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 $Email::MIME::ContentType::STRICT_PARAMS = 0;
16 use base qw(PublicInbox::Search);
17 use PublicInbox::MID qw/mid_clean id_compress mid_mime/;
18 use PublicInbox::MsgIter;
20 require PublicInbox::Git;
21 *xpfx = *PublicInbox::Search::xpfx;
23 use constant MAX_MID_SIZE => 244; # max term size - 1 in Xapian
27 OLD_PERM_EVERYBODY => 2,
29 PERM_EVERYBODY => 0664,
33 my ($class, $inbox, $creat) = @_;
37 $git_dir = $inbox->{mainrepo};
38 $altid = $inbox->{altid};
40 require PublicInbox::AltId;
42 PublicInbox::AltId->new($inbox, $_);
46 require Search::Xapian::WritableDatabase;
47 my $self = bless { git_dir => $git_dir, -altid => $altid }, $class;
48 my $perm = $self->_git_config_perm;
49 my $umask = _umask_for($perm);
50 $self->{umask} = $umask;
51 $self->{lock_path} = "$git_dir/ssoma.lock";
52 $self->{git} = PublicInbox::Git->new($git_dir);
53 $self->{creat} = ($creat || 0) == 1;
59 my $xdb = delete $self->{xdb} or croak 'not acquired';
61 _lock_release($self) if $self->{creat};
67 croak 'already acquired' if $self->{xdb};
68 my $dir = PublicInbox::Search->xdir($self->{git_dir});
69 my $flag = Search::Xapian::DB_OPEN;
73 File::Path::mkpath($dir);
74 $self->{batch_size} = 100;
75 $flag = Search::Xapian::DB_CREATE_OR_OPEN;
77 $self->{xdb} = Search::Xapian::WritableDatabase->new($dir, $flag);
80 # we only acquire the flock if creating or reindexing;
81 # PublicInbox::Import already has the lock on its own.
84 croak 'already locked' if $self->{lockfh};
85 sysopen(my $lockfh, $self->{lock_path}, O_WRONLY|O_CREAT) or
86 die "failed to open lock $self->{lock_path}: $!\n";
87 flock($lockfh, LOCK_EX) or die "lock failed: $!\n";
88 $self->{lockfh} = $lockfh;
93 my $lockfh = delete $self->{lockfh} or croak 'not locked';
94 flock($lockfh, LOCK_UN) or die "unlock failed: $!\n";
95 close $lockfh or die "close failed: $!\n";
99 my ($doc, $col, $num) = @_;
100 $num = Search::Xapian::sortable_serialise($num);
101 $doc->add_value($col, $num);
105 my ($self, $mime, $bytes, $num, $blob) = @_; # mime = Email::MIME object
106 my $db = $self->{xdb};
108 my ($doc_id, $old_tid);
109 my $mid = mid_clean(mid_mime($mime));
110 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
113 die 'Message-ID too long' if length($mid) > MAX_MID_SIZE;
114 my $smsg = $self->lookup_message($mid);
116 # convert a ghost to a regular message
117 # it will also clobber any existing regular message
118 $doc_id = $smsg->doc_id;
119 $old_tid = $smsg->thread_id;
121 $smsg = PublicInbox::SearchMsg->new($mime);
122 my $doc = $smsg->{doc};
123 $doc->add_term(xpfx('mid') . $mid);
125 my $subj = $smsg->subject;
127 my $path = $self->subject_path($subj);
128 $doc->add_term(xpfx('path') . id_compress($path));
131 add_val($doc, &PublicInbox::Search::TS, $smsg->ts);
134 add_val($doc, &PublicInbox::Search::NUM, $num);
137 add_val($doc, &PublicInbox::Search::BYTES, $bytes);
139 add_val($doc, &PublicInbox::Search::LINES,
140 $mime->body_raw =~ tr!\n!\n!);
142 my $tg = $self->term_generator;
144 $tg->set_document($doc);
145 $tg->index_text($subj, 1, 'S') if $subj;
146 $tg->increase_termpos;
147 $tg->index_text($subj) if $subj;
148 $tg->increase_termpos;
150 $tg->index_text($smsg->from);
151 $tg->increase_termpos;
153 msg_iter($mime, sub {
154 my ($part, $depth, @idx) = @{$_[0]};
155 my $ct = $part->content_type || $ct_msg;
157 # account for filter bugs...
