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 $U = 'OpenILS::Application::AppUtils';
17 if ($value =~ /^\d/) { # may have to use non-$ quoting
19 $value =~ s/\\/\\\\/g;
22 return "\$_$$\$$value\$_$$\$";
25 sub quote_phrase_value {
29 my $left_anchored = $value =~ m/^\^/;
30 my $right_anchored = $value =~ m/\$$/;
31 $value =~ s/\^// if $left_anchored;
32 $value =~ s/\$$// if $right_anchored;
33 $value = quotemeta($value);
34 $value = '^' . $value if $left_anchored;
35 $value = "$value\$" if $right_anchored;
36 return $self->quote_value($value);
43 sub default_preferred_language {
47 $self->custom_data->{default_preferred_language} = $lang if ($lang);
48 return $self->custom_data->{default_preferred_language};
51 sub default_preferred_language_multiplier {
55 $self->custom_data->{default_preferred_language_multiplier} = $lang if ($lang);
56 return $self->custom_data->{default_preferred_language_multiplier};
62 return 0 unless $self->parse_tree;
63 return 0 if @{$self->parse_tree->filters};
64 return 0 if @{$self->parse_tree->modifiers};
65 for my $node ( @{ $self->parse_tree->query_nodes } ) {
66 return 0 if (!ref($node) && $node eq '|');
67 next unless (ref($node));
68 return 0 if ($node->isa('QueryParser::query_plan'));
76 return $self->parse_tree->toSQL;
83 $self->custom_data->{dynamic_filters} ||= [];
84 push(@{$self->custom_data->{dynamic_filters}}, $new) if ($new);
85 return $self->custom_data->{dynamic_filters};
92 $self->custom_data->{dynamic_sorters} ||= [];
93 push(@{$self->custom_data->{dynamic_sorters}}, $new) if ($new);
94 return $self->custom_data->{dynamic_sorters};
97 sub facet_field_id_map {
101 $self->custom_data->{facet_field_id_map} ||= {};
102 $self->custom_data->{facet_field_id_map} = $map if ($map);
103 return $self->custom_data->{facet_field_id_map};
106 sub add_facet_field_id_map {
113 $self->add_facet_field( $class => $field );
114 $self->facet_field_id_map->{by_id}{$id} = { classname => $class, field => $field, weight => $weight };
115 $self->facet_field_id_map->{by_class}{$class}{$field} = $id;
118 by_id => { $id => { classname => $class, field => $field, weight => $weight } },
119 by_class => { $class => { $field => $id } }
123 sub facet_field_class_by_id {
127 return $self->facet_field_id_map->{by_id}{$id};
130 sub facet_field_ids_by_class {
135 return undef unless ($class);
138 return [$self->facet_field_id_map->{by_class}{$class}{$field}];
141 return [values( %{ $self->facet_field_id_map->{by_class}{$class} } )];
144 sub search_field_id_map {
148 $self->custom_data->{search_field_id_map} ||= {};
149 $self->custom_data->{search_field_id_map} = $map if ($map);
150 return $self->custom_data->{search_field_id_map};
153 sub add_search_field_id_map {
160 $self->add_search_field( $class => $field );
161 $self->search_field_id_map->{by_id}{$id} = { classname => $class, field => $field, weight => $weight };
162 $self->search_field_id_map->{by_class}{$class}{$field} = $id;
165 by_id => { $id => { classname => $class, field => $field, weight => $weight } },
166 by_class => { $class => { $field => $id } }
170 sub search_field_class_by_id {
174 return $self->search_field_id_map->{by_id}{$id};
177 sub search_field_ids_by_class {
182 return undef unless ($class);
185 return [$self->search_field_id_map->{by_class}{$class}{$field}];
188 return [values( %{ $self->search_field_id_map->{by_class}{$class} } )];
191 sub relevance_bumps {
195 $self->custom_data->{rel_bumps} ||= {};
