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;
18 use Carp qw(croak carp);
19 use POSIX qw(strftime);
20 use Time::Local qw(timegm);
21 use PublicInbox::OverIdx;
22 use PublicInbox::Spawn qw(spawn nodatacow_dir);
23 use PublicInbox::Git qw(git_unquote);
24 use PublicInbox::MsgTime qw(msg_timestamp msg_datestamp);
25 our @EXPORT_OK = qw(crlf_adjust log2stack is_ancestor check_size prepare_stack
26 index_text term_generator add_val is_bad_blob);
27 my $X = \%PublicInbox::Search::X;
28 our ($DB_CREATE_OR_OPEN, $DB_OPEN);
30 our $BATCH_BYTES = $ENV{XAPIAN_FLUSH_THRESHOLD} ? 0x7fffffff : 1_000_000;
31 use constant DEBUG => !!$ENV{DEBUG};
33 my $xapianlevels = qr/\A(?:full|medium)\z/;
35 my $OID = $hex .'{40,}';
36 our $INDEXLEVELS = qr/\A(?:full|medium|basic)\z/;
39 my ($class, $ibx, $creat, $shard) = @_;
40 ref $ibx or die "BUG: expected PublicInbox::Inbox object: $ibx";
41 my $inboxdir = $ibx->{inboxdir};
42 my $version = $ibx->version;
43 my $indexlevel = 'full';
44 my $altid = $ibx->{altid};
46 require PublicInbox::AltId;
47 $altid = [ map { PublicInbox::AltId->new($ibx, $_); } @$altid ];
49 if ($ibx->{indexlevel}) {
50 if ($ibx->{indexlevel} =~ $INDEXLEVELS) {
51 $indexlevel = $ibx->{indexlevel};
53 die("Invalid indexlevel $ibx->{indexlevel}\n");
56 $ibx = PublicInbox::InboxWritable->new($ibx);
59 xpfx => $inboxdir, # for xpfx_init
62 indexlevel => $indexlevel,
65 $self->{-set_indexlevel_once} = 1 if $indexlevel eq 'medium';
66 if ($ibx->{-skip_docdata}) {
67 $self->{-set_skip_docdata_once} = 1;
68 $self->{-skip_docdata} = 1;
72 $self->{lock_path} = "$inboxdir/ssoma.lock";
73 my $dir = $self->xdir;
74 $self->{oidx} = PublicInbox::OverIdx->new("$dir/over.sqlite3");
75 $self->{oidx}->{-no_fsync} = 1 if $ibx->{-no_fsync};
76 } elsif ($version == 2) {
77 defined $shard or die "shard is required for v2\n";
79 $self->{shard} = $shard;
80 $self->{lock_path} = undef;
82 die "unsupported inbox version=$version\n";
84 $self->{creat} = ($creat || 0) == 1;
88 sub need_xapian ($) { $_[0]->{indexlevel} =~ $xapianlevels }
91 my ($self, $wake) = @_;
92 if (need_xapian($self)) {
93 my $xdb = delete $self->{xdb} or croak 'not acquired';
96 $self->lock_release($wake) if $self->{creat};
100 sub load_xapian_writable () {
101 return 1 if $X->{WritableDatabase};
102 PublicInbox::Search::load_xapian() or return;
103 my $xap = $PublicInbox::Search::Xap;
104 for (qw(Document TermGenerator WritableDatabase)) {
105 $X->{$_} = $xap.'::'.$_;
107 eval 'require '.$X->{WritableDatabase} or die;
108 *sortable_serialise = $xap.'::sortable_serialise';
109 *sortable_unserialise = $xap.'::sortable_unserialise';
110 $DB_CREATE_OR_OPEN = eval($xap.'::DB_CREATE_OR_OPEN()');
111 $DB_OPEN = eval($xap.'::DB_OPEN()');
112 my $ver = (eval($xap.'::major_version()') << 16) |
113 (eval($xap.'::minor_version()') << 8);
114 $DB_NO_SYNC = 0x4 if $ver >= 0x10400;
121 my $dir = $self->xdir;
122 if (need_xapian($self)) {
123 croak 'already acquired' if $self->{xdb};
124 load_xapian_writable();
125 $flag = $self->{creat} ? $DB_CREATE_OR_OPEN : $DB_OPEN;
127 if ($self->{creat}) {
131 # don't create empty Xapian directories if we don't need Xapian
