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 base qw(PublicInbox::Search);
13 use PublicInbox::MID qw/mid_clean id_compress/;
14 require PublicInbox::Git;
15 *xpfx = *PublicInbox::Search::xpfx;
17 use constant MAX_MID_SIZE => 244; # max term size - 1 in Xapian
21 OLD_PERM_EVERYBODY => 2,
23 PERM_EVERYBODY => 0664,
27 my ($class, $git_dir, $writable) = @_;
28 my $dir = $class->xdir($git_dir);
29 require Search::Xapian::WritableDatabase;
30 my $flag = Search::Xapian::DB_OPEN;
31 my $self = bless { git_dir => $git_dir }, $class;
32 my $perm = $self->_git_config_perm;
33 my $umask = _umask_for($perm);
34 $self->{umask} = $umask;
35 $self->{xdb} = $self->with_umask(sub {
38 File::Path::mkpath($dir);
39 $flag = Search::Xapian::DB_CREATE_OR_OPEN;
41 Search::Xapian::WritableDatabase->new($dir, $flag);
47 my ($doc, $col, $num) = @_;
48 $num = Search::Xapian::sortable_serialise($num);
49 $doc->add_value($col, $num);
53 my ($self, $mime, $bytes, $num) = @_; # mime = Email::MIME object
54 my $db = $self->{xdb};
57 my $mid = mid_clean($mime->header('Message-ID'));
59 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
62 die 'Message-ID too long' if length($mid) > MAX_MID_SIZE;
63 my $smsg = $self->lookup_message($mid);
67 $smsg->ensure_metadata;
68 # convert a ghost to a regular message
69 # it will also clobber any existing regular message
73 my $type = xpfx('type');
75 $doc->remove_term($type . 'ghost');
79 # probably does not exist:
80 eval { $doc->remove_term($type . 'mail') };
81 $doc->add_term($type . 'mail');
83 $smsg = PublicInbox::SearchMsg->new($mime);
85 $doc->add_term(xpfx('mid') . $mid);
88 my $subj = $smsg->subject;
91 my $path = $self->subject_path($subj);
92 $doc->add_term(xpfx('path') . id_compress($path));
95 add_val($doc, &PublicInbox::Search::TS, $smsg->ts);
98 add_val($doc, &PublicInbox::Search::NUM, $num);
101 add_val($doc, &PublicInbox::Search::BYTES, $bytes);
103 add_val($doc, &PublicInbox::Search::LINES,
104 $mime->body_raw =~ tr!\n!\n!);
106 my $tg = $self->term_generator;
108 $tg->set_document($doc);
109 $tg->index_text($subj, 1, 'S') if $subj;
110 $tg->increase_termpos;
111 $tg->index_text($subj) if $subj;
112 $tg->increase_termpos;
114 $tg->index_text($smsg->from);
115 $tg->increase_termpos;
117 $mime->walk_parts(sub {
119 return if $part->subparts; # walk_parts already recurses
120 my $ct = $part->content_type || $ct_msg;
122 # account for filter bugs...
