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;
13 use Email::MIME::ContentType;
14 $Email::MIME::ContentType::STRICT_PARAMS = 0;
15 use base qw(PublicInbox::Search);
16 use PublicInbox::MID qw/mid_clean id_compress mid_mime/;
17 use PublicInbox::MsgIter;
18 require PublicInbox::Git;
19 *xpfx = *PublicInbox::Search::xpfx;
21 use constant MAX_MID_SIZE => 244; # max term size - 1 in Xapian
25 OLD_PERM_EVERYBODY => 2,
27 PERM_EVERYBODY => 0664,
31 my ($class, $git_dir, $writable) = @_;
32 my $dir = $class->xdir($git_dir);
33 require Search::Xapian::WritableDatabase;
34 my $flag = Search::Xapian::DB_OPEN;
35 my $self = bless { git_dir => $git_dir }, $class;
36 my $perm = $self->_git_config_perm;
37 my $umask = _umask_for($perm);
38 $self->{umask} = $umask;
39 $self->{xdb} = $self->with_umask(sub {
42 File::Path::mkpath($dir);
43 $flag = Search::Xapian::DB_CREATE_OR_OPEN;
45 Search::Xapian::WritableDatabase->new($dir, $flag);
51 my ($doc, $col, $num) = @_;
52 $num = Search::Xapian::sortable_serialise($num);
53 $doc->add_value($col, $num);
57 my ($self, $mime, $bytes, $num) = @_; # mime = Email::MIME object
58 my $db = $self->{xdb};
61 my $mid = mid_clean(mid_mime($mime));
63 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
66 die 'Message-ID too long' if length($mid) > MAX_MID_SIZE;
67 my $smsg = $self->lookup_message($mid);
71 $smsg->ensure_metadata;
72 # convert a ghost to a regular message
73 # it will also clobber any existing regular message
77 my $type = xpfx('type');
79 $doc->remove_term($type . 'ghost');
83 # probably does not exist:
84 eval { $doc->remove_term($type . 'mail') };
85 $doc->add_term($type . 'mail');
87 $smsg = PublicInbox::SearchMsg->new($mime);
89 $doc->add_term(xpfx('mid') . $mid);
92 my $subj = $smsg->subject;
95 my $path = $self->subject_path($subj);
96 $doc->add_term(xpfx('path') . id_compress($path));
99 add_val($doc, &PublicInbox::Search::TS, $smsg->ts);
102 add_val($doc, &PublicInbox::Search::NUM, $num);
105 add_val($doc, &PublicInbox::Search::BYTES, $bytes);
107 add_val($doc, &PublicInbox::Search::LINES,
108 $mime->body_raw =~ tr!\n!\n!);
110 my $tg = $self->term_generator;
112 $tg->set_document($doc);
113 $tg->index_text($subj, 1, 'S') if $subj;
114 $tg->increase_termpos;
115 $tg->index_text($subj) if $subj;
116 $tg->increase_termpos;
118 $tg->index_text($smsg->from);
119 $tg->increase_termpos;
121 msg_iter($mime, sub {
122 my ($part, $depth, @idx) = @{$_[0]};
123 my $ct = $part->content_type || $ct_msg;
125 # account for filter bugs...
