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);
176 $tg->index_text($subj, 1, 'S') if $subj;
177 $tg->increase_termpos;
178 $tg->index_text($subj) if $subj;
179 $tg->increase_termpos;
181 index_users($tg, $smsg);
183 msg_iter($mime, sub {
184 my ($part, $depth, @idx) = @{$_[0]};
185 my $ct = $part->content_type || $ct_msg;
187 # account for filter bugs...
188 $ct =~ m!\btext/plain\b!i or return;
191 my $body = $part->body;
193 my @lines = split(/\n/, $body);
194 while (defined(my $l = shift @lines)) {
202 $tg->index_text(join("\n", @quot), 0);
204 $tg->increase_termpos;
207 $tg->index_text(join("\n", @orig));
209 $tg->increase_termpos;
213 link_message($self, $smsg, $old_tid);
214 $tg->index_text($mid, 1);
215 $doc->set_data($smsg->to_doc_data($blob));
217 if (my $altid = $self->{-altid}) {
218 foreach my $alt (@$altid) {
219 my $id = $alt->mid2alt($mid);
220 next unless defined $id;
221 $doc->add_term($alt->{xprefix} . $id);
224 if (defined $doc_id) {
225 $db->replace_document($doc_id, $doc);
227 $doc_id = $db->add_document($doc);
232 warn "failed to index message <$mid>: $@\n";
238 # returns deleted doc_id on success, undef on missing
240 my ($self, $mid) = @_;
241 my $db = $self->{xdb};
243 $mid = mid_clean($mid);
246 $doc_id = $self->find_unique_doc_id('mid', $mid);
247 $db->delete_document($doc_id) if defined $doc_id;
251 warn "failed to remove message <$mid>: $@\n";
257 sub term_generator { # write-only
260 my $tg = $self->{term_generator};
263 $tg = Search::Xapian::TermGenerator->new;
264 $tg->set_stemmer($self->stemmer);
266 $self->{term_generator} = $tg;
269 # increments last_thread_id counter
270 # returns a 64-bit integer represented as a hex string
273 my $db = $self->{xdb};
274 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
276 $db->set_metadata('last_thread_id', ++$last_thread_id);
282 my ($self, $smsg, $old_tid) = @_;
283 my $doc = $smsg->{doc};
284 my $mid = $smsg->mid;
285 my $mime = $smsg->mime;
286 my $hdr = $mime->header_obj;
287 my $refs = $hdr->header_raw('References');
288 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
289 if (my $irt = $hdr->header_raw('In-Reply-To')) {
290 # last References should be $irt
291 # we will de-dupe later
292 push @refs, mid_clean($irt);
297 my %uniq = ($mid => 1);
298 my @orig_refs = @refs;
301 # prevent circular references via References: here:
302 foreach my $ref (@orig_refs) {
303 if (length($ref) > MAX_MID_SIZE) {
304 warn "References: <$ref> too long, ignoring\n";
312 $smsg->{references} = '<'.join('> <', @refs).'>';
314 # first ref *should* be the thread root,
315 # but we can never trust clients to do the right thing
316 my $ref = shift @refs;
317 $tid = $self->_resolve_mid_to_tid($ref);
318 $self->merge_threads($tid, $old_tid) if defined $old_tid;
320 # the rest of the refs should point to this tid:
321 foreach $ref (@refs) {
322 my $ptid = $self->_resolve_mid_to_tid($ref);
323 merge_threads($self, $tid, $ptid);
326 $tid = $self->next_thread_id;
328 $doc->add_term(xpfx('thread') . $tid);
332 my ($self, $mime, $bytes, $num, $blob) = @_;
333 $self->add_message($mime, $bytes, $num, $blob);
337 my ($self, $mime) = @_;
338 my $mid = eval { mid_clean(mid_mime($mime)) };
339 $self->remove_message($mid) if defined $mid;
343 my ($self, $mime) = @_;
344 $self->{mm}->mid_insert(mid_clean(mid_mime($mime)));
348 my ($self, $mime) = @_;
349 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
353 my ($self, $mime, $bytes, $blob) = @_;
354 my $num = $self->{mm}->num_for(mid_clean(mid_mime($mime)));
355 index_blob($self, $mime, $bytes, $num, $blob);
359 my ($self, $mime) = @_;
360 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
361 unindex_blob($self, $mime);
365 my ($self, $mime, $bytes, $blob) = @_;
366 my $num = index_mm($self, $mime);
367 index_blob($self, $mime, $bytes, $num, $blob);
371 my ($self, $mime) = @_;
372 unindex_blob($self, $mime);
373 unindex_mm($self, $mime);
377 my ($git, $blob, $sizeref) = @_;
379 my $str = $git->cat_file($blob, $sizeref);
380 # fixup bugs from import:
381 $$str =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
382 Email::MIME->new($str);
388 my ($self, $opts) = @_;
389 with_umask($self, sub { $self->_index_sync($opts) });
393 my ($self, $log, $add_cb, $del_cb, $batch_cb) = @_;
394 my $hex = '[a-f0-9]';
395 my $h40 = $hex .'{40}';
396 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
397 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
398 my $git = $self->{git};
401 my $max = $self->{batch_size}; # may be undef
404 while (defined($line = <$log>)) {
405 if ($line =~ /$addmsg/o) {
407 my $mime = do_cat_mail($git, $blob, \$bytes) or next;
408 $add_cb->($self, $mime, $bytes, $blob);
409 } elsif ($line =~ /$delmsg/o) {
