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 CREATE OR REPLACE FUNCTION search.pick_table (TEXT) RETURNS TEXT AS $$
19 WHEN $1 = 'author' THEN 'metabib.author_field_entry'
20 WHEN $1 = 'title' THEN 'metabib.title_field_entry'
21 WHEN $1 = 'subject' THEN 'metabib.subject_field_entry'
22 WHEN $1 = 'keyword' THEN 'metabib.keyword_field_entry'
23 WHEN $1 = 'series' THEN 'metabib.series_field_entry'
27 CREATE TYPE search.search_result AS ( id BIGINT, rel NUMERIC, record INT, total INT, checked INT, visible INT, deleted INT, excluded INT );
28 CREATE TYPE search.search_args AS ( id INT, field_class TEXT, field_name TEXT, table_alias TEXT, term TEXT, term_type TEXT );
30 CREATE OR REPLACE FUNCTION search.staged_fts (
34 param_searches TEXT, -- JSON hash, to be turned into a resultset via search.parse_search_args
36 param_audience TEXT[],
37 param_language TEXT[],
38 param_lit_form TEXT[],
41 param_vformats TEXT[],
42 param_bib_level TEXT[],
44 param_pref_lang_multiplier REAL,
53 ) RETURNS SETOF search.search_result AS $func$
56 current_res search.search_result%ROWTYPE;
57 query_part search.search_args%ROWTYPE;
58 phrase_query_part search.search_args%ROWTYPE;
63 rank_adjust search.relevance_adjustment%ROWTYPE;
69 query_table_alias TEXT;
70 from_alias_array TEXT[] := '{}';
71 used_ranks TEXT[] := '{}';
74 search_org_list INT[];
75 select_clause TEXT := 'SELECT';
76 from_clause TEXT := ' FROM metabib.metarecord_source_map m JOIN metabib.rec_descriptor mrd ON (m.source = mrd.record) ';
77 where_clause TEXT := ' WHERE 1=1 ';
78 mrd_used BOOL := FALSE;
79 sort_desc BOOL := FALSE;
82 core_cursor REFCURSOR;
85 inner_where_clause TEXT;
89 deleted_count INT := 0;
90 visible_count INT := 0;
91 excluded_count INT := 0;
95 core_rel_limit := COALESCE( param_rel_limit, 25000 );
96 core_chk_limit := COALESCE( param_chk_limit, 1000 );
97 core_skip_chk := COALESCE( param_skip_chk, 1 );
100 select_clause := select_clause || ' m.metarecord as id, array_accum(distinct m.source) as records,';
102 select_clause := select_clause || ' m.source as id, array_accum(distinct m.source) as records,';
105 -- first we need to construct the base query
106 FOR query_part IN SELECT * FROM search.parse_search_args(param_searches) WHERE term_type = 'fts_query' LOOP
108 inner_where_clause := 'index_vector @@ ' || query_part.term;
110 IF query_part.field_name IS NOT NULL THEN
112 SELECT id INTO mb_field
113 FROM config.metabib_field
114 WHERE field_class = query_part.field_class
115 AND name = query_part.field_name;
118 inner_where_clause := inner_where_clause ||
119 ' AND ' || 'field = ' || mb_field;
124 -- moving on to the rank ...