132 my $is_shard = defined($self->{shard});
133 if (!-d $dir && (!$is_shard ||
134 ($is_shard && need_xapian($self)))) {
135 File::Path::mkpath($dir);
137 $self->{-set_has_threadid_once} = 1;
140 return unless defined $flag;
141 $flag |= $DB_NO_SYNC if ($self->{ibx} // $self->{eidx})->{-no_fsync};
142 my $xdb = eval { ($X->{WritableDatabase})->new($dir, $flag) };
143 croak "Failed opening $dir: $@" if $@;
148 my ($doc, $col, $num) = @_;
149 $num = sortable_serialise($num);
150 $doc->add_value($col, $num);
153 sub term_generator ($) { # write-only
156 $self->{term_generator} //= do {
157 my $tg = $X->{TermGenerator}->new;
158 $tg->set_stemmer(PublicInbox::Search::stemmer($self));
163 sub index_text ($$$$) {
164 my ($self, $text, $wdf_inc, $prefix) = @_;
165 my $tg = term_generator($self); # man Search::Xapian::TermGenerator
167 if ($self->{indexlevel} eq 'full') {
168 $tg->index_text($text, $wdf_inc, $prefix);
169 $tg->increase_termpos;
171 $tg->index_text_without_positions($text, $wdf_inc, $prefix);
175 sub index_headers ($$) {
176 my ($self, $smsg) = @_;
177 my @x = (from => 'A', # Author
178 subject => 'S', to => 'XTO', cc => 'XCC');
179 while (my ($field, $pfx) = splice(@x, 0, 2)) {
180 my $val = $smsg->{$field};
181 index_text($self, $val, 1, $pfx) if $val ne '';
185 sub index_diff_inc ($$$$) {
186 my ($self, $text, $pfx, $xnq) = @_;
188 index_text($self, join("\n", @$xnq), 1, 'XNQ');
191 index_text($self, $text, 1, $pfx);
194 sub index_old_diff_fn {
195 my ($self, $seen, $fa, $fb, $xnq) = @_;
197 # no renames or space support for traditional diffs,
198 # find the number of leading common paths to strip:
199 my @fa = split('/', $fa);
200 my @fb = split('/', $fb);
201 while (scalar(@fa) && scalar(@fb)) {
202 $fa = join('/', @fa);
203 $fb = join('/', @fb);
205 unless ($seen->{$fa}++) {
206 index_diff_inc($self, $fa, 'XDFN', $xnq);
216 sub index_diff ($$$) {
217 my ($self, $txt, $doc) = @_;
222 foreach (split(/\n/, $txt)) {
223 if ($in_diff && s/^ //) { # diff context
224 index_diff_inc($self, $_, 'XDFCTX', $xnq);
225 } elsif (/^-- $/) { # email signature begins
227 } elsif (m!^diff --git "?[^/]+/.+ "?[^/]+/.+\z!) {
228 # wait until "---" and "+++" to capture filenames
231 } elsif (m/^diff -(.+) (\S+) (\S+)$/) {
232 my ($opt, $fa, $fb) = ($1, $2, $3);
234 # only support unified:
235 next unless $opt =~ /[uU]/;
236 $in_diff = index_old_diff_fn($self, \%seen, $fa, $fb,
238 } elsif (m!^--- ("?[^/]+/.+)!) {
240 $fn = (split('/', git_unquote($fn), 2))[1];
241 $seen{$fn}++ or index_diff_inc($self, $fn, 'XDFN', $xnq);
243 } elsif (m!^\+\+\+ ("?[^/]+/.+)!) {
245 $fn = (split('/', git_unquote($fn), 2))[1];
246 $seen{$fn}++ or index_diff_inc($self, $fn, 'XDFN', $xnq);
248 } elsif (/^--- (\S+)/) {
251 } elsif (defined $in_diff && /^\+\+\+ (\S+)/) {
252 $in_diff = index_old_diff_fn($self, \%seen, $in_diff,
254 } elsif ($in_diff && s/^\+//) { # diff added
255 index_diff_inc($self, $_, 'XDFB', $xnq);
256 } elsif ($in_diff && s/^-//) { # diff removed
257 index_diff_inc($self, $_, 'XDFA', $xnq);
258 } elsif (m!^index ([a-f0-9]+)\.\.([a-f0-9]+)!) {
259 my ($ba, $bb) = ($1, $2);
260 index_git_blob_id($doc, 'XDFPRE', $ba);
