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::Search;
7 use PublicInbox::SearchMsg;
8 use Search::Xapian qw/:standard/;
9 require PublicInbox::View;
11 use PublicInbox::MID qw/mid_clean mid_compressed/;
15 # SCHEMA_VERSION history
17 # 1 - subject_path is lower-cased
18 # 2 - subject_path is mid_compressed in the index, only
21 QP_FLAGS => FLAG_PHRASE|FLAG_BOOLEAN|FLAG_LOVEHATE|FLAG_WILDCARD,
25 my %bool_pfx_internal = (
26 type => 'T', # "mail" or "ghost"
27 mid => 'Q', # uniQue id (Message-ID or mid_compressed)
30 my %bool_pfx_external = (
32 thread => 'G', # newsGroup (or similar entity - e.g. a web forum name)
33 references => 'XREFS',
41 my %all_pfx = (%bool_pfx_internal, %bool_pfx_external, %prob_prefix);
43 sub xpfx { $all_pfx{$_[0]} }
46 while (my ($k, $v) = each %all_pfx) {
47 next if $prob_prefix{$k};
48 $PFX2TERM_RMAP{$v} = $k;
51 my $mail_query = Search::Xapian::Query->new(xpfx('type') . 'mail');
54 my ($class, $git_dir, $writable) = @_;
55 # allow concurrent versions for easier rollback:
56 my $dir = "$git_dir/public-inbox/xapian" . SCHEMA_VERSION;
59 if ($writable) { # not used by the WWW interface
60 require Search::Xapian::WritableDatabase;
62 File::Path::mkpath($dir);
63 $db = Search::Xapian::WritableDatabase->new($dir,
64 Search::Xapian::DB_CREATE_OR_OPEN);
66 $db = Search::Xapian::Database->new($dir);
68 bless { xdb => $db, git_dir => $git_dir }, $class;
71 sub reopen { $_[0]->{xdb}->reopen }
74 my ($self, $mime) = @_; # mime = Email::MIME object
75 my $db = $self->{xdb};
78 my $mid = mid_clean($mime->header_obj->header_raw('Message-ID'));
79 $mid = mid_compressed($mid);
81 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
82 my $enc_msg = PublicInbox::View::enc_for($ct_msg);
84 $db->begin_transaction;
86 my $smsg = $self->lookup_message($mid);
90 $smsg->ensure_metadata;
91 # convert a ghost to a regular message
92 # it will also clobber any existing regular message
96 my $type = xpfx('type');
98 $doc->remove_term($type . 'ghost');
102 # probably does not exist:
103 eval { $doc->remove_term($type . 'mail') };
104 $doc->add_term($type . 'mail');
106 $smsg = PublicInbox::SearchMsg->new($mime);
108 $doc->add_term(xpfx('mid') . $mid);
111 my $subj = $smsg->subject;
114 $doc->add_term(xpfx('subject') . $subj);
116 my $path = subject_path($subj);
117 $doc->add_term(xpfx('path') . mid_compressed($path));
120 my $from = $smsg->from_name;
121 my $date = $smsg->date;
122 my $ts = Search::Xapian::sortable_serialise($smsg->ts);
123 $doc->add_value(PublicInbox::Search::TS, $ts);
125 $doc->set_data($smsg->to_doc_data);
127 my $tg = $self->term_generator;
129 $tg->set_document($doc);
130 $tg->index_text($subj, 1, 'S') if $subj;
131 $tg->increase_termpos;
132 $tg->index_text($subj) if $subj;
133 $tg->increase_termpos;
135 $tg->index_text($smsg->from->format);
136 $tg->increase_termpos;
138 $mime->walk_parts(sub {
140 return if $part->subparts; # walk_parts already recurses
141 my $ct = $part->content_type || $ct_msg;
143 # account for filter bugs...