196 $self->custom_data->{rel_bumps} = $bumps if ($bumps);
197 return $self->custom_data->{rel_bumps};
200 sub find_relevance_bumps {
205 return $self->relevance_bumps->{$class}{$field};
208 sub add_relevance_bump {
213 my $multiplier = shift;
216 if (defined($active) and $active eq 'f') {
222 $self->relevance_bumps->{$class}{$field}{$type} = { multiplier => $multiplier, active => $active };
224 return { $class => { $field => { $type => { multiplier => $multiplier, active => $active } } } };
228 sub initialize_search_field_id_map {
230 my $cmf_list = shift;
232 for my $cmf (@$cmf_list) {
233 __PACKAGE__->add_search_field_id_map( $cmf->field_class, $cmf->name, $cmf->id, $cmf->weight ) if ($U->is_true($cmf->search_field));
234 __PACKAGE__->add_facet_field_id_map( $cmf->field_class, $cmf->name, $cmf->id, $cmf->weight ) if ($U->is_true($cmf->facet_field));
237 return $self->search_field_id_map;
240 sub initialize_aliases {
242 my $cmsa_list = shift;
244 for my $cmsa (@$cmsa_list) {
246 __PACKAGE__->add_search_class_alias( $cmsa->field_class, $cmsa->alias );
248 my $c = $self->search_field_class_by_id( $cmsa->field );
249 __PACKAGE__->add_search_field_alias( $cmsa->field_class, $c->{field}, $cmsa->alias );
254 sub initialize_relevance_bumps {
256 my $sra_list = shift;
258 for my $sra (@$sra_list) {
259 my $c = $self->search_field_class_by_id( $sra->field );
260 __PACKAGE__->add_relevance_bump( $c->{classname}, $c->{field}, $sra->bump_type, $sra->multiplier, $sra->active );
263 return $self->relevance_bumps;
266 sub initialize_query_normalizers {
268 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" }] }
270 for my $cmfinm ( @$tree ) {
271 my $field_info = $self->search_field_class_by_id( $cmfinm->field );
272 __PACKAGE__->add_query_normalizer( $field_info->{classname}, $field_info->{field}, $cmfinm->norm->func, OpenSRF::Utils::JSON->JSON2perl($cmfinm->params) );
276 sub initialize_dynamic_filters {
278 my $list = shift; # open-ils.cstore.direct.config.record_attr_definition.search.atomic { "id" : { "!=" : null } }
280 for my $crad ( @$list ) {
281 __PACKAGE__->dynamic_filters( __PACKAGE__->add_search_filter( $crad->name ) ) if ($U->is_true($crad->filter));
282 __PACKAGE__->dynamic_sorters( $crad->name ) if ($U->is_true($crad->sorter));
286 sub initialize_filter_normalizers {
288 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" }] }
290 for my $crainm ( @$tree ) {
291 __PACKAGE__->add_filter_normalizer( $crainm->attr, $crainm->norm->func, OpenSRF::Utils::JSON->JSON2perl($crainm->params) );
296 sub initialization_complete {
304 return $_complete if ($_complete);
306 # tsearch rank normalization adjustments. see http://www.postgresql.org/docs/9.0/interactive/textsearch-controls.html#TEXTSEARCH-RANKING for details
307 $self->custom_data->{rank_cd_weight_map} = {
308 CD_logDocumentLength => 1,
309 CD_documentLength => 2,
310 CD_meanHarmonic => 4,
312 CD_logUniqueWords => 16,
316 $self->add_search_modifier( $_ ) for (keys %{ $self->custom_data->{rank_cd_weight_map} });
318 $self->initialize_search_field_id_map( $args{config_metabib_field} )
319 if ($args{config_metabib_field});
321 $self->initialize_aliases( $args{config_metabib_search_alias} )
322 if ($args{config_metabib_search_alias});
324 $self->initialize_relevance_bumps( $args{search_relevance_adjustment} )
325 if ($args{search_relevance_adjustment});
327 $self->initialize_query_normalizers( $args{config_metabib_field_index_norm_map} )
328 if ($args{config_metabib_field_index_norm_map});