261 index_git_blob_id($doc, 'XDFPOST', $bb);
263 } elsif (/^@@ (?:\S+) (?:\S+) @@\s*$/) {
264 # traditional diff w/o -p
265 } elsif (/^@@ (?:\S+) (?:\S+) @@\s*(\S+.*)$/) {
266 # hunk header context
267 index_diff_inc($self, $1, 'XDFHH', $xnq);
268 # ignore the following lines:
269 } elsif (/^(?:dis)similarity index/ ||
270 /^(?:old|new) mode/ ||
271 /^(?:deleted|new) file mode/ ||
272 /^(?:copy|rename) (?:from|to) / ||
273 /^(?:dis)?similarity index / ||
274 /^\\ No newline at end of file/ ||
275 /^Binary files .* differ/) {
278 # possible to be in diff context, some mail may be
279 # stripped by MUA or even GNU diff(1). "git apply"
280 # treats a bare "\n" as diff context, too
283 warn "non-diff line: $_\n" if DEBUG && $_ ne '';
288 index_text($self, join("\n", @xnq), 1, 'XNQ');
291 sub index_xapian { # msg_iter callback
292 my $part = $_[0]->[0]; # ignore $depth and $idx
293 my ($self, $doc) = @{$_[1]};
294 my $ct = $part->content_type || 'text/plain';
295 my $fn = $part->filename;
296 if (defined $fn && $fn ne '') {
297 index_text($self, $fn, 1, 'XFN');
299 if ($part->{is_submsg}) {
300 my $mids = mids_for_index($part);
301 index_ids($self, $doc, $part, $mids);
302 my $smsg = bless {}, 'PublicInbox::Smsg';
303 $smsg->populate($part);
304 index_headers($self, $smsg);
307 my ($s, undef) = msg_part_text($part, $ct);
308 defined $s or return;
309 $_[0]->[0] = $part = undef; # free memory
311 # split off quoted and unquoted blocks:
312 my @sections = PublicInbox::MsgIter::split_quotes($s);
313 undef $s; # free memory
314 for my $txt (@sections) {
316 index_text($self, $txt, 0, 'XQUOT');
318 # does it look like a diff?
319 if ($txt =~ /^(?:diff|---|\+\+\+) /ms) {
320 index_diff($self, $txt, $doc);
322 index_text($self, $txt, 1, 'XNQ');
325 undef $txt; # free memory
329 sub index_list_id ($$$) {
330 my ($self, $doc, $hdr) = @_;
331 for my $l ($hdr->header_raw('List-Id')) {
332 $l =~ /<([^>]+)>/ or next;
334 $doc->add_boolean_term('G' . $lid);
335 index_text($self, $lid, 1, 'XL'); # probabilistic
339 sub index_ids ($$$$) {
340 my ($self, $doc, $hdr, $mids) = @_;
341 for my $mid (@$mids) {
342 index_text($self, $mid, 1, 'XM');
344 # because too many Message-IDs are prefixed with
346 if ($mid =~ /\w{12,}/) {
347 my @long = ($mid =~ /(\w{3,}+)/g);
348 index_text($self, join(' ', @long), 1, 'XM');
351 $doc->add_boolean_term('Q' . $_) for @$mids;
352 index_list_id($self, $doc, $hdr);
355 sub eml2doc ($$$;$) {
356 my ($self, $eml, $smsg, $mids) = @_;
357 $mids //= mids_for_index($eml);
358 my $doc = $X->{Document}->new;
359 add_val($doc, PublicInbox::Search::TS(), $smsg->{ts});
360 my @ds = gmtime($smsg->{ds});
361 my $yyyymmdd = strftime('%Y%m%d', @ds);
362 add_val($doc, PublicInbox::Search::YYYYMMDD(), $yyyymmdd);
363 my $dt = strftime('%Y%m%d%H%M%S', @ds);
364 add_val($doc, PublicInbox::Search::DT(), $dt);
365 add_val($doc, PublicInbox::Search::BYTES(), $smsg->{bytes});
366 add_val($doc, PublicInbox::Search::UID(), $smsg->{num});
367 add_val($doc, PublicInbox::Search::THREADID, $smsg->{tid});
369 my $tg = term_generator($self);
370 $tg->set_document($doc);
371 index_headers($self, $smsg);
