2 * Copyright (C) 2007-2008 Equinox Software, Inc.
3 * Mike Rylander <miker@esilibrary.com>
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License
7 * as published by the Free Software Foundation; either version 2
8 * of the License, or (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
18 DROP SCHEMA search CASCADE;
24 CREATE TABLE search.relevance_adjustment (
25 id SERIAL PRIMARY KEY,
26 active BOOL NOT NULL DEFAULT TRUE,
27 field INT NOT NULL REFERENCES config.metabib_field (id) DEFERRABLE INITIALLY DEFERRED,
28 bump_type TEXT NOT NULL CHECK (bump_type IN ('word_order','first_word','full_match')),
29 multiplier NUMERIC NOT NULL DEFAULT 1.0
31 CREATE UNIQUE INDEX bump_once_per_field_idx ON search.relevance_adjustment ( field, bump_type );
33 CREATE OR REPLACE FUNCTION search.pick_table (TEXT) RETURNS TEXT AS $$
35 WHEN $1 = 'author' THEN 'metabib.author_field_entry'
36 WHEN $1 = 'title' THEN 'metabib.title_field_entry'
37 WHEN $1 = 'subject' THEN 'metabib.subject_field_entry'
38 WHEN $1 = 'keyword' THEN 'metabib.keyword_field_entry'
39 WHEN $1 = 'series' THEN 'metabib.series_field_entry'
43 CREATE TYPE search.search_result AS ( id BIGINT, rel NUMERIC, record INT, total INT, checked INT, visible INT, deleted INT, excluded INT );
44 CREATE TYPE search.search_args AS ( id INT, field_class TEXT, field_name TEXT, table_alias TEXT, term TEXT, term_type TEXT );
46 CREATE OR REPLACE FUNCTION search.staged_fts (
50 param_searches TEXT, -- JSON hash, to be turned into a resultset via search.parse_search_args
52 param_locations INT[],
53 param_audience TEXT[],
54 param_language TEXT[],
55 param_lit_form TEXT[],
58 param_vformats TEXT[],
59 param_bib_level TEXT[],
61 param_pref_lang_multiplier REAL,
70 ) RETURNS SETOF search.search_result AS $func$
73 current_res search.search_result%ROWTYPE;
74 query_part search.search_args%ROWTYPE;
75 phrase_query_part search.search_args%ROWTYPE;
80 rank_adjust search.relevance_adjustment%ROWTYPE;
86 query_table_alias TEXT;
87 from_alias_array TEXT[] := '{}';
88 used_ranks TEXT[] := '{}';
91 search_org_list INT[];
92 select_clause TEXT := 'SELECT';
93 from_clause TEXT := ' FROM metabib.metarecord_source_map m JOIN metabib.rec_descriptor mrd ON (m.source = mrd.record) ';
94 where_clause TEXT := ' WHERE 1=1 ';
95 mrd_used BOOL := FALSE;
96 sort_desc BOOL := FALSE;
99 core_cursor REFCURSOR;
101 vis_limit_query TEXT;
102 inner_where_clause TEXT;
104 total_count INT := 0;
105 check_count INT := 0;
106 deleted_count INT := 0;
107 visible_count INT := 0;
108 excluded_count INT := 0;
112 core_rel_limit := COALESCE( param_rel_limit, 25000 );
113 core_chk_limit := COALESCE( param_chk_limit, 1000 );
114 core_skip_chk := COALESCE( param_skip_chk, 1 );
117 select_clause := select_clause || ' m.metarecord as id, array_accum(distinct m.source) as records,';
119 select_clause := select_clause || ' m.source as id, array_accum(distinct m.source) as records,';
122 -- first we need to construct the base query
123 FOR query_part IN SELECT * FROM search.parse_search_args(param_searches) WHERE term_type = 'fts_query' LOOP
125 inner_where_clause := 'index_vector @@ ' || query_part.term;
127 IF query_part.field_name IS NOT NULL THEN
129 SELECT id INTO mb_field
130 FROM config.metabib_field
131 WHERE field_class = query_part.field_class
132 AND name = query_part.field_name;
135 inner_where_clause := inner_where_clause ||
136 ' AND ' || 'field = ' || mb_field;
141 -- moving on to the rank ...
