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_audience TEXT[],
53 param_language TEXT[],
54 param_lit_form TEXT[],
57 param_vformats TEXT[],
58 param_bib_level TEXT[],
60 param_pref_lang_multiplier REAL,
69 ) RETURNS SETOF search.search_result AS $func$
72 current_res search.search_result%ROWTYPE;
73 query_part search.search_args%ROWTYPE;
74 phrase_query_part search.search_args%ROWTYPE;
79 rank_adjust search.relevance_adjustment%ROWTYPE;
85 query_table_alias TEXT;
86 from_alias_array TEXT[] := '{}';
87 used_ranks TEXT[] := '{}';
90 search_org_list INT[];
91 select_clause TEXT := 'SELECT';
92 from_clause TEXT := ' FROM metabib.metarecord_source_map m JOIN metabib.rec_descriptor mrd ON (m.source = mrd.record) ';
93 where_clause TEXT := ' WHERE 1=1 ';
94 mrd_used BOOL := FALSE;
95 sort_desc BOOL := FALSE;
98 core_cursor REFCURSOR;
100 vis_limit_query TEXT;
101 inner_where_clause TEXT;
103 total_count INT := 0;
104 check_count INT := 0;
105 deleted_count INT := 0;
106 visible_count INT := 0;
107 excluded_count INT := 0;
111 core_rel_limit := COALESCE( param_rel_limit, 25000 );
112 core_chk_limit := COALESCE( param_chk_limit, 1000 );
113 core_skip_chk := COALESCE( param_skip_chk, 1 );
116 select_clause := select_clause || ' m.metarecord as id, array_accum(distinct m.source) as records,';
118 select_clause := select_clause || ' m.source as id, array_accum(distinct m.source) as records,';
121 -- first we need to construct the base query
122 FOR query_part IN SELECT * FROM search.parse_search_args(param_searches) WHERE term_type = 'fts_query' LOOP
124 inner_where_clause := 'index_vector @@ ' || query_part.term;
126 IF query_part.field_name IS NOT NULL THEN
128 SELECT id INTO mb_field
129 FROM config.metabib_field
130 WHERE field_class = query_part.field_class
131 AND name = query_part.field_name;
134 inner_where_clause := inner_where_clause ||
135 ' AND ' || 'field = ' || mb_field;
140 -- moving on to the rank ...
141 SELECT * INTO query_part
142 FROM search.parse_search_args(param_searches)
143 WHERE term_type = 'fts_rank'
144 AND table_alias = query_part.table_alias;
146 current_rank := query_part.term || ' * ' || query_part.table_alias || '_weight.weight';
148 IF query_part.field_name IS NOT NULL THEN
150 SELECT array_accum(distinct id) INTO mb_field_list
151 FROM config.metabib_field
152 WHERE field_class = query_part.field_class
153 AND name = query_part.field_name;
157 SELECT array_accum(distinct id) INTO mb_field_list
158 FROM config.metabib_field
159 WHERE field_class = query_part.field_class;
163 FOR rank_adjust IN SELECT * FROM search.relevance_adjustment WHERE active AND field IN ( SELECT * FROM search.explode_array( mb_field_list ) ) LOOP
165 IF NOT rank_adjust.bump_type = ANY (used_ranks) THEN
167 IF rank_adjust.bump_type = 'first_word' THEN
168 SELECT term INTO tmp_text
169 FROM search.parse_search_args(param_searches)
170 WHERE table_alias = query_part.table_alias AND term_type = 'word'
174 tmp_text := query_part.table_alias || '.value ILIKE ' || quote_literal( tmp_text || '%' );
176 ELSIF rank_adjust.bump_type = 'word_order' THEN