144 $ct =~ m!\btext/plain\b!i or return;
146 my $enc = PublicInbox::View::enc_for($ct, $enc_msg);
148 foreach my $l (split(/\n/, $enc->decode($part->body))) {
156 $tg->index_text(join("\n", @quot), 0);
157 $tg->increase_termpos;
160 $tg->index_text(join("\n", @orig));
161 $tg->increase_termpos;
166 $doc_id = $smsg->doc_id;
167 $self->link_message($smsg, 0);
168 $db->replace_document($doc_id, $doc);
170 $self->link_message($smsg, 0);
171 $doc_id = $db->add_document($doc);
176 warn "failed to index message <$mid>: $@\n";
177 $db->cancel_transaction;
179 $db->commit_transaction;
184 # returns deleted doc_id on success, undef on missing
186 my ($self, $mid) = @_;
187 my $db = $self->{xdb};
189 $mid = mid_clean($mid);
190 $mid = mid_compressed($mid);
192 $db->begin_transaction;
194 $doc_id = $self->find_unique_doc_id('mid', $mid);
195 $db->delete_document($doc_id) if defined $doc_id;
199 warn "failed to remove message <$mid>: $@\n";
200 $db->cancel_transaction;
202 $db->commit_transaction;
209 my ($self, $query_string, $opts) = @_;
210 my $query = $self->qp->parse_query($query_string, QP_FLAGS);
212 $query = Search::Xapian::Query->new(OP_AND, $mail_query, $query);
213 $self->do_enquire($query, $opts);
216 sub get_subject_path {
217 my ($self, $path, $opts) = @_;
218 my $query = $self->qp->parse_query("path:".mid_compressed($path), 0);
219 $self->do_enquire($query);
222 # given a message ID, get followups to a message
224 my ($self, $mid, $opts) = @_;
225 $mid = mid_clean($mid);
226 $mid = mid_compressed($mid);
228 my $irt = $qp->parse_query("inreplyto:$mid", 0);
229 my $ref = $qp->parse_query("references:$mid", 0);
230 my $query = Search::Xapian::Query->new(OP_OR, $irt, $ref);
232 $self->do_enquire($query);
236 my ($self, $mid, $opts) = @_;
237 my $smsg = eval { $self->lookup_message($mid) };
239 return { count => 0, msgs => [] } unless $smsg;
241 my $qtid = $qp->parse_query('thread:'.$smsg->thread_id);
242 my $qsub = $qp->parse_query('path:'.mid_compressed($smsg->path));
243 my $query = Search::Xapian::Query->new(OP_OR, $qtid, $qsub);
244 $self->do_enquire($query);
250 my ($self, $query, $opts) = @_;
251 my $enquire = $self->enquire;
253 $enquire->set_query($query);
254 $enquire->set_sort_by_relevance_then_value(TS, 0);
256 my $offset = $opts->{offset} || 0;
257 my $limit = $opts->{limit} || 50;
258 my $mset = $enquire->get_mset($offset, $limit);
260 PublicInbox::SearchMsg->load_doc($_->get_document);
263 { count => $mset->get_matches_estimated, msgs => \@msgs }
267 sub stemmer { Search::Xapian::Stem->new(LANG) }
273 my $qp = $self->{query_parser};
277 $qp = Search::Xapian::QueryParser->new;
278 $qp->set_default_op(OP_AND);
279 $qp->set_database($self->{xdb});
280 $qp->set_stemmer($self->stemmer);
281 $qp->set_stemming_strategy(STEM_SOME);
282 $qp->add_valuerangeprocessor($self->ts_range_processor);
283 $qp->add_valuerangeprocessor($self->date_range_processor);
285 while (my ($name, $prefix) = each %bool_pfx_external) {
286 $qp->add_boolean_prefix($name, $prefix);
289 while (my ($name, $prefix) = each %prob_prefix) {
290 $qp->add_prefix($name, $prefix);
293 $self->{query_parser} = $qp;
296 sub term_generator { # write-only
299 my $tg = $self->{term_generator};
302 $tg = Search::Xapian::TermGenerator->new;
303 $tg->set_stemmer($self->stemmer);
305 $self->{term_generator} = $tg;
308 sub next_doc_id { $_[0]->{xdb}->get_lastdocid + 1 }
310 # increments last_thread_id counter
311 # returns a 64-bit integer represented as a hex string
314 my $db = $self->{xdb};
315 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
317 $db->set_metadata('last_thread_id', ++$last_thread_id);
322 sub ts_range_processor {
323 $_[0]->{tsrp} ||= Search::Xapian::NumberValueRangeProcessor->new(TS);
326 sub date_range_processor {
327 $_[0]->{drp} ||= Search::Xapian::DateValueRangeProcessor->new(TS);
331 my ($self, $smsg, $is_ghost) = @_;
334 $smsg->ensure_metadata;
336 $self->link_message_to_parents($smsg);
340 sub link_message_to_parents {
341 my ($self, $smsg) = @_;
342 my $doc = $smsg->{doc};
343 my $mid = mid_compressed($smsg->mid);
344 my $mime = $smsg->mime;
345 my $refs = $mime->header_obj->header_raw('References');
346 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
347 my $irt = $mime->header_obj->header_raw('In-Reply-To');
349 if ($irt =~ /<([^>]+)>/) {
