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 mid_mime/;
14 use PublicInbox::MsgIter;
15 require PublicInbox::Git;
16 *xpfx = *PublicInbox::Search::xpfx;
18 use constant MAX_MID_SIZE => 244; # max term size - 1 in Xapian
22 OLD_PERM_EVERYBODY => 2,
24 PERM_EVERYBODY => 0664,
28 my ($class, $git_dir, $writable) = @_;
29 my $dir = $class->xdir($git_dir);
30 require Search::Xapian::WritableDatabase;
31 my $flag = Search::Xapian::DB_OPEN;
32 my $self = bless { git_dir => $git_dir }, $class;
33 my $perm = $self->_git_config_perm;
34 my $umask = _umask_for($perm);
35 $self->{umask} = $umask;
36 $self->{xdb} = $self->with_umask(sub {
39 File::Path::mkpath($dir);
40 $flag = Search::Xapian::DB_CREATE_OR_OPEN;
42 Search::Xapian::WritableDatabase->new($dir, $flag);
48 my ($doc, $col, $num) = @_;
49 $num = Search::Xapian::sortable_serialise($num);
50 $doc->add_value($col, $num);
54 my ($self, $mime, $bytes, $num) = @_; # mime = Email::MIME object
55 my $db = $self->{xdb};
58 my $mid = mid_clean(mid_mime($mime));
60 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
63 die 'Message-ID too long' if length($mid) > MAX_MID_SIZE;
64 my $smsg = $self->lookup_message($mid);
68 $smsg->ensure_metadata;
69 # convert a ghost to a regular message
70 # it will also clobber any existing regular message
74 my $type = xpfx('type');
76 $doc->remove_term($type . 'ghost');
80 # probably does not exist:
81 eval { $doc->remove_term($type . 'mail') };
82 $doc->add_term($type . 'mail');
84 $smsg = PublicInbox::SearchMsg->new($mime);
86 $doc->add_term(xpfx('mid') . $mid);
89 my $subj = $smsg->subject;
92 my $path = $self->subject_path($subj);
93 $doc->add_term(xpfx('path') . id_compress($path));
96 add_val($doc, &PublicInbox::Search::TS, $smsg->ts);
99 add_val($doc, &PublicInbox::Search::NUM, $num);
102 add_val($doc, &PublicInbox::Search::BYTES, $bytes);
104 add_val($doc, &PublicInbox::Search::LINES,
105 $mime->body_raw =~ tr!\n!\n!);
107 my $tg = $self->term_generator;
109 $tg->set_document($doc);
110 $tg->index_text($subj, 1, 'S') if $subj;
111 $tg->increase_termpos;
112 $tg->index_text($subj) if $subj;
113 $tg->increase_termpos;
115 $tg->index_text($smsg->from);
116 $tg->increase_termpos;
118 msg_iter($mime, sub {
119 my ($part, $depth, @idx) = @{$_[0]};
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 $hdr = $mime->header_obj;
226 my $refs = $hdr->header_raw('References');
227 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
228 if (my $irt = $hdr->header_raw('In-Reply-To')) {
229 # last References should be $irt
230 # we will de-dupe later
231 push @refs, mid_clean($irt);
236 my %uniq = ($mid => 1);
237 my @orig_refs = @refs;
240 # prevent circular references via References: here:
241 foreach my $ref (@orig_refs) {
242 if (length($ref) > MAX_MID_SIZE) {
243 warn "References: <$ref> too long, ignoring\n";
251 $smsg->{references} = '<'.join('> <', @refs).'>';
253 # first ref *should* be the thread root,
254 # but we can never trust clients to do the right thing
255 my $ref = shift @refs;
256 $tid = $self->_resolve_mid_to_tid($ref);
258 # the rest of the refs should point to this tid:
259 foreach $ref (@refs) {
260 my $ptid = $self->_resolve_mid_to_tid($ref);
262 $self->merge_threads($tid, $ptid);
266 $tid = $self->next_thread_id;
268 $doc->add_term(xpfx('thread') . $tid);
272 my ($self, $git, $mime, $bytes, $num) = @_;
273 $self->add_message($mime, $bytes, $num);
277 my ($self, $git, $mime) = @_;
278 my $mid = eval { mid_clean(mid_mime($mime)) };
279 $self->remove_message($mid) if defined $mid;
283 my ($self, $git, $mime) = @_;
284 $self->{mm}->mid_insert(mid_clean(mid_mime($mime)));
288 my ($self, $git, $mime) = @_;
289 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
293 my ($self, $git, $mime, $bytes) = @_;
