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 use OpenILS::Utils::CStoreEditor;
10 my $U = 'OpenILS::Application::AppUtils';
13 sub subquery_callback {
14 my ($invocant, $self, $struct, $filter, $params, $negate) = @_;
21 OpenILS::Utils::CStoreEditor
23 ->search_actor_search_query({ id => $params })
32 if ($value =~ /^\d/) { # may have to use non-$ quoting
34 $value =~ s/\\/\\\\/g;
37 return "\$_$$\$$value\$_$$\$";
40 sub quote_phrase_value {
44 my $left_anchored = $value =~ m/^\^/;
45 my $right_anchored = $value =~ m/\$$/;
46 $value =~ s/\^// if $left_anchored;
47 $value =~ s/\$$// if $right_anchored;
48 $value = quotemeta($value);
49 $value = '^' . $value if $left_anchored;
50 $value = "$value\$" if $right_anchored;
51 return $self->quote_value($value);
58 sub default_preferred_language {
62 $self->custom_data->{default_preferred_language} = $lang if ($lang);
63 return $self->custom_data->{default_preferred_language};
66 sub default_preferred_language_multiplier {
70 $self->custom_data->{default_preferred_language_multiplier} = $lang if ($lang);
71 return $self->custom_data->{default_preferred_language_multiplier};
77 return 0 unless $self->parse_tree;
78 return 0 if @{$self->parse_tree->filters};
79 return 0 if @{$self->parse_tree->modifiers};
80 for my $node ( @{ $self->parse_tree->query_nodes } ) {
81 return 0 if (!ref($node) && $node eq '|');
82 next unless (ref($node));
83 return 0 if ($node->isa('QueryParser::query_plan'));
91 return $self->parse_tree->toSQL;
98 $self->custom_data->{dynamic_filters} ||= [];
99 push(@{$self->custom_data->{dynamic_filters}}, $new) if ($new);
100 return $self->custom_data->{dynamic_filters};
103 sub dynamic_sorters {
107 $self->custom_data->{dynamic_sorters} ||= [];
108 push(@{$self->custom_data->{dynamic_sorters}}, $new) if ($new);
109 return $self->custom_data->{dynamic_sorters};
112 sub facet_field_id_map {
116 $self->custom_data->{facet_field_id_map} ||= {};
117 $self->custom_data->{facet_field_id_map} = $map if ($map);
118 return $self->custom_data->{facet_field_id_map};
121 sub add_facet_field_id_map {
128 $self->add_facet_field( $class => $field );
129 $self->facet_field_id_map->{by_id}{$id} = { classname => $class, field => $field, weight => $weight };
130 $self->facet_field_id_map->{by_class}{$class}{$field} = $id;
133 by_id => { $id => { classname => $class, field => $field, weight => $weight } },
134 by_class => { $class => { $field => $id } }
138 sub facet_field_class_by_id {
142 return $self->facet_field_id_map->{by_id}{$id};
145 sub facet_field_ids_by_class {
150 return undef unless ($class);
153 return [$self->facet_field_id_map->{by_class}{$class}{$field}];
156 return [values( %{ $self->facet_field_id_map->{by_class}{$class} } )];
159 sub search_field_id_map {
163 $self->custom_data->{search_field_id_map} ||= {};
164 $self->custom_data->{search_field_id_map} = $map if ($map);
165 return $self->custom_data->{search_field_id_map};
168 sub add_search_field_id_map {
175 $self->add_search_field( $class => $field );
176 $self->search_field_id_map->{by_id}{$id} = { classname => $class, field => $field, weight => $weight };
177 $self->search_field_id_map->{by_class}{$class}{$field} = $id;
180 by_id => { $id => { classname => $class, field => $field, weight => $weight } },
181 by_class => { $class => { $field => $id } }
185 sub search_field_class_by_id {
189 return $self->search_field_id_map->{by_id}{$id};
192 sub search_field_ids_by_class {
197 return undef unless ($class);
200 return [$self->search_field_id_map->{by_class}{$class}{$field}];
203 return [values( %{ $self->search_field_id_map->{by_class}{$class} } )];
206 sub relevance_bumps {
210 $self->custom_data->{rel_bumps} ||= {};
211 $self->custom_data->{rel_bumps} = $bumps if ($bumps);
212 return $self->custom_data->{rel_bumps};