330 $self->initialize_dynamic_filters( $args{config_record_attr_definition} )
331 if ($args{config_record_attr_definition});
333 $self->initialize_filter_normalizers( $args{config_record_attr_index_norm_map} )
334 if ($args{config_record_attr_index_norm_map});
337 $args{config_metabib_field_index_norm_map} &&
338 $args{search_relevance_adjustment} &&
339 $args{config_metabib_search_alias} &&
340 $args{config_metabib_field} &&
341 $args{config_record_attr_definition}
349 __PACKAGE__->add_search_field_id_map( series => seriestitle => 1 => 1 );
351 __PACKAGE__->add_search_field_id_map( series => seriestitle => 1 => 1 );
352 __PACKAGE__->add_relevance_bump( series => seriestitle => first_word => 1.5 );
353 __PACKAGE__->add_relevance_bump( series => seriestitle => full_match => 20 );
355 __PACKAGE__->add_search_field_id_map( title => abbreviated => 2 => 1 );
356 __PACKAGE__->add_relevance_bump( title => abbreviated => first_word => 1.5 );
357 __PACKAGE__->add_relevance_bump( title => abbreviated => full_match => 20 );
359 __PACKAGE__->add_search_field_id_map( title => translated => 3 => 1 );
360 __PACKAGE__->add_relevance_bump( title => translated => first_word => 1.5 );
361 __PACKAGE__->add_relevance_bump( title => translated => full_match => 20 );
363 __PACKAGE__->add_search_field_id_map( title => proper => 6 => 1 );
364 __PACKAGE__->add_query_normalizer( title => proper => 'search_normalize' );
365 __PACKAGE__->add_relevance_bump( title => proper => first_word => 1.5 );
366 __PACKAGE__->add_relevance_bump( title => proper => full_match => 20 );
367 __PACKAGE__->add_relevance_bump( title => proper => word_order => 10 );
369 __PACKAGE__->add_search_field_id_map( author => corporate => 7 => 1 );
370 __PACKAGE__->add_relevance_bump( author => corporate => first_word => 1.5 );
371 __PACKAGE__->add_relevance_bump( author => corporate => full_match => 20 );
373 __PACKAGE__->add_facet_field_id_map( author => personal => 8 => 1 );
375 __PACKAGE__->add_search_field_id_map( author => personal => 8 => 1 );
376 __PACKAGE__->add_relevance_bump( author => personal => first_word => 1.5 );
377 __PACKAGE__->add_relevance_bump( author => personal => full_match => 20 );
378 __PACKAGE__->add_query_normalizer( author => personal => 'search_normalize' );
379 __PACKAGE__->add_query_normalizer( author => personal => 'split_date_range' );
381 __PACKAGE__->add_facet_field_id_map( subject => topic => 14 => 1 );
383 __PACKAGE__->add_search_field_id_map( subject => topic => 14 => 1 );
384 __PACKAGE__->add_relevance_bump( subject => topic => first_word => 1 );
385 __PACKAGE__->add_relevance_bump( subject => topic => full_match => 1 );
387 __PACKAGE__->add_search_field_id_map( subject => complete => 16 => 1 );
388 __PACKAGE__->add_relevance_bump( subject => complete => first_word => 1 );
389 __PACKAGE__->add_relevance_bump( subject => complete => full_match => 1 );
391 __PACKAGE__->add_search_field_id_map( keyword => keyword => 15 => 1 );
392 __PACKAGE__->add_relevance_bump( keyword => keyword => first_word => 1 );
393 __PACKAGE__->add_relevance_bump( keyword => keyword => full_match => 1 );
396 __PACKAGE__->add_search_class_alias( keyword => 'kw' );
397 __PACKAGE__->add_search_class_alias( title => 'ti' );
398 __PACKAGE__->add_search_class_alias( author => 'au' );
399 __PACKAGE__->add_search_class_alias( author => 'name' );
400 __PACKAGE__->add_search_class_alias( author => 'dc.contributor' );
401 __PACKAGE__->add_search_class_alias( subject => 'su' );