373 if (defined(my $eidx_key = $smsg->{eidx_key})) {
374 $doc->add_boolean_term('O'.$eidx_key);
376 msg_iter($eml, \&index_xapian, [ $self, $doc ]);
377 index_ids($self, $doc, $eml, $mids);
379 # by default, we maintain compatibility with v1.5.0 and earlier
380 # by writing to docdata.glass, users who never exect to downgrade can
382 if (!$self->{-skip_docdata}) {
383 # WWW doesn't need {to} or {cc}, only NNTP
384 $smsg->{to} = $smsg->{cc} = '';
385 PublicInbox::OverIdx::parse_references($smsg, $eml, $mids);
386 my $data = $smsg->to_doc_data;
387 $doc->set_data($data);
390 if (my $altid = $self->{-altid}) {
391 foreach my $alt (@$altid) {
392 my $pfx = $alt->{xprefix};
393 foreach my $mid (@$mids) {
394 my $id = $alt->mid2alt($mid);
395 next unless defined $id;
396 $doc->add_boolean_term($pfx . $id);
403 sub add_xapian ($$$$) {
404 my ($self, $eml, $smsg, $mids) = @_;
405 my $doc = eml2doc($self, $eml, $smsg, $mids);
406 $self->{xdb}->replace_document($smsg->{num}, $doc);
409 sub _msgmap_init ($) {
411 die "BUG: _msgmap_init is only for v1\n" if $self->{ibx_ver} != 1;
412 $self->{mm} //= eval {
413 require PublicInbox::Msgmap;
414 my $rw = $self->{ibx}->{-no_fsync} ? 2 : 1;
415 PublicInbox::Msgmap->new($self->{ibx}->{inboxdir}, $rw);
420 # mime = PublicInbox::Eml or Email::MIME object
421 my ($self, $mime, $smsg, $sync) = @_;
422 my $mids = mids_for_index($mime);
423 $smsg //= bless { blob => '' }, 'PublicInbox::Smsg'; # test-only compat
424 $smsg->{mid} //= $mids->[0]; # v1 compatibility
425 $smsg->{num} //= do { # v1
427 index_mm($self, $mime, $smsg->{blob}, $sync);
431 $smsg->populate($mime, $sync);
432 $smsg->{bytes} //= length($mime->as_string);
435 # order matters, overview stores every possible piece of
436 # data in doc_data (deflated). Xapian only stores a subset
437 # of the fields which exist in over.sqlite3. We may stop
438 # storing doc_data in Xapian sometime after we get multi-inbox
440 if (my $oidx = $self->{oidx}) { # v1 only
441 $oidx->add_overview($mime, $smsg);
443 if (need_xapian($self)) {
444 add_xapian($self, $mime, $smsg, $mids);
449 warn "failed to index message <".join('> <',@$mids).">: $@\n";
456 my ($self, $docid) = @_;
457 my $doc = eval { $self->{xdb}->get_document($docid) };
459 warn "E: $@\n" if $@;
460 warn "E: #$docid missing in Xapian\n";
466 my ($self, $docid, $eidx_key, $eml) = @_;
467 begin_txn_lazy($self);
468 my $doc = _get_doc($self, $docid) or return;
469 term_generator($self)->set_document($doc);
470 $doc->add_boolean_term('O'.$eidx_key);
471 index_list_id($self, $doc, $eml);
472 $self->{xdb}->replace_document($docid, $doc);
475 sub remove_eidx_info {
476 my ($self, $docid, $eidx_key, $eml) = @_;
477 begin_txn_lazy($self);
478 my $doc = _get_doc($self, $docid) or return;
479 eval { $doc->remove_term('O'.$eidx_key) };
480 warn "W: ->remove_term O$eidx_key: $@\n" if $@;
481 for my $l ($eml ? $eml->header_raw('List-Id') : ()) {
482 $l =~ /<([^>]+)>/ or next;
484 eval { $doc->remove_term('G' . $lid) };
485 warn "W: ->remove_term G$lid: $@\n" if $@;
487 # nb: we don't remove the XL probabilistic terms
488 # since terms may overlap if cross-posted.
