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 Exporter);
14 use PublicInbox::InboxWritable;
15 use PublicInbox::MID qw(mids_for_index mids);
16 use PublicInbox::MsgIter;
17 use PublicInbox::IdxStack;
19 use POSIX qw(strftime);
20 use PublicInbox::OverIdx;
21 use PublicInbox::Spawn qw(spawn);
22 use PublicInbox::Git qw(git_unquote);
23 use PublicInbox::MsgTime qw(msg_timestamp msg_datestamp);
24 our @EXPORT_OK = qw(crlf_adjust log2stack is_ancestor check_size nodatacow_dir);
25 my $X = \%PublicInbox::Search::X;
26 my ($DB_CREATE_OR_OPEN, $DB_OPEN);
28 our $BATCH_BYTES = defined($ENV{XAPIAN_FLUSH_THRESHOLD}) ?
29 0x7fffffff : 1_000_000;
30 use constant DEBUG => !!$ENV{DEBUG};
32 my $xapianlevels = qr/\A(?:full|medium)\z/;
34 my $OID = $hex .'{40,}';
37 my ($class, $ibx, $creat, $shard) = @_;
38 ref $ibx or die "BUG: expected PublicInbox::Inbox object: $ibx";
39 my $levels = qr/\A(?:full|medium|basic)\z/;
40 my $inboxdir = $ibx->{inboxdir};
41 my $version = $ibx->version;
42 my $indexlevel = 'full';
43 my $altid = $ibx->{altid};
45 require PublicInbox::AltId;
46 $altid = [ map { PublicInbox::AltId->new($ibx, $_); } @$altid ];
48 if ($ibx->{indexlevel}) {
49 if ($ibx->{indexlevel} =~ $levels) {
50 $indexlevel = $ibx->{indexlevel};
52 die("Invalid indexlevel $ibx->{indexlevel}\n");
55 $ibx = PublicInbox::InboxWritable->new($ibx);
58 xpfx => $inboxdir, # for xpfx_init
61 indexlevel => $indexlevel,
64 $self->{-set_indexlevel_once} = 1 if $indexlevel eq 'medium';
67 $self->{lock_path} = "$inboxdir/ssoma.lock";
68 my $dir = $self->xdir;
69 $self->{over} = PublicInbox::OverIdx->new("$dir/over.sqlite3");
70 $self->{over}->{-no_sync} = 1 if $ibx->{-no_sync};
71 $self->{index_max_size} = $ibx->{index_max_size};
72 } elsif ($version == 2) {
73 defined $shard or die "shard is required for v2\n";
75 $self->{shard} = $shard;
76 $self->{lock_path} = undef;
78 die "unsupported inbox version=$version\n";
80 $self->{creat} = ($creat || 0) == 1;
84 sub need_xapian ($) { $_[0]->{indexlevel} =~ $xapianlevels }
87 my ($self, $wake) = @_;
88 if (need_xapian($self)) {
89 my $xdb = delete $self->{xdb} or croak 'not acquired';
92 $self->lock_release($wake) if $self->{creat};
96 sub load_xapian_writable () {
97 return 1 if $X->{WritableDatabase};
98 PublicInbox::Search::load_xapian() or return;
99 my $xap = $PublicInbox::Search::Xap;
100 for (qw(Document TermGenerator WritableDatabase)) {
101 $X->{$_} = $xap.'::'.$_;
103 eval 'require '.$X->{WritableDatabase} or die;
104 *sortable_serialise = $xap.'::sortable_serialise';
105 $DB_CREATE_OR_OPEN = eval($xap.'::DB_CREATE_OR_OPEN()');
106 $DB_OPEN = eval($xap.'::DB_OPEN()');
107 my $ver = (eval($xap.'::major_version()') << 16) |
108 (eval($xap.'::minor_version()') << 8);
109 $DB_NO_SYNC = 0x4 if $ver >= 0x10400;
113 sub nodatacow_dir ($) {
115 opendir my $dh, $dir or die "opendir($dir): $!\n";
116 PublicInbox::Spawn::set_nodatacow(fileno($dh));
122 my $dir = $self->xdir;
123 if (need_xapian($self)) {
124 croak 'already acquired' if $self->{xdb};
125 load_xapian_writable();
126 $flag = $self->{creat} ? $DB_CREATE_OR_OPEN : $DB_OPEN;
