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;
132 $tg->index_text($to, 1, 'XTO') if $to ne '';
133 $tg->increase_termpos;
134 $tg->index_text($cc, 1, 'XCC') if $cc ne '';
135 $tg->increase_termpos;
138 sub index_body ($$$) {
139 my ($tg, $lines, $inc) = @_;
140 $tg->index_text(join("\n", @$lines), $inc, $inc ? 'XNQ' : 'XQUOT');
142 $tg->increase_termpos;
146 my ($self, $mime, $bytes, $num, $blob) = @_; # mime = Email::MIME object
147 my $db = $self->{xdb};
149 my ($doc_id, $old_tid);
150 my $mid = mid_clean(mid_mime($mime));
151 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
154 die 'Message-ID too long' if length($mid) > MAX_MID_SIZE;
155 my $smsg = $self->lookup_message($mid);
157 # convert a ghost to a regular message
158 # it will also clobber any existing regular message
159 $doc_id = $smsg->doc_id;
160 $old_tid = $smsg->thread_id;
162 $smsg = PublicInbox::SearchMsg->new($mime);
163 my $doc = $smsg->{doc};
164 $doc->add_term(xpfx('mid') . $mid);
166 my $subj = $smsg->subject;
168 my $path = $self->subject_path($subj);
169 $doc->add_term(xpfx('path') . id_compress($path));
172 add_values($smsg, $bytes, $num);
174 my $tg = $self->term_generator;
176 $tg->set_document($doc);
177 $tg->index_text($subj, 1, 'S') if $subj;
178 $tg->increase_termpos;
180 index_users($tg, $smsg);
182 msg_iter($mime, sub {
183 my ($part, $depth, @idx) = @{$_[0]};
184 my $ct = $part->content_type || $ct_msg;
186 # account for filter bugs...
187 $ct =~ m!\btext/plain\b!i or return;
190 my $body = $part->body;
192 my @lines = split(/\n/, $body);
193 while (defined(my $l = shift @lines)) {
195 index_body($tg, \@orig, 1) if @orig;
198 index_body($tg, \@quot, 0) if @quot;
202 index_body($tg, \@quot, 0) if @quot;
203 index_body($tg, \@orig, 1) if @orig;
206 link_message($self, $smsg, $old_tid);
207 $tg->index_text($mid, 1, 'XMID');
208 $doc->set_data($smsg->to_doc_data($blob));
210 if (my $altid = $self->{-altid}) {
211 foreach my $alt (@$altid) {
212 my $id = $alt->mid2alt($mid);
213 next unless defined $id;
214 $doc->add_term($alt->{xprefix} . $id);
217 if (defined $doc_id) {
218 $db->replace_document($doc_id, $doc);
220 $doc_id = $db->add_document($doc);
225 warn "failed to index message <$mid>: $@\n";
231 # returns deleted doc_id on success, undef on missing
233 my ($self, $mid) = @_;
234 my $db = $self->{xdb};
236 $mid = mid_clean($mid);
239 $doc_id = $self->find_unique_doc_id('mid', $mid);
240 $db->delete_document($doc_id) if defined $doc_id;
244 warn "failed to remove message <$mid>: $@\n";
250 sub term_generator { # write-only
253 my $tg = $self->{term_generator};
256 $tg = Search::Xapian::TermGenerator->new;
257 $tg->set_stemmer($self->stemmer);
259 $self->{term_generator} = $tg;
262 # increments last_thread_id counter
263 # returns a 64-bit integer represented as a hex string
266 my $db = $self->{xdb};
267 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
269 $db->set_metadata('last_thread_id', ++$last_thread_id);
275 my ($self, $smsg, $old_tid) = @_;
276 my $doc = $smsg->{doc};
277 my $mid = $smsg->mid;
278 my $mime = $smsg->mime;
279 my $hdr = $mime->header_obj;
280 my $refs = $hdr->header_raw('References');
281 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
282 if (my $irt = $hdr->header_raw('In-Reply-To')) {
283 # last References should be $irt
284 # we will de-dupe later
285 push @refs, mid_clean($irt);
290 my %uniq = ($mid => 1);
291 my @orig_refs = @refs;
294 # prevent circular references via References: here:
295 foreach my $ref (@orig_refs) {
