1 # Copyright (C) 2015 all contributors <meta@public-inbox.org>
2 # License: AGPLv3 or later (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 Mail::Thread. This writes to the search
9 package PublicInbox::SearchIdx;
12 use Fcntl qw(:flock :DEFAULT);
14 use Email::MIME::ContentType;
15 $Email::MIME::ContentType::STRICT_PARAMS = 0;
16 use base qw(PublicInbox::Search);
17 use PublicInbox::MID qw/mid_clean id_compress mid_mime/;
18 use PublicInbox::MsgIter;
20 use POSIX qw(strftime);
21 require PublicInbox::Git;
22 *xpfx = *PublicInbox::Search::xpfx;
24 use constant MAX_MID_SIZE => 244; # max term size - 1 in Xapian
28 OLD_PERM_EVERYBODY => 2,
30 PERM_EVERYBODY => 0664,
34 my ($class, $inbox, $creat) = @_;
38 $git_dir = $inbox->{mainrepo};
39 $altid = $inbox->{altid};
41 require PublicInbox::AltId;
43 PublicInbox::AltId->new($inbox, $_);
47 require Search::Xapian::WritableDatabase;
48 my $self = bless { git_dir => $git_dir, -altid => $altid }, $class;
49 my $perm = $self->_git_config_perm;
50 my $umask = _umask_for($perm);
51 $self->{umask} = $umask;
52 $self->{lock_path} = "$git_dir/ssoma.lock";
53 $self->{git} = PublicInbox::Git->new($git_dir);
54 $self->{creat} = ($creat || 0) == 1;
60 my $xdb = delete $self->{xdb} or croak 'not acquired';
62 _lock_release($self) if $self->{creat};
68 croak 'already acquired' if $self->{xdb};
69 my $dir = PublicInbox::Search->xdir($self->{git_dir});
70 my $flag = Search::Xapian::DB_OPEN;
74 File::Path::mkpath($dir);
75 $self->{batch_size} = 100;
76 $flag = Search::Xapian::DB_CREATE_OR_OPEN;
78 $self->{xdb} = Search::Xapian::WritableDatabase->new($dir, $flag);
81 # we only acquire the flock if creating or reindexing;
82 # PublicInbox::Import already has the lock on its own.
85 croak 'already locked' if $self->{lockfh};
86 sysopen(my $lockfh, $self->{lock_path}, O_WRONLY|O_CREAT) or
87 die "failed to open lock $self->{lock_path}: $!\n";
88 flock($lockfh, LOCK_EX) or die "lock failed: $!\n";
89 $self->{lockfh} = $lockfh;
94 my $lockfh = delete $self->{lockfh} or croak 'not locked';
95 flock($lockfh, LOCK_UN) or die "unlock failed: $!\n";
96 close $lockfh or die "close failed: $!\n";
100 my ($doc, $col, $num) = @_;
101 $num = Search::Xapian::sortable_serialise($num);
102 $doc->add_value($col, $num);
105 sub add_values ($$$) {
106 my ($smsg, $bytes, $num) = @_;
109 my $doc = $smsg->{doc};
110 add_val($doc, &PublicInbox::Search::TS, $ts);
112 defined($num) and add_val($doc, &PublicInbox::Search::NUM, $num);
114 defined($bytes) and add_val($doc, &PublicInbox::Search::BYTES, $bytes);
116 add_val($doc, &PublicInbox::Search::LINES,
117 $smsg->{mime}->body_raw =~ tr!\n!\n!);
119 my $yyyymmdd = strftime('%Y%m%d', gmtime($ts));
120 $doc->add_value(&PublicInbox::Search::YYYYMMDD, $yyyymmdd);
123 sub index_users ($$) {
124 my ($tg, $smsg) = @_;
126 my $from = $smsg->from;
130 $tg->index_text($from, 1, 'A'); # A - author
131 $tg->increase_termpos;
133 $tg->index_text($to, 1, 'XTO') if $to ne '';
134 $tg->index_text($cc, 1, 'XCC') if $cc ne '';
135 my $tc = join("\t", $to, $cc);
136 $tg->index_text($tc, 1, 'XTC') if $tc ne '';
137 my $tcf = join("\t", $tc, $from);
138 $tg->index_text($tcf, 1, 'XTCF') if $tcf ne '';
140 $tg->index_text($from);
141 $tg->increase_termpos;
145 my ($self, $mime, $bytes, $num, $blob) = @_; # mime = Email::MIME object
146 my $db = $self->{xdb};
148 my ($doc_id, $old_tid);
149 my $mid = mid_clean(mid_mime($mime));
