2 * Copyright (C) 2004-2008 Georgia Public Library Service
3 * Copyright (C) 2007-2008 Equinox Software, Inc.
4 * Mike Rylander <miker@esilibrary.com>
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public License
8 * as published by the Free Software Foundation; either version 2
9 * of the License, or (at your option) any later version.
11 * This program is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
18 DROP SCHEMA IF EXISTS metabib CASCADE;
21 CREATE SCHEMA metabib;
23 CREATE TABLE metabib.metarecord (
24 id BIGSERIAL PRIMARY KEY,
25 fingerprint TEXT NOT NULL,
29 CREATE INDEX metabib_metarecord_master_record_idx ON metabib.metarecord (master_record);
30 CREATE INDEX metabib_metarecord_fingerprint_idx ON metabib.metarecord (fingerprint);
32 CREATE TABLE metabib.identifier_field_entry (
33 id BIGSERIAL PRIMARY KEY,
34 source BIGINT NOT NULL,
37 index_vector tsvector NOT NULL
39 CREATE TRIGGER metabib_identifier_field_entry_fti_trigger
40 BEFORE UPDATE OR INSERT ON metabib.identifier_field_entry
41 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('identifier');
43 CREATE INDEX metabib_identifier_field_entry_index_vector_idx ON metabib.identifier_field_entry USING GIST (index_vector);
44 CREATE INDEX metabib_identifier_field_entry_value_idx ON metabib.identifier_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
45 CREATE INDEX metabib_identifier_field_entry_source_idx ON metabib.identifier_field_entry (source);
48 CREATE TABLE metabib.title_field_entry (
49 id BIGSERIAL PRIMARY KEY,
50 source BIGINT NOT NULL,
53 index_vector tsvector NOT NULL
55 CREATE TRIGGER metabib_title_field_entry_fti_trigger
56 BEFORE UPDATE OR INSERT ON metabib.title_field_entry
57 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('title');
59 CREATE INDEX metabib_title_field_entry_index_vector_idx ON metabib.title_field_entry USING GIST (index_vector);
60 CREATE INDEX metabib_title_field_entry_value_idx ON metabib.title_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
61 CREATE INDEX metabib_title_field_entry_source_idx ON metabib.title_field_entry (source);
64 CREATE TABLE metabib.author_field_entry (
65 id BIGSERIAL PRIMARY KEY,
66 source BIGINT NOT NULL,
69 index_vector tsvector NOT NULL
71 CREATE TRIGGER metabib_author_field_entry_fti_trigger
72 BEFORE UPDATE OR INSERT ON metabib.author_field_entry
73 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('author');
75 CREATE INDEX metabib_author_field_entry_index_vector_idx ON metabib.author_field_entry USING GIST (index_vector);
76 CREATE INDEX metabib_author_field_entry_value_idx ON metabib.author_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
77 CREATE INDEX metabib_author_field_entry_source_idx ON metabib.author_field_entry (source);
80 CREATE TABLE metabib.subject_field_entry (
81 id BIGSERIAL PRIMARY KEY,
82 source BIGINT NOT NULL,
85 index_vector tsvector NOT NULL
87 CREATE TRIGGER metabib_subject_field_entry_fti_trigger
88 BEFORE UPDATE OR INSERT ON metabib.subject_field_entry
89 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('subject');
91 CREATE INDEX metabib_subject_field_entry_index_vector_idx ON metabib.subject_field_entry USING GIST (index_vector);
92 CREATE INDEX metabib_subject_field_entry_value_idx ON metabib.subject_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
93 CREATE INDEX metabib_subject_field_entry_source_idx ON metabib.subject_field_entry (source);
96 CREATE TABLE metabib.keyword_field_entry (
97 id BIGSERIAL PRIMARY KEY,
98 source BIGINT NOT NULL,
101 index_vector tsvector NOT NULL
103 CREATE TRIGGER metabib_keyword_field_entry_fti_trigger
104 BEFORE UPDATE OR INSERT ON metabib.keyword_field_entry
105 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('keyword');
107 CREATE INDEX metabib_keyword_field_entry_index_vector_idx ON metabib.keyword_field_entry USING GIST (index_vector);
