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 parent qw(PublicInbox::Search PublicInbox::Lock);
14 use PublicInbox::InboxWritable;
15 use PublicInbox::MID qw(mid_mime mids_for_index mids);
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,
56 indexlevel => $indexlevel,
58 $self->{-set_indexlevel_once} = 1 if $indexlevel eq 'medium';
61 $self->{lock_path} = "$inboxdir/ssoma.lock";
62 my $dir = $self->xdir;
63 $self->{over} = PublicInbox::OverIdx->new("$dir/over.sqlite3");
64 $self->{index_max_size} = $ibx->{index_max_size};
65 } elsif ($version == 2) {
66 defined $shard or die "shard is required for v2\n";
68 $self->{shard} = $shard;
69 $self->{lock_path} = undef;
71 die "unsupported inbox version=$version\n";
73 $self->{creat} = ($creat || 0) == 1;
77 sub need_xapian ($) { $_[0]->{indexlevel} =~ $xapianlevels }
80 my ($self, $wake) = @_;
81 if (need_xapian($self)) {
82 my $xdb = delete $self->{xdb} or croak 'not acquired';
85 $self->lock_release($wake) if $self->{creat};
89 sub load_xapian_writable () {
90 return 1 if $X->{WritableDatabase};
91 PublicInbox::Search::load_xapian() or return;
92 my $xap = $PublicInbox::Search::Xap;
93 for (qw(Document TermGenerator WritableDatabase)) {
94 $X->{$_} = $xap.'::'.$_;
96 eval 'require '.$X->{WritableDatabase} or die;
97 *sortable_serialise = $xap.'::sortable_serialise';
98 $DB_CREATE_OR_OPEN = eval($xap.'::DB_CREATE_OR_OPEN()');
99 $DB_OPEN = eval($xap.'::DB_OPEN()');
106 my $dir = $self->xdir;
107 if (need_xapian($self)) {
108 croak 'already acquired' if $self->{xdb};
109 load_xapian_writable();
110 $flag = $self->{creat} ? $DB_CREATE_OR_OPEN : $DB_OPEN;
112 if ($self->{creat}) {
116 # don't create empty Xapian directories if we don't need Xapian
117 my $is_shard = defined($self->{shard});
118 if (!$is_shard || ($is_shard && need_xapian($self))) {
119 File::Path::mkpath($dir);
122 return unless defined $flag;
123 my $xdb = eval { ($X->{WritableDatabase})->new($dir, $flag) };
125 die "Failed opening $dir: ", $@;
131 my ($doc, $col, $num) = @_;
132 $num = sortable_serialise($num);
133 $doc->add_value($col, $num);
136 sub term_generator ($) { # write-only
139 $self->{term_generator} //= do {
140 my $tg = $X->{TermGenerator}->new;
141 $tg->set_stemmer($self->stemmer);
146 sub index_text ($$$$) {
147 my ($self, $text, $wdf_inc, $prefix) = @_;
148 my $tg = term_generator($self); # man Search::Xapian::TermGenerator
150 if ($self->{indexlevel} eq 'full') {
151 $tg->index_text($text, $wdf_inc, $prefix);
152 $tg->increase_termpos;
154 $tg->index_text_without_positions($text, $wdf_inc, $prefix);
158 sub index_headers ($$) {
159 my ($self, $smsg) = @_;
160 my @x = (from => 'A', # Author
161 subject => 'S', to => 'XTO', cc => 'XCC');
162 while (my ($field, $pfx) = splice(@x, 0, 2)) {
163 my $val = $smsg->{$field};
164 index_text($self, $val, 1, $pfx) if $val ne '';
168 sub index_diff_inc ($$$$) {
169 my ($self, $text, $pfx, $xnq) = @_;
171 index_text($self, join("\n", @$xnq), 1, 'XNQ');
174 index_text($self, $text, 1, $pfx);
177 sub index_old_diff_fn {
178 my ($self, $seen, $fa, $fb, $xnq) = @_;
180 # no renames or space support for traditional diffs,
181 # find the number of leading common paths to strip:
182 my @fa = split('/', $fa);
183 my @fb = split('/', $fb);
184 while (scalar(@fa) && scalar(@fb)) {
185 $fa = join('/', @fa);
186 $fb = join('/', @fb);
188 unless ($seen->{$fa}++) {
189 index_diff_inc($self, $fa, 'XDFN', $xnq);
199 sub index_diff ($$$) {
200 my ($self, $txt, $doc) = @_;
205 foreach (split(/\n/, $txt)) {
206 if ($in_diff && s/^ //) { # diff context
207 index_diff_inc($self, $_, 'XDFCTX', $xnq);
208 } elsif (/^-- $/) { # email signature begins
210 } elsif (m!^diff --git "?[^/]+/.+ "?[^/]+/.+\z!) {
211 # wait until "---" and "+++" to capture filenames
214 } elsif (m/^diff -(.+) (\S+) (\S+)$/) {
215 my ($opt, $fa, $fb) = ($1, $2, $3);
217 # only support unified:
218 next unless $opt =~ /[uU]/;
219 $in_diff = index_old_diff_fn($self, \%seen, $fa, $fb,
221 } elsif (m!^--- ("?[^/]+/.+)!) {
223 $fn = (split('/', git_unquote($fn), 2))[1];
224 $seen{$fn}++ or index_diff_inc($self, $fn, 'XDFN', $xnq);
226 } elsif (m!^\+\+\+ ("?[^/]+/.+)!) {
228 $fn = (split('/', git_unquote($fn), 2))[1];
229 $seen{$fn}++ or index_diff_inc($self, $fn, 'XDFN', $xnq);
231 } elsif (/^--- (\S+)/) {
234 } elsif (defined $in_diff && /^\+\+\+ (\S+)/) {
235 $in_diff = index_old_diff_fn($self, \%seen, $in_diff,
237 } elsif ($in_diff && s/^\+//) { # diff added
238 index_diff_inc($self, $_, 'XDFB', $xnq);
239 } elsif ($in_diff && s/^-//) { # diff removed
240 index_diff_inc($self, $_, 'XDFA', $xnq);
241 } elsif (m!^index ([a-f0-9]+)\.\.([a-f0-9]+)!) {
242 my ($ba, $bb) = ($1, $2);
243 index_git_blob_id($doc, 'XDFPRE', $ba);
244 index_git_blob_id($doc, 'XDFPOST', $bb);
246 } elsif (/^@@ (?:\S+) (?:\S+) @@\s*$/) {
247 # traditional diff w/o -p
248 } elsif (/^@@ (?:\S+) (?:\S+) @@\s*(\S+.*)$/) {
249 # hunk header context
250 index_diff_inc($self, $1, 'XDFHH', $xnq);
251 # ignore the following lines:
252 } elsif (/^(?:dis)similarity index/ ||
253 /^(?:old|new) mode/ ||
254 /^(?:deleted|new) file mode/ ||
255 /^(?:copy|rename) (?:from|to) / ||
256 /^(?:dis)?similarity index / ||
257 /^\\ No newline at end of file/ ||
258 /^Binary files .* differ/) {
261 # possible to be in diff context, some mail may be
262 # stripped by MUA or even GNU diff(1). "git apply"
263 # treats a bare "\n" as diff context, too
266 warn "non-diff line: $_\n" if DEBUG && $_ ne '';
271 index_text($self, join("\n", @xnq), 1, 'XNQ');
274 sub index_xapian { # msg_iter callback
275 my $part = $_[0]->[0]; # ignore $depth and $idx
276 my ($self, $doc) = @{$_[1]};
277 my $ct = $part->content_type || 'text/plain';
278 my $fn = $part->filename;
279 if (defined $fn && $fn ne '') {
280 index_text($self, $fn, 1, 'XFN');
282 if ($part->{is_submsg}) {
283 my $mids = mids_for_index($part);
284 index_ids($self, $doc, $part, $mids);
285 my $smsg = bless {}, 'PublicInbox::Smsg';
286 $smsg->populate($part);
287 index_headers($self, $smsg);
290 my ($s, undef) = msg_part_text($part, $ct);
291 defined $s or return;
292 $_[0]->[0] = $part = undef; # free memory
294 # split off quoted and unquoted blocks:
295 my @sections = PublicInbox::MsgIter::split_quotes($s);
296 undef $s; # free memory
297 for my $txt (@sections) {
299 index_text($self, $txt, 0, 'XQUOT');
301 # does it look like a diff?