296 if (length($ref) > MAX_MID_SIZE) {
297 warn "References: <$ref> too long, ignoring\n";
305 $smsg->{references} = '<'.join('> <', @refs).'>';
307 # first ref *should* be the thread root,
308 # but we can never trust clients to do the right thing
309 my $ref = shift @refs;
310 $tid = $self->_resolve_mid_to_tid($ref);
311 $self->merge_threads($tid, $old_tid) if defined $old_tid;
313 # the rest of the refs should point to this tid:
314 foreach $ref (@refs) {
315 my $ptid = $self->_resolve_mid_to_tid($ref);
316 merge_threads($self, $tid, $ptid);
319 $tid = $self->next_thread_id;
321 $doc->add_term(xpfx('thread') . $tid);
325 my ($self, $mime, $bytes, $num, $blob) = @_;
326 $self->add_message($mime, $bytes, $num, $blob);
330 my ($self, $mime) = @_;
331 my $mid = eval { mid_clean(mid_mime($mime)) };
332 $self->remove_message($mid) if defined $mid;
336 my ($self, $mime) = @_;
337 $self->{mm}->mid_insert(mid_clean(mid_mime($mime)));
341 my ($self, $mime) = @_;
342 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
346 my ($self, $mime, $bytes, $blob) = @_;
347 my $num = $self->{mm}->num_for(mid_clean(mid_mime($mime)));
348 index_blob($self, $mime, $bytes, $num, $blob);
352 my ($self, $mime) = @_;
353 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
354 unindex_blob($self, $mime);
358 my ($self, $mime, $bytes, $blob) = @_;
359 my $num = index_mm($self, $mime);
360 index_blob($self, $mime, $bytes, $num, $blob);
364 my ($self, $mime) = @_;
365 unindex_blob($self, $mime);
366 unindex_mm($self, $mime);
370 my ($git, $blob, $sizeref) = @_;
372 my $str = $git->cat_file($blob, $sizeref);
373 # fixup bugs from import:
374 $$str =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
375 Email::MIME->new($str);
381 my ($self, $opts) = @_;
382 with_umask($self, sub { $self->_index_sync($opts) });
386 my ($self, $log, $add_cb, $del_cb, $batch_cb) = @_;
387 my $hex = '[a-f0-9]';
388 my $h40 = $hex .'{40}';
389 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
390 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
391 my $git = $self->{git};
394 my $max = $self->{batch_size}; # may be undef
397 while (defined($line = <$log>)) {
398 if ($line =~ /$addmsg/o) {
400 my $mime = do_cat_mail($git, $blob, \$bytes) or next;
401 $add_cb->($self, $mime, $bytes, $blob);
402 } elsif ($line =~ /$delmsg/o) {
404 my $mime = do_cat_mail($git, $blob) or next;
405 $del_cb->($self, $mime);
406 } elsif ($line =~ /^commit ($h40)/o) {
407 if (defined $max && --$max <= 0) {
408 $max = $self->{batch_size};
409 $batch_cb->($latest, 1);
414 $batch_cb->($latest, 0);
420 require PublicInbox::Msgmap;
421 PublicInbox::Msgmap->new($self->{git_dir}, 1);
426 my ($self, $range) = @_;
427 $self->{git}->popen(qw/log --reverse --no-notes --no-color
428 --raw -r --no-abbrev/, $range);
431 # indexes all unindexed messages
433 my ($self, $opts) = @_;
434 my $tip = $opts->{ref} || 'HEAD';
435 my $reindex = $opts->{reindex};
436 my ($mkey, $last_commit, $lx, $xlog);
437 $self->{git}->batch_prepare;
438 my $xdb = _xdb_acquire($self);
439 $xdb->begin_transaction;
442 $mkey = 'last_commit';
443 $last_commit = $xdb->get_metadata('last_commit');
447 $mkey = undef if $last_commit ne '';
449 $xdb->cancel_transaction;
450 $xdb = _xdb_release($self);
452 # ensure we leak no FDs to "git log"
453 my $range = $lx eq '' ? $tip : "$lx..$tip";
454 $xlog = _git_log($self, $range);
456 $xdb = _xdb_acquire($self);
457 $xdb->begin_transaction;
458 } while ($xdb->get_metadata('last_commit') ne $last_commit);
