2 DROP SCHEMA search CASCADE;
8 CREATE TABLE search.relevance_adjustment (
10 active BOOL NOT NULL DEFAULT TRUE,
11 field INT NOT NULL REFERENCES config.metabib_field (id),
12 bump_type TEXT NOT NULL CHECK (bump_type IN ('word_order','first_word','full_match')),
13 multiplier NUMERIC NOT NULL DEFAULT 1.0
15 CREATE UNIQUE INDEX bump_once_per_field_idx ON search.relevance_adjustment ( field, bump_type );
17 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(1, 'first_word', 1.5);
18 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(1, 'full_match', 20);
19 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(2, 'first_word', 1.5);
20 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(2, 'word_order', 10);
21 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(2, 'full_match', 20);
22 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(3, 'first_word', 1.5);
23 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(3, 'word_order', 10);
24 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(3, 'full_match', 20);
25 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(4, 'first_word', 1.5);
26 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(4, 'word_order', 10);
27 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(4, 'full_match', 20);
28 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(5, 'first_word', 1.5);
29 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(5, 'word_order', 10);
30 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(5, 'full_match', 20);
31 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(6, 'first_word', 1.5);
32 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(7, 'first_word', 1.5);
33 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(8, 'first_word', 1.5);
34 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(9, 'first_word', 1.5);
35 INSERT INTO search.relevance_adjustment (field, bump_type, multiplier) VALUES(14, 'word_order', 10);
37 CREATE OR REPLACE FUNCTION search.pick_table (TEXT) RETURNS TEXT AS $$
39 WHEN $1 = 'author' THEN 'metabib.author_field_entry'
40 WHEN $1 = 'title' THEN 'metabib.title_field_entry'
41 WHEN $1 = 'subject' THEN 'metabib.subject_field_entry'
42 WHEN $1 = 'keyword' THEN 'metabib.keyword_field_entry'
43 WHEN $1 = 'series' THEN 'metabib.series_field_entry'
47 CREATE TYPE search.search_result AS ( id BIGINT, rel NUMERIC, record INT, total INT, checked INT, visible INT, deleted INT, excluded INT );
48 CREATE TYPE search.search_args AS ( id INT, field_class TEXT, field_name TEXT, table_alias TEXT, term TEXT, term_type TEXT );
50 CREATE OR REPLACE FUNCTION search.staged_fts (
54 param_searches TEXT, -- JSON hash, to be turned into a resultset via search.parse_search_args
56 param_audience TEXT[],
57 param_language TEXT[],
58 param_lit_form TEXT[],
61 param_vformats TEXT[],
62 param_bib_level TEXT[],
64 param_pref_lang_multiplier REAL,
73 ) RETURNS SETOF search.search_result AS $func$
76 current_res search.search_result%ROWTYPE;
77 query_part search.search_args%ROWTYPE;
78 phrase_query_part search.search_args%ROWTYPE;
83 rank_adjust search.relevance_adjustment%ROWTYPE;
89 query_table_alias TEXT;
90 from_alias_array TEXT[] := '{}';
91 used_ranks TEXT[] := '{}';
94 search_org_list INT[];
95 select_clause TEXT := 'SELECT';
96 from_clause TEXT := ' FROM metabib.metarecord_source_map m JOIN metabib.rec_descriptor mrd ON (m.source = mrd.record) ';
97 where_clause TEXT := ' WHERE 1=1 ';
98 mrd_used BOOL := FALSE;
99 sort_desc BOOL := FALSE;
102 core_cursor REFCURSOR;
104 vis_limit_query TEXT;
105 inner_where_clause TEXT;
107 total_count INT := 0;
108 check_count INT := 0;
109 deleted_count INT := 0;
110 visible_count INT := 0;
111 excluded_count INT := 0;
115 core_rel_limit := COALESCE( param_rel_limit, 25000 );
116 core_chk_limit := COALESCE( param_chk_limit, 1000 );
117 core_skip_chk := COALESCE( param_skip_chk, 1 );
120 select_clause := select_clause || ' m.metarecord as id, array_accum(distinct m.source) as records,';
122 select_clause := select_clause || ' m.source as id, array_accum(distinct m.source) as records,';
125 -- first we need to construct the base query
126 FOR query_part IN SELECT * FROM search.parse_search_args(param_searches) WHERE term_type = 'fts_query' LOOP
128 inner_where_clause := 'index_vector @@ ' || query_part.term;
130 IF query_part.field_name IS NOT NULL THEN
132 SELECT id INTO mb_field
133 FROM config.metabib_field
134 WHERE field_class = query_part.field_class
135 AND name = query_part.field_name;
138 inner_where_clause := inner_where_clause ||
139 ' AND ' || 'field = ' || mb_field;
144 -- moving on to the rank ...