177 SELECT array_to_string( array_accum( term ), '%' ) INTO tmp_text
178 FROM search.parse_search_args(param_searches)
179 WHERE table_alias = query_part.table_alias AND term_type = 'word';
181 tmp_text := query_part.table_alias || '.value ILIKE ' || quote_literal( '%' || tmp_text || '%' );
183 ELSIF rank_adjust.bump_type = 'full_match' THEN
184 SELECT array_to_string( array_accum( term ), E'\\s+' ) INTO tmp_text
185 FROM search.parse_search_args(param_searches)
186 WHERE table_alias = query_part.table_alias AND term_type = 'word';
188 tmp_text := query_part.table_alias || '.value ~ ' || quote_literal( '^' || tmp_text || E'\\W*$' );
193 current_rank := current_rank || ' * ( CASE WHEN ' || tmp_text ||
194 ' THEN ' || rank_adjust.multiplier || '::REAL ELSE 1.0 END )';
196 used_ranks := array_append( used_ranks, rank_adjust.bump_type );
202 ranks := array_append( ranks, current_rank );
205 FOR phrase_query_part IN
207 FROM search.parse_search_args(param_searches)
208 WHERE term_type = 'phrase'
209 AND table_alias = query_part.table_alias LOOP
211 tmp_text := replace( phrase_query_part.term, '*', E'\\*' );
212 tmp_text := replace( tmp_text, '?', 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'\\]' );
220 inner_where_clause := inner_where_clause || ' AND ' || 'value ~* ' || quote_literal( E'(^|\\W+)' || regexp_replace(tmp_text, E'\\s+',E'\\\\s+','g') || E'(\\W+|\$)' );
224 query_table := search.pick_table(query_part.field_class);
226 from_clause := from_clause ||
227 ' JOIN ( SELECT * FROM ' || query_table || ' WHERE ' || inner_where_clause ||
228 CASE WHEN core_rel_limit > 0 THEN ' LIMIT ' || core_rel_limit::TEXT ELSE '' END || ' ) AS ' || query_part.table_alias ||
229 ' ON ( m.source = ' || query_part.table_alias || '.source )' ||
230 ' JOIN config.metabib_field AS ' || query_part.table_alias || '_weight' ||
231 ' ON ( ' || query_part.table_alias || '.field = ' || query_part.table_alias || '_weight.id AND ' || query_part.table_alias || '_weight.search_field)';
233 from_alias_array := array_append(from_alias_array, query_part.table_alias);
237 IF param_pref_lang IS NOT NULL AND param_pref_lang_multiplier IS NOT NULL THEN
238 current_rank := ' CASE WHEN mrd.item_lang = ' || quote_literal( param_pref_lang ) ||
239 ' THEN ' || param_pref_lang_multiplier || '::REAL ELSE 1.0 END ';
241 --ranks := array_append( ranks, current_rank );
244 current_rank := ' AVG( ( (' || array_to_string( ranks, ') + (' ) || ') ) * ' || current_rank || ' ) ';
245 select_clause := select_clause || current_rank || ' AS rel,';
247 sort_desc = param_sort_desc;
249 IF param_sort = 'pubdate' THEN
251 tmp_text := '999999';
252 IF param_sort_desc THEN tmp_text := '0'; END IF;
256 SELECT SUBSTRING(frp.value FROM E'\\d{4}')
257 FROM metabib.full_rec frp
258 WHERE frp.record = m.source
260 AND frp.subfield = 'c'
262 )), $$ || quote_literal(tmp_text) || $$ )::INT )
265 ELSIF param_sort = 'title' THEN
267 tmp_text := 'zzzzzz';
268 IF param_sort_desc THEN tmp_text := ' '; END IF;
272 SELECT LTRIM(SUBSTR( frt.value, COALESCE(SUBSTRING(frt.ind2 FROM E'\\d+'),'0')::INT + 1 ))