128 if ($self->{creat}) {
132 # don't create empty Xapian directories if we don't need Xapian
133 my $is_shard = defined($self->{shard});
134 if (!-d $dir && (!$is_shard ||
135 ($is_shard && need_xapian($self)))) {
136 File::Path::mkpath($dir);
140 return unless defined $flag;
141 $flag |= $DB_NO_SYNC if $self->{ibx}->{-no_sync};
142 my $xdb = eval { ($X->{WritableDatabase})->new($dir, $flag) };
144 die "Failed opening $dir: ", $@;
150 my ($doc, $col, $num) = @_;
151 $num = sortable_serialise($num);
152 $doc->add_value($col, $num);
155 sub term_generator ($) { # write-only
158 $self->{term_generator} //= do {
159 my $tg = $X->{TermGenerator}->new;
160 $tg->set_stemmer($self->stemmer);
165 sub index_text ($$$$) {
166 my ($self, $text, $wdf_inc, $prefix) = @_;
167 my $tg = term_generator($self); # man Search::Xapian::TermGenerator
169 if ($self->{indexlevel} eq 'full') {
170 $tg->index_text($text, $wdf_inc, $prefix);
171 $tg->increase_termpos;
173 $tg->index_text_without_positions($text, $wdf_inc, $prefix);
177 sub index_headers ($$) {
178 my ($self, $smsg) = @_;
179 my @x = (from => 'A', # Author
180 subject => 'S', to => 'XTO', cc => 'XCC');
181 while (my ($field, $pfx) = splice(@x, 0, 2)) {
182 my $val = $smsg->{$field};
183 index_text($self, $val, 1, $pfx) if $val ne '';
187 sub index_diff_inc ($$$$) {
188 my ($self, $text, $pfx, $xnq) = @_;
190 index_text($self, join("\n", @$xnq), 1, 'XNQ');
193 index_text($self, $text, 1, $pfx);
196 sub index_old_diff_fn {
197 my ($self, $seen, $fa, $fb, $xnq) = @_;
199 # no renames or space support for traditional diffs,
200 # find the number of leading common paths to strip:
201 my @fa = split('/', $fa);
202 my @fb = split('/', $fb);
203 while (scalar(@fa) && scalar(@fb)) {
204 $fa = join('/', @fa);
205 $fb = join('/', @fb);
207 unless ($seen->{$fa}++) {
208 index_diff_inc($self, $fa, 'XDFN', $xnq);
218 sub index_diff ($$$) {
219 my ($self, $txt, $doc) = @_;
224 foreach (split(/\n/, $txt)) {
225 if ($in_diff && s/^ //) { # diff context
226 index_diff_inc($self, $_, 'XDFCTX', $xnq);
227 } elsif (/^-- $/) { # email signature begins
229 } elsif (m!^diff --git "?[^/]+/.+ "?[^/]+/.+\z!) {
230 # wait until "---" and "+++" to capture filenames
233 } elsif (m/^diff -(.+) (\S+) (\S+)$/) {
234 my ($opt, $fa, $fb) = ($1, $2, $3);
236 # only support unified:
237 next unless $opt =~ /[uU]/;
238 $in_diff = index_old_diff_fn($self, \%seen, $fa, $fb,
240 } elsif (m!^--- ("?[^/]+/.+)!) {
242 $fn = (split('/', git_unquote($fn), 2))[1];
243 $seen{$fn}++ or index_diff_inc($self, $fn, 'XDFN', $xnq);
245 } elsif (m!^\+\+\+ ("?[^/]+/.+)!) {
247 $fn = (split('/', git_unquote($fn), 2))[1];
248 $seen{$fn}++ or index_diff_inc($self, $fn, 'XDFN', $xnq);
250 } elsif (/^--- (\S+)/) {
253 } elsif (defined $in_diff && /^\+\+\+ (\S+)/) {
254 $in_diff = index_old_diff_fn($self, \%seen, $in_diff,
256 } elsif ($in_diff && s/^\+//) { # diff added
257 index_diff_inc($self, $_, 'XDFB', $xnq);
258 } elsif ($in_diff && s/^-//) { # diff removed
259 index_diff_inc($self, $_, 'XDFA', $xnq);
260 } elsif (m!^index ([a-f0-9]+)\.\.([a-f0-9]+)!) {