150 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
153 die 'Message-ID too long' if length($mid) > MAX_MID_SIZE;
154 my $smsg = $self->lookup_message($mid);
156 # convert a ghost to a regular message
157 # it will also clobber any existing regular message
158 $doc_id = $smsg->doc_id;
159 $old_tid = $smsg->thread_id;
161 $smsg = PublicInbox::SearchMsg->new($mime);
162 my $doc = $smsg->{doc};
163 $doc->add_term(xpfx('mid') . $mid);
165 my $subj = $smsg->subject;
167 my $path = $self->subject_path($subj);
168 $doc->add_term(xpfx('path') . id_compress($path));
171 add_values($smsg, $bytes, $num);
173 my $tg = $self->term_generator;
175 $tg->set_document($doc);
177 $tg->index_text($subj, 1, 'S');
178 $tg->index_text($subj, 1, 'XBS');
180 $tg->increase_termpos;
181 $tg->index_text($subj) if $subj;
182 $tg->increase_termpos;
184 index_users($tg, $smsg);
186 msg_iter($mime, sub {
187 my ($part, $depth, @idx) = @{$_[0]};
188 my $ct = $part->content_type || $ct_msg;
190 # account for filter bugs...
191 $ct =~ m!\btext/plain\b!i or return;
194 my $body = $part->body;
196 my @lines = split(/\n/, $body);
197 while (defined(my $l = shift @lines)) {
205 my $s = join("\n", @quot);
207 $tg->index_text($s, 1, 'XQUOT');
208 $tg->index_text($s, 0, 'XBS');
209 $tg->index_text($s, 0, 'XBODY');
210 $tg->index_text($s, 0);
211 $tg->increase_termpos;
214 my $s = join("\n", @orig);
216 $tg->index_text($s, 1, 'XNQ');
217 $tg->index_text($s, 1, 'XBS');
218 $tg->index_text($s, 1, 'XBODY');
220 $tg->increase_termpos;
224 link_message($self, $smsg, $old_tid);
225 $tg->index_text($mid, 1);
226 $doc->set_data($smsg->to_doc_data($blob));
228 if (my $altid = $self->{-altid}) {
229 foreach my $alt (@$altid) {
230 my $id = $alt->mid2alt($mid);
231 next unless defined $id;
232 $doc->add_term($alt->{xprefix} . $id);
235 if (defined $doc_id) {
236 $db->replace_document($doc_id, $doc);
238 $doc_id = $db->add_document($doc);
243 warn "failed to index message <$mid>: $@\n";
249 # returns deleted doc_id on success, undef on missing
251 my ($self, $mid) = @_;
252 my $db = $self->{xdb};
254 $mid = mid_clean($mid);
257 $doc_id = $self->find_unique_doc_id('mid', $mid);
258 $db->delete_document($doc_id) if defined $doc_id;
262 warn "failed to remove message <$mid>: $@\n";
268 sub term_generator { # write-only
271 my $tg = $self->{term_generator};
274 $tg = Search::Xapian::TermGenerator->new;
275 $tg->set_stemmer($self->stemmer);
277 $self->{term_generator} = $tg;
280 # increments last_thread_id counter
281 # returns a 64-bit integer represented as a hex string
284 my $db = $self->{xdb};
285 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
287 $db->set_metadata('last_thread_id', ++$last_thread_id);
293 my ($self, $smsg, $old_tid) = @_;
294 my $doc = $smsg->{doc};
295 my $mid = $smsg->mid;
296 my $mime = $smsg->mime;
297 my $hdr = $mime->header_obj;
298 my $refs = $hdr->header_raw('References');
299 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
300 if (my $irt = $hdr->header_raw('In-Reply-To')) {
301 # last References should be $irt
302 # we will de-dupe later
303 push @refs, mid_clean($irt);
308 my %uniq = ($mid => 1);
309 my @orig_refs = @refs;
312 # prevent circular references via References: here:
313 foreach my $ref (@orig_refs) {
314 if (length($ref) > MAX_MID_SIZE) {
315 warn "References: <$ref> too long, ignoring\n";
323 $smsg->{references} = '<'.join('> <', @refs).'>';
325 # first ref *should* be the thread root,