402 __PACKAGE__->add_search_class_alias( subject => 'bib.subject(?:Title|Place|Occupation)' );
403 __PACKAGE__->add_search_class_alias( series => 'se' );
404 __PACKAGE__->add_search_class_alias( keyword => 'dc.identifier' );
406 __PACKAGE__->add_query_normalizer( author => corporate => 'search_normalize' );
407 __PACKAGE__->add_query_normalizer( keyword => keyword => 'search_normalize' );
409 __PACKAGE__->add_search_field_alias( subject => name => 'bib.subjectName' );
413 __PACKAGE__->default_search_class( 'keyword' );
415 # will be retained simply for back-compat
416 __PACKAGE__->add_search_filter( 'format' );
418 # grumble grumble, special cases against date1 and date2
419 __PACKAGE__->add_search_filter( 'before' );
420 __PACKAGE__->add_search_filter( 'after' );
421 __PACKAGE__->add_search_filter( 'between' );
422 __PACKAGE__->add_search_filter( 'during' );
424 # used by layers above this
425 __PACKAGE__->add_search_filter( 'statuses' );
426 __PACKAGE__->add_search_filter( 'locations' );
427 __PACKAGE__->add_search_filter( 'location_groups' );
428 __PACKAGE__->add_search_filter( 'site' );
429 __PACKAGE__->add_search_filter( 'pref_ou' );
430 __PACKAGE__->add_search_filter( 'lasso' );
431 __PACKAGE__->add_search_filter( 'my_lasso' );
432 __PACKAGE__->add_search_filter( 'depth' );
433 __PACKAGE__->add_search_filter( 'language' );
434 __PACKAGE__->add_search_filter( 'offset' );
435 __PACKAGE__->add_search_filter( 'limit' );
436 __PACKAGE__->add_search_filter( 'check_limit' );
437 __PACKAGE__->add_search_filter( 'skip_check' );
438 __PACKAGE__->add_search_filter( 'superpage' );
439 __PACKAGE__->add_search_filter( 'superpage_size' );
440 __PACKAGE__->add_search_filter( 'estimation_strategy' );
441 __PACKAGE__->add_search_modifier( 'available' );
442 __PACKAGE__->add_search_modifier( 'staff' );
444 # Start from container data (bre, acn, acp): container(bre,bookbag,123,deadb33fdeadb33fdeadb33fdeadb33f)
445 __PACKAGE__->add_search_filter( 'container' );
447 # Start from a list of record ids, either bre or metarecords, depending on the #metabib modifier
448 __PACKAGE__->add_search_filter( 'record_list' );
450 # used internally, but generally not user-settable
451 __PACKAGE__->add_search_filter( 'preferred_language' );
452 __PACKAGE__->add_search_filter( 'preferred_language_weight' );
453 __PACKAGE__->add_search_filter( 'preferred_language_multiplier' );
454 __PACKAGE__->add_search_filter( 'core_limit' );
456 # XXX Valid values to be supplied by SVF
457 __PACKAGE__->add_search_filter( 'sort' );
459 # modifies core query, not configurable
460 __PACKAGE__->add_search_modifier( 'descending' );
461 __PACKAGE__->add_search_modifier( 'ascending' );
462 __PACKAGE__->add_search_modifier( 'nullsfirst' );
463 __PACKAGE__->add_search_modifier( 'nullslast' );
464 __PACKAGE__->add_search_modifier( 'metarecord' );
465 __PACKAGE__->add_search_modifier( 'metabib' );
468 #-------------------------------
469 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan;
470 use base 'QueryParser::query_plan';
471 use OpenSRF::Utils::Logger qw($logger);
473 use OpenILS::Application::AppUtils;
474 my $apputils = "OpenILS::Application::AppUtils";
481 my ($format) = $self->find_filter('format');
483 my ($t,$f) = split('-', $format->args->[0]);
484 $self->new_filter( item_type => [ split '', $t ] ) if ($t);
485 $self->new_filter( item_form => [ split '', $f ] ) if ($f);
488 for my $f ( qw/preferred_language preferred_language_multiplier preferred_language_weight core_limit check_limit skip_check superpage superpage_size/ ) {