261 my ($ba, $bb) = ($1, $2);
262 index_git_blob_id($doc, 'XDFPRE', $ba);
263 index_git_blob_id($doc, 'XDFPOST', $bb);
265 } elsif (/^@@ (?:\S+) (?:\S+) @@\s*$/) {
266 # traditional diff w/o -p
267 } elsif (/^@@ (?:\S+) (?:\S+) @@\s*(\S+.*)$/) {
268 # hunk header context
269 index_diff_inc($self, $1, 'XDFHH', $xnq);
270 # ignore the following lines:
271 } elsif (/^(?:dis)similarity index/ ||
272 /^(?:old|new) mode/ ||
273 /^(?:deleted|new) file mode/ ||
274 /^(?:copy|rename) (?:from|to) / ||
275 /^(?:dis)?similarity index / ||
276 /^\\ No newline at end of file/ ||
277 /^Binary files .* differ/) {
280 # possible to be in diff context, some mail may be
281 # stripped by MUA or even GNU diff(1). "git apply"
282 # treats a bare "\n" as diff context, too
285 warn "non-diff line: $_\n" if DEBUG && $_ ne '';
290 index_text($self, join("\n", @xnq), 1, 'XNQ');
293 sub index_xapian { # msg_iter callback
294 my $part = $_[0]->[0]; # ignore $depth and $idx
295 my ($self, $doc) = @{$_[1]};
296 my $ct = $part->content_type || 'text/plain';
297 my $fn = $part->filename;
298 if (defined $fn && $fn ne '') {
299 index_text($self, $fn, 1, 'XFN');
301 if ($part->{is_submsg}) {
302 my $mids = mids_for_index($part);
303 index_ids($self, $doc, $part, $mids);
304 my $smsg = bless {}, 'PublicInbox::Smsg';
305 $smsg->populate($part);
306 index_headers($self, $smsg);
309 my ($s, undef) = msg_part_text($part, $ct);
310 defined $s or return;
311 $_[0]->[0] = $part = undef; # free memory
313 # split off quoted and unquoted blocks:
314 my @sections = PublicInbox::MsgIter::split_quotes($s);
315 undef $s; # free memory
316 for my $txt (@sections) {
318 index_text($self, $txt, 0, 'XQUOT');
320 # does it look like a diff?
321 if ($txt =~ /^(?:diff|---|\+\+\+) /ms) {
322 index_diff($self, $txt, $doc);
324 index_text($self, $txt, 1, 'XNQ');
327 undef $txt; # free memory
331 sub index_ids ($$$$) {
332 my ($self, $doc, $hdr, $mids) = @_;
333 for my $mid (@$mids) {
334 index_text($self, $mid, 1, 'XM');
336 # because too many Message-IDs are prefixed with
338 if ($mid =~ /\w{12,}/) {
339 my @long = ($mid =~ /(\w{3,}+)/g);
340 index_text($self, join(' ', @long), 1, 'XM');
343 $doc->add_boolean_term('Q' . $_) for @$mids;
344 for my $l ($hdr->header_raw('List-Id')) {
345 $l =~ /<([^>]+)>/ or next;
347 $doc->add_boolean_term('G' . $lid);
348 index_text($self, $lid, 1, 'XL'); # probabilistic
352 sub add_xapian ($$$$) {
353 my ($self, $mime, $smsg, $mids) = @_;
354 my $hdr = $mime->header_obj;
355 my $doc = $X->{Document}->new;
356 add_val($doc, PublicInbox::Search::TS(), $smsg->{ts});
357 my @ds = gmtime($smsg->{ds});
358 my $yyyymmdd = strftime('%Y%m%d', @ds);
359 add_val($doc, PublicInbox::Search::YYYYMMDD(), $yyyymmdd);
360 my $dt = strftime('%Y%m%d%H%M%S', @ds);
361 add_val($doc, PublicInbox::Search::DT(), $dt);
362 add_val($doc, PublicInbox::Search::BYTES(), $smsg->{bytes});
363 add_val($doc, PublicInbox::Search::UID(), $smsg->{num});
365 my $tg = term_generator($self);
366 $tg->set_document($doc);
367 index_headers($self, $smsg);