357 @refs = map { mid_compressed($_) } @refs;
359 @refs = grep { !$uniq{$_}++ } @refs; # uniq
361 $doc->add_term(xpfx('inreplyto') . $refs[-1]);
363 my $ref_pfx = xpfx('references');
365 # first ref *should* be the thread root,
366 # but we can never trust clients to do the right thing
367 my $ref = shift @refs;
368 $doc->add_term($ref_pfx . $ref);
369 $tid = $self->_resolve_mid_to_tid($ref);
371 # the rest of the refs should point to this tid:
372 foreach $ref (@refs) {
373 $doc->add_term($ref_pfx . $ref);
374 my $ptid = $self->_resolve_mid_to_tid($ref);
376 $self->merge_threads($tid, $ptid);
380 $tid = $self->next_thread_id;
382 $doc->add_term(xpfx('thread') . $tid);
386 my ($self, $mid) = @_;
387 $mid = mid_clean($mid);
388 $mid = mid_compressed($mid);
390 my $doc_id = $self->find_unique_doc_id('mid', $mid);
392 if (defined $doc_id) {
394 my $doc = $self->{xdb}->get_document($doc_id);
395 $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
396 $smsg->doc_id($doc_id);
401 sub find_unique_doc_id {
402 my ($self, $term, $value) = @_;
404 my ($begin, $end) = $self->find_doc_ids($term, $value);
406 return undef if $begin->equal($end); # not found
408 my $rv = $begin->get_docid;
412 $begin->equal($end) or die "Term '$term:$value' is not unique\n";
416 # returns begin and end PostingIterator
418 my ($self, $term, $value) = @_;
420 $self->find_doc_ids_for_term(xpfx($term) . $value);
423 # returns begin and end PostingIterator
424 sub find_doc_ids_for_term {
425 my ($self, $term) = @_;
426 my $db = $self->{xdb};
428 ($db->postlist_begin($term), $db->postlist_end($term));
431 # this will create a ghost as necessary
432 sub _resolve_mid_to_tid {
433 my ($self, $mid) = @_;
435 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
440 my ($self, $mid, $tid) = @_;
442 $mid = mid_compressed($mid);
443 $tid = $self->next_thread_id unless defined $tid;
445 my $doc = Search::Xapian::Document->new;
446 $doc->add_term(xpfx('mid') . $mid);
447 $doc->add_term(xpfx('thread') . $tid);
448 $doc->add_term(xpfx('type') . 'ghost');
450 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
451 $self->link_message($smsg, 1);
452 $self->{xdb}->add_document($doc);
458 my ($self, $winner_tid, $loser_tid) = @_;
459 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
460 my $thread_pfx = xpfx('thread');
461 my $db = $self->{xdb};
463 for (; $head != $tail; $head->inc) {
464 my $docid = $head->get_docid;
465 my $doc = $db->get_document($docid);
466 $doc->remove_term($thread_pfx . $loser_tid);
467 $doc->add_term($thread_pfx . $winner_tid);
468 $db->replace_document($docid, $doc);
472 # normalize subjects so they are suitable as pathnames for URLs
478 $subj =~ s/^(?:re|aw):\s*//i; # remove reply prefix (aw: German)
479 $subj =~ s![^a-zA-Z0-9_\.~/\-]+!_!g;
484 my ($git, $blob) = @_;
486 my $str = $git->cat_file($blob);
487 Email::MIME->new($str);
493 my ($self, $git, $blob) = @_;
494 my $mime = do_cat_mail($git, $blob) or return;
495 eval { $self->add_message($mime) };
496 warn "W: index_blob $blob: $@\n" if $@;
500 my ($self, $git, $blob) = @_;
501 my $mime = do_cat_mail($git, $blob) or return;
502 my $mid = $mime->header_obj->header_raw('Message-ID');
503 eval { $self->remove_message($mid) } if defined $mid;
504 warn "W: unindex_blob $blob: $@\n" if $@;
509 $self->{enquire} ||= Search::Xapian::Enquire->new($self->{xdb});
512 # indexes all unindexed messages
514 my ($self, $git) = @_;
515 my $db = $self->{xdb};
516 my $latest = $db->get_metadata('last_commit');
517 my $range = length $latest ? "$latest..HEAD" : 'HEAD';
520 my $hex = '[a-f0-9]';
521 my $h40 = $hex .'{40}';
522 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
523 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
525 # get indexed messages
526 my @cmd = ('git', "--git-dir=$git->{git_dir}", "log",
527 qw/--reverse --no-notes --no-color --raw -r --no-abbrev/,
530 my $pid = open(my $log, '-|', @cmd) or
531 die('open` '.join(' ', @cmd) . " pipe failed: $!\n");
533 while (my $line = <$log>) {
534 if ($line =~ /$addmsg/o) {
535 $self->index_blob($git, $1);
536 } elsif ($line =~ /$delmsg/o) {
537 $self->unindex_blob($git, $1);
538 } elsif ($line =~ /^commit ($h40)/o) {
540 if (defined $latest) {
541 $db->set_metadata('last_commit', $latest)
547 $db->set_metadata('last_commit', $latest) if defined $latest;