490 $col = 'preferred_language_multiplier' if ($f eq 'preferred_language_weight');
491 my ($filter) = $self->find_filter($f);
492 if ($filter and @{$filter->args}) {
493 $filters{$col} = $filter->args->[0];
497 $self->QueryParser->superpage($filters{superpage}) if ($filters{superpage});
498 $self->QueryParser->superpage_size($filters{superpage_size}) if ($filters{superpage_size});
499 $self->QueryParser->core_limit($filters{core_limit}) if ($filters{core_limit});
501 $logger->debug("Query plan:\n".Dumper($self));
503 my $flat_plan = $self->flatten;
505 # generate the relevance ranking
506 my $rel = "AVG(\n${spc}${spc}(" . join(")+\n${spc}${spc}(", @{$$flat_plan{rank_list}}) . ")\n${spc})+1";
508 # find any supplied sort option
509 my ($sort_filter) = $self->find_filter('sort');
511 $sort_filter = $sort_filter->args->[0];
513 $sort_filter = 'rel';
516 if (($filters{preferred_language} || $self->QueryParser->default_preferred_language) && ($filters{preferred_language_multiplier} || $self->QueryParser->default_preferred_language_multiplier)) {
517 my $pl = $self->QueryParser->quote_value( $filters{preferred_language} ? $filters{preferred_language} : $self->QueryParser->default_preferred_language );
518 my $plw = $filters{preferred_language_multiplier} ? $filters{preferred_language_multiplier} : $self->QueryParser->default_preferred_language_multiplier;
519 $rel = "($rel * COALESCE( NULLIF( FIRST(mrd.attrs \@> hstore('item_lang', $pl)), FALSE )::INT * $plw, 1))";
521 $rel = "1.0/($rel)::NUMERIC";
523 my %dyn_filters = ( '' => [] ); # the "catch-all" key
524 for my $f ( @{ $self->QueryParser->dynamic_filters } ) {
526 $col = 'item_lang' if ($f eq 'language'); #XXX filter aliases would address this ... booo ... later
528 my ($filter) = $self->find_filter($f);
530 my @fargs = @{$filter->args};
532 if (@fargs > 1 || $filter->negate) {
533 my $NOT = $filter->negate ? 'NOT' : '';
534 $dyn_filters{$f} = "$NOT( " .
537 map { "mrd.attrs \@> hstore('$col', " . $self->QueryParser->quote_value($_) . ")" } @fargs
541 push(@{$dyn_filters{''}}, "hstore('$col', " . $self->QueryParser->quote_value($fargs[0]) . ")");
546 my $combined_dyn_filters = '';
547 $combined_dyn_filters .= 'AND mrd.attrs @> (' . join(' || ', @{$dyn_filters{''}}) . ') ' if (@{$dyn_filters{''}});
548 delete($dyn_filters{''});
550 my @dyn_filter_list = values(%dyn_filters);
551 $combined_dyn_filters .= 'AND ' . join(' AND ', @dyn_filter_list) if (@dyn_filter_list);
556 $desc = 'DESC' if ($self->find_modifier('descending'));
558 my $nullpos = 'NULLS LAST';
559 $nullpos = 'NULLS FIRST' if ($self->find_modifier('nullsfirst'));
561 if (grep {$_ eq $sort_filter} @{$self->QueryParser->dynamic_sorters}) {
562 $rank = "FIRST(mrd.attrs->'$sort_filter')"
563 } elsif ($sort_filter eq 'create_date') {
564 $rank = "FIRST((SELECT create_date FROM biblio.record_entry rbr WHERE rbr.id = m.source))";
565 } elsif ($sort_filter eq 'edit_date') {
566 $rank = "FIRST((SELECT edit_date FROM biblio.record_entry rbr WHERE rbr.id = m.source))";
568 # default to rel ranking
572 my $key = 'm.source';
573 $key = 'm.metarecord' if (grep {$_->name eq 'metarecord' or $_->name eq 'metabib'} @{$self->modifiers});
575 my ($before) = $self->find_filter('before');
576 my ($after) = $self->find_filter('after');
577 my ($during) = $self->find_filter('during');
578 my ($between) = $self->find_filter('between');