145 SELECT * INTO query_part
146 FROM search.parse_search_args(param_searches)
147 WHERE term_type = 'fts_rank'
148 AND table_alias = query_part.table_alias;
150 current_rank := query_part.term || ' * ' || query_part.table_alias || '_weight.weight';
152 IF query_part.field_name IS NOT NULL THEN
154 SELECT array_accum(distinct id) INTO mb_field_list
155 FROM config.metabib_field
156 WHERE field_class = query_part.field_class
157 AND name = query_part.field_name;
161 SELECT array_accum(distinct id) INTO mb_field_list
162 FROM config.metabib_field
163 WHERE field_class = query_part.field_class;
167 FOR rank_adjust IN SELECT * FROM search.relevance_adjustment WHERE active AND field IN ( SELECT * FROM search.explode_array( mb_field_list ) ) LOOP
169 IF NOT rank_adjust.bump_type = ANY (used_ranks) THEN
171 IF rank_adjust.bump_type = 'first_word' THEN
172 SELECT term INTO tmp_text
173 FROM search.parse_search_args(param_searches)
174 WHERE table_alias = query_part.table_alias AND term_type = 'word'
178 tmp_text := query_part.table_alias || '.value ILIKE ' || quote_literal( tmp_text || '%' );
180 ELSIF rank_adjust.bump_type = 'word_order' THEN
181 SELECT array_to_string( array_accum( term ), '%' ) INTO tmp_text
182 FROM search.parse_search_args(param_searches)
183 WHERE table_alias = query_part.table_alias AND term_type = 'word';
185 tmp_text := query_part.table_alias || '.value ILIKE ' || quote_literal( '%' || tmp_text || '%' );
187 ELSIF rank_adjust.bump_type = 'full_match' THEN
188 SELECT array_to_string( array_accum( term ), E'\\s+' ) INTO tmp_text
189 FROM search.parse_search_args(param_searches)
190 WHERE table_alias = query_part.table_alias AND term_type = 'word';
192 tmp_text := query_part.table_alias || '.value ~ ' || quote_literal( '^' || tmp_text || E'\\W*$' );
197 current_rank := current_rank || ' * ( CASE WHEN ' || tmp_text ||
198 ' THEN ' || rank_adjust.multiplier || '::REAL ELSE 1.0 END )';
200 used_ranks := array_append( used_ranks, rank_adjust.bump_type );
206 ranks := array_append( ranks, current_rank );
209 FOR phrase_query_part IN
211 FROM search.parse_search_args(param_searches)
212 WHERE term_type = 'phrase'
213 AND table_alias = query_part.table_alias LOOP
215 tmp_text := replace( phrase_query_part.term, '*', 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'\\(' );
220 tmp_text := replace( tmp_text, ')', E'\\)' );
221 tmp_text := replace( tmp_text, '[', E'\\[' );
222 tmp_text := replace( tmp_text, ']', E'\\]' );
224 inner_where_clause := inner_where_clause || ' AND ' || 'value ~* ' || quote_literal( E'(^|\\W+)' || regexp_replace(tmp_text, E'\\s+',E'\\\\s+','g') || E'(\\W+|\$)' );
228 query_table := search.pick_table(query_part.field_class);
230 from_clause := from_clause ||
231 ' JOIN ( SELECT * FROM ' || query_table || ' WHERE ' || inner_where_clause ||
232 CASE WHEN core_rel_limit > 0 THEN ' LIMIT ' || core_rel_limit::TEXT ELSE '' END || ' ) AS ' || query_part.table_alias ||
233 ' ON ( m.source = ' || query_part.table_alias || '.source )' ||
234 ' JOIN config.metabib_field AS ' || query_part.table_alias || '_weight' ||
235 ' ON ( ' || query_part.table_alias || '.field = ' || query_part.table_alias || '_weight.id AND ' || query_part.table_alias || '_weight.search_field)';
237 from_alias_array := array_append(from_alias_array, query_part.table_alias);
241 IF param_pref_lang IS NOT NULL AND param_pref_lang_multiplier IS NOT NULL THEN
242 current_rank := ' CASE WHEN mrd.item_lang = ' || quote_literal( param_pref_lang ) ||
243 ' THEN ' || param_pref_lang_multiplier || '::REAL ELSE 1.0 END ';
245 --ranks := array_append( ranks, current_rank );
248 current_rank := ' AVG( ( (' || array_to_string( ranks, ') + (' ) || ') ) * ' || current_rank || ' ) ';