125 SELECT * INTO query_part
126 FROM search.parse_search_args(param_searches)
127 WHERE term_type = 'fts_rank'
128 AND table_alias = query_part.table_alias;
130 current_rank := query_part.term || ' * ' || query_part.table_alias || '_weight.weight';
132 IF query_part.field_name IS NOT NULL THEN
134 SELECT array_accum(distinct id) INTO mb_field_list
135 FROM config.metabib_field
136 WHERE field_class = query_part.field_class
137 AND name = query_part.field_name;
141 SELECT array_accum(distinct id) INTO mb_field_list
142 FROM config.metabib_field
143 WHERE field_class = query_part.field_class;
147 FOR rank_adjust IN SELECT * FROM search.relevance_adjustment WHERE active AND field IN ( SELECT * FROM search.explode_array( mb_field_list ) ) LOOP
149 IF NOT rank_adjust.bump_type = ANY (used_ranks) THEN
151 IF rank_adjust.bump_type = 'first_word' THEN
152 SELECT term INTO tmp_text
153 FROM search.parse_search_args(param_searches)
154 WHERE table_alias = query_part.table_alias AND term_type = 'word'
158 tmp_text := query_part.table_alias || '.value ILIKE ' || quote_literal( tmp_text || '%' );
160 ELSIF rank_adjust.bump_type = 'word_order' THEN
161 SELECT array_to_string( array_accum( term ), '%' ) INTO tmp_text
162 FROM search.parse_search_args(param_searches)
163 WHERE table_alias = query_part.table_alias AND term_type = 'word';
165 tmp_text := query_part.table_alias || '.value ILIKE ' || quote_literal( '%' || tmp_text || '%' );
167 ELSIF rank_adjust.bump_type = 'full_match' THEN
168 SELECT array_to_string( array_accum( term ), E'\\s+' ) INTO tmp_text
169 FROM search.parse_search_args(param_searches)
170 WHERE table_alias = query_part.table_alias AND term_type = 'word';
172 tmp_text := query_part.table_alias || '.value ~ ' || quote_literal( '^' || tmp_text || E'\\W*$' );
177 current_rank := current_rank || ' * ( CASE WHEN ' || tmp_text ||
178 ' THEN ' || rank_adjust.multiplier || '::REAL ELSE 1.0 END )';
180 used_ranks := array_append( used_ranks, rank_adjust.bump_type );
186 ranks := array_append( ranks, current_rank );
189 FOR phrase_query_part IN
191 FROM search.parse_search_args(param_searches)
192 WHERE term_type = 'phrase'
193 AND table_alias = query_part.table_alias LOOP
195 tmp_text := replace( phrase_query_part.term, '*', E'\\*' );
196 tmp_text := replace( tmp_text, '?', E'\\?' );
197 tmp_text := replace( tmp_text, '+', E'\\+' );
198 tmp_text := replace( tmp_text, '|', E'\\|' );
199 tmp_text := replace( tmp_text, '(', E'\\(' );
200 tmp_text := replace( tmp_text, ')', E'\\)' );
201 tmp_text := replace( tmp_text, '[', E'\\[' );
202 tmp_text := replace( tmp_text, ']', E'\\]' );
204 inner_where_clause := inner_where_clause || ' AND ' || 'value ~* ' || quote_literal( E'(^|\\W+)' || regexp_replace(tmp_text, E'\\s+',E'\\\\s+','g') || E'(\\W+|\$)' );
208 query_table := search.pick_table(query_part.field_class);
210 from_clause := from_clause ||
211 ' JOIN ( SELECT * FROM ' || query_table || ' WHERE ' || inner_where_clause ||
212 CASE WHEN core_rel_limit > 0 THEN ' LIMIT ' || core_rel_limit::TEXT ELSE '' END || ' ) AS ' || query_part.table_alias ||
213 ' ON ( m.source = ' || query_part.table_alias || '.source )' ||
214 ' JOIN config.metabib_field AS ' || query_part.table_alias || '_weight' ||
215 ' ON ( ' || query_part.table_alias || '.field = ' || query_part.table_alias || '_weight.id AND ' || query_part.table_alias || '_weight.search_field)';
217 from_alias_array := array_append(from_alias_array, query_part.table_alias);
221 IF param_pref_lang IS NOT NULL AND param_pref_lang_multiplier IS NOT NULL THEN
222 current_rank := ' CASE WHEN mrd.item_lang = ' || quote_literal( param_pref_lang ) ||
223 ' THEN ' || param_pref_lang_multiplier || '::REAL ELSE 1.0 END ';
225 --ranks := array_append( ranks, current_rank );