158 $ct =~ m!\btext/plain\b!i or return;
161 my $body = $part->body;
163 my @lines = split(/\n/, $body);
164 while (defined(my $l = shift @lines)) {
172 $tg->index_text(join("\n", @quot), 0);
174 $tg->increase_termpos;
177 $tg->index_text(join("\n", @orig));
179 $tg->increase_termpos;
183 link_message($self, $smsg, $old_tid);
184 $tg->index_text($mid, 1);
185 $doc->set_data($smsg->to_doc_data($blob));
187 if (my $altid = $self->{-altid}) {
188 foreach my $alt (@$altid) {
189 my $id = $alt->mid2alt($mid);
190 next unless defined $id;
191 $doc->add_term($alt->{xprefix} . $id);
194 if (defined $doc_id) {
195 $db->replace_document($doc_id, $doc);
197 $doc_id = $db->add_document($doc);
202 warn "failed to index message <$mid>: $@\n";
208 # returns deleted doc_id on success, undef on missing
210 my ($self, $mid) = @_;
211 my $db = $self->{xdb};
213 $mid = mid_clean($mid);
216 $doc_id = $self->find_unique_doc_id('mid', $mid);
217 $db->delete_document($doc_id) if defined $doc_id;
221 warn "failed to remove message <$mid>: $@\n";
227 sub term_generator { # write-only
230 my $tg = $self->{term_generator};
233 $tg = Search::Xapian::TermGenerator->new;
234 $tg->set_stemmer($self->stemmer);
236 $self->{term_generator} = $tg;
239 # increments last_thread_id counter
240 # returns a 64-bit integer represented as a hex string
243 my $db = $self->{xdb};
244 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
246 $db->set_metadata('last_thread_id', ++$last_thread_id);
252 my ($self, $smsg, $old_tid) = @_;
253 my $doc = $smsg->{doc};
254 my $mid = $smsg->mid;
255 my $mime = $smsg->mime;
256 my $hdr = $mime->header_obj;
257 my $refs = $hdr->header_raw('References');
258 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
259 if (my $irt = $hdr->header_raw('In-Reply-To')) {
260 # last References should be $irt
261 # we will de-dupe later
262 push @refs, mid_clean($irt);
267 my %uniq = ($mid => 1);
268 my @orig_refs = @refs;
271 # prevent circular references via References: here:
272 foreach my $ref (@orig_refs) {
273 if (length($ref) > MAX_MID_SIZE) {
274 warn "References: <$ref> too long, ignoring\n";
282 $smsg->{references} = '<'.join('> <', @refs).'>';
284 # first ref *should* be the thread root,
285 # but we can never trust clients to do the right thing
286 my $ref = shift @refs;
287 $tid = $self->_resolve_mid_to_tid($ref);
288 $self->merge_threads($tid, $old_tid) if defined $old_tid;
290 # the rest of the refs should point to this tid:
291 foreach $ref (@refs) {
292 my $ptid = $self->_resolve_mid_to_tid($ref);
293 merge_threads($self, $tid, $ptid);
296 $tid = $self->next_thread_id;
298 $doc->add_term(xpfx('thread') . $tid);
302 my ($self, $mime, $bytes, $num, $blob) = @_;
303 $self->add_message($mime, $bytes, $num, $blob);
307 my ($self, $mime) = @_;
308 my $mid = eval { mid_clean(mid_mime($mime)) };
309 $self->remove_message($mid) if defined $mid;
313 my ($self, $mime) = @_;
314 $self->{mm}->mid_insert(mid_clean(mid_mime($mime)));
318 my ($self, $mime) = @_;
319 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
323 my ($self, $mime, $bytes, $blob) = @_;
324 my $num = $self->{mm}->num_for(mid_clean(mid_mime($mime)));
325 index_blob($self, $mime, $bytes, $num, $blob);
329 my ($self, $mime) = @_;
330 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
331 unindex_blob($self, $mime);
335 my ($self, $mime, $bytes, $blob) = @_;
336 my $num = index_mm($self, $mime);
337 index_blob($self, $mime, $bytes, $num, $blob);
341 my ($self, $mime) = @_;
342 unindex_blob($self, $mime);
343 unindex_mm($self, $mime);
347 my ($git, $blob, $sizeref) = @_;
349 my $str = $git->cat_file($blob, $sizeref);
350 # fixup bugs from import:
351 $$str =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
352 Email::MIME->new($str);
358 my ($self, $opts) = @_;
359 with_umask($self, sub { $self->_index_sync($opts) });
363 my ($self, $log, $add_cb, $del_cb, $batch_cb) = @_;
364 my $hex = '[a-f0-9]';
365 my $h40 = $hex .'{40}';
366 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
367 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
368 my $git = $self->{git};
371 my $max = $self->{batch_size}; # may be undef
374 while (defined($line = <$log>)) {
375 if ($line =~ /$addmsg/o) {
377 my $mime = do_cat_mail($git, $blob, \$bytes) or next;
378 $add_cb->($self, $mime, $bytes, $blob);
379 } elsif ($line =~ /$delmsg/o) {
381 my $mime = do_cat_mail($git, $blob) or next;
382 $del_cb->($self, $mime);
383 } elsif ($line =~ /^commit ($h40)/o) {
384 if (defined $max && --$max <= 0) {
385 $max = $self->{batch_size};