369 msg_iter($mime, \&index_xapian, [ $self, $doc ]);
370 index_ids($self, $doc, $hdr, $mids);
371 $smsg->{to} = $smsg->{cc} = ''; # WWW doesn't need these, only NNTP
372 PublicInbox::OverIdx::parse_references($smsg, $hdr, $mids);
373 my $data = $smsg->to_doc_data;
374 $doc->set_data($data);
375 if (my $altid = $self->{-altid}) {
376 foreach my $alt (@$altid) {
377 my $pfx = $alt->{xprefix};
378 foreach my $mid (@$mids) {
379 my $id = $alt->mid2alt($mid);
380 next unless defined $id;
381 $doc->add_boolean_term($pfx . $id);
385 $self->{xdb}->replace_document($smsg->{num}, $doc);
388 sub _msgmap_init ($) {
390 die "BUG: _msgmap_init is only for v1\n" if $self->{ibx_ver} != 1;
391 $self->{mm} //= eval {
392 require PublicInbox::Msgmap;
393 my $rw = $self->{ibx}->{-no_sync} ? 2 : 1;
394 PublicInbox::Msgmap->new($self->{ibx}->{inboxdir}, $rw);
399 # mime = PublicInbox::Eml or Email::MIME object
400 my ($self, $mime, $smsg, $sync) = @_;
401 my $hdr = $mime->header_obj;
402 my $mids = mids_for_index($hdr);
403 $smsg //= bless { blob => '' }, 'PublicInbox::Smsg'; # test-only compat
404 $smsg->{mid} //= $mids->[0]; # v1 compatibility
405 $smsg->{num} //= do { # v1
407 index_mm($self, $mime, $smsg->{blob}, $sync);
411 $smsg->populate($hdr, $sync);
412 $smsg->{bytes} //= length($mime->as_string);
415 # order matters, overview stores every possible piece of
416 # data in doc_data (deflated). Xapian only stores a subset
417 # of the fields which exist in over.sqlite3. We may stop
418 # storing doc_data in Xapian sometime after we get multi-inbox
420 if (my $over = $self->{over}) { # v1 only
421 $over->add_overview($mime, $smsg);
423 if (need_xapian($self)) {
424 add_xapian($self, $mime, $smsg, $mids);
429 warn "failed to index message <".join('> <',@$mids).">: $@\n";
436 my ($self, $oid, @removed) = @_;
437 my $xdb = $self->{xdb} or return;
438 for my $num (@removed) {
439 my $doc = eval { $xdb->get_document($num) };
441 warn "E: $@\n" if $@;
442 warn "E: #$num $oid missing in Xapian\n";
445 my $smsg = bless {}, 'PublicInbox::Smsg';
446 $smsg->load_expand($doc);
447 my $blob = $smsg->{blob} // '(unset)';
449 $xdb->delete_document($num);
451 warn "E: #$num $oid != $blob in Xapian\n";
457 my ($self, $oid, $num) = @_;
458 die "BUG: remove_by_oid is v2-only\n" if $self->{over};
459 $self->begin_txn_lazy;
460 xdb_remove($self, $oid, $num) if need_xapian($self);
463 sub index_git_blob_id {
464 my ($doc, $pfx, $objid) = @_;
466 my $len = length($objid);
467 for (my $len = length($objid); $len >= 7; ) {
468 $doc->add_term($pfx.$objid);
469 $objid = substr($objid, 0, --$len);
475 my ($self, $oid, $eml) = @_;
476 my $mids = mids($eml);
479 for my $mid (@$mids) {
480 my @removed = eval { $self->{over}->remove_oid($oid, $mid) };
482 warn "E: failed to remove <$mid> from overview: $@\n";
484 $nr += scalar @removed;
485 $tmp{$_}++ for @removed;
489 $mids = join('> <', @$mids);
490 warn "W: <$mids> missing for removal from overview\n";
492 while (my ($num, $nr) = each %tmp) {
493 warn "BUG: $num appears >1 times ($nr) for $oid\n" if $nr != 1;
496 $self->{mm}->num_delete($_) for (keys %tmp);