126 $ct =~ m!\btext/plain\b!i or return;
129 my $body = $part->body;
131 my @lines = split(/\n/, $body);
132 while (defined(my $l = shift @lines)) {
140 $tg->index_text(join("\n", @quot), 0);
142 $tg->increase_termpos;
145 $tg->index_text(join("\n", @orig));
147 $tg->increase_termpos;
152 $doc_id = $smsg->doc_id;
153 $self->link_message($smsg, $smsg->thread_id);
154 $doc->set_data($smsg->to_doc_data);
155 $db->replace_document($doc_id, $doc);
157 $self->link_message($smsg);
158 $doc->set_data($smsg->to_doc_data);
159 $doc_id = $db->add_document($doc);
164 warn "failed to index message <$mid>: $@\n";
170 # returns deleted doc_id on success, undef on missing
172 my ($self, $mid) = @_;
173 my $db = $self->{xdb};
175 $mid = mid_clean($mid);
178 $doc_id = $self->find_unique_doc_id('mid', $mid);
179 $db->delete_document($doc_id) if defined $doc_id;
183 warn "failed to remove message <$mid>: $@\n";
189 sub term_generator { # write-only
192 my $tg = $self->{term_generator};
195 $tg = Search::Xapian::TermGenerator->new;
196 $tg->set_stemmer($self->stemmer);
198 $self->{term_generator} = $tg;
201 # increments last_thread_id counter
202 # returns a 64-bit integer represented as a hex string
205 my $db = $self->{xdb};
206 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
208 $db->set_metadata('last_thread_id', ++$last_thread_id);
214 my ($self, $smsg, $old_tid) = @_;
215 my $doc = $smsg->{doc};
216 my $mid = $smsg->mid;
217 my $mime = $smsg->mime;
218 my $hdr = $mime->header_obj;
219 my $refs = $hdr->header_raw('References');
220 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
221 if (my $irt = $hdr->header_raw('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);
250 $self->merge_threads($tid, $old_tid) if defined $old_tid;
252 # the rest of the refs should point to this tid:
253 foreach $ref (@refs) {
254 my $ptid = $self->_resolve_mid_to_tid($ref);
256 $self->merge_threads($tid, $ptid);
260 $tid = $self->next_thread_id;
262 $doc->add_term(xpfx('thread') . $tid);
266 my ($self, $git, $mime, $bytes, $num) = @_;
267 $self->add_message($mime, $bytes, $num);
271 my ($self, $git, $mime) = @_;
272 my $mid = eval { mid_clean(mid_mime($mime)) };
273 $self->remove_message($mid) if defined $mid;
277 my ($self, $git, $mime) = @_;
278 $self->{mm}->mid_insert(mid_clean(mid_mime($mime)));
282 my ($self, $git, $mime) = @_;
283 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
287 my ($self, $git, $mime, $bytes) = @_;
288 my $num = $self->{mm}->num_for(mid_clean(mid_mime($mime)));
289 index_blob($self, $git, $mime, $bytes, $num);
293 my ($self, $git, $mime) = @_;
294 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
295 unindex_blob($self, $git, $mime);
299 my ($self, $git, $mime, $bytes) = @_;
300 my $num = index_mm($self, $git, $mime);
301 index_blob($self, $git, $mime, $bytes, $num);
305 my ($self, $git, $mime) = @_;
306 unindex_blob($self, $git, $mime);
307 unindex_mm($self, $git, $mime);
311 my ($git, $blob, $sizeref) = @_;
313 my $str = $git->cat_file($blob, $sizeref);
314 Email::MIME->new($str);
320 my ($self, $head) = @_;
321 $self->with_umask(sub { $self->_index_sync($head) });
325 my ($self, $range, $add_cb, $del_cb) = @_;
326 my $hex = '[a-f0-9]';
327 my $h40 = $hex .'{40}';
328 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
329 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
330 my $git = PublicInbox::Git->new($self->{git_dir});
331 my $log = $git->popen(qw/log --reverse --no-notes --no-color
332 --raw -r --no-abbrev/, $range);
336 while (defined(my $line = <$log>)) {
337 if ($line =~ /$addmsg/o) {
