4 package OpenILS::Application::Storage::Driver::Pg::QueryParser;
5 use OpenILS::Application::Storage::QueryParser;
6 use base 'QueryParser';
7 use OpenSRF::Utils::JSON;
8 use OpenILS::Application::AppUtils;
9 my $U = 'OpenILS::Application::AppUtils';
15 if ($value =~ /^\d/) { # may have to use non-$ quoting
17 $value =~ s/\\/\\\\/g;
20 return "\$_$$\$$value\$_$$\$";
23 sub quote_phrase_value {
27 my $left_anchored = $value =~ m/^\^/;
28 my $right_anchored = $value =~ m/\$$/;
29 $value =~ s/\^// if $left_anchored;
30 $value =~ s/\$$// if $right_anchored;
31 $value = quotemeta($value);
32 $value = '^' . $value if $left_anchored;
33 $value = "$value\$" if $right_anchored;
34 return $self->quote_value($value);
41 sub default_preferred_language {
45 $self->custom_data->{default_preferred_language} = $lang if ($lang);
46 return $self->custom_data->{default_preferred_language};
49 sub default_preferred_language_multiplier {
53 $self->custom_data->{default_preferred_language_multiplier} = $lang if ($lang);
54 return $self->custom_data->{default_preferred_language_multiplier};
60 return 0 unless $self->parse_tree;
61 return 0 if @{$self->parse_tree->filters};
62 return 0 if @{$self->parse_tree->modifiers};
63 for my $node ( @{ $self->parse_tree->query_nodes } ) {
64 return 0 if (!ref($node) && $node eq '|');
65 next unless (ref($node));
66 return 0 if ($node->isa('QueryParser::query_plan'));
74 return $self->parse_tree->toSQL;
81 $self->custom_data->{dynamic_filters} ||= [];
82 push(@{$self->custom_data->{dynamic_filters}}, $new) if ($new);
83 return $self->custom_data->{dynamic_filters};
90 $self->custom_data->{dynamic_sorters} ||= [];
91 push(@{$self->custom_data->{dynamic_sorters}}, $new) if ($new);
92 return $self->custom_data->{dynamic_sorters};
95 sub facet_field_id_map {
99 $self->custom_data->{facet_field_id_map} ||= {};
100 $self->custom_data->{facet_field_id_map} = $map if ($map);
101 return $self->custom_data->{facet_field_id_map};
104 sub add_facet_field_id_map {
111 $self->add_facet_field( $class => $field );
112 $self->facet_field_id_map->{by_id}{$id} = { classname => $class, field => $field, weight => $weight };
113 $self->facet_field_id_map->{by_class}{$class}{$field} = $id;
116 by_id => { $id => { classname => $class, field => $field, weight => $weight } },
117 by_class => { $class => { $field => $id } }
121 sub facet_field_class_by_id {
125 return $self->facet_field_id_map->{by_id}{$id};
128 sub facet_field_ids_by_class {
133 return undef unless ($class);
136 return [$self->facet_field_id_map->{by_class}{$class}{$field}];
139 return [values( %{ $self->facet_field_id_map->{by_class}{$class} } )];
142 sub search_field_id_map {
146 $self->custom_data->{search_field_id_map} ||= {};
147 $self->custom_data->{search_field_id_map} = $map if ($map);
148 return $self->custom_data->{search_field_id_map};
151 sub add_search_field_id_map {
158 $self->add_search_field( $class => $field );
159 $self->search_field_id_map->{by_id}{$id} = { classname => $class, field => $field, weight => $weight };
160 $self->search_field_id_map->{by_class}{$class}{$field} = $id;
163 by_id => { $id => { classname => $class, field => $field, weight => $weight } },
164 by_class => { $class => { $field => $id } }
168 sub search_field_class_by_id {
172 return $self->search_field_id_map->{by_id}{$id};
175 sub search_field_ids_by_class {
180 return undef unless ($class);
183 return [$self->search_field_id_map->{by_class}{$class}{$field}];
186 return [values( %{ $self->search_field_id_map->{by_class}{$class} } )];
189 sub relevance_bumps {
193 $self->custom_data->{rel_bumps} ||= {};