497 } else { # just in case msgmap and over.sqlite3 become desynched:
498 $self->{mm}->mid_delete($mids->[0]);
500 xdb_remove($self, $oid, keys %tmp) if need_xapian($self);
504 my ($self, $mime, $oid, $sync) = @_;
505 my $mids = mids($mime);
506 my $mm = $self->{mm};
507 if ($sync->{reindex}) {
508 my $over = $self->{over};
509 for my $mid (@$mids) {
510 my ($num, undef) = $over->num_mid0_for_oid($oid, $mid);
511 return $num if defined $num;
513 $mm->num_for($mids->[0]) // $mm->mid_insert($mids->[0]);
515 # fallback to num_for since filters like RubyLang set the number
516 $mm->mid_insert($mids->[0]) // $mm->num_for($mids->[0]);
520 # returns the number of bytes to add if given a non-CRLF arg
521 sub crlf_adjust ($) {
522 if (index($_[0], "\r\n") < 0) {
523 # common case is LF-only, every \n needs an \r;
524 # so favor a cheap tr// over an expensive m//g
526 } else { # count number of '\n' w/o '\r', expensive:
527 scalar(my @n = ($_[0] =~ m/(?<!\r)\n/g));
531 sub index_both { # git->cat_async callback
532 my ($bref, $oid, $type, $size, $sync) = @_;
533 my ($nr, $max) = @$sync{qw(nr max)};
536 $size += crlf_adjust($$bref);
537 my $smsg = bless { bytes => $size, blob => $oid }, 'PublicInbox::Smsg';
538 my $self = $sync->{sidx};
539 my $eml = PublicInbox::Eml->new($bref);
540 $smsg->{num} = index_mm($self, $eml, $oid, $sync) or
541 die "E: could not generate NNTP article number for $oid";
542 add_message($self, $eml, $smsg, $sync);
545 sub unindex_both { # git->cat_async callback
546 my ($bref, $oid, $type, $size, $self) = @_;
547 unindex_eml($self, $oid, PublicInbox::Eml->new($bref));
550 # called by public-inbox-index
552 my ($self, $opts) = @_;
553 delete $self->{lock_path} if $opts->{-skip_lock};
554 $self->{ibx}->with_umask(\&_index_sync, $self, $opts);
555 if ($opts->{reindex}) {
557 delete @again{qw(rethread reindex)};
558 index_sync($self, \%again);
562 sub check_size { # check_async cb for -index --max-size=...
563 my ($oid, $type, $size, $arg, $git) = @_;
564 (($type // '') eq 'blob') or die "E: bad $oid in $git->{git_dir}";
565 if ($size <= $arg->{index_max_size}) {
566 $git->cat_async($oid, $arg->{index_oid}, $arg);
568 warn "W: skipping $oid ($size > $arg->{index_max_size})\n";
572 sub v1_checkpoint ($$;$) {
573 my ($self, $sync, $stk) = @_;
574 $self->{ibx}->git->check_async_wait;
575 $self->{ibx}->git->cat_async_wait;
577 # latest_cmt may be undef
578 my $newest = $stk ? $stk->{latest_cmt} : undef;
580 my $cur = $self->{mm}->last_commit || '';
581 if (need_update($self, $cur, $newest)) {
582 $self->{mm}->last_commit($newest);
585 ${$sync->{max}} = $BATCH_BYTES;
588 $self->{mm}->{dbh}->commit;
589 if ($newest && need_xapian($self)) {
590 my $cur = $self->{xdb}->get_metadata('last_commit');
591 if (need_update($self, $cur, $newest)) {
592 $self->{xdb}->set_metadata('last_commit', $newest);
596 $self->{over}->rethread_done($sync->{-opt}) if $newest; # all done
597 commit_txn_lazy($self);
598 $self->{ibx}->git->cleanup;
599 my $nr = ${$sync->{nr}};
600 idx_release($self, $nr);
601 # let another process do some work...
