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;
9 use Search::Xapian qw/:standard/;
10 require PublicInbox::View;
11 use Date::Parse qw/str2time/;
14 use PublicInbox::MID qw/mid_clean mid_compressed/;
16 our @EXPORT = qw/xpfx/;
22 QP_FLAGS => FLAG_PHRASE|FLAG_BOOLEAN|FLAG_LOVEHATE|FLAG_WILDCARD,
26 my %bool_pfx_internal = (
27 type => 'T', # "mail" or "ghost"
28 mid => 'Q', # uniQue id (Message-ID or mid_compressed)
31 my %bool_pfx_external = (
33 thread => 'G', # newsGroup (or similar entity - e.g. a web forum name)
34 references => 'XREFS',
42 my %all_pfx = (%bool_pfx_internal, %bool_pfx_external, %prob_prefix);
44 sub xpfx { $all_pfx{$_[0]} }
47 while (my ($k, $v) = each %all_pfx) {
48 next if $prob_prefix{$k};
49 $PFX2TERM_RMAP{$v} = $k;
52 my $mail_query = Search::Xapian::Query->new(xpfx('type') . 'mail');
55 my ($class, $git_dir, $writable) = @_;
56 # allow concurrent versions for easier rollback:
57 my $dir = "$git_dir/public-inbox/xapian" . SCHEMA_VERSION;
60 if ($writable) { # not used by the WWW interface
61 require Search::Xapian::WritableDatabase;
63 File::Path::mkpath($dir);
64 $db = Search::Xapian::WritableDatabase->new($dir,
65 Search::Xapian::DB_CREATE_OR_OPEN);
67 $db = Search::Xapian::Database->new($dir);
69 bless { xdb => $db, git_dir => $git_dir }, $class;
72 sub reopen { $_[0]->{xdb}->reopen }
75 my ($self, $mime) = @_; # mime = Email::MIME object
76 my $db = $self->{xdb};
79 my $mid = mid_clean($mime->header('Message-ID'));
80 $mid = mid_compressed($mid);
82 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
83 my $enc_msg = PublicInbox::View::enc_for($ct_msg);
85 $db->begin_transaction;
87 my $smsg = $self->lookup_message($mid);
91 $smsg->ensure_metadata;
92 # convert a ghost to a regular message
93 # it will also clobber any existing regular message
97 my $type = xpfx('type');
99 $doc->remove_term($type . 'ghost');
103 # probably does not exist:
104 eval { $doc->remove_term($type . 'mail') };
105 $doc->add_term($type . 'mail');
107 $smsg = PublicInbox::SearchMsg->new($mime);
109 $doc->add_term(xpfx('mid') . $mid);
112 my $subj = $mime->header('Subject');
113 $subj = '' unless defined $subj;
116 $doc->add_term(xpfx('subject') . $subj);
118 my $path = subject_path($subj);
119 $doc->add_term(xpfx('path') . $path);
122 my $from = $mime->header('From') || '';
126 @from = Email::Address->parse($from);
127 $from = $from[0]->name;
130 my $ts = eval { str2time($mime->header('Date')) } || 0;
131 my $date = POSIX::strftime('%Y-%m-%d %H:%M', gmtime($ts));
132 $ts = Search::Xapian::sortable_serialise($ts);
133 $doc->add_value(PublicInbox::Search::TS, $ts);
135 # this is what we show in index results:
138 $doc->set_data("$mid\n$subj\n$from\n$date");
140 my $tg = $self->term_generator;
142 $tg->set_document($doc);
143 $tg->index_text($subj, 1, 'S') if $subj;
144 $tg->increase_termpos;
145 $tg->index_text($subj) if $subj;
146 $tg->increase_termpos;
149 $tg->index_text($from[0]->format);
150 $tg->increase_termpos;
153 $mime->walk_parts(sub {
155 return if $part->subparts; # walk_parts already recurses
156 my $ct = $part->content_type || $ct_msg;
158 # account for filter bugs...
