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.
21 ALTER TABLE config.rule_max_fine ADD COLUMN is_percent BOOL NOT NULL DEFAULT FALSE;
23 CREATE OR REPLACE FUNCTION biblio.next_autogen_tcn_value () RETURNS TEXT AS $$
24 BEGIN RETURN 'AUTOGENERATED-' || nextval('biblio.autogen_tcn_value_seq'::TEXT); END;
28 CREATE OR REPLACE FUNCTION search.staged_fts (
32 param_searches TEXT, -- JSON hash, to be turned into a resultset via search.parse_search_args
34 param_locations INT[],
35 param_audience TEXT[],
36 param_language TEXT[],
37 param_lit_form TEXT[],
40 param_vformats TEXT[],
42 param_pref_lang_multiplier REAL,
51 ) RETURNS SETOF search.search_result AS $func$
54 current_res search.search_result%ROWTYPE;
55 query_part search.search_args%ROWTYPE;
56 phrase_query_part search.search_args%ROWTYPE;
61 rank_adjust search.relevance_adjustment%ROWTYPE;
67 query_table_alias TEXT;
68 from_alias_array TEXT[] := '{}';
69 used_ranks TEXT[] := '{}';
72 search_org_list INT[];
73 select_clause TEXT := 'SELECT';
74 from_clause TEXT := ' FROM metabib.metarecord_source_map m JOIN metabib.rec_descriptor mrd ON (m.source = mrd.record) ';
75 where_clause TEXT := ' WHERE 1=1 ';
76 mrd_used BOOL := FALSE;
77 sort_desc BOOL := FALSE;
80 core_cursor REFCURSOR;
83 inner_where_clause TEXT;
87 deleted_count INT := 0;
88 visible_count INT := 0;
89 excluded_count INT := 0;
93 core_rel_limit := COALESCE( param_rel_limit, 25000 );
94 core_chk_limit := COALESCE( param_chk_limit, 1000 );
95 core_skip_chk := COALESCE( param_skip_chk, 1 );
98 select_clause := select_clause || ' m.metarecord as id, array_accum(distinct m.source) as records,';
100 select_clause := select_clause || ' m.source as id, array_accum(distinct m.source) as records,';
103 -- first we need to construct the base query
104 FOR query_part IN SELECT * FROM search.parse_search_args(param_searches) WHERE term_type = 'fts_query' LOOP
106 inner_where_clause := 'index_vector @@ ' || query_part.term;
108 IF query_part.field_name IS NOT NULL THEN
110 SELECT id INTO mb_field
111 FROM config.metabib_field
112 WHERE field_class = query_part.field_class
113 AND name = query_part.field_name;
116 inner_where_clause := inner_where_clause ||
117 ' AND ' || 'field = ' || mb_field;
122 -- moving on to the rank ...
123 SELECT * INTO query_part
124 FROM search.parse_search_args(param_searches)
125 WHERE term_type = 'fts_rank'
126 AND table_alias = query_part.table_alias;
128 current_rank := query_part.term || ' * ' || query_part.table_alias || '_weight.weight';
130 IF query_part.field_name IS NOT NULL THEN
132 SELECT array_accum(distinct id) INTO mb_field_list
133 FROM config.metabib_field
134 WHERE field_class = query_part.field_class
135 AND name = query_part.field_name;
139 SELECT array_accum(distinct id) INTO mb_field_list
140 FROM config.metabib_field
141 WHERE field_class = query_part.field_class;
145 FOR rank_adjust IN SELECT * FROM search.relevance_adjustment WHERE active AND field IN ( SELECT * FROM search.explode_array( mb_field_list ) ) LOOP
147 IF NOT rank_adjust.bump_type = ANY (used_ranks) THEN
149 IF rank_adjust.bump_type = 'first_word' THEN
150 SELECT term INTO tmp_text
151 FROM search.parse_search_args(param_searches)
152 WHERE table_alias = query_part.table_alias AND term_type = 'word'
156 tmp_text := query_part.table_alias || '.value ILIKE ' || quote_literal( tmp_text || '%' );
158 ELSIF rank_adjust.bump_type = 'word_order' THEN
159 SELECT array_to_string( array_accum( term ), '%' ) INTO tmp_text
160 FROM search.parse_search_args(param_searches)
161 WHERE table_alias = query_part.table_alias AND term_type = 'word';
163 tmp_text := query_part.table_alias || '.value ILIKE ' || quote_literal( '%' || tmp_text || '%' );
165 ELSIF rank_adjust.bump_type = 'full_match' THEN
166 SELECT array_to_string( array_accum( term ), E'\\s+' ) INTO tmp_text
167 FROM search.parse_search_args(param_searches)
168 WHERE table_alias = query_part.table_alias AND term_type = 'word';
170 tmp_text := query_part.table_alias || '.value ~ ' || quote_literal( '^' || tmp_text || E'\\W*$' );
175 IF tmp_text IS NOT NULL THEN
176 current_rank := current_rank || ' * ( CASE WHEN ' || tmp_text ||
177 ' 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 core_rel_query := select_clause || from_clause || where_clause ||
318 ' GROUP BY 1 ORDER BY 4' || CASE WHEN sort_desc THEN ' DESC' ELSE ' ASC' END || ';';