294 my $num = $self->{mm}->num_for(mid_clean(mid_mime($mime)));
295 index_blob($self, $git, $mime, $bytes, $num);
299 my ($self, $git, $mime) = @_;
300 $self->{mm}->mid_delete(mid_clean(mid_mime($mime)));
301 unindex_blob($self, $git, $mime);
305 my ($self, $git, $mime, $bytes) = @_;
306 my $num = index_mm($self, $git, $mime);
307 index_blob($self, $git, $mime, $bytes, $num);
311 my ($self, $git, $mime) = @_;
312 unindex_blob($self, $git, $mime);
313 unindex_mm($self, $git, $mime);
317 my ($git, $blob, $sizeref) = @_;
319 my $str = $git->cat_file($blob, $sizeref);
320 Email::MIME->new($str);
326 my ($self, $head) = @_;
327 $self->with_umask(sub { $self->_index_sync($head) });
331 my ($self, $range, $add_cb, $del_cb) = @_;
332 my $hex = '[a-f0-9]';
333 my $h40 = $hex .'{40}';
334 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
335 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
336 my $git = PublicInbox::Git->new($self->{git_dir});
337 my $log = $git->popen(qw/log --reverse --no-notes --no-color
338 --raw -r --no-abbrev/, $range);
342 while (defined(my $line = <$log>)) {
343 if ($line =~ /$addmsg/o) {
344 my $mime = do_cat_mail($git, $1, \$bytes) or next;
345 $add_cb->($self, $git, $mime, $bytes);
346 } elsif ($line =~ /$delmsg/o) {
347 my $mime = do_cat_mail($git, $1) or next;
348 $del_cb->($self, $git, $mime);
349 } elsif ($line =~ /^commit ($h40)/o) {
356 # indexes all unindexed messages
358 my ($self, $head) = @_;
359 my $db = $self->{xdb};
361 my $mm = $self->{mm} = eval {
362 require PublicInbox::Msgmap;
363 PublicInbox::Msgmap->new($self->{git_dir}, 1);
366 $db->begin_transaction;
367 my $lx = $db->get_metadata('last_commit');
368 my $range = $lx eq '' ? $head : "$lx..$head";
370 $mm->{dbh}->begin_work;
371 my $lm = $mm->last_commit || '';
373 # Common case is the indexes are synced,
374 # we only need to run git-log once:
375 $lx = $self->rlog($range, *index_both, *unindex_both);
378 $db->set_metadata('last_commit', $lx);
379 $mm->last_commit($lx);
382 # dumb case, msgmap and xapian are out-of-sync
383 # do not care for performance:
384 my $r = $lm eq '' ? $head : "$lm..$head";
385 $lm = $self->rlog($r, *index_mm, *unindex_mm);
387 $mm->last_commit($lm) if defined $lm;
389 $lx = $self->rlog($range, *index_mm2, *unindex_mm2);
390 $db->set_metadata('last_commit', $lx) if defined $lx;
393 # user didn't install DBD::SQLite and DBI
394 $lx = $self->rlog($range, *index_blob, *unindex_blob);
395 $db->set_metadata('last_commit', $lx) if defined $lx;
398 $db->cancel_transaction;
399 $mm->{dbh}->rollback if $mm;
401 $db->commit_transaction;
405 # this will create a ghost as necessary
406 sub _resolve_mid_to_tid {
407 my ($self, $mid) = @_;
409 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
414 my ($self, $mid, $tid) = @_;
416 $tid = $self->next_thread_id unless defined $tid;
418 my $doc = Search::Xapian::Document->new;
419 $doc->add_term(xpfx('mid') . $mid);
420 $doc->add_term(xpfx('thread') . $tid);
421 $doc->add_term(xpfx('type') . 'ghost');
423 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
424 $self->link_message($smsg, 1);
425 $self->{xdb}->add_document($doc);
431 my ($self, $winner_tid, $loser_tid) = @_;
432 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
433 my $thread_pfx = xpfx('thread');
434 my $db = $self->{xdb};
436 for (; $head != $tail; $head->inc) {
437 my $docid = $head->get_docid;
438 my $doc = $db->get_document($docid);
439 $doc->remove_term($thread_pfx . $loser_tid);
440 $doc->add_term($thread_pfx . $winner_tid);
441 $db->replace_document($docid, $doc);
445 sub _read_git_config_perm {
447 my @cmd = qw(config core.sharedRepository);
448 my $fh = PublicInbox::Git->new($self->{git_dir})->popen(@cmd);
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->() };