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/;
13 # This is English-only, everything else is non-standard and may be confused as
14 # a prefix common in patch emails
15 our $REPLY_RE = qr/^re:\s+/i;
19 # SCHEMA_VERSION history
21 # 1 - subject_path is lower-cased
22 # 2 - subject_path is mid_compressed in the index, only
23 # 3 - message-ID is compressed if it includes '%' (hack!)
26 QP_FLAGS => FLAG_PHRASE|FLAG_BOOLEAN|FLAG_LOVEHATE|FLAG_WILDCARD,
30 my %bool_pfx_internal = (
31 type => 'T', # "mail" or "ghost"
32 mid => 'Q', # uniQue id (Message-ID or mid_compressed)
35 my %bool_pfx_external = (
37 thread => 'G', # newsGroup (or similar entity - e.g. a web forum name)
38 references => 'XREFS',
46 my %all_pfx = (%bool_pfx_internal, %bool_pfx_external, %prob_prefix);
48 sub xpfx { $all_pfx{$_[0]} }
51 while (my ($k, $v) = each %all_pfx) {
52 next if $prob_prefix{$k};
53 $PFX2TERM_RMAP{$v} = $k;
56 my $mail_query = Search::Xapian::Query->new(xpfx('type') . 'mail');
59 my ($class, $git_dir, $writable) = @_;
60 # allow concurrent versions for easier rollback:
61 my $dir = "$git_dir/public-inbox/xapian" . SCHEMA_VERSION;
64 if ($writable) { # not used by the WWW interface
65 require Search::Xapian::WritableDatabase;
66 my $flag = Search::Xapian::DB_OPEN;
69 File::Path::mkpath($dir);
70 $flag = Search::Xapian::DB_CREATE_OR_OPEN;
72 $db = Search::Xapian::WritableDatabase->new($dir, $flag);
74 $db = Search::Xapian::Database->new($dir);
76 bless { xdb => $db, git_dir => $git_dir }, $class;
79 sub reopen { $_[0]->{xdb}->reopen }
82 my ($self, $mime) = @_; # mime = Email::MIME object
83 my $db = $self->{xdb};
86 my $mid_orig = mid_clean($mime->header_obj->header_raw('Message-ID'));
87 my $mid = mid_compressed($mid_orig);
89 my $ct_msg = $mime->header('Content-Type') || 'text/plain';
90 my $enc_msg = PublicInbox::View::enc_for($ct_msg);
93 my $smsg = $self->lookup_message($mid);
97 $smsg->ensure_metadata;
98 # convert a ghost to a regular message
99 # it will also clobber any existing regular message
103 my $type = xpfx('type');
105 $doc->remove_term($type . 'ghost');
109 # probably does not exist:
110 eval { $doc->remove_term($type . 'mail') };
111 $doc->add_term($type . 'mail');
113 $smsg = PublicInbox::SearchMsg->new($mime);
115 $doc->add_term(xpfx('mid') . $mid);
118 my $subj = $smsg->subject;
121 $doc->add_term(xpfx('subject') . $subj);
123 my $path = subject_path($subj);
124 $doc->add_term(xpfx('path') . mid_compressed($path));
127 my $from = $smsg->from_name;
128 my $date = $smsg->date;
129 my $ts = Search::Xapian::sortable_serialise($smsg->ts);
130 $doc->add_value(PublicInbox::Search::TS, $ts);
132 $doc->set_data($smsg->to_doc_data);
134 my $tg = $self->term_generator;
136 $tg->set_document($doc);
137 $tg->index_text($subj, 1, 'S') if $subj;
138 $tg->increase_termpos;
139 $tg->index_text($subj) if $subj;
140 $tg->increase_termpos;
142 $tg->index_text($smsg->from->format);
143 $tg->increase_termpos;
145 $mime->walk_parts(sub {
147 return if $part->subparts; # walk_parts already recurses
148 my $ct = $part->content_type || $ct_msg;
150 # account for filter bugs...