411 my $mime = do_cat_mail($git, $blob) or next;
412 $del_cb->($self, $mime);
413 } elsif ($line =~ /^commit ($h40)/o) {
414 if (defined $max && --$max <= 0) {
415 $max = $self->{batch_size};
416 $batch_cb->($latest, 1);
421 $batch_cb->($latest, 0);
427 require PublicInbox::Msgmap;
428 PublicInbox::Msgmap->new($self->{git_dir}, 1);
433 my ($self, $range) = @_;
434 $self->{git}->popen(qw/log --reverse --no-notes --no-color
435 --raw -r --no-abbrev/, $range);
438 # indexes all unindexed messages
440 my ($self, $opts) = @_;
441 my $tip = $opts->{ref} || 'HEAD';
442 my $reindex = $opts->{reindex};
443 my ($mkey, $last_commit, $lx, $xlog);
444 $self->{git}->batch_prepare;
445 my $xdb = _xdb_acquire($self);
446 $xdb->begin_transaction;
449 $mkey = 'last_commit';
450 $last_commit = $xdb->get_metadata('last_commit');
454 $mkey = undef if $last_commit ne '';
456 $xdb->cancel_transaction;
457 $xdb = _xdb_release($self);
459 # ensure we leak no FDs to "git log"
460 my $range = $lx eq '' ? $tip : "$lx..$tip";
461 $xlog = _git_log($self, $range);
463 $xdb = _xdb_acquire($self);
464 $xdb->begin_transaction;
465 } while ($xdb->get_metadata('last_commit') ne $last_commit);
467 my $mm = _msgmap_init($self);
468 my $dbh = $mm->{dbh} if $mm;
471 my ($commit, $more) = @_;
473 $mm->last_commit($commit) if $commit;
477 $xdb->set_metadata($mkey, $commit) if $mkey && $commit;
478 $xdb->commit_transaction;
479 $xdb = _xdb_release($self);
481 # let another process do some work... <
484 $xdb = _xdb_acquire($self);
485 $xdb->begin_transaction;
487 $dbh->begin_work if $dbh;
493 my $lm = $mm->last_commit || '';
495 # Common case is the indexes are synced,
496 # we only need to run git-log once:
497 rlog($self, $xlog, *index_both, *unindex_both, $cb);
499 # Uncommon case, msgmap and xapian are out-of-sync
500 # do not care for performance (but git is fast :>)
501 # This happens if we have to reindex Xapian since
502 # msgmap is a frozen format and our Xapian format
504 my $r = $lm eq '' ? $tip : "$lm..$tip";
506 # first, ensure msgmap is up-to-date:
507 my $mkey_prev = $mkey;
508 $mkey = undef; # ignore xapian, for now
509 my $mlog = _git_log($self, $r);
511 rlog($self, $mlog, *index_mm, *unindex_mm, $cb);
512 $mm_only = $mlog = undef;
514 # now deal with Xapian
517 rlog($self, $xlog, *index_mm2, *unindex_mm2, $cb);
520 # user didn't install DBD::SQLite and DBI
521 rlog($self, $xlog, *index_blob, *unindex_blob, $cb);
525 # this will create a ghost as necessary
526 sub _resolve_mid_to_tid {
527 my ($self, $mid) = @_;
529 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
534 my ($self, $mid) = @_;
536 my $tid = $self->next_thread_id;
537 my $doc = Search::Xapian::Document->new;
538 $doc->add_term(xpfx('mid') . $mid);
539 $doc->add_term(xpfx('thread') . $tid);
540 $doc->add_term(xpfx('type') . 'ghost');
542 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
543 $self->{xdb}->add_document($doc);
549 my ($self, $winner_tid, $loser_tid) = @_;
550 return if $winner_tid == $loser_tid;
551 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
552 my $thread_pfx = xpfx('thread');
553 my $db = $self->{xdb};
555 for (; $head != $tail; $head->inc) {
556 my $docid = $head->get_docid;
557 my $doc = $db->get_document($docid);
558 $doc->remove_term($thread_pfx . $loser_tid);
559 $doc->add_term($thread_pfx . $winner_tid);
560 $db->replace_document($docid, $doc);
564 sub _read_git_config_perm {
566 my @cmd = qw(config core.sharedRepository);
567 my $fh = PublicInbox::Git->new($self->{git_dir})->popen(@cmd);
570 chomp $perm if defined $perm;
574 sub _git_config_perm {
576 my $perm = scalar @_ ? $_[0] : _read_git_config_perm($self);
577 return PERM_GROUP if (!defined($perm) || $perm eq '');
578 return PERM_UMASK if ($perm eq 'umask');
579 return PERM_GROUP if ($perm eq 'group');
580 if ($perm =~ /\A(?:all|world|everybody)\z/) {
581 return PERM_EVERYBODY;
583 return PERM_GROUP if ($perm =~ /\A(?:true|yes|on|1)\z/);
584 return PERM_UMASK if ($perm =~ /\A(?:false|no|off|0)\z/);
587 return PERM_UMASK if ($i == PERM_UMASK);
588 return PERM_GROUP if ($i == OLD_PERM_GROUP);
589 return PERM_EVERYBODY if ($i == OLD_PERM_EVERYBODY);
591 if (($i & 0600) != 0600) {
592 die "core.sharedRepository mode invalid: ".
593 sprintf('%.3o', $i) . "\nOwner must have permissions\n";
599 my ($perm) = @_; # _git_config_perm return value
601 return umask if $rv == 0;
603 # set +x bit if +r or +w were set
604 $rv |= 0100 if ($rv & 0600);
605 $rv |= 0010 if ($rv & 0060);
606 $rv |= 0001 if ($rv & 0006);
611 my ($self, $cb) = @_;
612 my $old = umask $self->{umask};
613 my $rv = eval { $cb->() };
621 # order matters for unlocking
622 $_[0]->{xdb} = undef;
623 $_[0]->{lockfh} = undef;