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
4 package PublicInbox::SearchIdx;
7 use base qw(PublicInbox::Search);
8 use PublicInbox::MID qw/mid_clean id_compress/;
9 *xpfx = *PublicInbox::Search::xpfx;
11 use constant MAX_MID_SIZE => 244; # max term size - 1 in Xapian
15 OLD_PERM_EVERYBODY => 2,
17 PERM_EVERYBODY => 0664,
21 my ($class, $git_dir, $writable) = @_;
22 my $dir = $class->xdir($git_dir);
23 require Search::Xapian::WritableDatabase;
24 my $flag = Search::Xapian::DB_OPEN;
25 my $self = bless { git_dir => $git_dir }, $class;
26 my $perm = $self->_git_config_perm;
27 my $umask = _umask_for($perm);
28 $self->{umask} = $umask;
29 $self->{xdb} = $self->with_umask(sub {
32 File::Path::mkpath($dir);
33 $flag = Search::Xapian::DB_CREATE_OR_OPEN;
35 Search::Xapian::WritableDatabase->new($dir, $flag);
41 my ($doc, $col, $num) = @_;
42 $num = Search::Xapian::sortable_serialise($num);
43 $doc->add_value($col, $num);
47 my ($self, $mime, $bytes, $num) = @_; # mime = Email::MIME object
48 my $db = $self->{xdb};
51 my $mid = mid_clean($mime->header('Message-ID'));
53 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
56 die 'Message-ID too long' if length($mid) > MAX_MID_SIZE;
57 my $smsg = $self->lookup_message($mid);
61 $smsg->ensure_metadata;
62 # convert a ghost to a regular message
63 # it will also clobber any existing regular message
67 my $type = xpfx('type');
69 $doc->remove_term($type . 'ghost');
73 # probably does not exist:
74 eval { $doc->remove_term($type . 'mail') };
75 $doc->add_term($type . 'mail');
77 $smsg = PublicInbox::SearchMsg->new($mime);
79 $doc->add_term(xpfx('mid') . $mid);
82 my $subj = $smsg->subject;
85 my $path = $self->subject_path($subj);
86 $doc->add_term(xpfx('path') . id_compress($path));
89 add_val($doc, &PublicInbox::Search::TS, $smsg->ts);
92 add_val($doc, &PublicInbox::Search::NUM, $num);
95 add_val($doc, &PublicInbox::Search::BYTES, $bytes);
97 add_val($doc, &PublicInbox::Search::LINES,
98 $mime->body_raw =~ tr!\n!\n!);
100 my $tg = $self->term_generator;
102 $tg->set_document($doc);
103 $tg->index_text($subj, 1, 'S') if $subj;
104 $tg->increase_termpos;
105 $tg->index_text($subj) if $subj;
106 $tg->increase_termpos;
108 $tg->index_text($smsg->from);
109 $tg->increase_termpos;
111 $mime->walk_parts(sub {
113 return if $part->subparts; # walk_parts already recurses
114 my $ct = $part->content_type || $ct_msg;
116 # account for filter bugs...