108 CREATE INDEX metabib_keyword_field_entry_value_idx ON metabib.keyword_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
109 CREATE INDEX metabib_keyword_field_entry_source_idx ON metabib.keyword_field_entry (source);
112 CREATE TABLE metabib.series_field_entry (
113 id BIGSERIAL PRIMARY KEY,
114 source BIGINT NOT NULL,
117 index_vector tsvector NOT NULL
119 CREATE TRIGGER metabib_series_field_entry_fti_trigger
120 BEFORE UPDATE OR INSERT ON metabib.series_field_entry
121 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('series');
123 CREATE INDEX metabib_series_field_entry_index_vector_idx ON metabib.series_field_entry USING GIST (index_vector);
124 CREATE INDEX metabib_series_field_entry_value_idx ON metabib.series_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
125 CREATE INDEX metabib_series_field_entry_source_idx ON metabib.series_field_entry (source);
128 CREATE TABLE metabib.facet_entry (
129 id BIGSERIAL PRIMARY KEY,
130 source BIGINT NOT NULL,
134 CREATE INDEX metabib_facet_entry_field_idx ON metabib.facet_entry (field);
135 CREATE INDEX metabib_facet_entry_value_idx ON metabib.facet_entry (SUBSTRING(value,1,1024));
136 CREATE INDEX metabib_facet_entry_source_idx ON metabib.facet_entry (source);
138 CREATE TABLE metabib.browse_entry (
139 id BIGSERIAL PRIMARY KEY,
141 index_vector tsvector
143 CREATE INDEX metabib_browse_entry_index_vector_idx ON metabib.browse_entry USING GIN (index_vector);
144 CREATE TRIGGER metabib_browse_entry_fti_trigger
145 BEFORE INSERT OR UPDATE ON metabib.browse_entry
146 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('keyword');
149 CREATE TABLE metabib.browse_entry_def_map (
150 id BIGSERIAL PRIMARY KEY,
151 entry BIGINT REFERENCES metabib.browse_entry (id),
152 def INT REFERENCES config.metabib_field (id),
153 source BIGINT REFERENCES biblio.record_entry (id)
155 CREATE INDEX browse_entry_def_map_def_idx ON metabib.browse_entry_def_map (def);
156 CREATE INDEX browse_entry_def_map_entry_idx ON metabib.browse_entry_def_map (entry);
157 CREATE INDEX browse_entry_def_map_source_idx ON metabib.browse_entry_def_map (source);
161 CREATE OR REPLACE FUNCTION metabib.facet_normalize_trigger () RETURNS TRIGGER AS $$
166 facet_text := NEW.value;
169 SELECT n.func AS func,
170 n.param_count AS param_count,
172 FROM config.index_normalizer n
173 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
174 WHERE m.field = NEW.field AND m.pos < 0
177 EXECUTE 'SELECT ' || normalizer.func || '(' ||
178 quote_literal( facet_text ) ||
180 WHEN normalizer.param_count > 0
181 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
188 NEW.value = facet_text;
194 CREATE TRIGGER facet_normalize_tgr
195 BEFORE UPDATE OR INSERT ON metabib.facet_entry
196 FOR EACH ROW EXECUTE PROCEDURE metabib.facet_normalize_trigger();
198 CREATE OR REPLACE FUNCTION evergreen.facet_force_nfc() RETURNS TRIGGER AS $$
200 NEW.value := force_unicode_normal_form(NEW.value,'NFC');
205 CREATE TRIGGER facet_force_nfc_tgr
206 BEFORE UPDATE OR INSERT ON metabib.facet_entry
207 FOR EACH ROW EXECUTE PROCEDURE evergreen.facet_force_nfc();
209 CREATE TABLE metabib.record_attr (
210 id BIGINT PRIMARY KEY REFERENCES biblio.record_entry (id) ON DELETE CASCADE,
211 attrs HSTORE NOT NULL DEFAULT ''::HSTORE
213 CREATE INDEX metabib_svf_attrs_idx ON metabib.record_attr USING GIST (attrs);
214 CREATE INDEX metabib_svf_date1_idx ON metabib.record_attr ((attrs->'date1'));
215 CREATE INDEX metabib_svf_dates_idx ON metabib.record_attr ((attrs->'date1'),(attrs->'date2'));
217 -- Back-compat view ... we're moving to an HSTORE world
218 CREATE TYPE metabib.rec_desc_type AS (
236 CREATE VIEW metabib.rec_descriptor AS
239 (populate_record(NULL::metabib.rec_desc_type, attrs)).*
240 FROM metabib.record_attr;
242 -- Use a sequence that matches previous version, for easier upgrading.
