->{indexlevel} =~ $xapianlevels }
82 if (need_xapian($self)) {
83 my $xdb = delete $self->{xdb} or croak 'not acquired';
86 $self->lock_release if $self->{creat};
90 sub load_xapian_writable () {
91 return 1 if $X->{WritableDatabase};
92 PublicInbox::Search::load_xapian() or return;
93 my $xap = $PublicInbox::Search::Xap;
94 for (qw(Document TermGenerator WritableDatabase)) {
95 $X->{$_} = $xap.'::'.$_;
97 eval 'require '.$X->{WritableDatabase} or die;
98 *sortable_serialise = $xap.'::sortable_serialise';
99 $DB_CREATE_OR_OPEN = eval($xap.'::DB_CREATE_OR_OPEN()');
100 $DB_OPEN = eval($xap.'::DB_OPEN()');
107 my $dir = $self->xdir;
108 if (need_xapian($self)) {
109 croak 'already acquired' if $self->{xdb};
110 load_xapian_writable();
111 $flag = $self->{creat} ? $DB_CREATE_OR_OPEN : $DB_OPEN;
113 if ($self->{creat}) {
117 # don't create empty Xapian directories if we don't need Xapian
118 my $is_shard = defined($self->{shard});
119 if (!$is_shard || ($is_shard && need_xapian($self))) {
120 File::Path::mkpath($dir);
123 return unless defined $flag;
124 my $xdb = eval { ($X->{WritableDatabase})->new($dir, $flag) };
126 die "Failed opening $dir: ", $@;
132 my ($doc, $col, $num) = @_;
133 $num = sortable_serialise($num);
134 $doc->add_value($col, $num);
137 sub term_generator ($) { # write-only
140 $self->{term_generator} //= do {
141 my $tg = $X->{TermGenerator}->new;
142 $tg->set_stemmer($self->stemmer);
147 sub index_text ($$$$) {
148 my ($self, $text, $wdf_inc, $prefix) = @_;
149 my $tg = term_generator($self); # man Search::Xapian::TermGenerator
151 if ($self->{indexlevel} eq 'full') {
152 $tg->index_text($text, $wdf_inc, $prefix);
153 $tg->increase_termpos;
155 $tg->index_text_without_positions($text, $wdf_inc, $prefix);
159 sub index_headers ($$) {
160 my ($self, $smsg) = @_;
161 my @x = (from => 'A', # Author
162 subject => 'S', to => 'XTO', cc => 'XCC');
163 while (my ($field, $pfx) = splice(@x, 0, 2)) {
164 my $val = $smsg->{$field};
165 index_text($self, $val, 1, $pfx) if $val ne '';
169 sub index_diff_inc ($$$$) {
170 my ($self, $text, $pfx, $xnq) = @_;
172 index_text($self, join("\n", @$xnq), 1, 'XNQ');
175 index_text($self, $text, 1, $pfx);
178 sub index_old_diff_fn {
179 my ($self, $seen, $fa, $fb, $xnq) = @_;
181 # no renames or space support for traditional diffs,
182 # find the number of leading common paths to strip:
183 my @fa = split('/', $fa);
184 my @fb = split('/', $fb);
185 while (scalar(@fa) && scalar(@fb)) {
186 $fa = join('/', @fa);
187 $fb = join('/', @fb);
189 unless ($seen->{$fa}++) {
190 index_diff_inc($self, $fa, 'XDFN', $xnq);
200 sub index_diff ($$$) {
201 my ($self, $txt, $doc) = @_;
206 foreach (split(/\n/, $txt)) {
207 if ($in_diff && s/^ //) { # diff context
208 index_diff_inc($self, $_, 'XDFCTX', $xnq);