159 $ct =~ m!\btext/plain\b!i or return;
161 my $enc = PublicInbox::View::enc_for($ct, $enc_msg);
163 foreach my $l (split(/\n/, $enc->decode($part->body))) {
171 $tg->index_text(join("\n", @quot), 0);
172 $tg->increase_termpos;
175 $tg->index_text(join("\n", @orig));
176 $tg->increase_termpos;
181 $doc_id = $smsg->doc_id;
182 $self->link_message($smsg, 0);
183 $db->replace_document($doc_id, $doc);
185 $self->link_message($smsg, 0);
186 $doc_id = $db->add_document($doc);
191 warn "failed to index message <$mid>: $@\n";
192 $db->cancel_transaction;
194 $db->commit_transaction;
199 # returns deleted doc_id on success, undef on missing
201 my ($self, $mid) = @_;
202 my $db = $self->{xdb};
204 $mid = mid_clean($mid);
205 $mid = mid_compressed($mid);
207 $db->begin_transaction;
209 $doc_id = $self->find_unique_doc_id('mid', $mid);
210 $db->delete_document($doc_id) if defined $doc_id;
214 warn "failed to remove message <$mid>: $@\n";
215 $db->cancel_transaction;
217 $db->commit_transaction;
224 my ($self, $query_string, $opts) = @_;
225 my $query = $self->qp->parse_query($query_string, QP_FLAGS);
227 $query = Search::Xapian::Query->new(OP_AND, $mail_query, $query);
228 $self->do_enquire($query, $opts);
231 # given a message ID, get replies to a message
233 my ($self, $mid, $opts) = @_;
234 $mid = mid_clean($mid);
235 $mid = mid_compressed($mid);
237 my $irt = $qp->parse_query("inreplyto:$mid", 0);
238 my $ref = $qp->parse_query("references:$mid", 0);
239 my $query = Search::Xapian::Query->new(OP_OR, $irt, $ref);
241 $self->do_enquire($query);
245 my ($self, $mid, $opts) = @_;
246 my $smsg = eval { $self->lookup_message($mid) };
248 return { count => 0, msgs => [] } unless $smsg;
250 my $qtid = $qp->parse_query('thread:'.$smsg->thread_id);
251 my $qsub = $qp->parse_query('path:'.$smsg->path);
252 my $query = Search::Xapian::Query->new(OP_OR, $qtid, $qsub);
253 $self->do_enquire($query);
259 my ($self, $query, $opts) = @_;
260 my $enquire = $self->enquire;
262 $enquire->set_query($query);
263 $enquire->set_sort_by_relevance_then_value(TS, 0);
265 my $offset = $opts->{offset} || 0;
266 my $limit = $opts->{limit} || 50;
267 my $mset = $enquire->get_mset($offset, $limit);
268 my @msgs = map { $_->get_document->get_data } $mset->items;
270 { count => $mset->get_matches_estimated, msgs => \@msgs }
274 sub stemmer { Search::Xapian::Stem->new(LANG) }
280 my $qp = $self->{query_parser};
284 $qp = Search::Xapian::QueryParser->new;
285 $qp->set_default_op(OP_AND);
286 $qp->set_database($self->{xdb});
287 $qp->set_stemmer($self->stemmer);
288 $qp->set_stemming_strategy(STEM_SOME);
289 $qp->add_valuerangeprocessor($self->ts_range_processor);
290 $qp->add_valuerangeprocessor($self->date_range_processor);
292 while (my ($name, $prefix) = each %bool_pfx_external) {
293 $qp->add_boolean_prefix($name, $prefix);
296 while (my ($name, $prefix) = each %prob_prefix) {
297 $qp->add_prefix($name, $prefix);
300 $self->{query_parser} = $qp;
303 sub term_generator { # write-only
306 my $tg = $self->{term_generator};
309 $tg = Search::Xapian::TermGenerator->new;
310 $tg->set_stemmer($self->stemmer);
312 $self->{term_generator} = $tg;
315 sub next_doc_id { $_[0]->{xdb}->get_lastdocid + 1 }
317 # increments last_thread_id counter
318 # returns a 64-bit integer represented as a hex string
321 my $db = $self->{xdb};
322 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
324 $db->set_metadata('last_thread_id', ++$last_thread_id);
329 sub ts_range_processor {
330 $_[0]->{tsrp} ||= Search::Xapian::NumberValueRangeProcessor->new(TS);
333 sub date_range_processor {
334 $_[0]->{drp} ||= Search::Xapian::DateValueRangeProcessor->new(TS);
338 my ($self, $smsg, $is_ghost) = @_;
341 $smsg->ensure_metadata;
343 $self->link_message_to_parents($smsg);
347 sub link_message_to_parents {
348 my ($self, $smsg) = @_;
349 my $doc = $smsg->{doc};
350 my $mid = mid_compressed($smsg->mid);
351 my $mime = $smsg->mime;
352 my $refs = $mime->header('References');
353 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
354 my $irt = $mime->header('In-Reply-To');
356 if ($irt =~ /<([^>]+)>/) {
364 @refs = map { mid_compressed($_) } @refs;
366 @refs = grep { !$uniq{$_}++ } @refs; # uniq