243 CREATE SEQUENCE metabib.full_rec_id_seq;
245 CREATE TABLE metabib.real_full_rec (
246 id BIGINT NOT NULL DEFAULT NEXTVAL('metabib.full_rec_id_seq'::REGCLASS),
247 record BIGINT NOT NULL,
248 tag CHAR(3) NOT NULL,
253 index_vector tsvector NOT NULL
255 ALTER TABLE metabib.real_full_rec ADD PRIMARY KEY (id);
257 CREATE INDEX metabib_full_rec_tag_subfield_idx ON metabib.real_full_rec (tag,subfield);
258 CREATE INDEX metabib_full_rec_value_idx ON metabib.real_full_rec (substring(value,1,1024));
259 /* Enable LIKE to use an index for database clusters with locales other than C or POSIX */
260 CREATE INDEX metabib_full_rec_value_tpo_index ON metabib.real_full_rec (substring(value,1,1024) text_pattern_ops);
261 CREATE INDEX metabib_full_rec_record_idx ON metabib.real_full_rec (record);
262 CREATE INDEX metabib_full_rec_index_vector_idx ON metabib.real_full_rec USING GIST (index_vector);
263 CREATE INDEX metabib_full_rec_isxn_caseless_idx
264 ON metabib.real_full_rec (LOWER(value))
265 WHERE tag IN ('020', '022', '024');
268 CREATE TRIGGER metabib_full_rec_fti_trigger
269 BEFORE UPDATE OR INSERT ON metabib.real_full_rec
270 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('default');
272 CREATE OR REPLACE VIEW metabib.full_rec AS
279 SUBSTRING(value,1,1024) AS value,
281 FROM metabib.real_full_rec;
283 CREATE OR REPLACE RULE metabib_full_rec_insert_rule
284 AS ON INSERT TO metabib.full_rec
286 INSERT INTO metabib.real_full_rec VALUES (
287 COALESCE(NEW.id, NEXTVAL('metabib.full_rec_id_seq'::REGCLASS)),
297 CREATE OR REPLACE RULE metabib_full_rec_update_rule
298 AS ON UPDATE TO metabib.full_rec
300 UPDATE metabib.real_full_rec SET
306 subfield = NEW.subfield,
308 index_vector = NEW.index_vector
311 CREATE OR REPLACE RULE metabib_full_rec_delete_rule
312 AS ON DELETE TO metabib.full_rec
314 DELETE FROM metabib.real_full_rec WHERE id = OLD.id;
316 CREATE TABLE metabib.metarecord_source_map (
317 id BIGSERIAL PRIMARY KEY,
318 metarecord BIGINT NOT NULL,
319 source BIGINT NOT NULL
321 CREATE INDEX metabib_metarecord_source_map_metarecord_idx ON metabib.metarecord_source_map (metarecord);
322 CREATE INDEX metabib_metarecord_source_map_source_record_idx ON metabib.metarecord_source_map (source);
324 CREATE TYPE metabib.field_entry_template AS (
335 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
337 bib biblio.record_entry%ROWTYPE;
338 idx config.metabib_field%ROWTYPE;
339 xfrm config.xml_transform%ROWTYPE;
341 transformed_xml TEXT;
343 xml_node_list TEXT[];
348 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
349 output_row metabib.field_entry_template%ROWTYPE;
353 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
355 -- Loop over the indexing entries
356 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
358 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
360 -- See if we can skip the XSLT ... it's expensive
361 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
362 -- Can't skip the transform
363 IF xfrm.xslt <> '---' THEN
364 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
366 transformed_xml := bib.marc;
369 prev_xfrm := xfrm.name;
372 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
375 FOR xml_node IN SELECT x FROM unnest(xml_node_list) AS x LOOP
376 CONTINUE WHEN xml_node !~ E'^\\s*<';
378 curr_text := ARRAY_TO_STRING(
379 oils_xpath( '//text()',
380 REGEXP_REPLACE( -- This escapes all &s not followed by "amp;". Data ise returned from oils_xpath (above) in UTF-8, not entity encoded
381 REGEXP_REPLACE( -- This escapes embeded <s
383 $re$(>[^<]+)(<)([^>]+<)$re$,
395 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
397 IF raw_text IS NOT NULL THEN