302 if ($txt =~ /^(?:diff|---|\+\+\+) /ms) {
303 index_diff($self, $txt, $doc);
305 index_text($self, $txt, 1, 'XNQ');
308 undef $txt; # free memory
312 sub index_ids ($$$$) {
313 my ($self, $doc, $hdr, $mids) = @_;
314 for my $mid (@$mids) {
315 index_text($self, $mid, 1, 'XM');
317 # because too many Message-IDs are prefixed with
319 if ($mid =~ /\w{12,}/) {
320 my @long = ($mid =~ /(\w{3,}+)/g);
321 index_text($self, join(' ', @long), 1, 'XM');
324 $doc->add_boolean_term('Q' . $_) for @$mids;
325 for my $l ($hdr->header_raw('List-Id')) {
326 $l =~ /<([^>]+)>/ or next;
328 $doc->add_boolean_term('G' . $lid);
329 index_text($self, $lid, 1, 'XL'); # probabilistic
333 sub add_xapian ($$$$) {
334 my ($self, $mime, $smsg, $mids) = @_;
335 my $hdr = $mime->header_obj;
336 my $doc = $X->{Document}->new;
337 add_val($doc, PublicInbox::Search::TS(), $smsg->{ts});
338 my @ds = gmtime($smsg->{ds});
339 my $yyyymmdd = strftime('%Y%m%d', @ds);
340 add_val($doc, PublicInbox::Search::YYYYMMDD(), $yyyymmdd);
341 my $dt = strftime('%Y%m%d%H%M%S', @ds);
342 add_val($doc, PublicInbox::Search::DT(), $dt);
343 add_val($doc, PublicInbox::Search::BYTES(), $smsg->{bytes});
344 add_val($doc, PublicInbox::Search::UID(), $smsg->{num});
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, $sync) = @_;
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, $sync);
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";
416 my ($self, $oid, @removed) = @_;
417 my $xdb = $self->{xdb} or return;
418 for my $num (@removed) {
419 my $doc = eval { $xdb->get_document($num) };
421 warn "E: $@\n" if $@;
422 warn "E: #$num $oid missing in Xapian\n";
425 my $smsg = bless {}, 'PublicInbox::Smsg';
426 $smsg->load_expand($doc);
427 my $blob = $smsg->{blob} // '(unset)';
429 $xdb->delete_document($num);
431 warn "E: #$num $oid != $blob in Xapian\n";
437 my ($self, $oid, $num) = @_;
438 die "BUG: remove_by_oid is v2-only\n" if $self->{over};
439 $self->begin_txn_lazy;
440 xdb_remove($self, $oid, $num) if need_xapian($self);
443 sub index_git_blob_id {
444 my ($doc, $pfx, $objid) = @_;
446 my $len = length($objid);
447 for (my $len = length($objid); $len >= 7; ) {
448 $doc->add_term($pfx.$objid);
449 $objid = substr($objid, 0, --$len);
455 my ($self, $oid, $eml) = @_;
456 my $mids = mids($eml);
459 for my $mid (@$mids) {
460 my @removed = eval { $self->{over}->remove_oid($oid, $mid) };
462 warn "E: failed to remove <$mid> from overview: $@\n";
464 $nr += scalar @removed;
465 $tmp{$_}++ for @removed;
469 $mids = join('> <', @$mids);
470 warn "W: <$mids> missing for removal from overview\n";
472 while (my ($num, $nr) = each %tmp) {
473 warn "BUG: $num appears >1 times ($nr) for $oid\n" if $nr != 1;
475 xdb_remove($self, $oid, keys %tmp) if need_xapian($self);
479 my ($self, $mime) = @_;