460 my $mm = _msgmap_init($self);
461 my $dbh = $mm->{dbh} if $mm;
464 my ($commit, $more) = @_;
466 $mm->last_commit($commit) if $commit;
470 $xdb->set_metadata($mkey, $commit) if $mkey && $commit;
471 $xdb->commit_transaction;
472 $xdb = _xdb_release($self);
474 # let another process do some work... <
477 $xdb = _xdb_acquire($self);
478 $xdb->begin_transaction;
480 $dbh->begin_work if $dbh;
486 my $lm = $mm->last_commit || '';
488 # Common case is the indexes are synced,
489 # we only need to run git-log once:
490 rlog($self, $xlog, *index_both, *unindex_both, $cb);
492 # Uncommon case, msgmap and xapian are out-of-sync
493 # do not care for performance (but git is fast :>)
494 # This happens if we have to reindex Xapian since
495 # msgmap is a frozen format and our Xapian format
497 my $r = $lm eq '' ? $tip : "$lm..$tip";
499 # first, ensure msgmap is up-to-date:
500 my $mkey_prev = $mkey;
501 $mkey = undef; # ignore xapian, for now
502 my $mlog = _git_log($self, $r);
504 rlog($self, $mlog, *index_mm, *unindex_mm, $cb);
505 $mm_only = $mlog = undef;
507 # now deal with Xapian
510 rlog($self, $xlog, *index_mm2, *unindex_mm2, $cb);
513 # user didn't install DBD::SQLite and DBI
514 rlog($self, $xlog, *index_blob, *unindex_blob, $cb);
518 # this will create a ghost as necessary
519 sub _resolve_mid_to_tid {
520 my ($self, $mid) = @_;
522 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
527 my ($self, $mid) = @_;
529 my $tid = $self->next_thread_id;
530 my $doc = Search::Xapian::Document->new;
531 $doc->add_term(xpfx('mid') . $mid);
532 $doc->add_term(xpfx('thread') . $tid);
533 $doc->add_term(xpfx('type') . 'ghost');
535 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
536 $self->{xdb}->add_document($doc);
542 my ($self, $winner_tid, $loser_tid) = @_;
543 return if $winner_tid == $loser_tid;
544 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
545 my $thread_pfx = xpfx('thread');
546 my $db = $self->{xdb};
548 for (; $head != $tail; $head->inc) {
549 my $docid = $head->get_docid;
550 my $doc = $db->get_document($docid);
551 $doc->remove_term($thread_pfx . $loser_tid);
552 $doc->add_term($thread_pfx . $winner_tid);
553 $db->replace_document($docid, $doc);
557 sub _read_git_config_perm {
559 my @cmd = qw(config core.sharedRepository);
560 my $fh = PublicInbox::Git->new($self->{git_dir})->popen(@cmd);
563 chomp $perm if defined $perm;
567 sub _git_config_perm {
569 my $perm = scalar @_ ? $_[0] : _read_git_config_perm($self);
570 return PERM_GROUP if (!defined($perm) || $perm eq '');
571 return PERM_UMASK if ($perm eq 'umask');
572 return PERM_GROUP if ($perm eq 'group');
573 if ($perm =~ /\A(?:all|world|everybody)\z/) {
574 return PERM_EVERYBODY;
576 return PERM_GROUP if ($perm =~ /\A(?:true|yes|on|1)\z/);
577 return PERM_UMASK if ($perm =~ /\A(?:false|no|off|0)\z/);
580 return PERM_UMASK if ($i == PERM_UMASK);
581 return PERM_GROUP if ($i == OLD_PERM_GROUP);
582 return PERM_EVERYBODY if ($i == OLD_PERM_EVERYBODY);
584 if (($i & 0600) != 0600) {
585 die "core.sharedRepository mode invalid: ".
586 sprintf('%.3o', $i) . "\nOwner must have permissions\n";
592 my ($perm) = @_; # _git_config_perm return value
594 return umask if $rv == 0;
596 # set +x bit if +r or +w were set
597 $rv |= 0100 if ($rv & 0600);
598 $rv |= 0010 if ($rv & 0060);
599 $rv |= 0001 if ($rv & 0006);
604 my ($self, $cb) = @_;
605 my $old = umask $self->{umask};
606 my $rv = eval { $cb->() };
614 # order matters for unlocking
615 $_[0]->{xdb} = undef;
616 $_[0]->{lockfh} = undef;