273 FROM metabib.full_rec frt
274 WHERE frt.record = m.source
276 AND frt.subfield = 'a'
278 )),$$ || quote_literal(tmp_text) || $$))
281 ELSIF param_sort = 'author' THEN
283 tmp_text := 'zzzzzz';
284 IF param_sort_desc THEN tmp_text := ' '; END IF;
288 SELECT LTRIM(fra.value)
289 FROM metabib.full_rec fra
290 WHERE fra.record = m.source
291 AND fra.tag LIKE '1%'
292 AND fra.subfield = 'a'
293 ORDER BY fra.tag::text::int
295 )),$$ || quote_literal(tmp_text) || $$))
298 ELSIF param_sort = 'create_date' THEN
299 current_rank := $$( FIRST (( SELECT create_date FROM biblio.record_entry rbr WHERE rbr.id = m.source)) )$$;
300 ELSIF param_sort = 'edit_date' THEN
301 current_rank := $$( FIRST (( SELECT edit_date FROM biblio.record_entry rbr WHERE rbr.id = m.source)) )$$;
303 sort_desc := NOT COALESCE(param_sort_desc, FALSE);
306 select_clause := select_clause || current_rank || ' AS rank';
308 -- now add the other qualifiers
309 IF param_audience IS NOT NULL AND array_upper(param_audience, 1) > 0 THEN
310 where_clause = where_clause || $$ AND mrd.audience IN ('$$ || array_to_string(param_audience, $$','$$) || $$') $$;
313 IF param_language IS NOT NULL AND array_upper(param_language, 1) > 0 THEN
314 where_clause = where_clause || $$ AND mrd.item_lang IN ('$$ || array_to_string(param_language, $$','$$) || $$') $$;
317 IF param_lit_form IS NOT NULL AND array_upper(param_lit_form, 1) > 0 THEN
318 where_clause = where_clause || $$ AND mrd.lit_form IN ('$$ || array_to_string(param_lit_form, $$','$$) || $$') $$;
321 IF param_types IS NOT NULL AND array_upper(param_types, 1) > 0 THEN
322 where_clause = where_clause || $$ AND mrd.item_type IN ('$$ || array_to_string(param_types, $$','$$) || $$') $$;
325 IF param_forms IS NOT NULL AND array_upper(param_forms, 1) > 0 THEN
326 where_clause = where_clause || $$ AND mrd.item_form IN ('$$ || array_to_string(param_forms, $$','$$) || $$') $$;
329 IF param_vformats IS NOT NULL AND array_upper(param_vformats, 1) > 0 THEN
330 where_clause = where_clause || $$ AND mrd.vr_format IN ('$$ || array_to_string(param_vformats, $$','$$) || $$') $$;
333 IF param_bib_level IS NOT NULL AND array_upper(param_bib_level, 1) > 0 THEN
334 where_clause = where_clause || $$ AND mrd.bib_level IN ('$$ || array_to_string(param_bib_level, $$','$$) || $$') $$;
337 core_rel_query := select_clause || from_clause || where_clause ||
338 ' GROUP BY 1 ORDER BY 4' || CASE WHEN sort_desc THEN ' DESC' ELSE ' ASC' END || ';';
339 --RAISE NOTICE 'Base Query: %', core_rel_query;
341 IF param_search_ou > 0 THEN
342 IF param_depth IS NOT NULL THEN
343 SELECT array_accum(distinct id) INTO search_org_list FROM actor.org_unit_descendants( param_search_ou, param_depth );
345 SELECT array_accum(distinct id) INTO search_org_list FROM actor.org_unit_descendants( param_search_ou );
347 ELSIF param_search_ou < 0 THEN
348 SELECT array_accum(distinct org_unit) INTO search_org_list FROM actor.org_lasso_map WHERE lasso = -param_search_ou;
349 ELSIF param_search_ou = 0 THEN
350 -- reserved for user lassos (ou_buckets/type='lasso') with ID passed in depth ... hack? sure.