194 $self->custom_data->{rel_bumps} = $bumps if ($bumps);
195 return $self->custom_data->{rel_bumps};
198 sub find_relevance_bumps {
203 return $self->relevance_bumps->{$class}{$field};
206 sub add_relevance_bump {
211 my $multiplier = shift;
214 if (defined($active) and $active eq 'f') {
220 $self->relevance_bumps->{$class}{$field}{$type} = { multiplier => $multiplier, active => $active };
222 return { $class => { $field => { $type => { multiplier => $multiplier, active => $active } } } };
226 sub initialize_search_field_id_map {
228 my $cmf_list = shift;
230 for my $cmf (@$cmf_list) {
231 __PACKAGE__->add_search_field_id_map( $cmf->field_class, $cmf->name, $cmf->id, $cmf->weight ) if ($U->is_true($cmf->search_field));
232 __PACKAGE__->add_facet_field_id_map( $cmf->field_class, $cmf->name, $cmf->id, $cmf->weight ) if ($U->is_true($cmf->facet_field));
235 return $self->search_field_id_map;
238 sub initialize_aliases {
240 my $cmsa_list = shift;
242 for my $cmsa (@$cmsa_list) {
244 __PACKAGE__->add_search_class_alias( $cmsa->field_class, $cmsa->alias );
246 my $c = $self->search_field_class_by_id( $cmsa->field );
247 __PACKAGE__->add_search_field_alias( $cmsa->field_class, $c->{field}, $cmsa->alias );
252 sub initialize_relevance_bumps {
254 my $sra_list = shift;
256 for my $sra (@$sra_list) {
257 my $c = $self->search_field_class_by_id( $sra->field );
258 __PACKAGE__->add_relevance_bump( $c->{classname}, $c->{field}, $sra->bump_type, $sra->multiplier, $sra->active );
261 return $self->relevance_bumps;
264 sub initialize_query_normalizers {
266 my $tree = shift; # open-ils.cstore.direct.config.metabib_field_index_norm_map.search.atomic { "id" : { "!=" : null } }, { "flesh" : 1, "flesh_fields" : { "cmfinm" : ["norm"] }, "order_by" : [{ "class" : "cmfinm", "field" : "pos" }] }
268 for my $cmfinm ( @$tree ) {
269 my $field_info = $self->search_field_class_by_id( $cmfinm->field );
270 __PACKAGE__->add_query_normalizer( $field_info->{classname}, $field_info->{field}, $cmfinm->norm->func, OpenSRF::Utils::JSON->JSON2perl($cmfinm->params) );
274 sub initialize_dynamic_filters {
276 my $list = shift; # open-ils.cstore.direct.config.record_attr_definition.search.atomic { "id" : { "!=" : null } }
278 for my $crad ( @$list ) {
279 __PACKAGE__->dynamic_filters( __PACKAGE__->add_search_filter( $crad->name ) ) if ($U->is_true($crad->filter));
280 __PACKAGE__->dynamic_sorters( $crad->name ) if ($U->is_true($crad->sorter));
284 sub initialize_filter_normalizers {
286 my $tree = shift; # open-ils.cstore.direct.config.record_attr_index_norm_map.search.atomic { "id" : { "!=" : null } }, { "flesh" : 1, "flesh_fields" : { "crainm" : ["norm"] }, "order_by" : [{ "class" : "crainm", "field" : "pos" }] }
288 for my $crainm ( @$tree ) {
289 __PACKAGE__->add_filter_normalizer( $crainm->attr, $crainm->norm->func, OpenSRF::Utils::JSON->JSON2perl($crainm->params) );
294 sub initialization_complete {
302 return $_complete if ($_complete);
304 # tsearch rank normalization adjustments. see http://www.postgresql.org/docs/9.0/interactive/textsearch-controls.html#TEXTSEARCH-RANKING for details
305 $self->custom_data->{rank_cd_weight_map} = {
306 CD_logDocumentLength => 1,
307 CD_documentLength => 2,
308 CD_meanHarmonic => 4,
310 CD_logUniqueWords => 16,
314 $self->add_search_modifier( $_ ) for (keys %{ $self->custom_data->{rank_cd_weight_map} });
316 $self->initialize_search_field_id_map( $args{config_metabib_field} )
317 if ($args{config_metabib_field});
319 $self->initialize_aliases( $args{config_metabib_search_alias} )
320 if ($args{config_metabib_search_alias});
322 $self->initialize_relevance_bumps( $args{search_relevance_adjustment} )