123 $ct =~ m!\btext/plain\b!i or return;
126 my $body = $part->body;
128 my @lines = split(/\n/, $body);
129 while (defined(my $l = shift @lines)) {
137 $tg->index_text(join("\n", @quot), 0);
139 $tg->increase_termpos;
142 $tg->index_text(join("\n", @orig));
144 $tg->increase_termpos;
149 $doc_id = $smsg->doc_id;
150 $self->link_message($smsg, 0);
151 $doc->set_data($smsg->to_doc_data);
152 $db->replace_document($doc_id, $doc);
154 $self->link_message($smsg, 0);
155 $doc->set_data($smsg->to_doc_data);
156 $doc_id = $db->add_document($doc);
161 warn "failed to index message <$mid>: $@\n";
167 # returns deleted doc_id on success, undef on missing
169 my ($self, $mid) = @_;
170 my $db = $self->{xdb};
172 $mid = mid_clean($mid);
175 $doc_id = $self->find_unique_doc_id('mid', $mid);
176 $db->delete_document($doc_id) if defined $doc_id;
180 warn "failed to remove message <$mid>: $@\n";
186 sub term_generator { # write-only
189 my $tg = $self->{term_generator};
192 $tg = Search::Xapian::TermGenerator->new;
193 $tg->set_stemmer($self->stemmer);
195 $self->{term_generator} = $tg;
198 # increments last_thread_id counter
199 # returns a 64-bit integer represented as a hex string
202 my $db = $self->{xdb};
203 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
205 $db->set_metadata('last_thread_id', ++$last_thread_id);
211 my ($self, $smsg, $is_ghost) = @_;
214 $smsg->ensure_metadata;
216 $self->link_message_to_parents($smsg);
220 sub link_message_to_parents {
221 my ($self, $smsg) = @_;
222 my $doc = $smsg->{doc};
223 my $mid = $smsg->mid;
224 my $mime = $smsg->mime;
225 my $refs = $mime->header('References');
226 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
227 if (my $irt = $mime->header('In-Reply-To')) {
228 # last References should be $irt
229 # we will de-dupe later
230 push @refs, mid_clean($irt);
235 my %uniq = ($mid => 1);
236 my @orig_refs = @refs;
239 # prevent circular references via References: here:
240 foreach my $ref (@orig_refs) {
241 if (length($ref) > MAX_MID_SIZE) {
242 warn "References: <$ref> too long, ignoring\n";
250 $smsg->{references} = '<'.join('> <', @refs).'>';
252 # first ref *should* be the thread root,
253 # but we can never trust clients to do the right thing
254 my $ref = shift @refs;
255 $tid = $self->_resolve_mid_to_tid($ref);
257 # the rest of the refs should point to this tid:
258 foreach $ref (@refs) {
259 my $ptid = $self->_resolve_mid_to_tid($ref);
261 $self->merge_threads($tid, $ptid);
265 $tid = $self->next_thread_id;
267 $doc->add_term(xpfx('thread') . $tid);
271 my ($self, $git, $mime, $bytes, $num) = @_;
272 $self->add_message($mime, $bytes, $num);
276 my ($self, $git, $mime) = @_;
277 my $mid = mid_clean($mime->header('Message-ID'));
278 $self->remove_message($mid) if defined $mid;
282 my ($self, $git, $mime) = @_;
283 $self->{mm}->mid_insert(mid_clean($mime->header('Message-ID')));
287 my ($self, $git, $mime) = @_;
288 $self->{mm}->mid_delete(mid_clean($mime->header('Message-ID')));
292 my ($self, $git, $mime, $bytes) = @_;
293 my $num = $self->{mm}->num_for(mid_clean($mime->header('Message-ID')));
294 index_blob($self, $git, $mime, $bytes, $num);
298 my ($self, $git, $mime) = @_;
299 $self->{mm}->mid_delete(mid_clean($mime->header('Message-ID')));
300 unindex_blob($self, $git, $mime);
304 my ($self, $git, $mime, $bytes) = @_;
305 my $num = index_mm($self, $git, $mime);
306 index_blob($self, $git, $mime, $bytes, $num);
310 my ($self, $git, $mime) = @_;
311 unindex_blob($self, $git, $mime);
312 unindex_mm($self, $git, $mime);
316 my ($git, $blob, $sizeref) = @_;
318 my $str = $git->cat_file($blob, $sizeref);
319 Email::MIME->new($str);
325 my ($self, $head) = @_;
326 $self->with_umask(sub { $self->_index_sync($head) });
330 my ($self, $range, $add_cb, $del_cb) = @_;
331 my $hex = '[a-f0-9]';
332 my $h40 = $hex .'{40}';
333 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