326 # but we can never trust clients to do the right thing
327 my $ref = shift @refs;
328 $tid = $self->_resolve_mid_to_tid($ref);
329 $self->merge_threads($tid, $old_tid) if defined $old_tid;
331 # the rest of the refs should point to this tid:
332 foreach $ref (@refs) {
333 my $ptid = $self->_resolve_mid_to_tid($ref);
334 merge_threads($self, $tid, $ptid);
337 $tid = $self->next_thread_id;
339 $doc->add_term(xpfx('thread') . $tid);
343 my ($self, $mime, $bytes, $num, $blob) = @_;
344 $self->add_message($mime, $bytes, $num, $blob);
348 my ($self, $mime) = @_;
349 my $mid = eval { mid_clean(mid_mime($mime)) };
350 $self->remove_message($mid) if defined $mid;
354 my ($self, $mime) = @_;
355 $self->{mm}->mid_insert(mid_clean(mid_mime($mime)));
359 my ($self, $mime) = @_;
360 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
364 my ($self, $mime, $bytes, $blob) = @_;
365 my $num = $self->{mm}->num_for(mid_clean(mid_mime($mime)));
366 index_blob($self, $mime, $bytes, $num, $blob);
370 my ($self, $mime) = @_;
371 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
372 unindex_blob($self, $mime);
376 my ($self, $mime, $bytes, $blob) = @_;
377 my $num = index_mm($self, $mime);
378 index_blob($self, $mime, $bytes, $num, $blob);
382 my ($self, $mime) = @_;
383 unindex_blob($self, $mime);
384 unindex_mm($self, $mime);
388 my ($git, $blob, $sizeref) = @_;
390 my $str = $git->cat_file($blob, $sizeref);
391 # fixup bugs from import:
392 $$str =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
393 Email::MIME->new($str);
399 my ($self, $opts) = @_;
400 with_umask($self, sub { $self->_index_sync($opts) });
404 my ($self, $log, $add_cb, $del_cb, $batch_cb) = @_;
405 my $hex = '[a-f0-9]';
406 my $h40 = $hex .'{40}';
407 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
408 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
409 my $git = $self->{git};
412 my $max = $self->{batch_size}; # may be undef
415 while (defined($line = <$log>)) {
416 if ($line =~ /$addmsg/o) {
418 my $mime = do_cat_mail($git, $blob, \$bytes) or next;
419 $add_cb->($self, $mime, $bytes, $blob);
420 } elsif ($line =~ /$delmsg/o) {
422 my $mime = do_cat_mail($git, $blob) or next;
423 $del_cb->($self, $mime);
424 } elsif ($line =~ /^commit ($h40)/o) {
425 if (defined $max && --$max <= 0) {
426 $max = $self->{batch_size};
427 $batch_cb->($latest, 1);
432 $batch_cb->($latest, 0);
438 require PublicInbox::Msgmap;
439 PublicInbox::Msgmap->new($self->{git_dir}, 1);
444 my ($self, $range) = @_;
445 $self->{git}->popen(qw/log --reverse --no-notes --no-color
446 --raw -r --no-abbrev/, $range);
449 # indexes all unindexed messages
451 my ($self, $opts) = @_;
452 my $tip = $opts->{ref} || 'HEAD';
453 my $reindex = $opts->{reindex};
454 my ($mkey, $last_commit, $lx, $xlog);
455 $self->{git}->batch_prepare;
456 my $xdb = _xdb_acquire($self);
457 $xdb->begin_transaction;
460 $mkey = 'last_commit';
461 $last_commit = $xdb->get_metadata('last_commit');
465 $mkey = undef if $last_commit ne '';
467 $xdb->cancel_transaction;
468 $xdb = _xdb_release($self);
470 # ensure we leak no FDs to "git log"
471 my $range = $lx eq '' ? $tip : "$lx..$tip";
472 $xlog = _git_log($self, $range);
474 $xdb = _xdb_acquire($self);
475 $xdb->begin_transaction;
476 } while ($xdb->get_metadata('last_commit') ne $last_commit);
478 my $mm = _msgmap_init($self);
479 my $dbh = $mm->{dbh} if $mm;
482 my ($commit, $more) = @_;