323 if ($args{search_relevance_adjustment});
325 $self->initialize_query_normalizers( $args{config_metabib_field_index_norm_map} )
326 if ($args{config_metabib_field_index_norm_map});
328 $self->initialize_dynamic_filters( $args{config_record_attr_definition} )
329 if ($args{config_record_attr_definition});
331 $self->initialize_filter_normalizers( $args{config_record_attr_index_norm_map} )
332 if ($args{config_record_attr_index_norm_map});
335 $args{config_metabib_field_index_norm_map} &&
336 $args{search_relevance_adjustment} &&
337 $args{config_metabib_search_alias} &&
338 $args{config_metabib_field} &&
339 $args{config_record_attr_definition}
347 __PACKAGE__->add_search_field_id_map( series => seriestitle => 1 => 1 );
349 __PACKAGE__->add_search_field_id_map( series => seriestitle => 1 => 1 );
350 __PACKAGE__->add_relevance_bump( series => seriestitle => first_word => 1.5 );
351 __PACKAGE__->add_relevance_bump( series => seriestitle => full_match => 20 );
353 __PACKAGE__->add_search_field_id_map( title => abbreviated => 2 => 1 );
354 __PACKAGE__->add_relevance_bump( title => abbreviated => first_word => 1.5 );
355 __PACKAGE__->add_relevance_bump( title => abbreviated => full_match => 20 );
357 __PACKAGE__->add_search_field_id_map( title => translated => 3 => 1 );
358 __PACKAGE__->add_relevance_bump( title => translated => first_word => 1.5 );
359 __PACKAGE__->add_relevance_bump( title => translated => full_match => 20 );
361 __PACKAGE__->add_search_field_id_map( title => proper => 6 => 1 );
362 __PACKAGE__->add_query_normalizer( title => proper => 'search_normalize' );
363 __PACKAGE__->add_relevance_bump( title => proper => first_word => 1.5 );
364 __PACKAGE__->add_relevance_bump( title => proper => full_match => 20 );
365 __PACKAGE__->add_relevance_bump( title => proper => word_order => 10 );
367 __PACKAGE__->add_search_field_id_map( author => corporate => 7 => 1 );
368 __PACKAGE__->add_relevance_bump( author => corporate => first_word => 1.5 );
369 __PACKAGE__->add_relevance_bump( author => corporate => full_match => 20 );
371 __PACKAGE__->add_facet_field_id_map( author => personal => 8 => 1 );
373 __PACKAGE__->add_search_field_id_map( author => personal => 8 => 1 );
374 __PACKAGE__->add_relevance_bump( author => personal => first_word => 1.5 );
375 __PACKAGE__->add_relevance_bump( author => personal => full_match => 20 );
376 __PACKAGE__->add_query_normalizer( author => personal => 'search_normalize' );
377 __PACKAGE__->add_query_normalizer( author => personal => 'split_date_range' );
379 __PACKAGE__->add_facet_field_id_map( subject => topic => 14 => 1 );
381 __PACKAGE__->add_search_field_id_map( subject => topic => 14 => 1 );
382 __PACKAGE__->add_relevance_bump( subject => topic => first_word => 1 );
383 __PACKAGE__->add_relevance_bump( subject => topic => full_match => 1 );
385 __PACKAGE__->add_search_field_id_map( subject => complete => 16 => 1 );
386 __PACKAGE__->add_relevance_bump( subject => complete => first_word => 1 );
387 __PACKAGE__->add_relevance_bump( subject => complete => full_match => 1 );
389 __PACKAGE__->add_search_field_id_map( keyword => keyword => 15 => 1 );
390 __PACKAGE__->add_relevance_bump( keyword => keyword => first_word => 1 );
391 __PACKAGE__->add_relevance_bump( keyword => keyword => full_match => 1 );
394 __PACKAGE__->add_search_class_alias( keyword => 'kw' );
395 __PACKAGE__->add_search_class_alias( title => 'ti' );
396 __PACKAGE__->add_search_class_alias( author => 'au' );
397 __PACKAGE__->add_search_class_alias( author => 'name' );
398 __PACKAGE__->add_search_class_alias( author => 'dc.contributor' );
399 __PACKAGE__->add_search_class_alias( subject => 'su' );