209 } elsif (/^-- $/) { # email signature begins
211 } elsif (m!^diff --git "?[^/]+/.+ "?[^/]+/.+\z!) {
212 # wait until "---" and "+++" to capture filenames
215 } elsif (m/^diff -(.+) (\S+) (\S+)$/) {
216 my ($opt, $fa, $fb) = ($1, $2, $3);
218 # only support unified:
219 next unless $opt =~ /[uU]/;
220 $in_diff = index_old_diff_fn($self, \%seen, $fa, $fb,
222 } elsif (m!^--- ("?[^/]+/.+)!) {
224 $fn = (split('/', git_unquote($fn), 2))[1];
225 $seen{$fn}++ or index_diff_inc($self, $fn, 'XDFN', $xnq);
227 } elsif (m!^\+\+\+ ("?[^/]+/.+)!) {
229 $fn = (split('/', git_unquote($fn), 2))[1];
230 $seen{$fn}++ or index_diff_inc($self, $fn, 'XDFN', $xnq);
232 } elsif (/^--- (\S+)/) {
235 } elsif (defined $in_diff && /^\+\+\+ (\S+)/) {
236 $in_diff = index_old_diff_fn($self, \%seen, $in_diff,
238 } elsif ($in_diff && s/^\+//) { # diff added
239 index_diff_inc($self, $_, 'XDFB', $xnq);
240 } elsif ($in_diff && s/^-//) { # diff removed
241 index_diff_inc($self, $_, 'XDFA', $xnq);
242 } elsif (m!^index ([a-f0-9]+)\.\.([a-f0-9]+)!) {
243 my ($ba, $bb) = ($1, $2);
244 index_git_blob_id($doc, 'XDFPRE', $ba);
245 index_git_blob_id($doc, 'XDFPOST', $bb);
247 } elsif (/^@@ (?:\S+) (?:\S+) @@\s*$/) {
248 # traditional diff w/o -p
249 } elsif (/^@@ (?:\S+) (?:\S+) @@\s*(\S+.*)$/) {
250 # hunk header context
251 index_diff_inc($self, $1, 'XDFHH', $xnq);
252 # ignore the following lines:
253 } elsif (/^(?:dis)similarity index/ ||
254 /^(?:old|new) mode/ ||
255 /^(?:deleted|new) file mode/ ||
256 /^(?:copy|rename) (?:from|to) / ||
257 /^(?:dis)?similarity index / ||
258 /^\\ No newline at end of file/ ||
259 /^Binary files .* differ/) {
262 # possible to be in diff context, some mail may be
263 # stripped by MUA or even GNU diff(1). "git apply"
264 # treats a bare "\n" as diff context, too
267 warn "non-diff line: $_\n" if DEBUG && $_ ne '';
272 index_text($self, join("\n", @xnq), 1, 'XNQ');
275 sub index_xapian { # msg_iter callback
276 my $part = $_[0]->[0]; # ignore $depth and $idx
277 my ($self, $doc) = @{$_[1]};
278 my $ct = $part->content_type || 'text/plain';
279 my $fn = $part->filename;
280 if (defined $fn && $fn ne '') {
281 index_text($self, $fn, 1, 'XFN');
283 if ($part->{is_submsg}) {
284 my $mids = mids_for_index($part);
285 index_ids($self, $doc, $part, $mids);
286 my $smsg = bless {}, 'PublicInbox::Smsg';
287 $smsg->populate($part);
288 index_headers($self, $smsg);
291 my ($s, undef) = msg_part_text($part, $ct);
292 defined $s or return;
293 $_[0]->[0] = $part = undef; # free memory
295 # split off quoted and unquoted blocks:
296 my @sections = PublicInbox::MsgIter::split_quotes($s);
297 undef $s; # free memory
298 for my $txt (@sections) {
300 index_text($self, $txt, 0, 'XQUOT');
302 # does it look like a diff?