602 if (my $pr = $sync->{-opt}->{-progress}) {
603 $pr->("indexed $nr/$sync->{ntodo}\n") if $nr;
605 if (!$stk) { # more to come
606 begin_txn_lazy($self);
607 $self->{mm}->{dbh}->begin_work;
613 my ($self, $sync, $stk) = @_;
614 my $git = $self->{ibx}->git;
615 my $max = $BATCH_BYTES;
618 $sync->{max} = \$max;
619 $sync->{sidx} = $self;
621 $self->{mm}->{dbh}->begin_work;
622 if (my @leftovers = keys %{delete($sync->{D}) // {}}) {
623 warn('W: unindexing '.scalar(@leftovers)." leftovers\n");
624 for my $oid (@leftovers) {
625 $oid = unpack('H*', $oid);
626 $git->cat_async($oid, \&unindex_both, $self);
629 if ($sync->{index_max_size} = $self->{ibx}->{index_max_size}) {
630 $sync->{index_oid} = \&index_both;
632 while (my ($f, $at, $ct, $oid) = $stk->pop_rec) {
634 my $arg = { %$sync, autime => $at, cotime => $ct };
635 if ($sync->{index_max_size}) {
636 $git->check_async($oid, \&check_size, $arg);
638 $git->cat_async($oid, \&index_both, $arg);
640 v1_checkpoint($self, $sync) if $max <= 0;
641 } elsif ($f eq 'd') {
642 $git->cat_async($oid, \&unindex_both, $self);
645 v1_checkpoint($self, $sync, $stk);
648 sub log2stack ($$$$) {
649 my ($sync, $git, $range, $ibx) = @_;
650 my $D = $sync->{D}; # OID_BIN => NR (if reindexing, undef otherwise)
652 if ($ibx->version == 1) {
653 my $path = $hex.'{2}/'.$hex.'{38}';
654 $add = qr!\A:000000 100644 \S+ ($OID) A\t$path$!;
655 $del = qr!\A:100644 000000 ($OID) \S+ D\t$path$!;
657 $del = qr!\A:\d{6} 100644 $OID ($OID) [AM]\td$!;
658 $add = qr!\A:\d{6} 100644 $OID ($OID) [AM]\tm$!;
661 # Count the new files so they can be added newest to oldest
662 # and still have numbers increasing from oldest to newest
663 my $fh = $git->popen(qw(log --raw -r --pretty=tformat:%at-%ct-%H
664 --no-notes --no-color --no-renames --no-abbrev),
668 if (/\A([0-9]+)-([0-9]+)-($OID)$/o) {
669 ($at, $ct) = ($1 + 0, $2 + 0);
670 $stk //= PublicInbox::IdxStack->new($3);
673 if ($D) { # reindex case
674 $D->{pack('H*', $oid)}++;
675 } else { # non-reindex case:
676 $stk->push_rec('d', $at, $ct, $oid);
681 my $oid_bin = pack('H*', $oid);
682 my $nr = --$D->{$oid_bin};
683 delete($D->{$oid_bin}) if $nr <= 0;
685 # nr < 0 (-1) means it never existed
686 $stk->push_rec('m', $at, $ct, $oid) if $nr < 0;
688 $stk->push_rec('m', $at, $ct, $oid);
692 close $fh or die "git log failed: \$?=$?";
693 $stk //= PublicInbox::IdxStack->new;
697 sub prepare_stack ($$$) {
698 my ($self, $sync, $range) = @_;
699 my $git = $self->{ibx}->git;
701 if (index($range, '..') < 0) {
702 # don't show annoying git errors to users who run -index
704 $git->qx(qw(rev-parse -q --verify), "$range^0");
705 return PublicInbox::IdxStack->new->read_prepare if $?;
707 $sync->{D} = $sync->{reindex} ? {} : undef; # OID_BIN => NR
708 log2stack($sync, $git, $range, $self->{ibx});
711 # --is-ancestor requires git 1.8.0+
712 sub is_ancestor ($$$) {
713 my ($git, $cur, $tip) = @_;
714 return 0 unless $git->check($cur);
715 my $cmd = [ 'git', "--git-dir=$git->{git_dir}",
716 qw(merge-base --is-ancestor), $cur, $tip ];
717 my $pid = spawn($cmd);
718 waitpid($pid, 0) == $pid or die join(' ', @$cmd) .' did not finish';
722 sub need_update ($$$) {
723 my ($self, $cur, $new) = @_;
724 my $git = $self->{ibx}->git;
725 return 1 if $cur && !is_ancestor($git, $cur, $new);
726 my $range = $cur eq '' ? $new : "$cur..$new";
727 chomp(my $n = $git->qx(qw(rev-list --count), $range));
728 ($n eq '' || $n > 0);
731 # The last git commit we indexed with Xapian or SQLite (msgmap)
732 # This needs to account for cases where Xapian or SQLite is
733 # out-of-date with respect to the other.