490 # IOW, a message which has both <foo.example.com>
491 # and <bar.example.com> would have overlapping
492 # "XLexample" and "XLcom" as terms and which we
493 # wouldn't know if they're safe to remove if we just
494 # unindex <foo.example.com> while preserving
497 # In any case, this entire sub is will likely never
498 # be needed and users using the "l:" prefix are probably
501 $self->{xdb}->replace_document($docid, $doc);
505 my ($doc, $col) = @_;
506 my $val = $doc->get_value($col) or return; # undefined is '' in Xapian
507 sortable_unserialise($val) + 0; # PV => IV conversion
510 sub smsg_from_doc ($) {
512 my $data = $doc->get_data or return;
513 my $smsg = bless {}, 'PublicInbox::Smsg';
514 $smsg->{ts} = int_val($doc, PublicInbox::Search::TS());
515 my $dt = int_val($doc, PublicInbox::Search::DT());
516 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $dt);
517 $smsg->{ds} = timegm($ss, $mm, $hh, $dd, $mon - 1, $yyyy);
518 $smsg->load_from_data($data);
523 my ($self, @docids) = @_;
524 my $xdb = $self->{xdb} or return;
525 for my $docid (@docids) {
526 eval { $xdb->delete_document($docid) };
527 warn "E: #$docid not in in Xapian? $@\n" if $@;
531 sub remove_by_docid {
532 my ($self, $num) = @_;
533 die "BUG: remove_by_docid is v2-only\n" if $self->{oidx};
534 $self->begin_txn_lazy;
535 xdb_remove($self, $num) if need_xapian($self);
538 sub index_git_blob_id {
539 my ($doc, $pfx, $objid) = @_;
541 my $len = length($objid);
542 for (my $len = length($objid); $len >= 7; ) {
543 $doc->add_term($pfx.$objid);
544 $objid = substr($objid, 0, --$len);
550 my ($self, $oid, $eml) = @_;
551 my $mids = mids($eml);
554 for my $mid (@$mids) {
555 my @removed = $self->{oidx}->remove_oid($oid, $mid);
556 $nr += scalar @removed;
557 $tmp{$_}++ for @removed;
560 my $m = join('> <', @$mids);
561 warn "W: <$m> missing for removal from overview\n";
563 while (my ($num, $nr) = each %tmp) {
564 warn "BUG: $num appears >1 times ($nr) for $oid\n" if $nr != 1;
567 $self->{mm}->num_delete($_) for (keys %tmp);
568 } else { # just in case msgmap and over.sqlite3 become desynched:
569 $self->{mm}->mid_delete($mids->[0]);
571 xdb_remove($self, keys %tmp) if need_xapian($self);
575 my ($self, $mime, $oid, $sync) = @_;
576 my $mids = mids($mime);
577 my $mm = $self->{mm};
578 if ($sync->{reindex}) {
579 my $oidx = $self->{oidx};
580 for my $mid (@$mids) {
581 my ($num, undef) = $oidx->num_mid0_for_oid($oid, $mid);
582 return $num if defined $num;
584 $mm->num_for($mids->[0]) // $mm->mid_insert($mids->[0]);
586 # fallback to num_for since filters like RubyLang set the number
587 $mm->mid_insert($mids->[0]) // $mm->num_for($mids->[0]);
591 # returns the number of bytes to add if given a non-CRLF arg
592 sub crlf_adjust ($) {
593 if (index($_[0], "\r\n") < 0) {
594 # common case is LF-only, every \n needs an \r;
595 # so favor a cheap tr// over an expensive m//g
597 } else { # count number of '\n' w/o '\r', expensive:
598 scalar(my @n = ($_[0] =~ m/(?<!\r)\n/g));
602 sub is_bad_blob ($$$$) {
603 my ($oid, $type, $size, $expect_oid) = @_;
604 if ($type ne 'blob') {
605 carp "W: $expect_oid is not a blob (type=$type)";
608 croak "BUG: $oid != $expect_oid" if $oid ne $expect_oid;
609 $size == 0 ? 1 : 0; # size == 0 means purged
612 sub index_both { # git->cat_async callback
613 my ($bref, $oid, $type, $size, $sync) = @_;
614 return if is_bad_blob($oid, $type, $size, $sync->{oid});
615 my ($nr, $max) = @$sync{qw(nr max)};
618 $size += crlf_adjust($$bref);
619 my $smsg = bless { bytes => $size, blob => $oid }, 'PublicInbox::Smsg';
620 my $self = $sync->{sidx};
621 local $self->{current_info} = "$self->{current_info}: $oid";
622 my $eml = PublicInbox::Eml->new($bref);
623 $smsg->{num} = index_mm($self, $eml, $oid, $sync) or
624 die "E: could not generate NNTP article number for $oid";
625 add_message($self, $eml, $smsg, $sync);
626 my $cur_cmt = $sync->{cur_cmt} // die 'BUG: {cur_cmt} missing';
627 ${$sync->{latest_cmt}} = $cur_cmt;
630 sub unindex_both { # git->cat_async callback
631 my ($bref, $oid, $type, $size, $sync) = @_;
632 return if is_bad_blob($oid, $type, $size, $sync->{oid});
633 my $self = $sync->{sidx};
634 local $self->{current_info} = "$self->{current_info}: $oid";
635 unindex_eml($self, $oid, PublicInbox::Eml->new($bref));
636 # may be undef if leftover
637 if (defined(my $cur_cmt = $sync->{cur_cmt})) {
638 ${$sync->{latest_cmt}} = $cur_cmt;
644 ($self->{ibx} // $self->{eidx})->with_umask(@_);
647 # called by public-inbox-index
649 my ($self, $opt) = @_;
650 delete $self->{lock_path} if $opt->{-skip_lock};
651 $self->with_umask(\&_index_sync, $self, $opt);
652 if ($opt->{reindex} && !$opt->{quit}) {
654 delete @again{qw(rethread reindex)};
655 index_sync($self, \%again);
656 $opt->{quit} = $again{quit}; # propagate to caller
660 sub check_size { # check_async cb for -index --max-size=...