151 $ct =~ m!\btext/plain\b!i or return;
153 my $enc = PublicInbox::View::enc_for($ct, $enc_msg);
155 foreach my $l (split(/\n/, $enc->decode($part->body))) {
163 $tg->index_text(join("\n", @quot), 0);
164 $tg->increase_termpos;
167 $tg->index_text(join("\n", @orig));
168 $tg->increase_termpos;
173 $doc_id = $smsg->doc_id;
174 $self->link_message($smsg, 0);
175 $db->replace_document($doc_id, $doc);
177 $self->link_message($smsg, 0);
178 $doc_id = $db->add_document($doc);
183 warn "failed to index message <$mid_orig>: $@\n";
189 # returns deleted doc_id on success, undef on missing
191 my ($self, $mid_orig) = @_;
192 my $db = $self->{xdb};
194 $mid_orig = mid_clean($mid_orig);
195 my $mid = mid_compressed($mid_orig);
198 $doc_id = $self->find_unique_doc_id('mid', $mid);
199 $db->delete_document($doc_id) if defined $doc_id;
203 warn "failed to remove message <$mid_orig>: $@\n";
211 my ($self, $query_string, $opts) = @_;
212 my $query = $self->qp->parse_query($query_string, QP_FLAGS);
214 $query = Search::Xapian::Query->new(OP_AND, $mail_query, $query);
215 $self->do_enquire($query, $opts);
218 sub get_subject_path {
219 my ($self, $path, $opts) = @_;
220 my $query = $self->qp->parse_query("path:".mid_compressed($path), 0);
221 $self->do_enquire($query);
224 # given a message ID, get followups to a message
226 my ($self, $mid, $opts) = @_;
227 $mid = mid_clean($mid);
228 $mid = mid_compressed($mid);
230 my $irt = $qp->parse_query("inreplyto:$mid", 0);
231 my $ref = $qp->parse_query("references:$mid", 0);
232 my $query = Search::Xapian::Query->new(OP_OR, $irt, $ref);
234 $self->do_enquire($query);
238 my ($self, $mid, $opts) = @_;
239 my $smsg = eval { $self->lookup_message($mid) };
241 return { count => 0, msgs => [] } unless $smsg;
243 my $qtid = $qp->parse_query('thread:'.$smsg->thread_id);
244 my $qsub = $qp->parse_query('path:'.mid_compressed($smsg->path));
245 my $query = Search::Xapian::Query->new(OP_OR, $qtid, $qsub);
246 $self->do_enquire($query);
252 my ($self, $query, $opts) = @_;
253 my $enquire = $self->enquire;
255 $enquire->set_query($query);
256 $enquire->set_sort_by_relevance_then_value(TS, 0);
258 my $offset = $opts->{offset} || 0;
259 my $limit = $opts->{limit} || 50;
260 my $mset = $enquire->get_mset($offset, $limit);
262 PublicInbox::SearchMsg->load_doc($_->get_document);
265 { count => $mset->get_matches_estimated, msgs => \@msgs }
269 sub stemmer { Search::Xapian::Stem->new(LANG) }
275 my $qp = $self->{query_parser};
279 $qp = Search::Xapian::QueryParser->new;
280 $qp->set_default_op(OP_AND);
281 $qp->set_database($self->{xdb});
282 $qp->set_stemmer($self->stemmer);
283 $qp->set_stemming_strategy(STEM_SOME);
284 $qp->add_valuerangeprocessor($self->ts_range_processor);
285 $qp->add_valuerangeprocessor($self->date_range_processor);
287 while (my ($name, $prefix) = each %bool_pfx_external) {
288 $qp->add_boolean_prefix($name, $prefix);
291 while (my ($name, $prefix) = each %prob_prefix) {
292 $qp->add_prefix($name, $prefix);
295 $self->{query_parser} = $qp;
298 sub term_generator { # write-only
301 my $tg = $self->{term_generator};
304 $tg = Search::Xapian::TermGenerator->new;
305 $tg->set_stemmer($self->stemmer);
307 $self->{term_generator} = $tg;
310 sub next_doc_id { $_[0]->{xdb}->get_lastdocid + 1 }
312 # increments last_thread_id counter
313 # returns a 64-bit integer represented as a hex string
316 my $db = $self->{xdb};
317 my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
319 $db->set_metadata('last_thread_id', ++$last_thread_id);
324 sub ts_range_processor {
325 $_[0]->{tsrp} ||= Search::Xapian::NumberValueRangeProcessor->new(TS);
328 sub date_range_processor {
329 $_[0]->{drp} ||= Search::Xapian::DateValueRangeProcessor->new(TS);
333 my ($self, $smsg, $is_ghost) = @_;
336 $smsg->ensure_metadata;
338 $self->link_message_to_parents($smsg);
342 sub link_message_to_parents {
343 my ($self, $smsg) = @_;
344 my $doc = $smsg->{doc};
345 my $mid = mid_compressed($smsg->mid);
346 my $mime = $smsg->mime;
347 my $refs = $mime->header_obj->header_raw('References');
348 my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
349 my $irt = $mime->header_obj->header_raw('In-Reply-To');
351 if ($irt =~ /<([^>]+)>/) {
355 # maybe some crazies will try to make a circular reference:
365 my @crefs = map { mid_compressed($_) } @refs;
366 my %uniq = ($mid => 1);
368 # prevent circular references via References: here:
370 foreach my $ref (@crefs) {