215 sub find_relevance_bumps {
220 return $self->relevance_bumps->{$class}{$field};
223 sub add_relevance_bump {
228 my $multiplier = shift;
231 if (defined($active) and $active eq 'f') {
237 $self->relevance_bumps->{$class}{$field}{$type} = { multiplier => $multiplier, active => $active };
239 return { $class => { $field => { $type => { multiplier => $multiplier, active => $active } } } };
243 sub initialize_search_field_id_map {
245 my $cmf_list = shift;
247 for my $cmf (@$cmf_list) {
248 __PACKAGE__->add_search_field_id_map( $cmf->field_class, $cmf->name, $cmf->id, $cmf->weight ) if ($U->is_true($cmf->search_field));
249 __PACKAGE__->add_facet_field_id_map( $cmf->field_class, $cmf->name, $cmf->id, $cmf->weight ) if ($U->is_true($cmf->facet_field));
252 return $self->search_field_id_map;
255 sub initialize_aliases {
257 my $cmsa_list = shift;
259 for my $cmsa (@$cmsa_list) {
261 __PACKAGE__->add_search_class_alias( $cmsa->field_class, $cmsa->alias );
263 my $c = $self->search_field_class_by_id( $cmsa->field );
264 __PACKAGE__->add_search_field_alias( $cmsa->field_class, $c->{field}, $cmsa->alias );
269 sub initialize_relevance_bumps {
271 my $sra_list = shift;
273 for my $sra (@$sra_list) {
274 my $c = $self->search_field_class_by_id( $sra->field );
275 __PACKAGE__->add_relevance_bump( $c->{classname}, $c->{field}, $sra->bump_type, $sra->multiplier, $sra->active );
278 return $self->relevance_bumps;
281 sub initialize_query_normalizers {
283 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" }] }
285 for my $cmfinm ( @$tree ) {
286 my $field_info = $self->search_field_class_by_id( $cmfinm->field );
287 __PACKAGE__->add_query_normalizer( $field_info->{classname}, $field_info->{field}, $cmfinm->norm->func, OpenSRF::Utils::JSON->JSON2perl($cmfinm->params) );
291 sub initialize_dynamic_filters {
293 my $list = shift; # open-ils.cstore.direct.config.record_attr_definition.search.atomic { "id" : { "!=" : null } }
295 for my $crad ( @$list ) {
296 __PACKAGE__->dynamic_filters( __PACKAGE__->add_search_filter( $crad->name ) ) if ($U->is_true($crad->filter));
297 __PACKAGE__->dynamic_sorters( $crad->name ) if ($U->is_true($crad->sorter));
301 sub initialize_filter_normalizers {
303 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" }] }
305 for my $crainm ( @$tree ) {
306 __PACKAGE__->add_filter_normalizer( $crainm->attr, $crainm->norm->func, OpenSRF::Utils::JSON->JSON2perl($crainm->params) );
311 sub initialization_complete {
319 return $_complete if ($_complete);
321 # tsearch rank normalization adjustments. see http://www.postgresql.org/docs/9.0/interactive/textsearch-controls.html#TEXTSEARCH-RANKING for details
322 $self->custom_data->{rank_cd_weight_map} = {
323 CD_logDocumentLength => 1,
324 CD_documentLength => 2,
325 CD_meanHarmonic => 4,
327 CD_logUniqueWords => 16,
331 $self->add_search_modifier( $_ ) for (keys %{ $self->custom_data->{rank_cd_weight_map} });
333 $self->initialize_search_field_id_map( $args{config_metabib_field} )
334 if ($args{config_metabib_field});
336 $self->initialize_aliases( $args{config_metabib_search_alias} )
337 if ($args{config_metabib_search_alias});
339 $self->initialize_relevance_bumps( $args{search_relevance_adjustment} )
340 if ($args{search_relevance_adjustment});
342 $self->initialize_query_normalizers( $args{config_metabib_field_index_norm_map} )
343 if ($args{config_metabib_field_index_norm_map});
345 $self->initialize_dynamic_filters( $args{config_record_attr_definition} )
346 if ($args{config_record_attr_definition});
348 $self->initialize_filter_normalizers( $args{config_record_attr_index_norm_map} )
349 if ($args{config_record_attr_index_norm_map});
352 $args{config_metabib_field_index_norm_map} &&
353 $args{search_relevance_adjustment} &&