480 my $mid = mid_mime($mime);
481 my $mm = $self->{mm};
484 if (defined $self->{regen_down}) {
485 $num = $mm->num_for($mid) and return $num;
487 while (($num = $self->{regen_down}--) > 0) {
488 if ($mm->mid_set($num, $mid) != 0) {
492 } elsif (defined $self->{regen_up}) {
493 $num = $mm->num_for($mid) and return $num;
495 # this is to fixup old bugs due to add-remove-add
496 while (($num = ++$self->{regen_up})) {
497 if ($mm->mid_set($num, $mid) != 0) {
503 $num = $mm->mid_insert($mid) and return $num;
505 # fallback to num_for since filters like RubyLang set the number
510 my ($self, $mime) = @_;
511 $self->{mm}->mid_delete(mid_mime($mime));
514 # returns the number of bytes to add if given a non-CRLF arg
515 sub crlf_adjust ($) {
516 if (index($_[0], "\r\n") < 0) {
517 # common case is LF-only, every \n needs an \r;
518 # so favor a cheap tr// over an expensive m//g
520 } else { # count number of '\n' w/o '\r', expensive:
521 scalar(my @n = ($_[0] =~ m/(?<!\r)\n/g));
525 sub index_both { # git->cat_async callback
526 my ($bref, $oid, $type, $size, $sync) = @_;
527 my ($nr, $max) = @$sync{qw(nr max)};
530 $size += crlf_adjust($$bref);
531 my $smsg = bless { bytes => $size, blob => $oid }, 'PublicInbox::Smsg';
532 my $self = $sync->{sidx};
533 my $eml = PublicInbox::Eml->new($bref);
534 my $num = index_mm($self, $eml);
536 add_message($self, $eml, $smsg, $sync);
539 sub unindex_both { # git->cat_async callback
540 my ($bref, $oid, $type, $size, $self) = @_;
541 my $eml = PublicInbox::Eml->new($bref);
542 unindex_eml($self, $oid, $eml);
543 unindex_mm($self, $eml);
546 # called by public-inbox-index
548 my ($self, $opts) = @_;
549 delete $self->{lock_path} if $opts->{-skip_lock};
550 $self->{ibx}->with_umask(\&_index_sync, $self, $opts);
554 my ($self, $oid) = @_;
555 my $max_size = $self->{index_max_size} or return;
556 my (undef, undef, $size) = $self->{ibx}->git->check($oid);
557 die "E: bad $oid in $self->{ibx}->{inboxdir}\n" if !defined($size);
558 return if $size <= $max_size;
559 warn "W: skipping $oid ($size > $max_size)\n";
565 my ($self, $log, $batch_cb) = @_;
566 my $hex = '[a-f0-9]';
567 my $h40 = $hex .'{40}';
568 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
569 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
570 my $git = $self->{ibx}->git;
572 my $max = $BATCH_BYTES;
578 my $sync = { sidx => $self, nr => \$nr, max => \$max };
579 while (defined($line = <$log>)) {
580 if ($line =~ /$addmsg/o) {
582 if (delete $D{$blob}) {
583 # make sure pending index writes are done
584 # before writing to ->mm
585 $git->cat_async_wait;
587 if (defined $self->{regen_down}) {
588 my $num = $self->{regen_down}--;
589 $self->{mm}->num_highwater($num);
593 next if too_big($self, $blob);
594 $git->cat_async($blob, \&index_both, { %$sync });