579 my ($container) = $self->find_filter('container');
580 my ($record_list) = $self->find_filter('record_list');
582 if ($before and @{$before->args} == 1) {
583 $before = "AND (mrd.attrs->'date1') <= " . $self->QueryParser->quote_value($before->args->[0]);
588 if ($after and @{$after->args} == 1) {
589 $after = "AND (mrd.attrs->'date1') >= " . $self->QueryParser->quote_value($after->args->[0]);
594 if ($during and @{$during->args} == 1) {
595 $during = "AND " . $self->QueryParser->quote_value($during->args->[0]) . " BETWEEN (mrd.attrs->'date1') AND (mrd.attrs->'date2')";
600 if ($between and @{$between->args} == 2) {
601 $between = "AND (mrd.attrs->'date1') BETWEEN " . $self->QueryParser->quote_value($between->args->[0]) . " AND " . $self->QueryParser->quote_value($between->args->[1]);
606 if ($container and @{$container->args} >= 3) {
607 my ($class, $ctype, $cid, $token) = @{ $container->args };
611 my $rec_field = 'ci.target_biblio_record_entry';
613 if ($class eq 'bre') {
614 $class = 'biblio_record_entry';
615 } elsif ($class eq 'acn') {
616 $class = 'call_number';
617 $rec_field = 'cn.record';
618 $rec_join = 'JOIN asset.call_number cn ON (ci.target_call_number = cn.id)';
619 } elsif ($class eq 'acp') {
621 $rec_field = 'cn.record';
622 $rec_join = 'JOIN asset.copy cp ON (ci.target_copy = cp.id) JOIN asset.call_number cn ON (cp.call_number = cn.id)';
623 # } elsif ($class eq 'au') {
625 # $rec_field = 'cn.record';
626 # $rec_join = 'JOIN asset.call_number cn ON ci.target_call_number = cn.id';
627 } else { $class = undef };
630 my ($u,$e) = $apputils->checksesperm($token) if ($token);
631 $perm_join = 'OR c.owner = ' . $u->id if ($u && !$e);
634 JOIN ( SELECT $rec_field AS container_item
635 FROM container.${class}_bucket_item ci
636 JOIN container.${class}_bucket c ON (c.id = ci.bucket)
638 WHERE c.btype = > . $self->QueryParser->quote_value($ctype) .
639 qq< AND c.id = > . $self->QueryParser->quote_value($cid) .
640 qq< AND (c.pub IS TRUE $perm_join)) container ON (container.container_item = mrd.id) >;
641 } else {$container = ''};
646 if ($record_list and @{$record_list->args} > 0) {
647 $record_list = 'JOIN (VALUES (' . join('),(', map { $self->QueryParser->quote_value($_) } @{ $record_list->args}) . ")) record_list(id) ON (record_list.id::BIGINT = $key)"
652 my $core_limit = $self->QueryParser->core_limit || 25000;
654 my $flat_where = $$flat_plan{where};
655 if ($flat_where eq '()') {
658 $flat_where = "AND $flat_where";
660 my $with = $$flat_plan{with};
661 $with= "\nWITH $with" if $with;
663 # Need an array for query parser db function; this gives a better plan
664 # than the ARRAY_AGG(DISTINCT m.source) option as of PostgreSQL 9.1
665 my $agg_records = 'ARRAY[m.source] AS records';
666 if ($key =~ /metarecord/) {
667 # metarecord searches still require the ARRAY_AGG approach
668 $agg_records = 'ARRAY_AGG(DISTINCT m.source) AS records';
677 FIRST(mrd.attrs->'date1') AS tie_break
678 FROM metabib.metarecord_source_map m
679 JOIN metabib.record_attr mrd ON (m.source = mrd.id)
688 $combined_dyn_filters
691 ORDER BY 4 $desc $nullpos, 5 DESC $nullpos, 3 DESC
695 warn $sql if $self->QueryParser->debug;
705 my $multiplier = shift;
707 my $only_atoms = $node->only_atoms;
708 return '' if (!@$only_atoms);
710 if ($bump eq 'first_word') {
711 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)";
712 } elsif ($bump eq 'full_match') {
713 return " /* full_match */ COALESCE(NULLIF( (search_normalize(".$node->table_alias.".value) ~ ('^'||".