228 current_rank := ' AVG( ( (' || array_to_string( ranks, ') + (' ) || ') ) * ' || current_rank || ' ) ';
229 select_clause := select_clause || current_rank || ' AS rel,';
231 sort_desc = param_sort_desc;
233 IF param_sort = 'pubdate' THEN
235 tmp_text := '999999';
236 IF param_sort_desc THEN tmp_text := '0'; END IF;
240 SELECT SUBSTRING(frp.value FROM E'\\d{4}')
241 FROM metabib.full_rec frp
242 WHERE frp.record = m.source
244 AND frp.subfield = 'c'
246 )), $$ || quote_literal(tmp_text) || $$ )::INT )
249 ELSIF param_sort = 'title' THEN
251 tmp_text := 'zzzzzz';
252 IF param_sort_desc THEN tmp_text := ' '; END IF;
256 SELECT LTRIM(SUBSTR( frt.value, COALESCE(SUBSTRING(frt.ind2 FROM E'\\d+'),'0')::INT + 1 ))
257 FROM metabib.full_rec frt
258 WHERE frt.record = m.source
260 AND frt.subfield = 'a'
262 )),$$ || quote_literal(tmp_text) || $$))
265 ELSIF param_sort = 'author' THEN
267 tmp_text := 'zzzzzz';
268 IF param_sort_desc THEN tmp_text := ' '; END IF;
272 SELECT LTRIM(fra.value)
273 FROM metabib.full_rec fra
274 WHERE fra.record = m.source
275 AND fra.tag LIKE '1%'
276 AND fra.subfield = 'a'
277 ORDER BY fra.tag::text::int
279 )),$$ || quote_literal(tmp_text) || $$))
282 ELSIF param_sort = 'create_date' THEN
283 current_rank := $$( FIRST (( SELECT create_date FROM biblio.record_entry rbr WHERE rbr.id = m.source)) )$$;
284 ELSIF param_sort = 'edit_date' THEN
285 current_rank := $$( FIRST (( SELECT edit_date FROM biblio.record_entry rbr WHERE rbr.id = m.source)) )$$;
287 sort_desc := NOT COALESCE(param_sort_desc, FALSE);
290 select_clause := select_clause || current_rank || ' AS rank';
292 -- now add the other qualifiers
293 IF param_audience IS NOT NULL AND array_upper(param_audience, 1) > 0 THEN
294 where_clause = where_clause || $$ AND mrd.audience IN ('$$ || array_to_string(param_audience, $$','$$) || $$') $$;
297 IF param_language IS NOT NULL AND array_upper(param_language, 1) > 0 THEN
298 where_clause = where_clause || $$ AND mrd.item_lang IN ('$$ || array_to_string(param_language, $$','$$) || $$') $$;
301 IF param_lit_form IS NOT NULL AND array_upper(param_lit_form, 1) > 0 THEN
302 where_clause = where_clause || $$ AND mrd.lit_form IN ('$$ || array_to_string(param_lit_form, $$','$$) || $$') $$;
305 IF param_types IS NOT NULL AND array_upper(param_types, 1) > 0 THEN
306 where_clause = where_clause || $$ AND mrd.item_type IN ('$$ || array_to_string(param_types, $$','$$) || $$') $$;
309 IF param_forms IS NOT NULL AND array_upper(param_forms, 1) > 0 THEN
310 where_clause = where_clause || $$ AND mrd.item_form IN ('$$ || array_to_string(param_forms, $$','$$) || $$') $$;
313 IF param_vformats IS NOT NULL AND array_upper(param_vformats, 1) > 0 THEN
314 where_clause = where_clause || $$ AND mrd.vr_format IN ('$$ || array_to_string(param_vformats, $$','$$) || $$') $$;
317 IF param_bib_level IS NOT NULL AND array_upper(param_bib_level, 1) > 0 THEN
318 where_clause = where_clause || $$ AND mrd.bib_level IN ('$$ || array_to_string(param_bib_level, $$','$$) || $$') $$;
321 core_rel_query := select_clause || from_clause || where_clause ||
322 ' GROUP BY 1 ORDER BY 4' || CASE WHEN sort_desc THEN ' DESC' ELSE ' ASC' END || ';';
323 --RAISE NOTICE 'Base Query: %', core_rel_query;
325 IF param_search_ou > 0 THEN
326 IF param_depth IS NOT NULL THEN
327 SELECT array_accum(distinct id) INTO search_org_list FROM actor.org_unit_descendants( param_search_ou, param_depth );
329 SELECT array_accum(distinct id) INTO search_org_list FROM actor.org_unit_descendants( param_search_ou );
331 ELSIF param_search_ou < 0 THEN
332 SELECT array_accum(distinct org_unit) INTO search_org_list FROM actor.org_lasso_map WHERE lasso = -param_search_ou;
333 ELSIF param_search_ou = 0 THEN
334 -- reserved for user lassos (ou_buckets/type='lasso') with ID passed in depth ... hack? sure.