354 $args{config_metabib_search_alias} &&
355 $args{config_metabib_field} &&
356 $args{config_record_attr_definition}
364 __PACKAGE__->add_search_field_id_map( series => seriestitle => 1 => 1 );
366 __PACKAGE__->add_search_field_id_map( series => seriestitle => 1 => 1 );
367 __PACKAGE__->add_relevance_bump( series => seriestitle => first_word => 1.5 );
368 __PACKAGE__->add_relevance_bump( series => seriestitle => full_match => 20 );
370 __PACKAGE__->add_search_field_id_map( title => abbreviated => 2 => 1 );
371 __PACKAGE__->add_relevance_bump( title => abbreviated => first_word => 1.5 );
372 __PACKAGE__->add_relevance_bump( title => abbreviated => full_match => 20 );
374 __PACKAGE__->add_search_field_id_map( title => translated => 3 => 1 );
375 __PACKAGE__->add_relevance_bump( title => translated => first_word => 1.5 );
376 __PACKAGE__->add_relevance_bump( title => translated => full_match => 20 );
378 __PACKAGE__->add_search_field_id_map( title => proper => 6 => 1 );
379 __PACKAGE__->add_query_normalizer( title => proper => 'search_normalize' );
380 __PACKAGE__->add_relevance_bump( title => proper => first_word => 1.5 );
381 __PACKAGE__->add_relevance_bump( title => proper => full_match => 20 );
382 __PACKAGE__->add_relevance_bump( title => proper => word_order => 10 );
384 __PACKAGE__->add_search_field_id_map( author => corporate => 7 => 1 );
385 __PACKAGE__->add_relevance_bump( author => corporate => first_word => 1.5 );
386 __PACKAGE__->add_relevance_bump( author => corporate => full_match => 20 );
388 __PACKAGE__->add_facet_field_id_map( author => personal => 8 => 1 );
390 __PACKAGE__->add_search_field_id_map( author => personal => 8 => 1 );
391 __PACKAGE__->add_relevance_bump( author => personal => first_word => 1.5 );
392 __PACKAGE__->add_relevance_bump( author => personal => full_match => 20 );
393 __PACKAGE__->add_query_normalizer( author => personal => 'search_normalize' );
394 __PACKAGE__->add_query_normalizer( author => personal => 'split_date_range' );
396 __PACKAGE__->add_facet_field_id_map( subject => topic => 14 => 1 );
398 __PACKAGE__->add_search_field_id_map( subject => topic => 14 => 1 );
399 __PACKAGE__->add_relevance_bump( subject => topic => first_word => 1 );
400 __PACKAGE__->add_relevance_bump( subject => topic => full_match => 1 );
402 __PACKAGE__->add_search_field_id_map( subject => complete => 16 => 1 );
403 __PACKAGE__->add_relevance_bump( subject => complete => first_word => 1 );
404 __PACKAGE__->add_relevance_bump( subject => complete => full_match => 1 );
406 __PACKAGE__->add_search_field_id_map( keyword => keyword => 15 => 1 );
407 __PACKAGE__->add_relevance_bump( keyword => keyword => first_word => 1 );
408 __PACKAGE__->add_relevance_bump( keyword => keyword => full_match => 1 );
411 __PACKAGE__->add_search_class_alias( keyword => 'kw' );
412 __PACKAGE__->add_search_class_alias( title => 'ti' );
413 __PACKAGE__->add_search_class_alias( author => 'au' );
414 __PACKAGE__->add_search_class_alias( author => 'name' );
415 __PACKAGE__->add_search_class_alias( author => 'dc.contributor' );
416 __PACKAGE__->add_search_class_alias( subject => 'su' );
417 __PACKAGE__->add_search_class_alias( subject => 'bib.subject(?:Title|Place|Occupation)' );
418 __PACKAGE__->add_search_class_alias( series => 'se' );
419 __PACKAGE__->add_search_class_alias( keyword => 'dc.identifier' );
421 __PACKAGE__->add_query_normalizer( author => corporate => 'search_normalize' );
422 __PACKAGE__->add_query_normalizer( keyword => keyword => 'search_normalize' );
424 __PACKAGE__->add_search_field_alias( subject => name => 'bib.subjectName' );
428 __PACKAGE__->default_search_class( 'keyword' );
430 # implements EG-specific stored subqueries
431 __PACKAGE__->add_search_filter( 'saved_query', sub { return __PACKAGE__->subquery_callback(@_) } );