142 SELECT * INTO query_part
143 FROM search.parse_search_args(param_searches)
144 WHERE term_type = 'fts_rank'
145 AND table_alias = query_part.table_alias;
147 current_rank := query_part.term || ' * ' || query_part.table_alias || '_weight.weight';
149 IF query_part.field_name IS NOT NULL THEN
151 SELECT array_accum(distinct id) INTO mb_field_list
152 FROM config.metabib_field
153 WHERE field_class = query_part.field_class
154 AND name = query_part.field_name;
158 SELECT array_accum(distinct id) INTO mb_field_list
159 FROM config.metabib_field
160 WHERE field_class = query_part.field_class;
164 FOR rank_adjust IN SELECT * FROM search.relevance_adjustment WHERE active AND field IN ( SELECT * FROM search.explode_array( mb_field_list ) ) LOOP
166 IF NOT rank_adjust.bump_type = ANY (used_ranks) THEN
168 IF rank_adjust.bump_type = 'first_word' THEN
169 SELECT term INTO tmp_text
170 FROM search.parse_search_args(param_searches)
171 WHERE table_alias = query_part.table_alias AND term_type = 'word'
175 tmp_text := query_part.table_alias || '.value ILIKE ' || quote_literal( tmp_text || '%' );
177 ELSIF rank_adjust.bump_type = 'word_order' THEN
178 SELECT array_to_string( array_accum( term ), '%' ) INTO tmp_text
179 FROM search.parse_search_args(param_searches)
180 WHERE table_alias = query_part.table_alias AND term_type = 'word';
182 tmp_text := query_part.table_alias || '.value ILIKE ' || quote_literal( '%' || tmp_text || '%' );
184 ELSIF rank_adjust.bump_type = 'full_match' THEN
185 SELECT array_to_string( array_accum( term ), E'\\s+' ) INTO tmp_text
186 FROM search.parse_search_args(param_searches)
187 WHERE table_alias = query_part.table_alias AND term_type = 'word';
189 tmp_text := query_part.table_alias || '.value ~ ' || quote_literal( '^' || tmp_text || E'\\W*$' );
194 current_rank := current_rank || ' * ( CASE WHEN ' || tmp_text ||
195 ' THEN ' || rank_adjust.multiplier || '::REAL ELSE 1.0 END )';
197 used_ranks := array_append( used_ranks, rank_adjust.bump_type );
203 ranks := array_append( ranks, current_rank );
206 FOR phrase_query_part IN
208 FROM search.parse_search_args(param_searches)
209 WHERE term_type = 'phrase'
210 AND table_alias = query_part.table_alias LOOP
212 tmp_text := replace( phrase_query_part.term, '*', E'\\*' );
213 tmp_text := replace( tmp_text, '?', E'\\?' );
214 tmp_text := replace( tmp_text, '+', E'\\+' );
215 tmp_text := replace( tmp_text, '|', E'\\|' );
216 tmp_text := replace( tmp_text, '(', E'\\(' );
217 tmp_text := replace( tmp_text, ')', E'\\)' );
218 tmp_text := replace( tmp_text, '[', E'\\[' );
219 tmp_text := replace( tmp_text, ']', E'\\]' );
221 inner_where_clause := inner_where_clause || ' AND ' || 'value ~* ' || quote_literal( E'(^|\\W+)' || regexp_replace(tmp_text, E'\\s+',E'\\\\s+','g') || E'(\\W+|\$)' );
225 query_table := search.pick_table(query_part.field_class);
227 from_clause := from_clause ||
228 ' JOIN ( SELECT * FROM ' || query_table || ' WHERE ' || inner_where_clause ||
229 CASE WHEN core_rel_limit > 0 THEN ' LIMIT ' || core_rel_limit::TEXT ELSE '' END || ' ) AS ' || query_part.table_alias ||
230 ' ON ( m.source = ' || query_part.table_alias || '.source )' ||