303 if ($txt =~ /^(?:diff|---|\+\+\+) /ms) {
304 index_diff($self, $txt, $doc);
306 index_text($self, $txt, 1, 'XNQ');
309 undef $txt; # free memory
313 sub index_ids ($$$$) {
314 my ($self, $doc, $hdr, $mids) = @_;
315 for my $mid (@$mids) {
316 index_text($self, $mid, 1, 'XM');
318 # because too many Message-IDs are prefixed with
320 if ($mid =~ /\w{12,}/) {
321 my @long = ($mid =~ /(\w{3,}+)/g);
322 index_text($self, join(' ', @long), 1, 'XM');
325 $doc->add_boolean_term('Q' . $_) for @$mids;
326 for my $l ($hdr->header_raw('List-Id')) {
327 $l =~ /<([^>]+)>/ or next;
329 $doc->add_boolean_term('G' . $lid);
330 index_text($self, $lid, 1, 'XL'); # probabilistic
334 sub add_xapian ($$$$) {
335 my ($self, $mime, $smsg, $mids) = @_;
336 my $hdr = $mime->header_obj;
337 my $doc = $X->{Document}->new;
338 add_val($doc, PublicInbox::Search::TS(), $smsg->{ts});
339 my @ds = gmtime($smsg->{ds});
340 my $yyyymmdd = strftime('%Y%m%d', @ds);
341 add_val($doc, PublicInbox::Search::YYYYMMDD(), $yyyymmdd);
342 my $dt = strftime('%Y%m%d%H%M%S', @ds);
343 add_val($doc, PublicInbox::Search::DT(), $dt);
344 add_val($doc, PublicInbox::Search::BYTES(), $smsg->{bytes});
346 my $tg = term_generator($self);
347 $tg->set_document($doc);
348 index_headers($self, $smsg);
350 msg_iter($mime, \&index_xapian, [ $self, $doc ]);
351 index_ids($self, $doc, $hdr, $mids);
352 $smsg->{to} = $smsg->{cc} = ''; # WWW doesn't need these, only NNTP
353 PublicInbox::OverIdx::parse_references($smsg, $hdr, $mids);
354 my $data = $smsg->to_doc_data;
355 $doc->set_data($data);
356 if (my $altid = $self->{-altid}) {
357 foreach my $alt (@$altid) {
358 my $pfx = $alt->{xprefix};
359 foreach my $mid (@$mids) {
360 my $id = $alt->mid2alt($mid);
361 next unless defined $id;
362 $doc->add_boolean_term($pfx . $id);
366 $self->{xdb}->replace_document($smsg->{num}, $doc);
369 sub _msgmap_init ($) {
371 die "BUG: _msgmap_init is only for v1\n" if $self->{ibx_ver} != 1;
372 $self->{mm} //= eval {
373 require PublicInbox::Msgmap;
374 PublicInbox::Msgmap->new($self->{inboxdir}, 1);
379 # mime = PublicInbox::Eml or Email::MIME object
380 my ($self, $mime, $smsg) = @_;
381 my $hdr = $mime->header_obj;
382 my $mids = mids_for_index($hdr);
383 $smsg //= bless { blob => '' }, 'PublicInbox::Smsg'; # test-only compat
384 $smsg->{mid} //= $mids->[0]; # v1 compatibility
385 $smsg->{num} //= do { # v1
387 index_mm($self, $mime);
391 $smsg->populate($hdr, $self);
392 $smsg->{bytes} //= length($mime->as_string);
395 # order matters, overview stores every possible piece of
396 # data in doc_data (deflated). Xapian only stores a subset
397 # of the fields which exist in over.sqlite3. We may stop
398 # storing doc_data in Xapian sometime after we get multi-inbox
400 if (my $over = $self->{over}) { # v1 only
401 $over->add_overview($mime, $smsg);
403 if (need_xapian($self)) {
404 add_xapian($self, $mime, $smsg, $mids);
409 warn "failed to index message <".join('> <',@$mids).">: $@\n";
415 # returns begin and end PostingIterator
417 my ($self, $termval) = @_;
418 my $db = $self->{xdb};
420 ($db->postlist_begin($termval), $db->postlist_end($termval));
425 my ($self, $termval, $cb) = @_;
426 my $batch_size = 1000; # don't let @ids grow too large to avoid OOM
428 my ($head, $tail) = $self->find_doc_ids($termval);
429 return if $head == $tail;
431 for (; $head != $tail && @ids < $batch_size; $head++) {
432 push @ids, $head->get_docid;
438 # v1 only, where $mid is unique
440 my ($self, $mid) = @_;
441 $mid = mid_clean($mid);
443 if (my $over = $self->{over}) {
444 my $nr = eval { $over->remove_oid(undef, $mid) };
446 warn "failed to remove <$mid> from overview: $@\n";
448 warn "<$mid> missing for removal from overview\n";
451 return unless need_xapian($self);
452 my $db = $self->{xdb};
455 batch_do($self, 'Q' . $mid, sub {
457 $db->delete_document($_) for @$ids;
462 warn "failed to remove <$mid> from Xapian: $@\n";
464 warn "<$mid> missing for removal from Xapian\n";
468 # MID is a hint in V2
470 my ($self, $oid, $mid) = @_;
472 $self->{over}->remove_oid($oid, $mid) if $self->{over};
474 return unless need_xapian($self);
475 my $db = $self->{xdb};
477 # XXX careful, we cannot use batch_do here since we conditionally
478 # delete documents based on other factors, so we cannot call
479 # find_doc_ids twice.