117 $ct =~ m!\btext/plain\b!i or return;
120 my $body = $part->body;
122 my @lines = split(/\n/, $body);
123 while (defined(my $l = shift @lines)) {
131 $tg->index_text(join("\n", @quot), 0);
133 $tg->increase_termpos;
136 $tg->index_text(join("\n", @orig));
138 $tg->increase_termpos;
143 $doc_id = $smsg->doc_id;
144 $self->link_message($smsg, 0);
145 $doc->set_data($smsg->to_doc_data);
146 $db->replace_document($doc_id, $doc);
148 $self->link_message($smsg, 0);
149 $doc->set_data($smsg->to_doc_data);
150 $doc_id = $db->add_document($doc);
155 warn "failed to index message <$mid>: $@\n";
161 # returns deleted doc_id on success, undef on missing
163 my ($self, $mid) = @_;
164 my $db = $self->{xdb};
166 $mid = mid_clean($mid);
169 $doc_id = $self->find_unique_doc_id('mid', $mid);
170 $db->delete_document($doc_id) if defined $doc_id;
174 warn "failed to remove message <$mid>: $@\n";
180 sub term_generator { # write-only
183 my $tg = $self->{term_generator};
186 $tg = Search::Xapian::TermGenerator->new;
187 $tg->set_stemmer($self->stemmer);
189 $self->{term_generator} = $tg;
192 # increments last_thread_id counter
193 # returns a 64-bit integer represented as a hex string
196 my $db = $self->{xdb};
197 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
199 $db->set_metadata('last_thread_id', ++$last_thread_id);
205 my ($self, $smsg, $is_ghost) = @_;
208 $smsg->ensure_metadata;
210 $self->link_message_to_parents($smsg);
214 sub link_message_to_parents {
215 my ($self, $smsg) = @_;
216 my $doc = $smsg->{doc};
217 my $mid = $smsg->mid;
218 my $mime = $smsg->mime;
219 my $refs = $mime->header('References');
220 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
221 if (my $irt = $mime->header('In-Reply-To')) {
222 # last References should be $irt
223 # we will de-dupe later
224 push @refs, mid_clean($irt);
229 my %uniq = ($mid => 1);
230 my @orig_refs = @refs;
233 # prevent circular references via References: here:
234 foreach my $ref (@orig_refs) {
235 if (length($ref) > MAX_MID_SIZE) {
236 warn "References: <$ref> too long, ignoring\n";
244 $smsg->{references} = '<'.join('> <', @refs).'>';
246 # first ref *should* be the thread root,
247 # but we can never trust clients to do the right thing
248 my $ref = shift @refs;
249 $tid = $self->_resolve_mid_to_tid($ref);
251 # the rest of the refs should point to this tid:
252 foreach $ref (@refs) {
253 my $ptid = $self->_resolve_mid_to_tid($ref);
255 $self->merge_threads($tid, $ptid);
259 $tid = $self->next_thread_id;
261 $doc->add_term(xpfx('thread') . $tid);
265 my ($self, $git, $mime, $bytes, $num) = @_;
266 $self->add_message($mime, $bytes, $num);
270 my ($self, $git, $mime) = @_;
271 my $mid = mid_clean($mime->header('Message-ID'));
272 $self->remove_message($mid) if defined $mid;
276 my ($self, $git, $mime) = @_;
277 $self->{mm}->mid_insert(mid_clean($mime->header('Message-ID')));
281 my ($self, $git, $mime) = @_;
282 $self->{mm}->mid_delete(mid_clean($mime->header('Message-ID')));
286 my ($self, $git, $mime, $bytes) = @_;
287 my $num = $self->{mm}->num_for(mid_clean($mime->header('Message-ID')));
288 index_blob($self, $git, $mime, $bytes, $num);
292 my ($self, $git, $mime) = @_;
293 $self->{mm}->mid_delete(mid_clean($mime->header('Message-ID')));
294 unindex_blob($self, $git, $mime);
298 my ($self, $git, $mime, $bytes) = @_;
299 my $num = index_mm($self, $git, $mime);
300 index_blob($self, $git, $mime, $bytes, $num);
304 my ($self, $git, $mime) = @_;
305 unindex_blob($self, $git, $mime);
306 unindex_mm($self, $git, $mime);
310 my ($git, $blob, $sizeref) = @_;
312 my $str = $git->cat_file($blob, $sizeref);
313 Email::MIME->new($str);
319 my ($self, $head) = @_;
320 $self->with_umask(sub { $self->_index_sync($head) });
324 my ($self, $range, $add_cb, $del_cb) = @_;
325 my $hex = '[a-f0-9]';
326 my $h40 = $hex .'{40}';
327 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
328 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
329 my $git_dir = $self->{git_dir};
330 require PublicInbox::GitCatFile;