596 $git->cat_async_wait;
598 $batch_cb->($nr, $latest);
600 } elsif ($line =~ /$delmsg/o) {
602 $D{$blob} = 1 unless too_big($self, $blob);
603 } elsif ($line =~ /^commit ($h40)/o) {
606 } elsif ($line =~ /^author .*? ([0-9]+) [\-\+][0-9]+$/) {
607 $sync->{autime} = $1;
608 } elsif ($line =~ /^committer .*? ([0-9]+) [\-\+][0-9]+$/) {
609 $sync->{cotime} = $1;
612 close($log) or die "git log failed: \$?=$?";
614 foreach my $blob (keys %D) {
615 $git->cat_async($blob, \&unindex_both, $self);
617 $git->cat_async_wait;
618 $batch_cb->($nr, $latest, $newest);
622 my ($self, $opts, $range) = @_;
623 my $git = $self->{ibx}->git;
625 if (index($range, '..') < 0) {
626 # don't show annoying git errors to users who run -index
628 $git->qx(qw(rev-parse -q --verify), "$range^0");
630 open my $fh, '<', '/dev/null' or
631 die "failed to open /dev/null: $!\n";
636 # Count the new files so they can be added newest to oldest
637 # and still have numbers increasing from oldest to newest
639 my $pr = $opts->{-progress};
640 $pr->("counting changes\n\t$range ... ") if $pr;
641 # can't use 'rev-list --count' if we use --diff-filter
642 my $fh = $git->popen(qw(log --pretty=tformat:%h
643 --no-notes --no-color --no-renames
644 --diff-filter=AM), $range);
645 ++$fcount while <$fh>;
646 close $fh or die "git log failed: \$?=$?";
647 my $high = $self->{mm}->num_highwater;
648 $pr->("$fcount\n") if $pr; # continue previous line
649 $self->{ntodo} = $fcount;
651 if (index($range, '..') < 0) {
652 if ($high && $high == $fcount) {
653 # fix up old bugs in full indexes which caused messages to
654 # not appear in Msgmap
655 $self->{regen_up} = $high;
657 # normal regen is for for fresh data
658 $self->{regen_down} = $fcount;
659 $self->{regen_down} += $high unless $opts->{reindex};
662 # Give oldest messages the smallest numbers
663 $self->{regen_down} = $high + $fcount;
666 $git->popen(qw/log --pretty=raw --no-notes --no-color --no-renames
667 --raw -r --no-abbrev/, $range);
670 # --is-ancestor requires git 1.8.0+
671 sub is_ancestor ($$$) {
672 my ($git, $cur, $tip) = @_;
673 return 0 unless $git->check($cur);
674 my $cmd = [ 'git', "--git-dir=$git->{git_dir}",
675 qw(merge-base --is-ancestor), $cur, $tip ];
676 my $pid = spawn($cmd);
677 waitpid($pid, 0) == $pid or die join(' ', @$cmd) .' did not finish';
681 sub need_update ($$$) {
682 my ($self, $cur, $new) = @_;
683 my $git = $self->{ibx}->git;
684 return 1 if $cur && !is_ancestor($git, $cur, $new);
685 my $range = $cur eq '' ? $new : "$cur..$new";
686 chomp(my $n = $git->qx(qw(rev-list --count), $range));
687 ($n eq '' || $n > 0);
690 # The last git commit we indexed with Xapian or SQLite (msgmap)
691 # This needs to account for cases where Xapian or SQLite is
692 # out-of-date with respect to the other.