319 --RAISE NOTICE 'Base Query: %', core_rel_query;
321 IF param_depth IS NOT NULL THEN
322 SELECT array_accum(distinct id) INTO search_org_list FROM actor.org_unit_descendants( param_search_ou, param_depth );
324 SELECT array_accum(distinct id) INTO search_org_list FROM actor.org_unit_descendants( param_search_ou );
327 OPEN core_cursor FOR EXECUTE core_rel_query;
331 FETCH core_cursor INTO core_result;
335 IF total_count % 1000 = 0 THEN
336 -- RAISE NOTICE ' % total, % checked so far ... ', total_count, check_count;
339 IF core_chk_limit > 0 AND total_count - core_skip_chk + 1 >= core_chk_limit THEN
340 total_count := total_count + 1;
344 total_count := total_count + 1;
346 CONTINUE WHEN param_skip_chk IS NOT NULL and total_count < param_skip_chk;
348 check_count := check_count + 1;
350 PERFORM 1 FROM biblio.record_entry b WHERE NOT b.deleted AND b.id IN ( SELECT * FROM search.explode_array( core_result.records ) );
352 -- RAISE NOTICE ' % were all deleted ... ', core_result.records;
353 deleted_count := deleted_count + 1;
358 FROM biblio.record_entry b
359 JOIN config.bib_source s ON (b.source = s.id)
361 AND b.id IN ( SELECT * FROM search.explode_array( core_result.records ) );
364 -- RAISE NOTICE ' % were all transcendant ... ', core_result.records;
365 visible_count := visible_count + 1;
367 current_res.id = core_result.id;
368 current_res.rel = core_result.rel;
372 SELECT COUNT(DISTINCT s.source) INTO tmp_int FROM metabib.metarecord_source_map s WHERE s.metarecord = core_result.id;
376 current_res.record = core_result.records[1];
378 current_res.record = NULL;
381 RETURN NEXT current_res;
386 IF param_statuses IS NOT NULL AND array_upper(param_statuses, 1) > 0 THEN
389 FROM asset.call_number cn
390 JOIN asset.copy cp ON (cp.call_number = cn.id)
393 AND cp.status IN ( SELECT * FROM search.explode_array( param_statuses ) )
394 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
395 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
399 -- RAISE NOTICE ' % were all status-excluded ... ', core_result.records;
400 excluded_count := excluded_count + 1;
406 IF param_locations IS NOT NULL AND array_upper(param_locations, 1) > 0 THEN
409 FROM asset.call_number cn
410 JOIN asset.copy cp ON (cp.call_number = cn.id)
413 AND cp.location IN ( SELECT * FROM search.explode_array( param_locations ) )
414 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
415 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
419 -- RAISE NOTICE ' % were all copy_location-excluded ... ', core_result.records;
420 excluded_count := excluded_count + 1;
426 IF staff IS NULL OR NOT staff THEN
429 FROM asset.call_number cn
430 JOIN asset.copy cp ON (cp.call_number = cn.id)
431 JOIN actor.org_unit a ON (cp.circ_lib = a.id)
432 JOIN asset.copy_location cl ON (cp.location = cl.id)
433 JOIN config.copy_status cs ON (cp.status = cs.id)
440 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
441 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
445 -- RAISE NOTICE ' % were all visibility-excluded ... ', core_result.records;
446 excluded_count := excluded_count + 1;
453 FROM asset.call_number cn
454 JOIN asset.copy cp ON (cp.call_number = cn.id)
455 JOIN actor.org_unit a ON (cp.circ_lib = a.id)
456 JOIN asset.copy_location cl ON (cp.location = cl.id)
457 JOIN config.copy_status cs ON (cp.status = cs.id)
460 AND cp.circ_lib IN ( SELECT * FROM search.explode_array( search_org_list ) )
461 AND cn.record IN ( SELECT * FROM search.explode_array( core_result.records ) )
467 FROM asset.call_number cn
468 WHERE 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;
481 visible_count := visible_count + 1;
483 current_res.id = core_result.id;
484 current_res.rel = core_result.rel;
488 SELECT COUNT(DISTINCT s.source) INTO tmp_int FROM metabib.metarecord_source_map s WHERE s.metarecord = core_result.id;
492 current_res.record = core_result.records[1];
494 current_res.record = NULL;
497 RETURN NEXT current_res;
499 IF visible_count % 1000 = 0 THEN
500 -- RAISE NOTICE ' % visible so far ... ', visible_count;
505 current_res.id = NULL;
506 current_res.rel = NULL;
507 current_res.record = NULL;
508 current_res.total = total_count;
509 current_res.checked = check_count;
510 current_res.deleted = deleted_count;
511 current_res.visible = visible_count;
512 current_res.excluded = excluded_count;
516 RETURN NEXT current_res;
519 $func$ LANGUAGE PLPGSQL;