735 my ($self, $mm) = @_;
736 my $lm = $mm->last_commit || '';
738 if (need_xapian($self)) {
739 $lx = $self->{xdb}->get_metadata('last_commit') || '';
743 # Use last_commit from msgmap if it is older or unset
744 if (!$lm || ($lx && $lm && is_ancestor($self->{ibx}->git, $lm, $lx))) {
750 sub reindex_from ($$) {
751 my ($reindex, $last_commit) = @_;
752 return $last_commit unless $reindex;
753 ref($reindex) eq 'HASH' ? $reindex->{from} : '';
756 # indexes all unindexed messages (v1 only)
758 my ($self, $opts) = @_;
759 my $tip = $opts->{ref} || 'HEAD';
760 my $git = $self->{ibx}->git;
762 my $pr = $opts->{-progress};
763 my $sync = { reindex => $opts->{reindex}, -opt => $opts };
764 my $xdb = $self->begin_txn_lazy;
765 $self->{over}->rethread_prepare($opts);
766 my $mm = _msgmap_init($self);
767 if ($sync->{reindex}) {
768 my $last = $mm->last_commit;
772 # somebody just blindly added --reindex when indexing
773 # for the first time, allow it:
774 undef $sync->{reindex};
777 my $last_commit = _last_x_commit($self, $mm);
778 my $lx = reindex_from($sync->{reindex}, $last_commit);
779 my $range = $lx eq '' ? $tip : "$lx..$tip";
780 $pr->("counting changes\n\t$range ... ") if $pr;
781 my $stk = prepare_stack($self, $sync, $range);
782 $sync->{ntodo} = $stk ? $stk->num_records : 0;
783 $pr->("$sync->{ntodo}\n") if $pr; # continue previous line
784 process_stack($self, $sync, $stk);
788 # order matters for unlocking
789 $_[0]->{xdb} = undef;
790 $_[0]->{lockfh} = undef;
793 # remote_* subs are only used by SearchIdxPart
796 if (my $w = $self->{w}) {
797 print $w "commit\n" or die "failed to write commit: $!";
799 $self->commit_txn_lazy;
805 if (my $w = delete $self->{w}) {
806 my $pid = delete $self->{pid} or die "no process to wait on\n";
807 print $w "close\n" or die "failed to write to pid:$pid: $!\n";
808 close $w or die "failed to close pipe for pid:$pid: $!\n";
809 waitpid($pid, 0) == $pid or die "remote process did not finish";
810 $? == 0 or die ref($self)." pid:$pid exited with: $?";
812 die "transaction in progress $self\n" if $self->{txn};
813 idx_release($self) if $self->{xdb};
818 my ($self, $oid, $num) = @_;
819 if (my $w = $self->{w}) {
820 # triggers remove_by_oid in a shard
821 print $w "D $oid $num\n" or die "failed to write remove $!";
823 $self->remove_by_oid($oid, $num);
829 my $xdb = $self->{xdb} || idx_acquire($self);
830 $self->{over}->begin_lazy if $self->{over};
831 $xdb->begin_transaction if $xdb;
838 $self->{ibx}->with_umask(\&_begin_txn, $self) if !$self->{txn};
841 # store 'indexlevel=medium' in v2 shard=0 and v1 (only one shard)
842 # This metadata is read by Admin::detect_indexlevel:
846 if (!$self->{shard} && # undef or 0, not >0
847 delete($self->{-set_indexlevel_once})) {
848 my $xdb = $self->{xdb};
849 my $level = $xdb->get_metadata('indexlevel');
850 if (!$level || $level ne 'medium') {
851 $xdb->set_metadata('indexlevel', 'medium');
858 if (my $xdb = $self->{xdb}) {
859 set_indexlevel($self);
860 $xdb->commit_transaction;
862 $self->{over}->commit_lazy if $self->{over};
865 sub commit_txn_lazy {
867 delete($self->{txn}) and
868 $self->{ibx}->with_umask(\&_commit_txn, $self);
873 if (need_xapian($self)) {
874 die "$$ $0 xdb not released\n" if $self->{xdb};
876 die "$$ $0 still in transaction\n" if $self->{txn};