334 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
335 my $git = PublicInbox::Git->new($self->{git_dir});
336 my $log = $git->popen(qw/log --reverse --no-notes --no-color
337 --raw -r --no-abbrev/, $range);
340 while (defined(my $line = <$log>)) {
341 if ($line =~ /$addmsg/o) {
342 my $mime = do_cat_mail($git, $1, \$bytes) or next;
343 $add_cb->($self, $git, $mime, $bytes);
344 } elsif ($line =~ /$delmsg/o) {
345 my $mime = do_cat_mail($git, $1) or next;
346 $del_cb->($self, $git, $mime);
347 } elsif ($line =~ /^commit ($h40)/o) {
355 # indexes all unindexed messages
357 my ($self, $head) = @_;
358 my $db = $self->{xdb};
360 my $mm = $self->{mm} = eval {
361 require PublicInbox::Msgmap;
362 PublicInbox::Msgmap->new($self->{git_dir}, 1);
365 $db->begin_transaction;
366 my $lx = $db->get_metadata('last_commit');
367 my $range = $lx eq '' ? $head : "$lx..$head";
369 $mm->{dbh}->begin_work;
370 my $lm = $mm->last_commit || '';
372 # Common case is the indexes are synced,
373 # we only need to run git-log once:
374 $lx = $self->rlog($range, *index_both, *unindex_both);
377 $db->set_metadata('last_commit', $lx);
378 $mm->last_commit($lx);
381 # dumb case, msgmap and xapian are out-of-sync
382 # do not care for performance:
383 my $r = $lm eq '' ? $head : "$lm..$head";
384 $lm = $self->rlog($r, *index_mm, *unindex_mm);
386 $mm->last_commit($lm) if defined $lm;
388 $lx = $self->rlog($range, *index_mm2, *unindex_mm2);
389 $db->set_metadata('last_commit', $lx) if defined $lx;
392 # user didn't install DBD::SQLite and DBI
393 $lx = $self->rlog($range, *index_blob, *unindex_blob);
394 $db->set_metadata('last_commit', $lx) if defined $lx;
397 $db->cancel_transaction;
398 $mm->{dbh}->rollback if $mm;
400 $db->commit_transaction;
404 # this will create a ghost as necessary
405 sub _resolve_mid_to_tid {
406 my ($self, $mid) = @_;
408 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
413 my ($self, $mid, $tid) = @_;
415 $tid = $self->next_thread_id unless defined $tid;
417 my $doc = Search::Xapian::Document->new;
418 $doc->add_term(xpfx('mid') . $mid);
419 $doc->add_term(xpfx('thread') . $tid);
420 $doc->add_term(xpfx('type') . 'ghost');
422 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
423 $self->link_message($smsg, 1);
424 $self->{xdb}->add_document($doc);
430 my ($self, $winner_tid, $loser_tid) = @_;
431 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
432 my $thread_pfx = xpfx('thread');
433 my $db = $self->{xdb};
435 for (; $head != $tail; $head->inc) {
436 my $docid = $head->get_docid;
437 my $doc = $db->get_document($docid);
438 $doc->remove_term($thread_pfx . $loser_tid);
439 $doc->add_term($thread_pfx . $winner_tid);
440 $db->replace_document($docid, $doc);
444 sub _read_git_config_perm {
446 my @cmd = qw(config core.sharedRepository);
447 my $fh = PublicInbox::Git->new($self->{git_dir})->popen(@cmd);
450 chomp $perm if defined $perm;
454 sub _git_config_perm {
456 my $perm = scalar @_ ? $_[0] : _read_git_config_perm($self);
457 return PERM_GROUP if (!defined($perm) || $perm eq '');
458 return PERM_UMASK if ($perm eq 'umask');
459 return PERM_GROUP if ($perm eq 'group');
460 if ($perm =~ /\A(?:all|world|everybody)\z/) {
461 return PERM_EVERYBODY;
463 return PERM_GROUP if ($perm =~ /\A(?:true|yes|on|1)\z/);
464 return PERM_UMASK if ($perm =~ /\A(?:false|no|off|0)\z/);
467 return PERM_UMASK if ($i == PERM_UMASK);
468 return PERM_GROUP if ($i == OLD_PERM_GROUP);
469 return PERM_EVERYBODY if ($i == OLD_PERM_EVERYBODY);
471 if (($i & 0600) != 0600) {
472 die "core.sharedRepository mode invalid: ".
473 sprintf('%.3o', $i) . "\nOwner must have permissions\n";
479 my ($perm) = @_; # _git_config_perm return value
481 return umask if $rv == 0;
483 # set +x bit if +r or +w were set
484 $rv |= 0100 if ($rv & 0600);
485 $rv |= 0010 if ($rv & 0060);
486 $rv |= 0001 if ($rv & 0006);
491 my ($self, $cb) = @_;
492 my $old = umask $self->{umask};
493 my $rv = eval { $cb->() };