433 # will be retained simply for back-compat
434 __PACKAGE__->add_search_filter( 'format' );
436 # grumble grumble, special cases against date1 and date2
437 __PACKAGE__->add_search_filter( 'before' );
438 __PACKAGE__->add_search_filter( 'after' );
439 __PACKAGE__->add_search_filter( 'between' );
440 __PACKAGE__->add_search_filter( 'during' );
442 # used by layers above this
443 __PACKAGE__->add_search_filter( 'statuses' );
444 __PACKAGE__->add_search_filter( 'locations' );
445 __PACKAGE__->add_search_filter( 'location_groups' );
446 __PACKAGE__->add_search_filter( 'site' );
447 __PACKAGE__->add_search_filter( 'pref_ou' );
448 __PACKAGE__->add_search_filter( 'lasso' );
449 __PACKAGE__->add_search_filter( 'my_lasso' );
450 __PACKAGE__->add_search_filter( 'depth' );
451 __PACKAGE__->add_search_filter( 'language' );
452 __PACKAGE__->add_search_filter( 'offset' );
453 __PACKAGE__->add_search_filter( 'limit' );
454 __PACKAGE__->add_search_filter( 'check_limit' );
455 __PACKAGE__->add_search_filter( 'skip_check' );
456 __PACKAGE__->add_search_filter( 'superpage' );
457 __PACKAGE__->add_search_filter( 'superpage_size' );
458 __PACKAGE__->add_search_filter( 'estimation_strategy' );
459 __PACKAGE__->add_search_modifier( 'available' );
460 __PACKAGE__->add_search_modifier( 'staff' );
462 # Start from container data (bre, acn, acp): container(bre,bookbag,123,deadb33fdeadb33fdeadb33fdeadb33f)
463 __PACKAGE__->add_search_filter( 'container' );
465 # Start from a list of record ids, either bre or metarecords, depending on the #metabib modifier
466 __PACKAGE__->add_search_filter( 'record_list' );
468 # used internally, but generally not user-settable
469 __PACKAGE__->add_search_filter( 'preferred_language' );
470 __PACKAGE__->add_search_filter( 'preferred_language_weight' );
471 __PACKAGE__->add_search_filter( 'preferred_language_multiplier' );
472 __PACKAGE__->add_search_filter( 'core_limit' );
474 # XXX Valid values to be supplied by SVF
475 __PACKAGE__->add_search_filter( 'sort' );
477 # modifies core query, not configurable
478 __PACKAGE__->add_search_modifier( 'descending' );
479 __PACKAGE__->add_search_modifier( 'ascending' );
480 __PACKAGE__->add_search_modifier( 'nullsfirst' );
481 __PACKAGE__->add_search_modifier( 'nullslast' );
482 __PACKAGE__->add_search_modifier( 'metarecord' );
483 __PACKAGE__->add_search_modifier( 'metabib' );
486 #-------------------------------
487 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan;
488 use base 'QueryParser::query_plan';
489 use OpenSRF::Utils::Logger qw($logger);
491 use OpenILS::Application::AppUtils;
492 my $apputils = "OpenILS::Application::AppUtils";
499 my ($format) = $self->find_filter('format');
501 my ($t,$f) = split('-', $format->args->[0]);
502 $self->new_filter( item_type => [ split '', $t ] ) if ($t);
503 $self->new_filter( item_form => [ split '', $f ] ) if ($f);
506 for my $f ( qw/preferred_language preferred_language_multiplier preferred_language_weight core_limit check_limit skip_check superpage superpage_size/ ) {
508 $col = 'preferred_language_multiplier' if ($f eq 'preferred_language_weight');
509 my ($filter) = $self->find_filter($f);
510 if ($filter and @{$filter->args}) {
511 $filters{$col} = $filter->args->[0];
515 $self->QueryParser->superpage($filters{superpage}) if ($filters{superpage});
516 $self->QueryParser->superpage_size($filters{superpage_size}) if ($filters{superpage_size});
517 $self->QueryParser->core_limit($filters{core_limit}) if ($filters{core_limit});
519 $logger->debug("Query plan:\n".Dumper($self));
521 my $flat_plan = $self->flatten;
523 # generate the relevance ranking
524 my $rel = "AVG(\n${spc}${spc}(" . join(")+\n${spc}${spc}(", @{$$flat_plan{rank_list}}) . ")\n${spc})+1";
526 # find any supplied sort option