400 __PACKAGE__->add_search_class_alias( subject => 'bib.subject(?:Title|Place|Occupation)' );
401 __PACKAGE__->add_search_class_alias( series => 'se' );
402 __PACKAGE__->add_search_class_alias( keyword => 'dc.identifier' );
404 __PACKAGE__->add_query_normalizer( author => corporate => 'search_normalize' );
405 __PACKAGE__->add_query_normalizer( keyword => keyword => 'search_normalize' );
407 __PACKAGE__->add_search_field_alias( subject => name => 'bib.subjectName' );
411 __PACKAGE__->default_search_class( 'keyword' );
413 # will be retained simply for back-compat
414 __PACKAGE__->add_search_filter( 'format' );
416 # grumble grumble, special cases against date1 and date2
417 __PACKAGE__->add_search_filter( 'before' );
418 __PACKAGE__->add_search_filter( 'after' );
419 __PACKAGE__->add_search_filter( 'between' );
420 __PACKAGE__->add_search_filter( 'during' );
422 # used by layers above this
423 __PACKAGE__->add_search_filter( 'statuses' );
424 __PACKAGE__->add_search_filter( 'locations' );
425 __PACKAGE__->add_search_filter( 'location_groups' );
426 __PACKAGE__->add_search_filter( 'site' );
427 __PACKAGE__->add_search_filter( 'pref_ou' );
428 __PACKAGE__->add_search_filter( 'lasso' );
429 __PACKAGE__->add_search_filter( 'my_lasso' );
430 __PACKAGE__->add_search_filter( 'depth' );
431 __PACKAGE__->add_search_filter( 'language' );
432 __PACKAGE__->add_search_filter( 'offset' );
433 __PACKAGE__->add_search_filter( 'limit' );
434 __PACKAGE__->add_search_filter( 'check_limit' );
435 __PACKAGE__->add_search_filter( 'skip_check' );
436 __PACKAGE__->add_search_filter( 'superpage' );
437 __PACKAGE__->add_search_filter( 'superpage_size' );
438 __PACKAGE__->add_search_filter( 'estimation_strategy' );
439 __PACKAGE__->add_search_modifier( 'available' );
440 __PACKAGE__->add_search_modifier( 'staff' );
442 # Start from container data (bre, acn, acp): container(bre,bookbag,123,deadb33fdeadb33fdeadb33fdeadb33f)
443 __PACKAGE__->add_search_filter( 'container' );
445 # Start from a list of record ids, either bre or metarecords, depending on the #metabib modifier
446 __PACKAGE__->add_search_filter( 'record_list' );
448 # used internally, but generally not user-settable
449 __PACKAGE__->add_search_filter( 'preferred_language' );
450 __PACKAGE__->add_search_filter( 'preferred_language_weight' );
451 __PACKAGE__->add_search_filter( 'preferred_language_multiplier' );
452 __PACKAGE__->add_search_filter( 'core_limit' );
454 # XXX Valid values to be supplied by SVF
455 __PACKAGE__->add_search_filter( 'sort' );
457 # modifies core query, not configurable
458 __PACKAGE__->add_search_modifier( 'descending' );
459 __PACKAGE__->add_search_modifier( 'ascending' );
460 __PACKAGE__->add_search_modifier( 'nullsfirst' );
461 __PACKAGE__->add_search_modifier( 'nullslast' );
462 __PACKAGE__->add_search_modifier( 'metarecord' );
463 __PACKAGE__->add_search_modifier( 'metabib' );
466 #-------------------------------
467 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan;
468 use base 'QueryParser::query_plan';
469 use OpenSRF::Utils::Logger qw($logger);
471 use OpenILS::Application::AppUtils;
472 my $apputils = "OpenILS::Application::AppUtils";
479 my ($format) = $self->find_filter('format');
481 my ($t,$f) = split('-', $format->args->[0]);
482 $self->new_filter( item_type => [ split '', $t ] ) if ($t);
483 $self->new_filter( item_form => [ split '', $f ] ) if ($f);
486 for my $f ( qw/preferred_language preferred_language_multiplier preferred_language_weight core_limit check_limit skip_check superpage superpage_size/ ) {
488 $col = 'preferred_language_multiplier' if ($f eq 'preferred_language_weight');