386 $batch_cb->($latest, 1);
391 $batch_cb->($latest, 0);
397 require PublicInbox::Msgmap;
398 PublicInbox::Msgmap->new($self->{git_dir}, 1);
403 my ($self, $range) = @_;
404 $self->{git}->popen(qw/log --reverse --no-notes --no-color
405 --raw -r --no-abbrev/, $range);
408 # indexes all unindexed messages
410 my ($self, $opts) = @_;
411 my $tip = $opts->{ref} || 'HEAD';
412 my $reindex = $opts->{reindex};
413 my ($mkey, $last_commit, $lx, $xlog);
414 $self->{git}->batch_prepare;
415 my $xdb = _xdb_acquire($self);
416 $xdb->begin_transaction;
419 $mkey = 'last_commit';
420 $last_commit = $xdb->get_metadata('last_commit');
424 $mkey = undef if $last_commit ne '';
426 $xdb->cancel_transaction;
427 $xdb = _xdb_release($self);
429 # ensure we leak no FDs to "git log"
430 my $range = $lx eq '' ? $tip : "$lx..$tip";
431 $xlog = _git_log($self, $range);
433 $xdb = _xdb_acquire($self);
434 $xdb->begin_transaction;
435 } while ($xdb->get_metadata('last_commit') ne $last_commit);
437 my $mm = _msgmap_init($self);
438 my $dbh = $mm->{dbh} if $mm;
440 my ($commit, $more) = @_;
442 $mm->last_commit($commit) if $commit;
445 $xdb->set_metadata($mkey, $commit) if $mkey && $commit;
446 $xdb->commit_transaction;
447 $xdb = _xdb_release($self);
448 # let another process do some work... <
450 $xdb = _xdb_acquire($self);
451 $xdb->begin_transaction;
452 $dbh->begin_work if $dbh;
458 my $lm = $mm->last_commit || '';
460 # Common case is the indexes are synced,
461 # we only need to run git-log once:
462 rlog($self, $xlog, *index_both, *unindex_both, $cb);
464 # Uncommon case, msgmap and xapian are out-of-sync
465 # do not care for performance (but git is fast :>)
466 # This happens if we have to reindex Xapian since
467 # msgmap is a frozen format and our Xapian format
469 my $r = $lm eq '' ? $tip : "$lm..$tip";
471 # first, ensure msgmap is up-to-date:
472 my $mkey_prev = $mkey;
473 $mkey = undef; # ignore xapian, for now
474 my $mlog = _git_log($self, $r);
475 rlog($self, $mlog, *index_mm, *unindex_mm, $cb);
478 # now deal with Xapian
481 $xdb = _xdb_acquire($self);
482 $xdb->begin_transaction;
483 rlog($self, $xlog, *index_mm2, *unindex_mm2, $cb);
486 # user didn't install DBD::SQLite and DBI
487 rlog($self, $xlog, *index_blob, *unindex_blob, $cb);
491 # this will create a ghost as necessary
492 sub _resolve_mid_to_tid {
493 my ($self, $mid) = @_;
495 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
500 my ($self, $mid) = @_;
502 my $tid = $self->next_thread_id;
503 my $doc = Search::Xapian::Document->new;
504 $doc->add_term(xpfx('mid') . $mid);
505 $doc->add_term(xpfx('thread') . $tid);
506 $doc->add_term(xpfx('type') . 'ghost');
508 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
509 $self->{xdb}->add_document($doc);
515 my ($self, $winner_tid, $loser_tid) = @_;
516 return if $winner_tid == $loser_tid;
517 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
518 my $thread_pfx = xpfx('thread');
519 my $db = $self->{xdb};
521 for (; $head != $tail; $head->inc) {
522 my $docid = $head->get_docid;
523 my $doc = $db->get_document($docid);
524 $doc->remove_term($thread_pfx . $loser_tid);
525 $doc->add_term($thread_pfx . $winner_tid);
526 $db->replace_document($docid, $doc);
530 sub _read_git_config_perm {
532 my @cmd = qw(config core.sharedRepository);
533 my $fh = PublicInbox::Git->new($self->{git_dir})->popen(@cmd);
536 chomp $perm if defined $perm;
540 sub _git_config_perm {
542 my $perm = scalar @_ ? $_[0] : _read_git_config_perm($self);
543 return PERM_GROUP if (!defined($perm) || $perm eq '');
544 return PERM_UMASK if ($perm eq 'umask');
545 return PERM_GROUP if ($perm eq 'group');
546 if ($perm =~ /\A(?:all|world|everybody)\z/) {
547 return PERM_EVERYBODY;
549 return PERM_GROUP if ($perm =~ /\A(?:true|yes|on|1)\z/);
550 return PERM_UMASK if ($perm =~ /\A(?:false|no|off|0)\z/);
553 return PERM_UMASK if ($i == PERM_UMASK);
554 return PERM_GROUP if ($i == OLD_PERM_GROUP);
555 return PERM_EVERYBODY if ($i == OLD_PERM_EVERYBODY);
557 if (($i & 0600) != 0600) {
558 die "core.sharedRepository mode invalid: ".
559 sprintf('%.3o', $i) . "\nOwner must have permissions\n";
565 my ($perm) = @_; # _git_config_perm return value
567 return umask if $rv == 0;
569 # set +x bit if +r or +w were set
570 $rv |= 0100 if ($rv & 0600);
571 $rv |= 0010 if ($rv & 0060);
572 $rv |= 0001 if ($rv & 0006);
577 my ($self, $cb) = @_;
578 my $old = umask $self->{umask};
579 my $rv = eval { $cb->() };
587 # order matters for unlocking
588 $_[0]->{xdb} = undef;
589 $_[0]->{lockfh} = undef;