231 ' JOIN config.metabib_field AS ' || query_part.table_alias || '_weight' ||
232 ' ON ( ' || query_part.table_alias || '.field = ' || query_part.table_alias || '_weight.id AND ' || query_part.table_alias || '_weight.search_field)';
234 from_alias_array := array_append(from_alias_array, query_part.table_alias);
238 IF param_pref_lang IS NOT NULL AND param_pref_lang_multiplier IS NOT NULL THEN
239 current_rank := ' CASE WHEN mrd.item_lang = ' || quote_literal( param_pref_lang ) ||
240 ' THEN ' || param_pref_lang_multiplier || '::REAL ELSE 1.0 END ';
242 --ranks := array_append( ranks, current_rank );
245 current_rank := ' AVG( ( (' || array_to_string( ranks, ') + (' ) || ') ) * ' || current_rank || ' ) ';
246 select_clause := select_clause || current_rank || ' AS rel,';
248 sort_desc = param_sort_desc;
250 IF param_sort = 'pubdate' THEN
252 tmp_text := '999999';
253 IF param_sort_desc THEN tmp_text := '0'; END IF;
257 SELECT SUBSTRING(frp.value FROM E'\\d{4}')
258 FROM metabib.full_rec frp
259 WHERE frp.record = m.source
261 AND frp.subfield = 'c'
263 )), $$ || quote_literal(tmp_text) || $$ )::INT )
266 ELSIF param_sort = 'title' THEN
268 tmp_text := 'zzzzzz';
269 IF param_sort_desc THEN tmp_text := ' '; END IF;
273 SELECT LTRIM(SUBSTR( frt.value, COALESCE(SUBSTRING(frt.ind2 FROM E'\\d+'),'0')::INT + 1 ))
274 FROM metabib.full_rec frt
275 WHERE frt.record = m.source
277 AND frt.subfield = 'a'
279 )),$$ || quote_literal(tmp_text) || $$))
282 ELSIF param_sort = 'author' THEN
284 tmp_text := 'zzzzzz';
285 IF param_sort_desc THEN tmp_text := ' '; END IF;
289 SELECT LTRIM(fra.value)
290 FROM metabib.full_rec fra
291 WHERE fra.record = m.source
292 AND fra.tag LIKE '1%'
293 AND fra.subfield = 'a'
294 ORDER BY fra.tag::text::int
296 )),$$ || quote_literal(tmp_text) || $$))
299 ELSIF param_sort = 'create_date' THEN
300 current_rank := $$( FIRST (( SELECT create_date FROM biblio.record_entry rbr WHERE rbr.id = m.source)) )$$;
301 ELSIF param_sort = 'edit_date' THEN
302 current_rank := $$( FIRST (( SELECT edit_date FROM biblio.record_entry rbr WHERE rbr.id = m.source)) )$$;
304 sort_desc := NOT COALESCE(param_sort_desc, FALSE);
307 select_clause := select_clause || current_rank || ' AS rank';
309 -- now add the other qualifiers
310 IF param_audience IS NOT NULL AND array_upper(param_audience, 1) > 0 THEN
311 where_clause = where_clause || $$ AND mrd.audience IN ('$$ || array_to_string(param_audience, $$','$$) || $$') $$;
314 IF param_language IS NOT NULL AND array_upper(param_language, 1) > 0 THEN
315 where_clause = where_clause || $$ AND mrd.item_lang IN ('$$ || array_to_string(param_language, $$','$$) || $$') $$;
318 IF param_lit_form IS NOT NULL AND array_upper(param_lit_form, 1) > 0 THEN
319 where_clause = where_clause || $$ AND mrd.lit_form IN ('$$ || array_to_string(param_lit_form, $$','$$) || $$') $$;
322 IF param_types IS NOT NULL AND array_upper(param_types, 1) > 0 THEN
323 where_clause = where_clause || $$ AND mrd.item_type IN ('$$ || array_to_string(param_types, $$','$$) || $$') $$;
326 IF param_forms IS NOT NULL AND array_upper(param_forms, 1) > 0 THEN