489 my ($filter) = $self->find_filter($f);
490 if ($filter and @{$filter->args}) {
491 $filters{$col} = $filter->args->[0];
495 $self->QueryParser->superpage($filters{superpage}) if ($filters{superpage});
496 $self->QueryParser->superpage_size($filters{superpage_size}) if ($filters{superpage_size});
497 $self->QueryParser->core_limit($filters{core_limit}) if ($filters{core_limit});
499 $logger->debug("Query plan:\n".Dumper($self));
501 my $flat_plan = $self->flatten;
503 # generate the relevance ranking
504 my $rel = "AVG(\n\t\t(" . join(")+\n\t\t(", @{$$flat_plan{rank_list}}) . ")\n\t)+1";
506 # find any supplied sort option
507 my ($sort_filter) = $self->find_filter('sort');
509 $sort_filter = $sort_filter->args->[0];
511 $sort_filter = 'rel';
514 if (($filters{preferred_language} || $self->QueryParser->default_preferred_language) && ($filters{preferred_language_multiplier} || $self->QueryParser->default_preferred_language_multiplier)) {
515 my $pl = $self->QueryParser->quote_value( $filters{preferred_language} ? $filters{preferred_language} : $self->QueryParser->default_preferred_language );
516 my $plw = $filters{preferred_language_multiplier} ? $filters{preferred_language_multiplier} : $self->QueryParser->default_preferred_language_multiplier;
517 $rel = "($rel * COALESCE( NULLIF( FIRST(mrd.attrs \@> hstore('item_lang', $pl)), FALSE )::INT * $plw, 1))";
519 $rel = "1.0/($rel)::NUMERIC";
521 my %dyn_filters = ( '' => [] ); # the "catch-all" key
522 for my $f ( @{ $self->QueryParser->dynamic_filters } ) {
524 $col = 'item_lang' if ($f eq 'language'); #XXX filter aliases would address this ... booo ... later
526 my ($filter) = $self->find_filter($f);
528 my @fargs = @{$filter->args};
530 if (@fargs > 1 || $filter->negate) {
531 my $NOT = $filter->negate ? 'NOT' : '';
532 $dyn_filters{$f} = "$NOT( " .
535 map { "mrd.attrs \@> hstore('$col', " . $self->QueryParser->quote_value($_) . ")" } @fargs
539 push(@{$dyn_filters{''}}, "hstore('$col', " . $self->QueryParser->quote_value($fargs[0]) . ")");
544 my $combined_dyn_filters = '';
545 $combined_dyn_filters .= 'AND mrd.attrs @> (' . join(' || ', @{$dyn_filters{''}}) . ') ' if (@{$dyn_filters{''}});
546 delete($dyn_filters{''});
548 my @dyn_filter_list = values(%dyn_filters);
549 $combined_dyn_filters .= 'AND ' . join(' AND ', @dyn_filter_list) if (@dyn_filter_list);
554 $desc = 'DESC' if ($self->find_modifier('descending'));
556 my $nullpos = 'NULLS LAST';
557 $nullpos = 'NULLS FIRST' if ($self->find_modifier('nullsfirst'));
559 if (grep {$_ eq $sort_filter} @{$self->QueryParser->dynamic_sorters}) {
560 $rank = "FIRST(mrd.attrs->'$sort_filter')"
561 } elsif ($sort_filter eq 'create_date') {
562 $rank = "FIRST((SELECT create_date FROM biblio.record_entry rbr WHERE rbr.id = m.source))";
563 } elsif ($sort_filter eq 'edit_date') {
564 $rank = "FIRST((SELECT edit_date FROM biblio.record_entry rbr WHERE rbr.id = m.source))";
566 # default to rel ranking
570 my $key = 'm.source';
571 $key = 'm.metarecord' if (grep {$_->name eq 'metarecord' or $_->name eq 'metabib'} @{$self->modifiers});
573 my ($before) = $self->find_filter('before');
574 my ($after) = $self->find_filter('after');
575 my ($during) = $self->find_filter('during');
576 my ($between) = $self->find_filter('between');
577 my ($container) = $self->find_filter('container');
578 my ($record_list) = $self->find_filter('record_list');
580 if ($before and @{$before->args} == 1) {
581 $before = "AND (mrd.attrs->'date1') <= " . $self->QueryParser->quote_value($before->args->[0]);
586 if ($after and @{$after->args} == 1) {