249 select_clause := select_clause || current_rank || ' AS rel,';
251 sort_desc = param_sort_desc;
253 IF param_sort = 'pubdate' THEN
255 tmp_text := '999999';
256 IF param_sort_desc THEN tmp_text := '0'; END IF;
260 SELECT SUBSTRING(frp.value FROM E'\\d{4}')
261 FROM metabib.full_rec frp
262 WHERE frp.record = m.source
264 AND frp.subfield = 'c'
266 )), $$ || quote_literal(tmp_text) || $$ )::INT )
269 ELSIF param_sort = 'title' THEN
271 tmp_text := 'zzzzzz';
272 IF param_sort_desc THEN tmp_text := ' '; END IF;
276 SELECT LTRIM(SUBSTR( frt.value, COALESCE(SUBSTRING(frt.ind2 FROM E'\\d+'),'0')::INT + 1 ))
277 FROM metabib.full_rec frt
278 WHERE frt.record = m.source
280 AND frt.subfield = 'a'
282 )),$$ || quote_literal(tmp_text) || $$))
285 ELSIF param_sort = 'author' THEN
287 tmp_text := 'zzzzzz';
288 IF param_sort_desc THEN tmp_text := ' '; END IF;
292 SELECT LTRIM(fra.value)
293 FROM metabib.full_rec fra
294 WHERE fra.record = m.source
295 AND fra.tag LIKE '1%'
296 AND fra.subfield = 'a'
297 ORDER BY fra.tag::text::int
299 )),$$ || quote_literal(tmp_text) || $$))
302 ELSIF param_sort = 'create_date' THEN
303 current_rank := $$( FIRST (( SELECT create_date FROM biblio.record_entry rbr WHERE rbr.id = m.source)) )$$;
304 ELSIF param_sort = 'edit_date' THEN
305 current_rank := $$( FIRST (( SELECT edit_date FROM biblio.record_entry rbr WHERE rbr.id = m.source)) )$$;
307 sort_desc := NOT COALESCE(param_sort_desc, FALSE);
310 select_clause := select_clause || current_rank || ' AS rank';
312 -- now add the other qualifiers
313 IF param_audience IS NOT NULL AND array_upper(param_audience, 1) > 0 THEN
314 where_clause = where_clause || $$ AND mrd.audience IN ('$$ || array_to_string(param_audience, $$','$$) || $$') $$;
317 IF param_language IS NOT NULL AND array_upper(param_language, 1) > 0 THEN
318 where_clause = where_clause || $$ AND mrd.item_lang IN ('$$ || array_to_string(param_language, $$','$$) || $$') $$;
321 IF param_lit_form IS NOT NULL AND array_upper(param_lit_form, 1) > 0 THEN
322 where_clause = where_clause || $$ AND mrd.lit_form IN ('$$ || array_to_string(param_lit_form, $$','$$) || $$') $$;
325 IF param_types IS NOT NULL AND array_upper(param_types, 1) > 0 THEN
326 where_clause = where_clause || $$ AND mrd.item_type IN ('$$ || array_to_string(param_types, $$','$$) || $$') $$;
329 IF param_forms IS NOT NULL AND array_upper(param_forms, 1) > 0 THEN
330 where_clause = where_clause || $$ AND mrd.item_form IN ('$$ || array_to_string(param_forms, $$','$$) || $$') $$;
333 IF param_vformats IS NOT NULL AND array_upper(param_vformats, 1) > 0 THEN
334 where_clause = where_clause || $$ AND mrd.vr_format IN ('$$ || array_to_string(param_vformats, $$','$$) || $$') $$;
337 IF param_bib_level IS NOT NULL AND array_upper(param_bib_level, 1) > 0 THEN
338 where_clause = where_clause || $$ AND mrd.bib_level IN ('$$ || array_to_string(param_bib_level, $$','$$) || $$') $$;
341 core_rel_query := select_clause || from_clause || where_clause ||
342 ' GROUP BY 1 ORDER BY 4' || CASE WHEN sort_desc THEN ' DESC' ELSE ' ASC' END || ';';
343 --RAISE NOTICE 'Base Query: %', core_rel_query;
345 IF param_search_ou > 0 THEN
346 IF param_depth IS NOT NULL THEN
347 SELECT array_accum(distinct id) INTO search_org_list FROM actor.org_unit_descendants( param_search_ou, param_depth );
349 SELECT array_accum(distinct id) INTO search_org_list FROM actor.org_unit_descendants( param_search_ou );
351 ELSIF param_search_ou < 0 THEN
352 SELECT array_accum(distinct org_unit) INTO search_org_list FROM actor.org_lasso_map WHERE lasso = -param_search_ou;
353 ELSIF param_search_ou = 0 THEN
354 -- reserved for user lassos (ou_buckets/type='lasso') with ID passed in depth ... hack? sure.