327 where_clause = where_clause || $$ AND mrd.item_form IN ('$$ || array_to_string(param_forms, $$','$$) || $$') $$;
330 IF param_vformats IS NOT NULL AND array_upper(param_vformats, 1) > 0 THEN
331 where_clause = where_clause || $$ AND mrd.vr_format IN ('$$ || array_to_string(param_vformats, $$','$$) || $$') $$;
334 IF param_bib_level IS NOT NULL AND array_upper(param_bib_level, 1) > 0 THEN
335 where_clause = where_clause || $$ AND mrd.bib_level IN ('$$ || array_to_string(param_bib_level, $$','$$) || $$') $$;
338 core_rel_query := select_clause || from_clause || where_clause ||
339 ' GROUP BY 1 ORDER BY 4' || CASE WHEN sort_desc THEN ' DESC' ELSE ' ASC' END || ';';
340 --RAISE NOTICE 'Base Query: %', core_rel_query;
342 IF param_search_ou > 0 THEN
343 IF param_depth IS NOT NULL THEN
344 SELECT array_accum(distinct id) INTO search_org_list FROM actor.org_unit_descendants( param_search_ou, param_depth );
346 SELECT array_accum(distinct id) INTO search_org_list FROM actor.org_unit_descendants( param_search_ou );
348 ELSIF param_search_ou < 0 THEN
349 SELECT array_accum(distinct org_unit) INTO search_org_list FROM actor.org_lasso_map WHERE lasso = -param_search_ou;
350 ELSIF param_search_ou = 0 THEN
351 -- reserved for user lassos (ou_buckets/type='lasso') with ID passed in depth ... hack? sure.
354 OPEN core_cursor FOR EXECUTE core_rel_query;
358 FETCH core_cursor INTO core_result;
362 IF total_count % 1000 = 0 THEN
363 -- RAISE NOTICE ' % total, % checked so far ... ', total_count, check_count;
366 IF core_chk_limit > 0 AND total_count - core_skip_chk + 1 >= core_chk_limit THEN
367 total_count := total_count + 1;
371 total_count := total_count + 1;
373 CONTINUE WHEN param_skip_chk IS NOT NULL and total_count < param_skip_chk;
375 check_count := check_count + 1;
377 PERFORM 1 FROM biblio.record_entry b WHERE NOT b.deleted AND b.id IN ( SELECT * FROM search.explode_array( core_result.records ) );
379 -- RAISE NOTICE ' % were all deleted ... ', core_result.records;
380 deleted_count := deleted_count + 1;
385 FROM biblio.record_entry b
386 JOIN config.bib_source s ON (b.source = s.id)
388 AND b.id IN ( SELECT * FROM search.explode_array( core_result.records ) );
391 -- RAISE NOTICE ' % were all transcendant ... ', core_result.records;
392 visible_count := visible_count + 1;
394 current_res.id = core_result.id;
395 current_res.rel = core_result.rel;
399 SELECT COUNT(DISTINCT s.source) INTO tmp_int FROM metabib.metarecord_source_map s WHERE s.metarecord = core_result.id;
403 current_res.record = core_result.records[1];
405 current_res.record = NULL;
408 RETURN NEXT current_res;
413 IF param_statuses IS NOT NULL AND array_upper(param_statuses, 1) > 0 THEN
416 FROM asset.call_number cn
417 JOIN asset.copy cp ON (cp.call_number = cn.id)
420 AND cp.status IN ( SELECT * FROM search.explode_array( param_statuses ) )
421 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
422 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
426 -- RAISE NOTICE ' % were all status-excluded ... ', core_result.records;
427 excluded_count := excluded_count + 1;
433 IF param_locations IS NOT NULL AND array_upper(param_locations, 1) > 0 THEN