527 my ($sort_filter) = $self->find_filter('sort');
529 $sort_filter = $sort_filter->args->[0];
531 $sort_filter = 'rel';
534 if (($filters{preferred_language} || $self->QueryParser->default_preferred_language) && ($filters{preferred_language_multiplier} || $self->QueryParser->default_preferred_language_multiplier)) {
535 my $pl = $self->QueryParser->quote_value( $filters{preferred_language} ? $filters{preferred_language} : $self->QueryParser->default_preferred_language );
536 my $plw = $filters{preferred_language_multiplier} ? $filters{preferred_language_multiplier} : $self->QueryParser->default_preferred_language_multiplier;
537 $rel = "($rel * COALESCE( NULLIF( FIRST(mrd.attrs \@> hstore('item_lang', $pl)), FALSE )::INT * $plw, 1))";
539 $rel = "1.0/($rel)::NUMERIC";
541 my $mra_join = 'INNER JOIN metabib.record_attr mrd ON (m.source = mrd.id';
542 $mra_join .= ' AND '. $flat_plan->{fwhere} if $flat_plan->{fwhere};
548 $desc = 'DESC' if ($self->find_modifier('descending'));
550 my $nullpos = 'NULLS LAST';
551 $nullpos = 'NULLS FIRST' if ($self->find_modifier('nullsfirst'));
553 if (grep {$_ eq $sort_filter} @{$self->QueryParser->dynamic_sorters}) {
554 $rank = "FIRST(mrd.attrs->'$sort_filter')"
555 } elsif ($sort_filter eq 'create_date') {
556 $rank = "FIRST((SELECT create_date FROM biblio.record_entry rbr WHERE rbr.id = m.source))";
557 } elsif ($sort_filter eq 'edit_date') {
558 $rank = "FIRST((SELECT edit_date FROM biblio.record_entry rbr WHERE rbr.id = m.source))";
560 # default to rel ranking
564 my $key = 'm.source';
565 $key = 'm.metarecord' if (grep {$_->name eq 'metarecord' or $_->name eq 'metabib'} @{$self->modifiers});
567 my ($before) = $self->find_filter('before');
568 my ($after) = $self->find_filter('after');
569 my ($during) = $self->find_filter('during');
570 my ($between) = $self->find_filter('between');
571 my ($container) = $self->find_filter('container');
572 my ($record_list) = $self->find_filter('record_list');
574 if ($before and @{$before->args} == 1) {
575 $before = "AND (mrd.attrs->'date1') <= " . $self->QueryParser->quote_value($before->args->[0]);
580 if ($after and @{$after->args} == 1) {
581 $after = "AND (mrd.attrs->'date1') >= " . $self->QueryParser->quote_value($after->args->[0]);
586 if ($during and @{$during->args} == 1) {
587 $during = "AND " . $self->QueryParser->quote_value($during->args->[0]) . " BETWEEN (mrd.attrs->'date1') AND (mrd.attrs->'date2')";
592 if ($between and @{$between->args} == 2) {
593 $between = "AND (mrd.attrs->'date1') BETWEEN " . $self->QueryParser->quote_value($between->args->[0]) . " AND " . $self->QueryParser->quote_value($between->args->[1]);
598 if ($container and @{$container->args} >= 3) {
599 my ($class, $ctype, $cid, $token) = @{ $container->args };
603 my $rec_field = 'ci.target_biblio_record_entry';
605 if ($class eq 'bre') {
606 $class = 'biblio_record_entry';
607 } elsif ($class eq 'acn') {
608 $class = 'call_number';
609 $rec_field = 'cn.record';
610 $rec_join = 'JOIN asset.call_number cn ON (ci.target_call_number = cn.id)';
611 } elsif ($class eq 'acp') {
613 $rec_field = 'cn.record';
614 $rec_join = 'JOIN asset.copy cp ON (ci.target_copy = cp.id) JOIN asset.call_number cn ON (cp.call_number = cn.id)';
615 # } elsif ($class eq 'au') {
617 # $rec_field = 'cn.record';
618 # $rec_join = 'JOIN asset.call_number cn ON ci.target_call_number = cn.id';
619 } else { $class = undef };
622 my ($u,$e) = $apputils->checksesperm($token) if ($token);
623 $perm_join = 'OR c.owner = ' . $u->id if ($u && !$e);
626 JOIN ( SELECT $rec_field AS container_item
627 FROM container.${class}_bucket_item ci
628 JOIN container.${class}_bucket c ON (c.id = ci.bucket)
630 WHERE c.btype = > . $self->QueryParser->quote_value($ctype) .
631 qq< AND c.id = > . $self->QueryParser->quote_value($cid) .