368 $doc->add_term(xpfx('inreplyto') . $refs[-1]);
370 my $ref_pfx = xpfx('references');
372 # first ref *should* be the thread root,
373 # but we can never trust clients to do the right thing
374 my $ref = shift @refs;
375 $doc->add_term($ref_pfx . $ref);
376 $tid = $self->_resolve_mid_to_tid($ref);
378 # the rest of the refs should point to this tid:
379 foreach $ref (@refs) {
380 $doc->add_term($ref_pfx . $ref);
381 my $ptid = $self->_resolve_mid_to_tid($ref);
383 $self->merge_threads($tid, $ptid);
387 $tid = $self->next_thread_id;
389 $doc->add_term(xpfx('thread') . $tid);
393 my ($self, $mid) = @_;
394 $mid = mid_clean($mid);
395 $mid = mid_compressed($mid);
397 my $doc_id = $self->find_unique_doc_id('mid', $mid);
399 if (defined $doc_id) {
401 my $doc = $self->{xdb}->get_document($doc_id);
402 $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
403 $smsg->doc_id($doc_id);
408 sub find_unique_doc_id {
409 my ($self, $term, $value) = @_;
411 my ($begin, $end) = $self->find_doc_ids($term, $value);
413 return undef if $begin->equal($end); # not found
415 my $rv = $begin->get_docid;
419 $begin->equal($end) or die "Term '$term:$value' is not unique\n";
423 # returns begin and end PostingIterator
425 my ($self, $term, $value) = @_;
427 $self->find_doc_ids_for_term(xpfx($term) . $value);
430 # returns begin and end PostingIterator
431 sub find_doc_ids_for_term {
432 my ($self, $term) = @_;
433 my $db = $self->{xdb};
435 ($db->postlist_begin($term), $db->postlist_end($term));
438 # this will create a ghost as necessary
439 sub _resolve_mid_to_tid {
440 my ($self, $mid) = @_;
442 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
447 my ($self, $mid, $tid) = @_;
449 $mid = mid_compressed($mid);
450 $tid = $self->next_thread_id unless defined $tid;
452 my $doc = Search::Xapian::Document->new;
453 $doc->add_term(xpfx('mid') . $mid);
454 $doc->add_term(xpfx('thread') . $tid);
455 $doc->add_term(xpfx('type') . 'ghost');
457 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
458 $self->link_message($smsg, 1);
459 $self->{xdb}->add_document($doc);
465 my ($self, $winner_tid, $loser_tid) = @_;
466 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
467 my $thread_pfx = xpfx('thread');
468 my $db = $self->{xdb};
470 for (; $head != $tail; $head->inc) {
471 my $docid = $head->get_docid;
472 my $doc = $db->get_document($docid);
473 $doc->remove_term($thread_pfx . $loser_tid);
474 $doc->add_term($thread_pfx . $winner_tid);
475 $db->replace_document($docid, $doc);
479 # normalize subjects so they are suitable as pathnames for URLs
485 $subj =~ s/^(?:re|aw):\s*//i; # remove reply prefix (aw: German)
486 $subj =~ s![^a-zA-Z0-9_\.~/\-]+!_!g;
491 my ($git, $blob) = @_;
493 my $str = $git->cat_file($blob);
494 Email::MIME->new($str);
500 my ($self, $git, $blob) = @_;
501 my $mime = do_cat_mail($git, $blob) or return;
502 eval { $self->add_message($mime) };
503 warn "W: index_blob $blob: $@\n" if $@;
507 my ($self, $git, $blob) = @_;
508 my $mime = do_cat_mail($git, $blob) or return;
509 my $mid = $mime->header('Message-ID');
510 eval { $self->remove_message($mid) } if defined $mid;
511 warn "W: unindex_blob $blob: $@\n" if $@;
516 $self->{enquire} ||= Search::Xapian::Enquire->new($self->{xdb});
519 # indexes all unindexed messages
521 my ($self, $git) = @_;
522 my $db = $self->{xdb};
523 my $latest = $db->get_metadata('last_commit');
524 my $range = length $latest ? "$latest..HEAD" : 'HEAD';
527 my $hex = '[a-f0-9]';
528 my $h40 = $hex .'{40}';
529 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
530 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
532 # get indexed messages
533 my @cmd = ('git', "--git-dir=$git->{git_dir}", "log",
534 qw/--reverse --no-notes --no-color --raw -r --no-abbrev/,
537 my $pid = open(my $log, '-|', @cmd) or
538 die('open` '.join(' ', @cmd) . " pipe failed: $!\n");
540 while (my $line = <$log>) {
541 if ($line =~ /$addmsg/o) {
542 $self->index_blob($git, $1);
543 } elsif ($line =~ /$delmsg/o) {
544 $self->unindex_blob($git, $1);
545 } elsif ($line =~ /^commit ($h40)/o) {
547 if (defined $latest) {
548 $db->set_metadata('last_commit', $latest)
554 $db->set_metadata('last_commit', $latest) if defined $latest;