587 $after = "AND (mrd.attrs->'date1') >= " . $self->QueryParser->quote_value($after->args->[0]);
592 if ($during and @{$during->args} == 1) {
593 $during = "AND " . $self->QueryParser->quote_value($during->args->[0]) . " BETWEEN (mrd.attrs->'date1') AND (mrd.attrs->'date2')";
598 if ($between and @{$between->args} == 2) {
599 $between = "AND (mrd.attrs->'date1') BETWEEN " . $self->QueryParser->quote_value($between->args->[0]) . " AND " . $self->QueryParser->quote_value($between->args->[1]);
604 if ($container and @{$container->args} >= 3) {
605 my ($class, $ctype, $cid, $token) = @{ $container->args };
609 my $rec_field = 'ci.target_biblio_record_entry';
611 if ($class eq 'bre') {
612 $class = 'biblio_record_entry';
613 } elsif ($class eq 'acn') {
614 $class = 'call_number';
615 $rec_field = 'cn.record';
616 $rec_join = 'JOIN asset.call_number cn ON (ci.target_call_number = cn.id)';
617 } elsif ($class eq 'acp') {
619 $rec_field = 'cn.record';
620 $rec_join = 'JOIN asset.copy cp ON (ci.target_copy = cp.id) JOIN asset.call_number cn ON (cp.call_number = cn.id)';
621 # } elsif ($class eq 'au') {
623 # $rec_field = 'cn.record';
624 # $rec_join = 'JOIN asset.call_number cn ON ci.target_call_number = cn.id';
625 } else { $class = undef };
628 my ($u,$e) = $apputils->checksesperm($token) if ($token);
629 $perm_join = 'OR c.owner = ' . $u->id if ($u && !$e);
632 JOIN ( SELECT $rec_field AS container_item
633 FROM container.${class}_bucket_item ci
634 JOIN container.${class}_bucket c ON (c.id = ci.bucket)
636 WHERE c.btype = > . $self->QueryParser->quote_value($ctype) .
637 qq< AND c.id = > . $self->QueryParser->quote_value($cid) .
638 qq< AND (c.pub IS TRUE $perm_join)) container ON (container.container_item = mrd.id) >;
639 } else {$container = ''};
644 if ($record_list and @{$record_list->args} > 0) {
645 $record_list = 'JOIN (VALUES (' . join('),(', map { $self->QueryParser->quote_value($_) } @{ $record_list->args}) . ")) record_list(id) ON (record_list.id::BIGINT = $key)"
650 my $core_limit = $self->QueryParser->core_limit || 25000;
652 my $flat_where = $$flat_plan{where};
653 if ($flat_where eq '()') {
656 $flat_where = "AND $flat_where";
658 my $with = $$flat_plan{with};
659 $with= "\n\t\t\tWITH $with" if $with;
661 # Need an array for query parser db function; this gives a better plan
662 # than the ARRAY_AGG(DISTINCT m.source) option as of PostgreSQL 9.1
663 my $agg_records = 'ARRAY[m.source] AS records';
664 if ($key =~ /metarecord/) {
665 # metarecord searches still require the ARRAY_AGG approach
666 $agg_records = 'ARRAY_AGG(DISTINCT m.source) AS records';
675 FIRST(mrd.attrs->'date1') AS tie_break
676 FROM metabib.metarecord_source_map m
677 JOIN metabib.record_attr mrd ON (m.source = mrd.id)
686 $combined_dyn_filters
689 ORDER BY 4 $desc $nullpos, 5 DESC $nullpos, 3 DESC
693 warn $sql if $self->QueryParser->debug;
703 my $multiplier = shift;
705 my $only_atoms = $node->only_atoms;
706 return '' if (!@$only_atoms);
708 if ($bump eq 'first_word') {
709 return " /* first_word */ COALESCE(NULLIF( (search_normalize(".$node->table_alias.".value) ~ ('^'||search_normalize(".$self->QueryParser->quote_phrase_value($only_atoms->[0]->content)."))), FALSE )::INT * $multiplier, 1)";
710 } elsif ($bump eq 'full_match') {
711 return " /* full_match */ COALESCE(NULLIF( (search_normalize(".$node->table_alias.".value) ~ ('^'||".
712 join( "||' '||", map { "search_normalize(".$self->QueryParser->quote_phrase_value($_->content).")" } @$only_atoms )."||'\$')), FALSE )::INT * $multiplier, 1)";
713 } elsif ($bump eq 'word_order') {
714 return " /* word_order */ COALESCE(NULLIF( (search_normalize(".$node->table_alias.".value) ~ (".