694 my ($self, $mm) = @_;
695 my $lm = $mm->last_commit || '';
697 if (need_xapian($self)) {
698 $lx = $self->{xdb}->get_metadata('last_commit') || '';
702 # Use last_commit from msgmap if it is older or unset
703 if (!$lm || ($lx && $lm && is_ancestor($self->{ibx}->git, $lm, $lx))) {
709 sub reindex_from ($$) {
710 my ($reindex, $last_commit) = @_;
711 return $last_commit unless $reindex;
712 ref($reindex) eq 'HASH' ? $reindex->{from} : '';
715 # indexes all unindexed messages (v1 only)
717 my ($self, $opts) = @_;
718 my $tip = $opts->{ref} || 'HEAD';
719 my ($last_commit, $lx, $xlog);
720 my $git = $self->{ibx}->git;
722 my $pr = $opts->{-progress};
724 my $xdb = $self->begin_txn_lazy;
725 $self->{over}->rethread_prepare($opts);
726 my $mm = _msgmap_init($self);
728 $xlog = undef; # stop previous git-log via SIGPIPE
729 $last_commit = _last_x_commit($self, $mm);
730 $lx = reindex_from($opts->{reindex}, $last_commit);
732 $self->{over}->rollback_lazy;
733 $self->{over}->disconnect;
736 $xdb->cancel_transaction if $xdb;
737 $xdb = _xdb_release($self);
739 # ensure we leak no FDs to "git log" with Xapian <= 1.2
740 my $range = $lx eq '' ? $tip : "$lx..$tip";
741 $xlog = _git_log($self, $opts, $range);
743 $xdb = $self->begin_txn_lazy;
744 } while (_last_x_commit($self, $mm) ne $last_commit);
746 my $dbh = $mm->{dbh} if $mm;
748 my ($nr, $commit, $newest) = @_;
751 my $cur = $mm->last_commit || '';
752 if (need_update($self, $cur, $newest)) {
753 $mm->last_commit($newest);
758 if ($newest && need_xapian($self)) {
759 my $cur = $xdb->get_metadata('last_commit');
760 if (need_update($self, $cur, $newest)) {
761 $xdb->set_metadata('last_commit', $newest);
765 $self->{over}->rethread_done($opts) if $newest; # all done
766 $self->commit_txn_lazy;
768 $xdb = _xdb_release($self, $nr);
769 # let another process do some work...
770 $pr->("indexed $nr/$self->{ntodo}\n") if $pr && $nr;
771 if (!$newest) { # more to come
772 $xdb = $self->begin_txn_lazy;
773 $dbh->begin_work if $dbh;
778 read_log($self, $xlog, $batch_cb);
782 # order matters for unlocking
783 $_[0]->{xdb} = undef;
784 $_[0]->{lockfh} = undef;
787 # remote_* subs are only used by SearchIdxPart
790 if (my $w = $self->{w}) {
791 print $w "commit\n" or die "failed to write commit: $!";
793 $self->commit_txn_lazy;
799 if (my $w = delete $self->{w}) {
800 my $pid = delete $self->{pid} or die "no process to wait on\n";
801 print $w "close\n" or die "failed to write to pid:$pid: $!\n";
802 close $w or die "failed to close pipe for pid:$pid: $!\n";
803 waitpid($pid, 0) == $pid or die "remote process did not finish";
804 $? == 0 or die ref($self)." pid:$pid exited with: $?";
806 die "transaction in progress $self\n" if $self->{txn};
807 $self->_xdb_release if $self->{xdb};
812 my ($self, $oid, $num) = @_;
813 if (my $w = $self->{w}) {
814 # triggers remove_by_oid in a shard
815 print $w "D $oid $num\n" or die "failed to write remove $!";
817 $self->remove_by_oid($oid, $num);
823 my $xdb = $self->{xdb} || $self->_xdb_acquire;
824 $self->{over}->begin_lazy if $self->{over};
825 $xdb->begin_transaction if $xdb;
832 $self->{ibx}->with_umask(\&_begin_txn, $self) if !$self->{txn};
835 # store 'indexlevel=medium' in v2 shard=0 and v1 (only one shard)
836 # This metadata is read by Admin::detect_indexlevel:
840 if (!$self->{shard} && # undef or 0, not >0
841 delete($self->{-set_indexlevel_once})) {
842 my $xdb = $self->{xdb};
843 my $level = $xdb->get_metadata('indexlevel');
844 if (!$level || $level ne 'medium') {
845 $xdb->set_metadata('indexlevel', 'medium');
852 if (my $xdb = $self->{xdb}) {
853 set_indexlevel($self);
854 $xdb->commit_transaction;
856 $self->{over}->commit_lazy if $self->{over};
859 sub commit_txn_lazy {
861 delete($self->{txn}) and
862 $self->{ibx}->with_umask(\&_commit_txn, $self);
867 if (need_xapian($self)) {
868 die "$$ $0 xdb not released\n" if $self->{xdb};
870 die "$$ $0 still in transaction\n" if $self->{txn};