357 OPEN core_cursor FOR EXECUTE core_rel_query;
361 FETCH core_cursor INTO core_result;
365 IF total_count % 1000 = 0 THEN
366 -- RAISE NOTICE ' % total, % checked so far ... ', total_count, check_count;
369 IF core_chk_limit > 0 AND total_count - core_skip_chk + 1 >= core_chk_limit THEN
370 total_count := total_count + 1;
374 total_count := total_count + 1;
376 CONTINUE WHEN param_skip_chk IS NOT NULL and total_count < param_skip_chk;
378 check_count := check_count + 1;
380 PERFORM 1 FROM biblio.record_entry b WHERE NOT b.deleted AND b.id IN ( SELECT * FROM search.explode_array( core_result.records ) );
382 -- RAISE NOTICE ' % were all deleted ... ', core_result.records;
383 deleted_count := deleted_count + 1;
388 FROM biblio.record_entry b
389 JOIN config.bib_source s ON (b.source = s.id)
391 AND b.id IN ( SELECT * FROM search.explode_array( core_result.records ) );
394 -- RAISE NOTICE ' % were all transcendant ... ', core_result.records;
395 visible_count := visible_count + 1;
397 current_res.id = core_result.id;
398 current_res.rel = core_result.rel;
402 SELECT COUNT(DISTINCT s.source) INTO tmp_int FROM metabib.metarecord_source_map s WHERE s.metarecord = core_result.id;
406 current_res.record = core_result.records[1];
408 current_res.record = NULL;
411 RETURN NEXT current_res;
416 IF param_statuses IS NOT NULL AND array_upper(param_statuses, 1) > 0 THEN
419 FROM asset.call_number cn
420 JOIN asset.copy cp ON (cp.call_number = cn.id)
423 AND cp.status IN ( SELECT * FROM search.explode_array( param_statuses ) )
424 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
425 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
429 -- RAISE NOTICE ' % were all status-excluded ... ', core_result.records;
430 excluded_count := excluded_count + 1;
436 IF staff IS NULL OR NOT staff THEN
439 FROM asset.call_number cn
440 JOIN asset.copy cp ON (cp.call_number = cn.id)
441 JOIN actor.org_unit a ON (cp.circ_lib = a.id)
442 JOIN asset.copy_location cl ON (cp.location = cl.id)
443 JOIN config.copy_status cs ON (cp.status = cs.id)
450 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
451 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
455 -- RAISE NOTICE ' % were all visibility-excluded ... ', core_result.records;
456 excluded_count := excluded_count + 1;
463 FROM asset.call_number cn
464 JOIN asset.copy cp ON (cp.call_number = cn.id)
465 JOIN actor.org_unit a ON (cp.circ_lib = a.id)
466 JOIN asset.copy_location cl ON (cp.location = cl.id)
467 JOIN config.copy_status cs ON (cp.status = cs.id)
470 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
471 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
477 FROM asset.call_number cn
478 WHERE cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
482 -- RAISE NOTICE ' % were all visibility-excluded ... ', core_result.records;
483 excluded_count := excluded_count + 1;
491 visible_count := visible_count + 1;
493 current_res.id = core_result.id;
494 current_res.rel = core_result.rel;
498 SELECT COUNT(DISTINCT s.source) INTO tmp_int FROM metabib.metarecord_source_map s WHERE s.metarecord = core_result.id;
502 current_res.record = core_result.records[1];
504 current_res.record = NULL;
507 RETURN NEXT current_res;
509 IF visible_count % 1000 = 0 THEN
510 -- RAISE NOTICE ' % visible so far ... ', visible_count;
515 current_res.id = NULL;
516 current_res.rel = NULL;
517 current_res.record = NULL;
518 current_res.total = total_count;
519 current_res.checked = check_count;
520 current_res.deleted = deleted_count;
521 current_res.visible = visible_count;
522 current_res.excluded = excluded_count;
526 RETURN NEXT current_res;
529 $func$ LANGUAGE PLPGSQL;
532 param_statuses INT[],
533 param_audience TEXT[], x
534 param_language TEXT[], x
535 param_lit_form TEXT[], x
536 param_types TEXT[], x
537 param_forms TEXT[], x
538 param_vformats TEXT[], x
541 CREATE OR REPLACE FUNCTION search.explode_array(anyarray) RETURNS SETOF anyelement AS $BODY$
542 SELECT ($1)[s] FROM generate_series(1, array_upper($1, 1)) AS s;
544 LANGUAGE 'sql' IMMUTABLE;
546 CREATE OR REPLACE FUNCTION search.parse_search_args (TEXT) RETURNS SETOF search.search_args AS $perlcode$
550 my $args = decode_json( $json );
554 for my $k ( keys %$args ) {
555 (my $alias = $k) =~ s/\|/_/gso;
556 my ($class, $field) = split /\|/, $k;
557 my $part = $args->{$k};
558 for my $p ( keys %$part ) {
559 my $data = $part->{$p};
560 $data = [$data] if (!ref($data));
561 for my $datum ( @$data ) {
563 { field_class => $class,
564 field_name => $field,
566 table_alias => $alias,
578 $perlcode$ LANGUAGE PLPERLU;