715 join( "||'.*'||", map { "search_normalize(".$self->QueryParser->quote_phrase_value($_->content).")" } @$only_atoms ).")), FALSE )::INT * $multiplier, 1)";
724 my $from = shift || '';
725 my $where = shift || '(';
729 for my $node ( @{$self->query_nodes} ) {
731 if ($node->isa( 'QueryParser::query_plan::node' )) {
733 unless (@{$node->only_atoms}) {
734 push @rank_list, '1';
739 my $table = $node->table;
740 my $talias = $node->table_alias;
742 my $node_rank = 'COALESCE(' . $node->rank . " * ${talias}.weight, 0.0)";
744 my $core_limit = $self->QueryParser->core_limit || 25000;
745 $from .= "\n\tLEFT JOIN (\n\t\tSELECT fe.*, fe_weight.weight, ${talias}_xq.tsq /* search */\n\t\t FROM $table AS fe";
746 $from .= "\n\t\t\tJOIN config.metabib_field AS fe_weight ON (fe_weight.id = fe.field)";
748 if ($node->dummy_count < @{$node->only_atoms} ) {
749 $with .= ",\n" if $with;
750 $with .= "${talias}_xq AS (SELECT ". $node->tsquery ." AS tsq )";
751 $from .= "\n\t\t\tJOIN ${talias}_xq ON (fe.index_vector @@ ${talias}_xq.tsq)";
753 $from .= "\n\t\t\t, (SELECT NULL::tsquery AS tsq ) AS x";
757 if (@{$node->fields} > 0) {
758 @bump_fields = @{$node->fields};
760 my @field_ids = grep defined, (
762 $self->QueryParser->search_field_ids_by_class(
768 $from .= "\n\t\t\tWHERE fe_weight.id IN (" .
769 join(',', @field_ids) . ")";
773 @bump_fields = @{$self->QueryParser->search_fields->{$node->classname}};
776 ###$from .= "\n\t\tLIMIT $core_limit";
777 $from .= "\n\t) AS $talias ON (m.source = ${talias}.source)";
781 for my $field ( @bump_fields ) {
782 my $bumps = $self->QueryParser->find_relevance_bumps( $node->classname => $field );
783 for my $b (keys %$bumps) {
784 next if (!$$bumps{$b}{active});
785 next if ($used_bumps{$b});
788 next if ($$bumps{$b}{multiplier} == 1); # optimization to remove unneeded bumps
790 my $bump_case = $self->rel_bump( $node, $b, $$bumps{$b}{multiplier} );
791 $node_rank .= "\n\t\t\t\t * " . $bump_case if ($bump_case);
795 $where .= '(' . $talias . ".id IS NOT NULL";
796 $where .= ' AND ' . join(' AND ', map {"${talias}.value ~* ".$self->QueryParser->quote_phrase_value($_)} @{$node->phrases}) if (@{$node->phrases});
797 $where .= ' AND ' . join(' AND ', map {"${talias}.value !~* ".$self->QueryParser->quote_phrase_value($_)} @{$node->unphrases}) if (@{$node->unphrases});
800 push @rank_list, $node_rank;
802 } elsif ($node->isa( 'QueryParser::query_plan::facet' )) {
804 my $table = $node->table;
805 my $talias = $node->table_alias;
808 if (@{$node->fields} > 0) {
809 push(@field_ids, $self->QueryParser->facet_field_ids_by_class( $node->classname, $_ )->[0]) for (@{$node->fields});
811 @field_ids = @{ $self->QueryParser->facet_field_ids_by_class( $node->classname ) };
814 my $join_type = $node->negate ? 'LEFT' : 'INNER';
815 $from .= "\n\t$join_type JOIN /* facet */ metabib.facet_entry $talias ON (\n\t\tm.source = ${talias}.source\n\t\t".
816 "AND SUBSTRING(${talias}.value,1,1024) IN (" . join(",", map { $self->QueryParser->quote_value($_) } @{$node->values}) . ")\n\t\t".
817 "AND ${talias}.field IN (". join(',', @field_ids) . ")\n\t)";
819 $where .= $node->negate ? "${talias}.id IS NULL" : 'TRUE';
822 my $subnode = $node->flatten;
824 push(@rank_list, @{$$subnode{rank_list}});
825 $from .= $$subnode{from};
826 $where .= "($$subnode{where})";
828 if ($$subnode{with}) {
829 $with .= ', ' if $with;
830 $with .= " " . $$subnode{with};
834 $where .= ' AND ' if ($node eq '&');
835 $where .= ' OR ' if ($node eq '|');
836 # ... stitching the WHERE together ...