337 OPEN core_cursor FOR EXECUTE core_rel_query;
341 FETCH core_cursor INTO core_result;
345 IF total_count % 1000 = 0 THEN
346 -- RAISE NOTICE ' % total, % checked so far ... ', total_count, check_count;
349 IF core_chk_limit > 0 AND total_count - core_skip_chk + 1 >= core_chk_limit THEN
350 total_count := total_count + 1;
354 total_count := total_count + 1;
356 CONTINUE WHEN param_skip_chk IS NOT NULL and total_count < param_skip_chk;
358 check_count := check_count + 1;
360 PERFORM 1 FROM biblio.record_entry b WHERE NOT b.deleted AND b.id IN ( SELECT * FROM search.explode_array( core_result.records ) );
362 -- RAISE NOTICE ' % were all deleted ... ', core_result.records;
363 deleted_count := deleted_count + 1;
368 FROM biblio.record_entry b
369 JOIN config.bib_source s ON (b.source = s.id)
371 AND b.id IN ( SELECT * FROM search.explode_array( core_result.records ) );
374 -- RAISE NOTICE ' % were all transcendant ... ', core_result.records;
375 visible_count := visible_count + 1;
377 current_res.id = core_result.id;
378 current_res.rel = core_result.rel;
382 SELECT COUNT(DISTINCT s.source) INTO tmp_int FROM metabib.metarecord_source_map s WHERE s.metarecord = core_result.id;
386 current_res.record = core_result.records[1];
388 current_res.record = NULL;
391 RETURN NEXT current_res;
396 IF param_statuses IS NOT NULL AND array_upper(param_statuses, 1) > 0 THEN
399 FROM asset.call_number cn
400 JOIN asset.copy cp ON (cp.call_number = cn.id)
403 AND cp.status IN ( SELECT * FROM search.explode_array( param_statuses ) )
404 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
405 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
409 -- RAISE NOTICE ' % were all status-excluded ... ', core_result.records;
410 excluded_count := excluded_count + 1;
416 IF staff IS NULL OR NOT staff THEN
419 FROM asset.call_number cn
420 JOIN asset.copy cp ON (cp.call_number = cn.id)
421 JOIN actor.org_unit a ON (cp.circ_lib = a.id)
422 JOIN asset.copy_location cl ON (cp.location = cl.id)
423 JOIN config.copy_status cs ON (cp.status = cs.id)
430 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
431 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
435 -- RAISE NOTICE ' % were all visibility-excluded ... ', core_result.records;
436 excluded_count := excluded_count + 1;
443 FROM asset.call_number cn
444 JOIN asset.copy cp ON (cp.call_number = cn.id)
445 JOIN actor.org_unit a ON (cp.circ_lib = a.id)
446 JOIN asset.copy_location cl ON (cp.location = cl.id)
447 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 ) )
457 FROM asset.call_number cn
458 WHERE cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
462 -- RAISE NOTICE ' % were all visibility-excluded ... ', core_result.records;
463 excluded_count := excluded_count + 1;
471 visible_count := visible_count + 1;
473 current_res.id = core_result.id;
474 current_res.rel = core_result.rel;
478 SELECT COUNT(DISTINCT s.source) INTO tmp_int FROM metabib.metarecord_source_map s WHERE s.metarecord = core_result.id;
482 current_res.record = core_result.records[1];
484 current_res.record = NULL;
487 RETURN NEXT current_res;
489 IF visible_count % 1000 = 0 THEN
490 -- RAISE NOTICE ' % visible so far ... ', visible_count;
495 current_res.id = NULL;
496 current_res.rel = NULL;
497 current_res.record = NULL;
498 current_res.total = total_count;
499 current_res.checked = check_count;
500 current_res.deleted = deleted_count;
501 current_res.visible = visible_count;
502 current_res.excluded = excluded_count;
506 RETURN NEXT current_res;
509 $func$ LANGUAGE PLPGSQL;
512 param_statuses INT[],
513 param_audience TEXT[], x
514 param_language TEXT[], x
515 param_lit_form TEXT[], x
516 param_types TEXT[], x
517 param_forms TEXT[], x
518 param_vformats TEXT[], x
521 CREATE OR REPLACE FUNCTION search.explode_array(anyarray) RETURNS SETOF anyelement AS $BODY$
522 SELECT ($1)[s] FROM generate_series(1, array_upper($1, 1)) AS s;
524 LANGUAGE 'sql' IMMUTABLE;
526 CREATE OR REPLACE FUNCTION search.parse_search_args (TEXT) RETURNS SETOF search.search_args AS $perlcode$
530 my $args = decode_json( $json );
534 for my $k ( keys %$args ) {
535 (my $alias = $k) =~ s/\|/_/gso;
536 my ($class, $field) = split /\|/, $k;
537 my $part = $args->{$k};
538 for my $p ( keys %$part ) {
539 my $data = $part->{$p};
540 $data = [$data] if (!ref($data));
541 for my $datum ( @$data ) {
543 { field_class => $class,
544 field_name => $field,
546 table_alias => $alias,
558 $perlcode$ LANGUAGE PLPERLU;