632 qq< AND (c.pub IS TRUE $perm_join)) container ON (container.container_item = mrd.id) >;
633 } else {$container = ''};
638 if ($record_list and @{$record_list->args} > 0) {
639 $record_list = 'JOIN (VALUES (' . join('),(', map { $self->QueryParser->quote_value($_) } @{ $record_list->args}) . ")) record_list(id) ON (record_list.id::BIGINT = $key)"
644 my $core_limit = $self->QueryParser->core_limit || 25000;
646 my $flat_where = $$flat_plan{where};
647 if ($flat_where eq '()') {
650 $flat_where = "AND $flat_where";
652 my $with = $$flat_plan{with};
653 $with= "\nWITH $with" if $with;
655 # Need an array for query parser db function; this gives a better plan
656 # than the ARRAY_AGG(DISTINCT m.source) option as of PostgreSQL 9.1
657 my $agg_records = 'ARRAY[m.source] AS records';
658 if ($key =~ /metarecord/) {
659 # metarecord searches still require the ARRAY_AGG approach
660 $agg_records = 'ARRAY_AGG(DISTINCT m.source) AS records';
669 FIRST(mrd.attrs->'date1') AS tie_break
670 FROM metabib.metarecord_source_map m
682 ORDER BY 4 $desc $nullpos, 5 DESC $nullpos, 3 DESC
686 warn $sql if $self->QueryParser->debug;
696 my $multiplier = shift;
698 my $only_atoms = $node->only_atoms;
699 return '' if (!@$only_atoms);
701 if ($bump eq 'first_word') {
702 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)";
703 } elsif ($bump eq 'full_match') {
704 return " /* full_match */ COALESCE(NULLIF( (search_normalize(".$node->table_alias.".value) ~ ('^'||".
705 join( "||' '||", map { "search_normalize(".$self->QueryParser->quote_phrase_value($_->content).")" } @$only_atoms )."||'\$')), FALSE )::INT * $multiplier, 1)";
706 } elsif ($bump eq 'word_order') {
707 return " /* word_order */ COALESCE(NULLIF( (search_normalize(".$node->table_alias.".value) ~ (".
708 join( "||'.*'||", map { "search_normalize(".$self->QueryParser->quote_phrase_value($_->content).")" } @$only_atoms ).")), FALSE )::INT * $multiplier, 1)";
717 my $from = shift || '';
718 my $where = shift || '(';
720 my $fwhere = shift || ''; # for joining dynamic filters (mra)
723 for my $filter (@{$self->filters}) {
724 push(@dyn_filters, $filter) if
725 grep { $_ eq $filter->name }
726 @{ $self->QueryParser->dynamic_filters };
730 for my $node ( @{$self->query_nodes} ) {
733 if ($node->isa( 'QueryParser::query_plan::node' )) {
735 unless (@{$node->only_atoms}) {
736 push @rank_list, '1';
741 my $table = $node->table;
742 my $talias = $node->table_alias;
744 my $node_rank = 'COALESCE(' . $node->rank . " * ${talias}.weight, 0.0)";
746 my $core_limit = $self->QueryParser->core_limit || 25000;
747 $from .= "\n${spc}LEFT JOIN (\n${spc}${spc}SELECT fe.*, fe_weight.weight, ${talias}_xq.tsq /* search */\n${spc}${spc} FROM $table AS fe";
748 $from .= "\n${spc}${spc}${spc}JOIN config.metabib_field AS fe_weight ON (fe_weight.id = fe.field)";
750 if ($node->dummy_count < @{$node->only_atoms} ) {
751 $with .= ",\n" if $with;
752 $with .= "${talias}_xq AS (SELECT ". $node->tsquery ." AS tsq )";
753 $from .= "\n${spc}${spc}${spc}JOIN ${talias}_xq ON (fe.index_vector @@ ${talias}_xq.tsq)";
755 $from .= "\n${spc}${spc}${spc}, (SELECT NULL::tsquery AS tsq ) AS x";
759 if (@{$node->fields} > 0) {
760 @bump_fields = @{$node->fields};
762 my @field_ids = grep defined, (
764 $self->QueryParser->search_field_ids_by_class(
770 $from .= "\n${spc}${spc}${spc}WHERE fe_weight.id IN (" .
771 join(',', @field_ids) . ")";
775 @bump_fields = @{$self->QueryParser->search_fields->{$node->classname}};
778 ###$from .= "\n${spc}${spc}LIMIT $core_limit";
779 $from .= "\n${spc}) AS $talias ON (m.source = ${talias}.source)";
783 for my $field ( @bump_fields ) {
784 my $bumps = $self->QueryParser->find_relevance_bumps( $node->classname => $field );
785 for my $b (keys %$bumps) {
786 next if (!$$bumps{$b}{active});
787 next if ($used_bumps{$b});
790 next if ($$bumps{$b}{multiplier} == 1); # optimization to remove unneeded bumps
792 my $bump_case = $self->rel_bump( $node, $b, $$bumps{$b}{multiplier} );
793 $node_rank .= "\n${spc}${spc}${spc}${spc} * " . $bump_case if ($bump_case);
798 my $twhere .= '(' . $talias . ".id IS NOT NULL";
799 $twhere .= ' AND ' . join(' AND ', map {"${talias}.value ~* ".$self->QueryParser->quote_phrase_value($_)} @{$node->phrases}) if (@{$node->phrases});
800 $twhere .= ' AND ' . join(' AND ', map {"${talias}.value !~* ".$self->QueryParser->quote_phrase_value($_)} @{$node->unphrases}) if (@{$node->unphrases});
803 if (@dyn_filters or !$self->top_plan) {
804 # if this WHERE is represented within the dynamic
805 # filter's ON clause, it's not also needed in the main WHERE.