840 return { rank_list => \@rank_list, from => $from, where => $where.')', with => $with };
845 #-------------------------------
846 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::filter;
847 use base 'QueryParser::query_plan::filter';
849 #-------------------------------
850 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::facet;
851 use base 'QueryParser::query_plan::facet';
855 my ($classname) = split '\|', $self->name;
861 return 'metabib.' . $self->classname . '_field_entry';
866 my ($classname,@fields) = split '\|', $self->name;
873 my $table_alias = "$self";
874 $table_alias =~ s/^.*\(0(x[0-9a-fA-F]+)\)$/$1/go;
875 $table_alias .= '_' . $self->name;
876 $table_alias =~ s/\|/_/go;
882 #-------------------------------
883 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::modifier;
884 use base 'QueryParser::query_plan::modifier';
886 #-------------------------------
887 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::node::atom;
888 use base 'QueryParser::query_plan::node::atom';
894 $self->{sql} = $sql if ($sql);
896 return $self->{sql} if ($self->{sql});
897 return $self->buildSQL;
903 my $classname = $self->node->classname;
905 return $self->sql("to_tsquery('$classname','')") if $self->{dummy};
907 my $normalizers = $self->node->plan->QueryParser->query_normalizers( $classname );
908 my $fields = $self->node->fields;
910 $fields = $self->node->plan->QueryParser->search_fields->{$classname} if (!@$fields);
914 for my $field (@$fields) {
915 for my $nfield (keys %$normalizers) {
916 for my $nizer ( @{$$normalizers{$nfield}} ) {
917 if ($field eq $nfield) {
918 my $param_string = OpenSRF::Utils::JSON->perl2JSON($nizer->{params});
919 if (!exists($norms{$nizer->{function}.$param_string})) {
920 $norms{$nizer->{function}.$param_string} = {p=>$pos++,n=>$nizer};
927 my $sql = $self->node->plan->QueryParser->quote_value($self->content);
929 for my $n ( map { $$_{n} } sort { $$a{p} <=> $$b{p} } values %norms ) {
930 $sql = join(', ', $sql, map { $self->node->plan->QueryParser->quote_value($_) } @{ $n->{params} });
931 $sql = $n->{function}."($sql)";
934 my $prefix = $self->prefix || '';
935 my $suffix = $self->suffix || '';
937 $prefix = "'$prefix' ||" if $prefix;
939 my $suffix_after = '';
941 $suffix_op = ":$suffix" if $suffix;
942 $suffix_after = "|| '$suffix_op'" if $suffix;
944 $sql = "to_tsquery('$classname', COALESCE(NULLIF($prefix '(' || btrim(regexp_replace($sql,E'(?:\\\\s+|:)','$suffix_op&','g'),'&|') $suffix_after || ')', '()'), ''))";
946 return $self->sql($sql);
949 #-------------------------------
950 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::node;
951 use base 'QueryParser::query_plan::node';
956 $self->{dummy_count} = 0;
958 my $atoms = $self->query_atoms;
960 for my $a (@$atoms) {
961 push(@only_atoms, $a) if (ref($a) && $a->isa('QueryParser::query_plan::node::atom'));
962 $self->{dummy_count}++ if (ref($a) && $a->{dummy});
970 return $self->{dummy_count};
976 $self->{table} = $table if ($table);
977 return $self->{table} if $self->{table};
978 return $self->table( 'metabib.' . $self->classname . '_field_entry' );
983 my $table_alias = shift;
984 $self->{table_alias} = $table_alias if ($table_alias);
985 return $self->{table_alias} if ($self->{table_alias});
987 $table_alias = "$self";
988 $table_alias =~ s/^.*\(0(x[0-9a-fA-F]+)\)$/$1/go;
989 $table_alias .= '_' . $self->requested_class;
990 $table_alias =~ s/\|/_/go;
992 return $self->table_alias( $table_alias );
997 return $self->{tsquery} if ($self->{tsquery});
999 for my $atom (@{$self->query_atoms}) {
1001 $self->{tsquery} .= "\n\t\t\t" .$atom->sql;
1003 $self->{tsquery} .= $atom x 2;
1007 return $self->{tsquery};
1013 my $rank_norm_map = $self->plan->QueryParser->custom_data->{rank_cd_weight_map};
1015 my $cover_density = 0;
1016 for my $norm ( keys %$rank_norm_map) {
1017 $cover_density += $$rank_norm_map{$norm} if ($self->plan->find_modifier($norm));
1020 return $self->{rank} if ($self->{rank});
1021 return $self->{rank} = 'ts_rank_cd(' . $self->table_alias . '.index_vector, ' . $self->table_alias . ".tsq, $cover_density)";