661 my ($oid, $type, $size, $arg, $git) = @_;
662 (($type // '') eq 'blob') or die "E: bad $oid in $git->{git_dir}";
663 if ($size <= $arg->{max_size}) {
664 $git->cat_async($oid, $arg->{index_oid}, $arg);
666 warn "W: skipping $oid ($size > $arg->{max_size})\n";
670 sub v1_checkpoint ($$;$) {
671 my ($self, $sync, $stk) = @_;
672 $self->{ibx}->git->async_wait_all;
674 # $newest may be undef
675 my $newest = $stk ? $stk->{latest_cmt} : ${$sync->{latest_cmt}};
676 if (defined($newest)) {
677 my $cur = $self->{mm}->last_commit || '';
678 if (need_update($self, $cur, $newest)) {
679 $self->{mm}->last_commit($newest);
682 ${$sync->{max}} = $self->{batch_bytes};
684 $self->{mm}->{dbh}->commit;
685 my $xdb = need_xapian($self) ? $self->{xdb} : undef;
686 if ($newest && $xdb) {
687 my $cur = $xdb->get_metadata('last_commit');
688 if (need_update($self, $cur, $newest)) {
689 $xdb->set_metadata('last_commit', $newest);
692 if ($stk) { # all done if $stk is passed
693 # let SearchView know a full --reindex was done so it can
694 # generate ->has_threadid-dependent links
695 if ($xdb && $sync->{reindex} && !ref($sync->{reindex})) {
696 my $n = $xdb->get_metadata('has_threadid');
697 $xdb->set_metadata('has_threadid', '1') if $n ne '1';
699 $self->{oidx}->rethread_done($sync->{-opt}); # all done
701 commit_txn_lazy($self);
702 $sync->{ibx}->git->cleanup;
703 my $nr = ${$sync->{nr}};
704 idx_release($self, $nr);
705 # let another process do some work...
706 if (my $pr = $sync->{-opt}->{-progress}) {
707 $pr->("indexed $nr/$sync->{ntodo}\n") if $nr;
709 if (!$stk && !$sync->{quit}) { # more to come
710 begin_txn_lazy($self);
711 $self->{mm}->{dbh}->begin_work;
717 my ($self, $sync, $stk) = @_;
718 my $git = $sync->{ibx}->git;
719 my $max = $self->{batch_bytes};
722 $sync->{max} = \$max;
723 $sync->{sidx} = $self;
724 $sync->{latest_cmt} = \(my $latest_cmt);
726 $self->{mm}->{dbh}->begin_work;
727 if (my @leftovers = keys %{delete($sync->{D}) // {}}) {
728 warn('W: unindexing '.scalar(@leftovers)." leftovers\n");
729 for my $oid (@leftovers) {
730 last if $sync->{quit};
731 $oid = unpack('H*', $oid);
732 $git->cat_async($oid, \&unindex_both, $sync);
735 if ($sync->{max_size} = $sync->{-opt}->{max_size}) {
736 $sync->{index_oid} = \&index_both;
738 while (my ($f, $at, $ct, $oid, $cur_cmt) = $stk->pop_rec) {
739 my $arg = { %$sync, cur_cmt => $cur_cmt, oid => $oid };
740 last if $sync->{quit};
742 $arg->{autime} = $at;
743 $arg->{cotime} = $ct;
744 if ($sync->{max_size}) {
745 $git->check_async($oid, \&check_size, $arg);
747 $git->cat_async($oid, \&index_both, $arg);
749 v1_checkpoint($self, $sync) if $max <= 0;