353 OPEN core_cursor FOR EXECUTE core_rel_query;
357 FETCH core_cursor INTO core_result;
361 IF total_count % 1000 = 0 THEN
362 -- RAISE NOTICE ' % total, % checked so far ... ', total_count, check_count;
365 IF core_chk_limit > 0 AND total_count - core_skip_chk + 1 >= core_chk_limit THEN
366 total_count := total_count + 1;
370 total_count := total_count + 1;
372 CONTINUE WHEN param_skip_chk IS NOT NULL and total_count < param_skip_chk;
374 check_count := check_count + 1;
376 PERFORM 1 FROM biblio.record_entry b WHERE NOT b.deleted AND b.id IN ( SELECT * FROM search.explode_array( core_result.records ) );
378 -- RAISE NOTICE ' % were all deleted ... ', core_result.records;
379 deleted_count := deleted_count + 1;
384 FROM biblio.record_entry b
385 JOIN config.bib_source s ON (b.source = s.id)
387 AND b.id IN ( SELECT * FROM search.explode_array( core_result.records ) );
390 -- RAISE NOTICE ' % were all transcendant ... ', core_result.records;
391 visible_count := visible_count + 1;
393 current_res.id = core_result.id;
394 current_res.rel = core_result.rel;
398 SELECT COUNT(DISTINCT s.source) INTO tmp_int FROM metabib.metarecord_source_map s WHERE s.metarecord = core_result.id;
402 current_res.record = core_result.records[1];
404 current_res.record = NULL;
407 RETURN NEXT current_res;
412 IF param_statuses IS NOT NULL AND array_upper(param_statuses, 1) > 0 THEN
415 FROM asset.call_number cn
416 JOIN asset.copy cp ON (cp.call_number = cn.id)
419 AND cp.status IN ( SELECT * FROM search.explode_array( param_statuses ) )
420 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
421 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
425 -- RAISE NOTICE ' % were all status-excluded ... ', core_result.records;
426 excluded_count := excluded_count + 1;
432 IF staff IS NULL OR NOT staff THEN
435 FROM asset.call_number cn
436 JOIN asset.copy cp ON (cp.call_number = cn.id)
437 JOIN actor.org_unit a ON (cp.circ_lib = a.id)
438 JOIN asset.copy_location cl ON (cp.location = cl.id)
439 JOIN config.copy_status cs ON (cp.status = cs.id)
446 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
447 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
451 -- RAISE NOTICE ' % were all visibility-excluded ... ', core_result.records;
452 excluded_count := excluded_count + 1;
459 FROM asset.call_number cn
460 JOIN asset.copy cp ON (cp.call_number = cn.id)
461 JOIN actor.org_unit a ON (cp.circ_lib = a.id)
462 JOIN asset.copy_location cl ON (cp.location = cl.id)
463 JOIN config.copy_status cs ON (cp.status = cs.id)
466 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
467 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
473 FROM asset.call_number cn
474 WHERE cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
478 -- RAISE NOTICE ' % were all visibility-excluded ... ', core_result.records;
479 excluded_count := excluded_count + 1;
487 visible_count := visible_count + 1;
489 current_res.id = core_result.id;
490 current_res.rel = core_result.rel;
494 SELECT COUNT(DISTINCT s.source) INTO tmp_int FROM metabib.metarecord_source_map s WHERE s.metarecord = core_result.id;
498 current_res.record = core_result.records[1];
500 current_res.record = NULL;
503 RETURN NEXT current_res;
505 IF visible_count % 1000 = 0 THEN
506 -- RAISE NOTICE ' % visible so far ... ', visible_count;
511 current_res.id = NULL;
512 current_res.rel = NULL;
513 current_res.record = NULL;
514 current_res.total = total_count;
515 current_res.checked = check_count;
516 current_res.deleted = deleted_count;
517 current_res.visible = visible_count;
518 current_res.excluded = excluded_count;
522 RETURN NEXT current_res;
525 $func$ LANGUAGE PLPGSQL;
528 param_statuses INT[],
529 param_audience TEXT[], x
530 param_language TEXT[], x
531 param_lit_form TEXT[], x
532 param_types TEXT[], x
533 param_forms TEXT[], x
534 param_vformats TEXT[], x
537 CREATE OR REPLACE FUNCTION search.explode_array(anyarray) RETURNS SETOF anyelement AS $BODY$
538 SELECT ($1)[s] FROM generate_series(1, array_upper($1, 1)) AS s;
540 LANGUAGE 'sql' IMMUTABLE;
542 CREATE OR REPLACE FUNCTION search.parse_search_args (TEXT) RETURNS SETOF search.search_args AS $perlcode$
546 my $args = decode_json( $json );
550 for my $k ( keys %$args ) {
551 (my $alias = $k) =~ s/\|/_/gso;
552 my ($class, $field) = split /\|/, $k;
553 my $part = $args->{$k};
554 for my $p ( keys %$part ) {
555 my $data = $part->{$p};
556 $data = [$data] if (!ref($data));
557 for my $datum ( @$data ) {
559 { field_class => $class,
560 field_name => $field,
562 table_alias => $alias,
574 $perlcode$ LANGUAGE PLPERLU;