811 push @rank_list, $node_rank;
813 } elsif ($node->isa( 'QueryParser::query_plan::facet' )) {
815 my $table = $node->table;
816 my $talias = $node->table_alias;
819 if (@{$node->fields} > 0) {
820 push(@field_ids, $self->QueryParser->facet_field_ids_by_class( $node->classname, $_ )->[0]) for (@{$node->fields});
822 @field_ids = @{ $self->QueryParser->facet_field_ids_by_class( $node->classname ) };
825 my $join_type = ($node->negate or @dyn_filters or !$self->top_plan) ? 'LEFT' : 'INNER';
826 $from .= "\n${spc}$join_type JOIN /* facet */ metabib.facet_entry $talias ON (\n${spc}${spc}m.source = ${talias}.source\n${spc}${spc}".
827 "AND SUBSTRING(${talias}.value,1,1024) IN (" . join(",", map { $self->QueryParser->quote_value($_) } @{$node->values}) . ")\n${spc}${spc}".
828 "AND ${talias}.field IN (". join(',', @field_ids) . ")\n${spc})";
830 if (@dyn_filters or !$self->top_plan) {
831 my $NOT = $node->negate ? '' : ' NOT';
832 $fwhere .= "${talias}.id IS$NOT NULL";
834 $where .= $node->negate ? "${talias}.id IS NULL" : 'TRUE';
838 my $subnode = $node->flatten;
840 # strip the trailing bool from the previous loop if there is
841 # nothing to add to the where/fwhere within this loop.
842 if ($$subnode{where} eq '()') {
843 $where =~ s/\s(AND|OR)\s$//;
845 } elsif ($$subnode{fwhere} eq '') {
846 $fwhere =~ s/\s(AND|OR)\s$//;
849 push(@rank_list, @{$$subnode{rank_list}});
850 $from .= $$subnode{from};
852 $where .= "($$subnode{where})" unless $$subnode{where} eq '()';
853 $fwhere .= "($$subnode{fwhere})" if $$subnode{fwhere};
855 if ($$subnode{with}) {
856 $with .= ', ' if $with;
857 $with .= " " . $$subnode{with};
862 warn "flatten(): appending WHERE bool to: $where\n" if $self->QueryParser->debug;
865 # bool joiner for inter-plan filters
866 $fwhere .= ' AND ' if ($node eq '&');
867 $fwhere .= ' OR ' if ($node eq '|');
869 } elsif ($where ne '(') {
871 $where .= ' AND ' if ($node eq '&');
872 $where .= ' OR ' if ($node eq '|');
877 # for each dynamic filter, build the ON clause for the JOIN
878 for my $filter (@dyn_filters) {
880 warn "flatten(): processing dynamic filter ". $filter->name ."\n"
881 if $self->QueryParser->debug;
883 # bool joiner for intra-plan nodes/filters
884 $fwhere .= sprintf(" %s ", ($self->joiner eq '&' ? 'AND' : 'OR')) if $fwhere;
886 my @fargs = @{$filter->args};
887 my $NOT = $filter->negate ? ' NOT' : '';
888 my $fname = $filter->name;
889 $fname = 'item_lang' if $fname eq 'language'; #XXX filter aliases
892 "attrs->'%s'$NOT IN (%s)", $fname,
893 join(',', map { $self->QueryParser->quote_value($_) } @fargs)
896 warn "flatten(): filter where => $fwhere\n"
897 if $self->QueryParser->debug;
899 warn "flatten(): full filter where => $fwhere\n" if $self->QueryParser->debug;
901 return { rank_list => \@rank_list, from => $from, where => $where.')', with => $with, fwhere => $fwhere };
905 #-------------------------------
906 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::filter;
907 use base 'QueryParser::query_plan::filter';
909 #-------------------------------
910 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::facet;
911 use base 'QueryParser::query_plan::facet';
915 my ($classname) = split '\|', $self->name;
921 return 'metabib.' . $self->classname . '_field_entry';
926 my ($classname,@fields) = split '\|', $self->name;
933 my $table_alias = "$self";
934 $table_alias =~ s/^.*\(0(x[0-9a-fA-F]+)\)$/$1/go;
935 $table_alias .= '_' . $self->name;
936 $table_alias =~ s/\|/_/go;