750 } elsif ($f eq 'd') {
751 $git->cat_async($oid, \&unindex_both, $arg);
754 v1_checkpoint($self, $sync, $sync->{quit} ? undef : $stk);
757 sub log2stack ($$$) {
758 my ($sync, $git, $range) = @_;
759 my $D = $sync->{D}; # OID_BIN => NR (if reindexing, undef otherwise)
761 if ($sync->{ibx}->version == 1) {
762 my $path = $hex.'{2}/'.$hex.'{38}';
763 $add = qr!\A:000000 100644 \S+ ($OID) A\t$path$!;
764 $del = qr!\A:100644 000000 ($OID) \S+ D\t$path$!;
766 $del = qr!\A:\d{6} 100644 $OID ($OID) [AM]\td$!;
767 $add = qr!\A:\d{6} 100644 $OID ($OID) [AM]\tm$!;
770 # Count the new files so they can be added newest to oldest
771 # and still have numbers increasing from oldest to newest
772 my $fh = $git->popen(qw(log --raw -r --pretty=tformat:%at-%ct-%H
773 --no-notes --no-color --no-renames --no-abbrev),
775 my ($at, $ct, $stk, $cmt);
777 return if $sync->{quit};
778 if (/\A([0-9]+)-([0-9]+)-($OID)$/o) {
779 ($at, $ct, $cmt) = ($1 + 0, $2 + 0, $3);
780 $stk //= PublicInbox::IdxStack->new($cmt);
783 if ($D) { # reindex case
784 $D->{pack('H*', $oid)}++;
785 } else { # non-reindex case:
786 $stk->push_rec('d', $at, $ct, $oid, $cmt);
791 my $oid_bin = pack('H*', $oid);
792 my $nr = --$D->{$oid_bin};
793 delete($D->{$oid_bin}) if $nr <= 0;
794 # nr < 0 (-1) means it never existed
797 $stk->push_rec('m', $at, $ct, $oid, $cmt);
800 close $fh or die "git log failed: \$?=$?";
801 $stk //= PublicInbox::IdxStack->new;
805 sub prepare_stack ($$) {
806 my ($sync, $range) = @_;
807 my $git = $sync->{ibx}->git;
809 if (index($range, '..') < 0) {
810 # don't show annoying git errors to users who run -index
812 $git->qx(qw(rev-parse -q --verify), "$range^0");
813 return PublicInbox::IdxStack->new->read_prepare if $?;
815 $sync->{D} = $sync->{reindex} ? {} : undef; # OID_BIN => NR
816 log2stack($sync, $git, $range);
819 # --is-ancestor requires git 1.8.0+
820 sub is_ancestor ($$$) {
821 my ($git, $cur, $tip) = @_;
822 return 0 unless $git->check($cur);
823 my $cmd = [ 'git', "--git-dir=$git->{git_dir}",
824 qw(merge-base --is-ancestor), $cur, $tip ];
825 my $pid = spawn($cmd);
826 waitpid($pid, 0) == $pid or die join(' ', @$cmd) .' did not finish';
830 sub need_update ($$$) {
831 my ($self, $cur, $new) = @_;
832 my $git = $self->{ibx}->git;
833 return 1 if $cur && !is_ancestor($git, $cur, $new);
834 my $range = $cur eq '' ? $new : "$cur..$new";
835 chomp(my $n = $git->qx(qw(rev-list --count), $range));
836 ($n eq '' || $n > 0);
839 # The last git commit we indexed with Xapian or SQLite (msgmap)
840 # This needs to account for cases where Xapian or SQLite is
841 # out-of-date with respect to the other.