331 my $git = PublicInbox::GitCatFile->new($git_dir);
332 my @cmd = ('git', "--git-dir=$git_dir", "log",
333 qw/--reverse --no-notes --no-color --raw -r --no-abbrev/,
337 my $pid = open(my $log, '-|', @cmd) or
338 die('open` '.join(' ', @cmd) . " pipe failed: $!\n");
339 while (my $line = <$log>) {
340 if ($line =~ /$addmsg/o) {
341 my $mime = do_cat_mail($git, $1, \$bytes) or next;
342 $add_cb->($self, $git, $mime, $bytes);
343 } elsif ($line =~ /$delmsg/o) {
344 my $mime = do_cat_mail($git, $1) or next;
345 $del_cb->($self, $git, $mime);
346 } elsif ($line =~ /^commit ($h40)/o) {
354 # indexes all unindexed messages
356 my ($self, $head) = @_;
357 my $db = $self->{xdb};
359 my $mm = $self->{mm} = eval {
360 require PublicInbox::Msgmap;
361 PublicInbox::Msgmap->new($self->{git_dir}, 1);
364 $db->begin_transaction;
365 my $lx = $db->get_metadata('last_commit');
366 my $range = $lx eq '' ? $head : "$lx..$head";
368 $mm->{dbh}->begin_work;
369 my $lm = $mm->last_commit || '';
371 # Common case is the indexes are synced,
372 # we only need to run git-log once:
373 $lx = $self->rlog($range, *index_both, *unindex_both);
376 $db->set_metadata('last_commit', $lx);
377 $mm->last_commit($lx);
380 # dumb case, msgmap and xapian are out-of-sync
381 # do not care for performance:
382 my $r = $lm eq '' ? $head : "$lm..$head";
383 $lm = $self->rlog($r, *index_mm, *unindex_mm);
385 $mm->last_commit($lm) if defined $lm;
387 $lx = $self->rlog($range, *index_mm2, *unindex_mm2);
388 $db->set_metadata('last_commit', $lx) if defined $lx;
391 # user didn't install DBD::SQLite and DBI
392 $lx = $self->rlog($range, *index_blob, *unindex_blob);
393 $db->set_metadata('last_commit', $lx) if defined $lx;
396 $db->cancel_transaction;
397 $mm->{dbh}->rollback if $mm;
399 $db->commit_transaction;
403 # this will create a ghost as necessary
404 sub _resolve_mid_to_tid {
405 my ($self, $mid) = @_;
407 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
412 my ($self, $mid, $tid) = @_;
414 $tid = $self->next_thread_id unless defined $tid;
416 my $doc = Search::Xapian::Document->new;
417 $doc->add_term(xpfx('mid') . $mid);
418 $doc->add_term(xpfx('thread') . $tid);
419 $doc->add_term(xpfx('type') . 'ghost');
421 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
422 $self->link_message($smsg, 1);
423 $self->{xdb}->add_document($doc);
429 my ($self, $winner_tid, $loser_tid) = @_;
430 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
431 my $thread_pfx = xpfx('thread');
432 my $db = $self->{xdb};
434 for (; $head != $tail; $head->inc) {
435 my $docid = $head->get_docid;
436 my $doc = $db->get_document($docid);
437 $doc->remove_term($thread_pfx . $loser_tid);
438 $doc->add_term($thread_pfx . $winner_tid);
439 $db->replace_document($docid, $doc);
443 sub _read_git_config_perm {
445 my @cmd = ('git', "--git-dir=$self->{git_dir}",
446 qw(config core.sharedRepository));
447 my $pid = open(my $fh, '-|', @cmd) or
448 die('open `'.join(' ', @cmd) . " pipe failed: $!\n");
451 chomp $perm if defined $perm;
455 sub _git_config_perm {
457 my $perm = scalar @_ ? $_[0] : _read_git_config_perm($self);
458 return PERM_GROUP if (!defined($perm) || $perm eq '');
459 return PERM_UMASK if ($perm eq 'umask');
460 return PERM_GROUP if ($perm eq 'group');
461 if ($perm =~ /\A(?:all|world|everybody)\z/) {
462 return PERM_EVERYBODY;
464 return PERM_GROUP if ($perm =~ /\A(?:true|yes|on|1)\z/);
465 return PERM_UMASK if ($perm =~ /\A(?:false|no|off|0)\z/);
468 return PERM_UMASK if ($i == PERM_UMASK);
469 return PERM_GROUP if ($i == OLD_PERM_GROUP);
470 return PERM_EVERYBODY if ($i == OLD_PERM_EVERYBODY);
472 if (($i & 0600) != 0600) {
473 die "core.sharedRepository mode invalid: ".
474 sprintf('%.3o', $i) . "\nOwner must have permissions\n";
480 my ($perm) = @_; # _git_config_perm return value
482 return umask if $rv == 0;
484 # set +x bit if +r or +w were set
485 $rv |= 0100 if ($rv & 0600);
486 $rv |= 0010 if ($rv & 0060);
487 $rv |= 0001 if ($rv & 0006);
492 my ($self, $cb) = @_;
493 my $old = umask $self->{umask};
494 my $rv = eval { $cb->() };