480 my ($head, $tail) = $self->find_doc_ids('Q' . $mid);
481 return if $head == $tail;
483 # there is only ONE element in @delete unless we
484 # have bugs in our v2writable deduplication check
486 for (; $head != $tail; $head++) {
487 my $docid = $head->get_docid;
488 my $doc = $db->get_document($docid);
489 my $smsg = bless { mid => $mid }, 'PublicInbox::Smsg';
490 $smsg->load_expand($doc);
491 if ($smsg->{blob} eq $oid) {
492 push(@delete, $docid);
495 $db->delete_document($_) foreach @delete;
499 sub index_git_blob_id {
500 my ($doc, $pfx, $objid) = @_;
502 my $len = length($objid);
503 for (my $len = length($objid); $len >= 7; ) {
504 $doc->add_term($pfx.$objid);
505 $objid = substr($objid, 0, --$len);
510 my ($self, $mime) = @_;
511 my $mid = eval { mid_mime($mime) };
512 $self->remove_message($mid) if defined $mid;
516 my ($self, $mime) = @_;
517 my $mid = mid_mime($mime);
518 my $mm = $self->{mm};
521 if (defined $self->{regen_down}) {
522 $num = $mm->num_for($mid) and return $num;
524 while (($num = $self->{regen_down}--) > 0) {
525 if ($mm->mid_set($num, $mid) != 0) {
529 } elsif (defined $self->{regen_up}) {
530 $num = $mm->num_for($mid) and return $num;
532 # this is to fixup old bugs due to add-remove-add
533 while (($num = ++$self->{regen_up})) {
534 if ($mm->mid_set($num, $mid) != 0) {
540 $num = $mm->mid_insert($mid) and return $num;
542 # fallback to num_for since filters like RubyLang set the number
547 my ($self, $mime) = @_;
548 $self->{mm}->mid_delete(mid_mime($mime));
552 my ($self, $mime, $smsg) = @_;
553 my $num = index_mm($self, $mime);
555 add_message($self, $mime, $smsg);
559 my ($self, $mime) = @_;
560 unindex_blob($self, $mime);
561 unindex_mm($self, $mime);
565 my ($git, $blob, $sizeref) = @_;
566 my $str = $git->cat_file($blob, $sizeref) or
567 die "BUG: $blob not found in $git->{git_dir}";
568 PublicInbox::Eml->new($str);
571 # called by public-inbox-index
573 my ($self, $opts) = @_;
574 delete $self->{lock_path} if $opts->{-skip_lock};
575 $self->{-inbox}->with_umask(sub { $self->_index_sync($opts) })
578 sub batch_adjust ($$$$$) {
579 my ($max, $bytes, $batch_cb, $latest, $nr) = @_;
582 $$max = $BATCH_BYTES;
583 $batch_cb->($nr, $latest);
588 my ($self, $git, $oid) = @_;
589 my $max_size = $self->{index_max_size} or return;
590 my (undef, undef, $size) = $git->check($oid);
591 die "E: bad $oid in $git->{git_dir}\n" if !defined($size);
592 return if $size <= $max_size;
593 warn "W: skipping $oid ($size > $max_size)\n";
599 my ($self, $log, $add_cb, $del_cb, $batch_cb) = @_;
600 my $hex = '[a-f0-9]';
601 my $h40 = $hex .'{40}';
602 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
603 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
604 my $git = $self->{git};
607 my $max = $BATCH_BYTES;
613 while (defined($line = <$log>)) {
614 if ($line =~ /$addmsg/o) {
616 if (delete $D{$blob}) {
617 if (defined $self->{regen_down}) {
618 my $num = $self->{regen_down}--;
619 $self->{mm}->num_highwater($num);