436 FROM asset.call_number cn
437 JOIN asset.copy cp ON (cp.call_number = cn.id)
440 AND cp.location IN ( SELECT * FROM search.explode_array( param_locations ) )
441 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
442 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
446 -- RAISE NOTICE ' % were all copy_location-excluded ... ', core_result.records;
447 excluded_count := excluded_count + 1;
453 IF staff IS NULL OR NOT staff THEN
456 FROM asset.call_number cn
457 JOIN asset.copy cp ON (cp.call_number = cn.id)
458 JOIN actor.org_unit a ON (cp.circ_lib = a.id)
459 JOIN asset.copy_location cl ON (cp.location = cl.id)
460 JOIN config.copy_status cs ON (cp.status = cs.id)
467 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
468 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
472 -- RAISE NOTICE ' % were all visibility-excluded ... ', core_result.records;
473 excluded_count := excluded_count + 1;
480 FROM asset.call_number cn
481 JOIN asset.copy cp ON (cp.call_number = cn.id)
482 JOIN actor.org_unit a ON (cp.circ_lib = a.id)
483 JOIN asset.copy_location cl ON (cp.location = cl.id)
486 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
487 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
493 FROM asset.call_number cn
494 WHERE cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
498 -- RAISE NOTICE ' % were all visibility-excluded ... ', core_result.records;
499 excluded_count := excluded_count + 1;
507 visible_count := visible_count + 1;
509 current_res.id = core_result.id;
510 current_res.rel = core_result.rel;
514 SELECT COUNT(DISTINCT s.source) INTO tmp_int FROM metabib.metarecord_source_map s WHERE s.metarecord = core_result.id;
518 current_res.record = core_result.records[1];
520 current_res.record = NULL;
523 RETURN NEXT current_res;
525 IF visible_count % 1000 = 0 THEN
526 -- RAISE NOTICE ' % visible so far ... ', visible_count;
531 current_res.id = NULL;
532 current_res.rel = NULL;
533 current_res.record = NULL;
534 current_res.total = total_count;
535 current_res.checked = check_count;
536 current_res.deleted = deleted_count;
537 current_res.visible = visible_count;
538 current_res.excluded = excluded_count;
542 RETURN NEXT current_res;
545 $func$ LANGUAGE PLPGSQL;
548 param_statuses INT[],
549 param_audience TEXT[], x
550 param_language TEXT[], x
551 param_lit_form TEXT[], x
552 param_types TEXT[], x
553 param_forms TEXT[], x
554 param_vformats TEXT[], x
557 CREATE OR REPLACE FUNCTION search.explode_array(anyarray) RETURNS SETOF anyelement AS $BODY$
558 SELECT ($1)[s] FROM generate_series(1, array_upper($1, 1)) AS s;
560 LANGUAGE 'sql' IMMUTABLE;
562 CREATE OR REPLACE FUNCTION search.parse_search_args (TEXT) RETURNS SETOF search.search_args AS $perlcode$
566 my $args = decode_json( $json );
570 for my $k ( keys %$args ) {
571 (my $alias = $k) =~ s/\|/_/gso;
572 my ($class, $field) = split /\|/, $k;
573 my $part = $args->{$k};
574 for my $p ( keys %$part ) {
575 my $data = $part->{$p};
576 $data = [$data] if (!ref($data));
577 for my $datum ( @$data ) {
579 { field_class => $class,
580 field_name => $field,
582 table_alias => $alias,
594 $perlcode$ LANGUAGE PLPERLU;