484 $mm->last_commit($commit) if $commit;
488 $xdb->set_metadata($mkey, $commit) if $mkey && $commit;
489 $xdb->commit_transaction;
490 $xdb = _xdb_release($self);
492 # let another process do some work... <
495 $xdb = _xdb_acquire($self);
496 $xdb->begin_transaction;
498 $dbh->begin_work if $dbh;
504 my $lm = $mm->last_commit || '';
506 # Common case is the indexes are synced,
507 # we only need to run git-log once:
508 rlog($self, $xlog, *index_both, *unindex_both, $cb);
510 # Uncommon case, msgmap and xapian are out-of-sync
511 # do not care for performance (but git is fast :>)
512 # This happens if we have to reindex Xapian since
513 # msgmap is a frozen format and our Xapian format
515 my $r = $lm eq '' ? $tip : "$lm..$tip";
517 # first, ensure msgmap is up-to-date:
518 my $mkey_prev = $mkey;
519 $mkey = undef; # ignore xapian, for now
520 my $mlog = _git_log($self, $r);
522 rlog($self, $mlog, *index_mm, *unindex_mm, $cb);
523 $mm_only = $mlog = undef;
525 # now deal with Xapian
528 rlog($self, $xlog, *index_mm2, *unindex_mm2, $cb);
531 # user didn't install DBD::SQLite and DBI
532 rlog($self, $xlog, *index_blob, *unindex_blob, $cb);
536 # this will create a ghost as necessary
537 sub _resolve_mid_to_tid {
538 my ($self, $mid) = @_;
540 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
545 my ($self, $mid) = @_;
547 my $tid = $self->next_thread_id;
548 my $doc = Search::Xapian::Document->new;
549 $doc->add_term(xpfx('mid') . $mid);
550 $doc->add_term(xpfx('thread') . $tid);
551 $doc->add_term(xpfx('type') . 'ghost');
553 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
554 $self->{xdb}->add_document($doc);
560 my ($self, $winner_tid, $loser_tid) = @_;
561 return if $winner_tid == $loser_tid;
562 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
563 my $thread_pfx = xpfx('thread');
564 my $db = $self->{xdb};
566 for (; $head != $tail; $head->inc) {
567 my $docid = $head->get_docid;
568 my $doc = $db->get_document($docid);
569 $doc->remove_term($thread_pfx . $loser_tid);
570 $doc->add_term($thread_pfx . $winner_tid);
571 $db->replace_document($docid, $doc);
575 sub _read_git_config_perm {
577 my @cmd = qw(config core.sharedRepository);
578 my $fh = PublicInbox::Git->new($self->{git_dir})->popen(@cmd);
581 chomp $perm if defined $perm;
585 sub _git_config_perm {
587 my $perm = scalar @_ ? $_[0] : _read_git_config_perm($self);
588 return PERM_GROUP if (!defined($perm) || $perm eq '');
589 return PERM_UMASK if ($perm eq 'umask');
590 return PERM_GROUP if ($perm eq 'group');
591 if ($perm =~ /\A(?:all|world|everybody)\z/) {
592 return PERM_EVERYBODY;
594 return PERM_GROUP if ($perm =~ /\A(?:true|yes|on|1)\z/);
595 return PERM_UMASK if ($perm =~ /\A(?:false|no|off|0)\z/);
598 return PERM_UMASK if ($i == PERM_UMASK);
599 return PERM_GROUP if ($i == OLD_PERM_GROUP);
600 return PERM_EVERYBODY if ($i == OLD_PERM_EVERYBODY);
602 if (($i & 0600) != 0600) {
603 die "core.sharedRepository mode invalid: ".
604 sprintf('%.3o', $i) . "\nOwner must have permissions\n";
610 my ($perm) = @_; # _git_config_perm return value
612 return umask if $rv == 0;
614 # set +x bit if +r or +w were set
615 $rv |= 0100 if ($rv & 0600);
616 $rv |= 0010 if ($rv & 0060);
617 $rv |= 0001 if ($rv & 0006);
622 my ($self, $cb) = @_;
623 my $old = umask $self->{umask};
624 my $rv = eval { $cb->() };
632 # order matters for unlocking
633 $_[0]->{xdb} = undef;
634 $_[0]->{lockfh} = undef;