942 #-------------------------------
943 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::modifier;
944 use base 'QueryParser::query_plan::modifier';
946 #-------------------------------
947 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::node::atom;
948 use base 'QueryParser::query_plan::node::atom';
954 $self->{sql} = $sql if ($sql);
956 return $self->{sql} if ($self->{sql});
957 return $self->buildSQL;
963 my $classname = $self->node->classname;
965 return $self->sql("to_tsquery('$classname','')") if $self->{dummy};
967 my $normalizers = $self->node->plan->QueryParser->query_normalizers( $classname );
968 my $fields = $self->node->fields;
970 $fields = $self->node->plan->QueryParser->search_fields->{$classname} if (!@$fields);
974 for my $field (@$fields) {
975 for my $nfield (keys %$normalizers) {
976 for my $nizer ( @{$$normalizers{$nfield}} ) {
977 if ($field eq $nfield) {
978 my $param_string = OpenSRF::Utils::JSON->perl2JSON($nizer->{params});
979 if (!exists($norms{$nizer->{function}.$param_string})) {
980 $norms{$nizer->{function}.$param_string} = {p=>$pos++,n=>$nizer};
987 my $sql = $self->node->plan->QueryParser->quote_value($self->content);
989 for my $n ( map { $$_{n} } sort { $$a{p} <=> $$b{p} } values %norms ) {
990 $sql = join(', ', $sql, map { $self->node->plan->QueryParser->quote_value($_) } @{ $n->{params} });
991 $sql = $n->{function}."($sql)";
994 my $prefix = $self->prefix || '';
995 my $suffix = $self->suffix || '';
997 $prefix = "'$prefix' ||" if $prefix;
999 my $suffix_after = '';
1001 $suffix_op = ":$suffix" if $suffix;
1002 $suffix_after = "|| '$suffix_op'" if $suffix;
1004 $sql = "to_tsquery('$classname', COALESCE(NULLIF($prefix '(' || btrim(regexp_replace($sql,E'(?:\\\\s+|:)','$suffix_op&','g'),'&|') $suffix_after || ')', '()'), ''))";
1006 return $self->sql($sql);
1009 #-------------------------------
1010 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::node;
1011 use base 'QueryParser::query_plan::node';
1016 $self->{dummy_count} = 0;
1018 my $atoms = $self->query_atoms;
1020 for my $a (@$atoms) {
1021 push(@only_atoms, $a) if (ref($a) && $a->isa('QueryParser::query_plan::node::atom'));
1022 $self->{dummy_count}++ if (ref($a) && $a->{dummy});
1025 return \@only_atoms;
1030 return $self->{dummy_count};
1036 $self->{table} = $table if ($table);
1037 return $self->{table} if $self->{table};
1038 return $self->table( 'metabib.' . $self->classname . '_field_entry' );
1043 my $table_alias = shift;
1044 $self->{table_alias} = $table_alias if ($table_alias);
1045 return $self->{table_alias} if ($self->{table_alias});
1047 $table_alias = "$self";
1048 $table_alias =~ s/^.*\(0(x[0-9a-fA-F]+)\)$/$1/go;
1049 $table_alias .= '_' . $self->requested_class;
1050 $table_alias =~ s/\|/_/go;
1052 return $self->table_alias( $table_alias );
1057 return $self->{tsquery} if ($self->{tsquery});
1059 for my $atom (@{$self->query_atoms}) {
1061 $self->{tsquery} .= "\n${spc}${spc}${spc}" .$atom->sql;
1063 $self->{tsquery} .= $atom x 2;
1067 return $self->{tsquery};
1073 my $rank_norm_map = $self->plan->QueryParser->custom_data->{rank_cd_weight_map};
1075 my $cover_density = 0;
1076 for my $norm ( keys %$rank_norm_map) {
1077 $cover_density += $$rank_norm_map{$norm} if ($self->plan->find_modifier($norm));
1080 return $self->{rank} if ($self->{rank});
1081 return $self->{rank} = 'ts_rank_cd(' . $self->table_alias . '.index_vector, ' . $self->table_alias . ".tsq, $cover_density)";