714 join( "||' '||", map { "search_normalize(".$self->QueryParser->quote_phrase_value($_->content).")" } @$only_atoms )."||'\$')), FALSE )::INT * $multiplier, 1)";
715 } elsif ($bump eq 'word_order') {
716 return " /* word_order */ COALESCE(NULLIF( (search_normalize(".$node->table_alias.".value) ~ (".
717 join( "||'.*'||", map { "search_normalize(".$self->QueryParser->quote_phrase_value($_->content).")" } @$only_atoms ).")), FALSE )::INT * $multiplier, 1)";
726 my $from = shift || '';
727 my $where = shift || '(';
731 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);
797 $where .= '(' . $talias . ".id IS NOT NULL";
798 $where .= ' AND ' . join(' AND ', map {"${talias}.value ~* ".$self->QueryParser->quote_phrase_value($_)} @{$node->phrases}) if (@{$node->phrases});
799 $where .= ' AND ' . join(' AND ', map {"${talias}.value !~* ".$self->QueryParser->quote_phrase_value($_)} @{$node->unphrases}) if (@{$node->unphrases});
802 push @rank_list, $node_rank;
804 } elsif ($node->isa( 'QueryParser::query_plan::facet' )) {
806 my $table = $node->table;
807 my $talias = $node->table_alias;
810 if (@{$node->fields} > 0) {
811 push(@field_ids, $self->QueryParser->facet_field_ids_by_class( $node->classname, $_ )->[0]) for (@{$node->fields});
813 @field_ids = @{ $self->QueryParser->facet_field_ids_by_class( $node->classname ) };
816 my $join_type = $node->negate ? 'LEFT' : 'INNER';
817 $from .= "\n${spc}$join_type JOIN /* facet */ metabib.facet_entry $talias ON (\n${spc}${spc}m.source = ${talias}.source\n${spc}${spc}".
818 "AND SUBSTRING(${talias}.value,1,1024) IN (" . join(",", map { $self->QueryParser->quote_value($_) } @{$node->values}) . ")\n${spc}${spc}".
819 "AND ${talias}.field IN (". join(',', @field_ids) . ")\n${spc})";
821 $where .= $node->negate ? "${talias}.id IS NULL" : 'TRUE';
824 my $subnode = $node->flatten;
826 push(@rank_list, @{$$subnode{rank_list}});
827 $from .= $$subnode{from};
828 $where .= "($$subnode{where})";
830 if ($$subnode{with}) {
831 $with .= ', ' if $with;
832 $with .= " " . $$subnode{with};
836 $where .= ' AND ' if ($node eq '&');
837 $where .= ' OR ' if ($node eq '|');
838 # ... stitching the WHERE together ...