623 next if too_big($self, $git, $blob);
624 my $mime = do_cat_mail($git, $blob, \$bytes);
625 my $smsg = bless {}, 'PublicInbox::Smsg';
626 batch_adjust(\$max, $bytes, $batch_cb, $latest, ++$nr);
627 $smsg->{blob} = $blob;
628 $smsg->{bytes} = $bytes;
629 $add_cb->($self, $mime, $smsg);
630 } elsif ($line =~ /$delmsg/o) {
632 $D{$blob} = 1 unless too_big($self, $git, $blob);
633 } elsif ($line =~ /^commit ($h40)/o) {
636 } elsif ($line =~ /^author .*? ([0-9]+) [\-\+][0-9]+$/) {
637 $self->{autime} = $1;
638 } elsif ($line =~ /^committer .*? ([0-9]+) [\-\+][0-9]+$/) {
639 $self->{cotime} = $1;
642 close($log) or die "git log failed: \$?=$?";
644 foreach my $blob (keys %D) {
645 my $mime = do_cat_mail($git, $blob, \$bytes);
646 $del_cb->($self, $mime);
648 delete @$self{qw(autime cotime)};
649 $batch_cb->($nr, $latest, $newest);
653 my ($self, $opts, $range) = @_;
654 my $git = $self->{git};
656 if (index($range, '..') < 0) {
657 # don't show annoying git errors to users who run -index
659 $git->qx(qw(rev-parse -q --verify), "$range^0");
661 open my $fh, '<', '/dev/null' or
662 die "failed to open /dev/null: $!\n";
667 # Count the new files so they can be added newest to oldest
668 # and still have numbers increasing from oldest to newest
670 my $pr = $opts->{-progress};
671 $pr->("counting changes\n\t$range ... ") if $pr;
672 # can't use 'rev-list --count' if we use --diff-filter
673 my $fh = $git->popen(qw(log --pretty=tformat:%h
674 --no-notes --no-color --no-renames
675 --diff-filter=AM), $range);
676 ++$fcount while <$fh>;
677 close $fh or die "git log failed: \$?=$?";
678 my $high = $self->{mm}->num_highwater;
679 $pr->("$fcount\n") if $pr; # continue previous line
680 $self->{ntodo} = $fcount;
682 if (index($range, '..') < 0) {
683 if ($high && $high == $fcount) {
684 # fix up old bugs in full indexes which caused messages to
685 # not appear in Msgmap
686 $self->{regen_up} = $high;
688 # normal regen is for for fresh data
689 $self->{regen_down} = $fcount;
692 # Give oldest messages the smallest numbers
693 $self->{regen_down} = $high + $fcount;
696 $git->popen(qw/log --pretty=raw --no-notes --no-color --no-renames
697 --raw -r --no-abbrev/, $range);
700 # --is-ancestor requires git 1.8.0+
701 sub is_ancestor ($$$) {
702 my ($git, $cur, $tip) = @_;
703 return 0 unless $git->check($cur);
704 my $cmd = [ 'git', "--git-dir=$git->{git_dir}",
705 qw(merge-base --is-ancestor), $cur, $tip ];
706 my $pid = spawn($cmd);
707 waitpid($pid, 0) == $pid or die join(' ', @$cmd) .' did not finish';
711 sub need_update ($$$) {
712 my ($self, $cur, $new) = @_;
713 my $git = $self->{git};
714 return 1 if $cur && !is_ancestor($git, $cur, $new);
715 my $range = $cur eq '' ? $new : "$cur..$new";
716 chomp(my $n = $git->qx(qw(rev-list --count), $range));
717 ($n eq '' || $n > 0);
720 # The last git commit we indexed with Xapian or SQLite (msgmap)
721 # This needs to account for cases where Xapian or SQLite is
722 # out-of-date with respect to the other.