338 my $mime = do_cat_mail($git, $1, \$bytes) or next;
339 $add_cb->($self, $git, $mime, $bytes);
340 } elsif ($line =~ /$delmsg/o) {
341 my $mime = do_cat_mail($git, $1) or next;
342 $del_cb->($self, $git, $mime);
343 } elsif ($line =~ /^commit ($h40)/o) {
350 # indexes all unindexed messages
352 my ($self, $head) = @_;
353 my $db = $self->{xdb};
355 my $mm = $self->{mm} = eval {
356 require PublicInbox::Msgmap;
357 PublicInbox::Msgmap->new($self->{git_dir}, 1);
360 $db->begin_transaction;
361 my $lx = $db->get_metadata('last_commit');
362 my $range = $lx eq '' ? $head : "$lx..$head";
364 $mm->{dbh}->begin_work;
365 my $lm = $mm->last_commit || '';
367 # Common case is the indexes are synced,
368 # we only need to run git-log once:
369 $lx = $self->rlog($range, *index_both, *unindex_both);
372 $db->set_metadata('last_commit', $lx);
373 $mm->last_commit($lx);
376 # dumb case, msgmap and xapian are out-of-sync
377 # do not care for performance:
378 my $r = $lm eq '' ? $head : "$lm..$head";
379 $lm = $self->rlog($r, *index_mm, *unindex_mm);
381 $mm->last_commit($lm) if defined $lm;
383 $lx = $self->rlog($range, *index_mm2, *unindex_mm2);
384 $db->set_metadata('last_commit', $lx) if defined $lx;
387 # user didn't install DBD::SQLite and DBI
388 $lx = $self->rlog($range, *index_blob, *unindex_blob);
389 $db->set_metadata('last_commit', $lx) if defined $lx;
392 $db->cancel_transaction;
393 $mm->{dbh}->rollback if $mm;
395 $db->commit_transaction;
399 # this will create a ghost as necessary
400 sub _resolve_mid_to_tid {
401 my ($self, $mid) = @_;
403 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
408 my ($self, $mid) = @_;
410 my $tid = $self->next_thread_id;
411 my $doc = Search::Xapian::Document->new;
412 $doc->add_term(xpfx('mid') . $mid);
413 $doc->add_term(xpfx('thread') . $tid);
414 $doc->add_term(xpfx('type') . 'ghost');
416 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
417 $self->{xdb}->add_document($doc);
423 my ($self, $winner_tid, $loser_tid) = @_;
424 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
425 my $thread_pfx = xpfx('thread');
426 my $db = $self->{xdb};
428 for (; $head != $tail; $head->inc) {
429 my $docid = $head->get_docid;
430 my $doc = $db->get_document($docid);
431 $doc->remove_term($thread_pfx . $loser_tid);
432 $doc->add_term($thread_pfx . $winner_tid);
433 $db->replace_document($docid, $doc);
437 sub _read_git_config_perm {
439 my @cmd = qw(config core.sharedRepository);
440 my $fh = PublicInbox::Git->new($self->{git_dir})->popen(@cmd);
443 chomp $perm if defined $perm;
447 sub _git_config_perm {
449 my $perm = scalar @_ ? $_[0] : _read_git_config_perm($self);
450 return PERM_GROUP if (!defined($perm) || $perm eq '');
451 return PERM_UMASK if ($perm eq 'umask');
452 return PERM_GROUP if ($perm eq 'group');
453 if ($perm =~ /\A(?:all|world|everybody)\z/) {
454 return PERM_EVERYBODY;
456 return PERM_GROUP if ($perm =~ /\A(?:true|yes|on|1)\z/);
457 return PERM_UMASK if ($perm =~ /\A(?:false|no|off|0)\z/);
460 return PERM_UMASK if ($i == PERM_UMASK);
461 return PERM_GROUP if ($i == OLD_PERM_GROUP);
462 return PERM_EVERYBODY if ($i == OLD_PERM_EVERYBODY);
464 if (($i & 0600) != 0600) {
465 die "core.sharedRepository mode invalid: ".
466 sprintf('%.3o', $i) . "\nOwner must have permissions\n";
472 my ($perm) = @_; # _git_config_perm return value
474 return umask if $rv == 0;
476 # set +x bit if +r or +w were set
477 $rv |= 0100 if ($rv & 0600);
478 $rv |= 0010 if ($rv & 0060);
479 $rv |= 0001 if ($rv & 0006);
484 my ($self, $cb) = @_;
485 my $old = umask $self->{umask};
486 my $rv = eval { $cb->() };