842 return { rank_list => \@rank_list, from => $from, where => $where.')', with => $with };
847 #-------------------------------
848 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::filter;
849 use base 'QueryParser::query_plan::filter';
851 #-------------------------------
852 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::facet;
853 use base 'QueryParser::query_plan::facet';
857 my ($classname) = split '\|', $self->name;
863 return 'metabib.' . $self->classname . '_field_entry';
868 my ($classname,@fields) = split '\|', $self->name;
875 my $table_alias = "$self";
876 $table_alias =~ s/^.*\(0(x[0-9a-fA-F]+)\)$/$1/go;
877 $table_alias .= '_' . $self->name;
878 $table_alias =~ s/\|/_/go;
884 #-------------------------------
885 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::modifier;
886 use base 'QueryParser::query_plan::modifier';
888 #-------------------------------
889 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::node::atom;
890 use base 'QueryParser::query_plan::node::atom';
896 $self->{sql} = $sql if ($sql);
898 return $self->{sql} if ($self->{sql});
899 return $self->buildSQL;
905 my $classname = $self->node->classname;
907 return $self->sql("to_tsquery('$classname','')") if $self->{dummy};
909 my $normalizers = $self->node->plan->QueryParser->query_normalizers( $classname );
910 my $fields = $self->node->fields;
912 $fields = $self->node->plan->QueryParser->search_fields->{$classname} if (!@$fields);
916 for my $field (@$fields) {
917 for my $nfield (keys %$normalizers) {
918 for my $nizer ( @{$$normalizers{$nfield}} ) {
919 if ($field eq $nfield) {
920 my $param_string = OpenSRF::Utils::JSON->perl2JSON($nizer->{params});
921 if (!exists($norms{$nizer->{function}.$param_string})) {
922 $norms{$nizer->{function}.$param_string} = {p=>$pos++,n=>$nizer};
929 my $sql = $self->node->plan->QueryParser->quote_value($self->content);
931 for my $n ( map { $$_{n} } sort { $$a{p} <=> $$b{p} } values %norms ) {
932 $sql = join(', ', $sql, map { $self->node->plan->QueryParser->quote_value($_) } @{ $n->{params} });
933 $sql = $n->{function}."($sql)";
936 my $prefix = $self->prefix || '';
937 my $suffix = $self->suffix || '';
939 $prefix = "'$prefix' ||" if $prefix;
941 my $suffix_after = '';
943 $suffix_op = ":$suffix" if $suffix;
944 $suffix_after = "|| '$suffix_op'" if $suffix;
946 $sql = "to_tsquery('$classname', COALESCE(NULLIF($prefix '(' || btrim(regexp_replace($sql,E'(?:\\\\s+|:)','$suffix_op&','g'),'&|') $suffix_after || ')', '()'), ''))";
948 return $self->sql($sql);
951 #-------------------------------
952 package OpenILS::Application::Storage::Driver::Pg::QueryParser::query_plan::node;
953 use base 'QueryParser::query_plan::node';
958 $self->{dummy_count} = 0;
960 my $atoms = $self->query_atoms;
962 for my $a (@$atoms) {
963 push(@only_atoms, $a) if (ref($a) && $a->isa('QueryParser::query_plan::node::atom'));
964 $self->{dummy_count}++ if (ref($a) && $a->{dummy});
972 return $self->{dummy_count};
978 $self->{table} = $table if ($table);
979 return $self->{table} if $self->{table};
980 return $self->table( 'metabib.' . $self->classname . '_field_entry' );
985 my $table_alias = shift;
986 $self->{table_alias} = $table_alias if ($table_alias);
987 return $self->{table_alias} if ($self->{table_alias});
989 $table_alias = "$self";
990 $table_alias =~ s/^.*\(0(x[0-9a-fA-F]+)\)$/$1/go;
991 $table_alias .= '_' . $self->requested_class;
992 $table_alias =~ s/\|/_/go;
994 return $self->table_alias( $table_alias );
999 return $self->{tsquery} if ($self->{tsquery});
1001 for my $atom (@{$self->query_atoms}) {
1003 $self->{tsquery} .= "\n${spc}${spc}${spc}" .$atom->sql;
1005 $self->{tsquery} .= $atom x 2;
1009 return $self->{tsquery};
1015 my $rank_norm_map = $self->plan->QueryParser->custom_data->{rank_cd_weight_map};
1017 my $cover_density = 0;
1018 for my $norm ( keys %$rank_norm_map) {
1019 $cover_density += $$rank_norm_map{$norm} if ($self->plan->find_modifier($norm));
1022 return $self->{rank} if ($self->{rank});
1023 return $self->{rank} = 'ts_rank_cd(' . $self->table_alias . '.index_vector, ' . $self->table_alias . ".tsq, $cover_density)";