724 my ($self, $mm) = @_;
725 my $lm = $mm->last_commit || '';
727 if (need_xapian($self)) {
728 $lx = $self->{xdb}->get_metadata('last_commit') || '';
732 # Use last_commit from msgmap if it is older or unset
733 if (!$lm || ($lx && $lm && is_ancestor($self->{git}, $lm, $lx))) {
739 sub reindex_from ($$) {
740 my ($reindex, $last_commit) = @_;
741 return $last_commit unless $reindex;
742 ref($reindex) eq 'HASH' ? $reindex->{from} : '';
745 # indexes all unindexed messages (v1 only)
747 my ($self, $opts) = @_;
748 my $tip = $opts->{ref} || 'HEAD';
749 my ($last_commit, $lx, $xlog);
750 my $git = $self->{git};
752 my $pr = $opts->{-progress};
754 my $xdb = $self->begin_txn_lazy;
755 my $mm = _msgmap_init($self);
757 $xlog = undef; # stop previous git-log via SIGPIPE
758 $last_commit = _last_x_commit($self, $mm);
759 $lx = reindex_from($opts->{reindex}, $last_commit);
761 $self->{over}->rollback_lazy;
762 $self->{over}->disconnect;
765 $xdb->cancel_transaction if $xdb;
766 $xdb = _xdb_release($self);
768 # ensure we leak no FDs to "git log" with Xapian <= 1.2
769 my $range = $lx eq '' ? $tip : "$lx..$tip";
770 $xlog = _git_log($self, $opts, $range);
772 $xdb = $self->begin_txn_lazy;
773 } while (_last_x_commit($self, $mm) ne $last_commit);
775 my $dbh = $mm->{dbh} if $mm;
777 my ($nr, $commit, $newest) = @_;
780 my $cur = $mm->last_commit || '';
781 if (need_update($self, $cur, $newest)) {
782 $mm->last_commit($newest);
787 if ($newest && need_xapian($self)) {
788 my $cur = $xdb->get_metadata('last_commit');
789 if (need_update($self, $cur, $newest)) {
790 $xdb->set_metadata('last_commit', $newest);
793 $self->commit_txn_lazy;
795 $xdb = _xdb_release($self);
796 # let another process do some work... <
797 $pr->("indexed $nr/$self->{ntodo}\n") if $pr && $nr;
799 $xdb = $self->begin_txn_lazy;
800 $dbh->begin_work if $dbh;
805 read_log($self, $xlog, *index_both, *unindex_both, $cb);
809 # order matters for unlocking
810 $_[0]->{xdb} = undef;
811 $_[0]->{lockfh} = undef;
814 # remote_* subs are only used by SearchIdxPart
817 if (my $w = $self->{w}) {
818 print $w "commit\n" or die "failed to write commit: $!";
820 $self->commit_txn_lazy;
826 if (my $w = delete $self->{w}) {
827 my $pid = delete $self->{pid} or die "no process to wait on\n";
828 print $w "close\n" or die "failed to write to pid:$pid: $!\n";
829 close $w or die "failed to close pipe for pid:$pid: $!\n";
830 waitpid($pid, 0) == $pid or die "remote process did not finish";
831 $? == 0 or die ref($self)." pid:$pid exited with: $?";
833 die "transaction in progress $self\n" if $self->{txn};
834 $self->_xdb_release if $self->{xdb};
839 my ($self, $oid, $mid) = @_;
840 if (my $w = $self->{w}) {
841 # triggers remove_by_oid in a shard
842 print $w "D $oid $mid\n" or die "failed to write remove $!";
844 $self->begin_txn_lazy;
845 $self->remove_by_oid($oid, $mid);
851 return if $self->{txn};
853 $self->{-inbox}->with_umask(sub {
854 my $xdb = $self->{xdb} || $self->_xdb_acquire;
855 $self->{over}->begin_lazy if $self->{over};
856 $xdb->begin_transaction if $xdb;
862 # store 'indexlevel=medium' in v2 shard=0 and v1 (only one shard)
863 # This metadata is read by Admin::detect_indexlevel:
867 if (!$self->{shard} && # undef or 0, not >0
868 delete($self->{-set_indexlevel_once})) {
869 my $xdb = $self->{xdb};
870 my $level = $xdb->get_metadata('indexlevel');
871 if (!$level || $level ne 'medium') {
872 $xdb->set_metadata('indexlevel', 'medium');
877 sub commit_txn_lazy {
879 delete $self->{txn} or return;
880 $self->{-inbox}->with_umask(sub {
881 if (my $xdb = $self->{xdb}) {
882 set_indexlevel($self);
883 $xdb->commit_transaction;
885 $self->{over}->commit_lazy if $self->{over};
891 if (need_xapian($self)) {
892 die "$$ $0 xdb not released\n" if $self->{xdb};
894 die "$$ $0 still in transaction\n" if $self->{txn};