843 my ($self, $mm) = @_;
844 my $lm = $mm->last_commit || '';
846 if (need_xapian($self)) {
847 $lx = $self->{xdb}->get_metadata('last_commit') || '';
851 # Use last_commit from msgmap if it is older or unset
852 if (!$lm || ($lx && $lm && is_ancestor($self->{ibx}->git, $lm, $lx))) {
858 sub reindex_from ($$) {
859 my ($reindex, $last_commit) = @_;
860 return $last_commit unless $reindex;
861 ref($reindex) eq 'HASH' ? $reindex->{from} : '';
867 # we set {-opt}->{quit} too, so ->index_sync callers
868 # can abort multi-inbox loops this way
869 $sync->{quit} = $sync->{-opt}->{quit} = 1;
870 warn "gracefully quitting\n";
874 # indexes all unindexed messages (v1 only)
876 my ($self, $opt) = @_;
877 my $tip = $opt->{ref} || 'HEAD';
878 my $ibx = $self->{ibx};
879 local $self->{current_info} = "$ibx->{inboxdir}";
880 $self->{batch_bytes} = $opt->{batch_size} // $BATCH_BYTES;
881 $ibx->git->batch_prepare;
882 my $pr = $opt->{-progress};
883 my $sync = { reindex => $opt->{reindex}, -opt => $opt, ibx => $ibx };
884 my $quit = quit_cb($sync);
885 local $SIG{QUIT} = $quit;
886 local $SIG{INT} = $quit;
887 local $SIG{TERM} = $quit;
888 my $xdb = $self->begin_txn_lazy;
889 $self->{oidx}->rethread_prepare($opt);
890 my $mm = _msgmap_init($self);
891 if ($sync->{reindex}) {
892 my $last = $mm->last_commit;
896 # somebody just blindly added --reindex when indexing
897 # for the first time, allow it:
898 undef $sync->{reindex};
901 my $last_commit = _last_x_commit($self, $mm);
902 my $lx = reindex_from($sync->{reindex}, $last_commit);
903 my $range = $lx eq '' ? $tip : "$lx..$tip";
904 $pr->("counting changes\n\t$range ... ") if $pr;
905 my $stk = prepare_stack($sync, $range);
906 $sync->{ntodo} = $stk ? $stk->num_records : 0;
907 $pr->("$sync->{ntodo}\n") if $pr; # continue previous line
908 process_stack($self, $sync, $stk) if !$sync->{quit};
912 # order matters for unlocking
913 $_[0]->{xdb} = undef;
914 $_[0]->{lockfh} = undef;
919 my $xdb = $self->{xdb} || idx_acquire($self);
920 $self->{oidx}->begin_lazy if $self->{oidx};
921 $xdb->begin_transaction if $xdb;
928 $self->with_umask(\&_begin_txn, $self) if !$self->{txn};
931 # store 'indexlevel=medium' in v2 shard=0 and v1 (only one shard)
932 # This metadata is read by Admin::detect_indexlevel:
933 sub set_metadata_once {
936 return if $self->{shard}; # only continue if undef or 0, not >0
937 my $xdb = $self->{xdb};
939 if (delete($self->{-set_has_threadid_once})) {
940 $xdb->set_metadata('has_threadid', '1');
942 if (delete($self->{-set_indexlevel_once})) {
943 my $level = $xdb->get_metadata('indexlevel');
944 if (!$level || $level ne 'medium') {
945 $xdb->set_metadata('indexlevel', 'medium');
948 if (delete($self->{-set_skip_docdata_once})) {
949 $xdb->get_metadata('skip_docdata') or
950 $xdb->set_metadata('skip_docdata', '1');
956 if (my $eidx = $self->{eidx}) {
957 $eidx->git->async_wait_all;
958 $eidx->{transact_bytes} = 0;
960 if (my $xdb = $self->{xdb}) {
961 set_metadata_once($self);
962 $xdb->commit_transaction;
964 $self->{oidx}->commit_lazy if $self->{oidx};
967 sub commit_txn_lazy {
969 delete($self->{txn}) and
970 $self->with_umask(\&_commit_txn, $self);
975 if (need_xapian($self)) {
976 die "$$ $0 xdb not released\n" if $self->{xdb};
978 die "$$ $0 still in transaction\n" if $self->{txn};
982 my ($class, $eidx, $shard) = @_;
985 xpfx => $eidx->{xpfx},
986 indexlevel => $eidx->{indexlevel},
991 $self->{-set_indexlevel_once} = 1 if $self->{indexlevel} eq 'medium';
995 # ensure there's no stale Xapian docs by treating $over as canonical
997 my ($self, $over) = @_;
998 begin_txn_lazy($self);
999 my $sth = $over->dbh->prepare(<<'');
1000 SELECT COUNT(*) FROM over WHERE num = ?
1002 my $xdb = $self->{xdb};
1003 my $cur = $xdb->postlist_begin('');
1004 my $end = $xdb->postlist_end('');
1005 my $xdir = $self->xdir;
1006 for (; $cur != $end; $cur++) {
1007 my $docid = $cur->get_docid;
1008 $sth->execute($docid);
1009 my $x = $sth->fetchrow_array;
1011 warn "I: removing $xdir #$docid, not in `over'\n";
1012 $xdb->delete_document($docid);