375 $irt = undef if (defined $irt && !$uniq{$irt});
379 $doc->add_term(xpfx('inreplyto') . $irt);
382 my $ref_pfx = xpfx('references');
384 # first ref *should* be the thread root,
385 # but we can never trust clients to do the right thing
386 my $ref = shift @refs;
387 $doc->add_term($ref_pfx . $ref);
388 $tid = $self->_resolve_mid_to_tid($ref);
390 # the rest of the refs should point to this tid:
391 foreach $ref (@refs) {
392 $doc->add_term($ref_pfx . $ref);
393 my $ptid = $self->_resolve_mid_to_tid($ref);
395 $self->merge_threads($tid, $ptid);
399 $tid = $self->next_thread_id;
401 $doc->add_term(xpfx('thread') . $tid);
405 my ($self, $mid) = @_;
406 $mid = mid_clean($mid);
407 $mid = mid_compressed($mid);
409 my $doc_id = $self->find_unique_doc_id('mid', $mid);
411 if (defined $doc_id) {
413 my $doc = $self->{xdb}->get_document($doc_id);
414 $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
415 $smsg->doc_id($doc_id);
420 sub find_unique_doc_id {
421 my ($self, $term, $value) = @_;
423 my ($begin, $end) = $self->find_doc_ids($term, $value);
425 return undef if $begin->equal($end); # not found
427 my $rv = $begin->get_docid;
431 $begin->equal($end) or die "Term '$term:$value' is not unique\n";
435 # returns begin and end PostingIterator
437 my ($self, $term, $value) = @_;
439 $self->find_doc_ids_for_term(xpfx($term) . $value);
442 # returns begin and end PostingIterator
443 sub find_doc_ids_for_term {
444 my ($self, $term) = @_;
445 my $db = $self->{xdb};
447 ($db->postlist_begin($term), $db->postlist_end($term));
450 # this will create a ghost as necessary
451 sub _resolve_mid_to_tid {
452 my ($self, $mid) = @_;
454 my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
459 my ($self, $mid, $tid) = @_;
461 $mid = mid_compressed($mid);
462 $tid = $self->next_thread_id unless defined $tid;
464 my $doc = Search::Xapian::Document->new;
465 $doc->add_term(xpfx('mid') . $mid);
466 $doc->add_term(xpfx('thread') . $tid);
467 $doc->add_term(xpfx('type') . 'ghost');
469 my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
470 $self->link_message($smsg, 1);
471 $self->{xdb}->add_document($doc);
477 my ($self, $winner_tid, $loser_tid) = @_;
478 my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
479 my $thread_pfx = xpfx('thread');
480 my $db = $self->{xdb};
482 for (; $head != $tail; $head->inc) {
483 my $docid = $head->get_docid;
484 my $doc = $db->get_document($docid);
485 $doc->remove_term($thread_pfx . $loser_tid);
486 $doc->add_term($thread_pfx . $winner_tid);
487 $db->replace_document($docid, $doc);
491 # normalize subjects so they are suitable as pathnames for URLs
497 $subj =~ s/$REPLY_RE//igo; # remove reply prefix
498 $subj =~ s![^a-zA-Z0-9_\.~/\-]+!_!g;
503 my ($git, $blob) = @_;
505 my $str = $git->cat_file($blob);
506 Email::MIME->new($str);
512 my ($self, $git, $blob) = @_;
513 my $mime = do_cat_mail($git, $blob) or return;
514 eval { $self->add_message($mime) };
515 warn "W: index_blob $blob: $@\n" if $@;
519 my ($self, $git, $blob) = @_;
520 my $mime = do_cat_mail($git, $blob) or return;
521 my $mid = $mime->header_obj->header_raw('Message-ID');
522 eval { $self->remove_message($mid) } if defined $mid;
523 warn "W: unindex_blob $blob: $@\n" if $@;
528 $self->{enquire} ||= Search::Xapian::Enquire->new($self->{xdb});
531 # indexes all unindexed messages
534 require PublicInbox::GitCatFile;
535 my $db = $self->{xdb};
536 my $hex = '[a-f0-9]';
537 my $h40 = $hex .'{40}';
538 my $addmsg = qr!^:000000 100644 \S+ ($h40) A\t${hex}{2}/${hex}{38}$!;
539 my $delmsg = qr!^:100644 000000 ($h40) \S+ D\t${hex}{2}/${hex}{38}$!;
541 $db->begin_transaction;
543 my $git = PublicInbox::GitCatFile->new($self->{git_dir});
545 my $latest = $db->get_metadata('last_commit');
546 my $range = length $latest ? "$latest..HEAD" : 'HEAD';
549 # get indexed messages
550 my @cmd = ('git', "--git-dir=$self->{git_dir}", "log",
551 qw/--reverse --no-notes --no-color --raw -r
552 --no-abbrev/, $range);
553 my $pid = open(my $log, '-|', @cmd) or
554 die('open` '.join(' ', @cmd) . " pipe failed: $!\n");
556 while (my $line = <$log>) {
557 if ($line =~ /$addmsg/o) {
558 $self->index_blob($git, $1);
559 } elsif ($line =~ /$delmsg/o) {
560 $self->unindex_blob($git, $1);
561 } elsif ($line =~ /^commit ($h40)/o) {
566 $db->set_metadata('last_commit', $latest) if defined $latest;
569 warn "indexing failed: $@\n";
570 $db->cancel_transaction;
572 $db->commit_transaction;