398 raw_text := raw_text || joiner;
401 raw_text := COALESCE(raw_text,'') || curr_text;
403 -- autosuggest/metabib.browse_entry
404 IF idx.browse_field THEN
406 IF idx.browse_xpath IS NOT NULL AND idx.browse_xpath <> '' THEN
407 browse_text := oils_xpath_string( idx.browse_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
409 browse_text := curr_text;
412 output_row.field_class = idx.field_class;
413 output_row.field = idx.id;
414 output_row.source = rid;
415 output_row.value = BTRIM(REGEXP_REPLACE(browse_text, E'\\s+', ' ', 'g'));
417 output_row.browse_field = TRUE;
418 RETURN NEXT output_row;
419 output_row.browse_field = FALSE;
422 -- insert raw node text for faceting
423 IF idx.facet_field THEN
425 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
426 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
428 facet_text := curr_text;
431 output_row.field_class = idx.field_class;
432 output_row.field = -1 * idx.id;
433 output_row.source = rid;
434 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
436 output_row.facet_field = TRUE;
437 RETURN NEXT output_row;
438 output_row.facet_field = FALSE;
443 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
445 -- insert combined node text for searching
446 IF idx.search_field THEN
447 output_row.field_class = idx.field_class;
448 output_row.field = idx.id;
449 output_row.source = rid;
450 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
452 output_row.search_field = TRUE;
453 RETURN NEXT output_row;
460 $func$ LANGUAGE PLPGSQL;
462 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries( bib_id BIGINT ) RETURNS VOID AS $func$
465 ind_data metabib.field_entry_template%ROWTYPE;
466 mbe_row metabib.browse_entry%ROWTYPE;
469 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
471 FOR fclass IN SELECT * FROM config.metabib_class LOOP
472 -- RAISE NOTICE 'Emptying out %', fclass.name;
473 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
475 DELETE FROM metabib.facet_entry WHERE source = bib_id;
476 DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
479 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
480 IF ind_data.field < 0 THEN
481 ind_data.field = -1 * ind_data.field;
484 IF ind_data.facet_field THEN
485 INSERT INTO metabib.facet_entry (field, source, value)
486 VALUES (ind_data.field, ind_data.source, ind_data.value);
489 IF ind_data.browse_field THEN
490 -- A caveat about this SELECT: this should take care of replacing
491 -- old mbe rows when data changes, but not if normalization (by
492 -- which I mean specifically the output of
493 -- evergreen.oils_tsearch2()) changes. It may or may not be
494 -- expensive to add a comparison of index_vector to index_vector
495 -- to the WHERE clause below.
496 SELECT INTO mbe_row * FROM metabib.browse_entry WHERE value = ind_data.value;
498 mbe_id := mbe_row.id;
500 INSERT INTO metabib.browse_entry (value) VALUES
501 (metabib.browse_normalize(ind_data.value, ind_data.field));
502 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
505 INSERT INTO metabib.browse_entry_def_map (entry, def, source)
506 VALUES (mbe_id, ind_data.field, ind_data.source);
509 IF ind_data.search_field THEN
511 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
513 quote_literal(ind_data.field) || $$, $$ ||
514 quote_literal(ind_data.source) || $$, $$ ||
515 quote_literal(ind_data.value) ||
523 $func$ LANGUAGE PLPGSQL;
525 -- default to a space joiner
526 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
527 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
530 CREATE OR REPLACE FUNCTION authority.flatten_marc ( rid BIGINT ) RETURNS SETOF authority.full_rec AS $func$
532 auth authority.record_entry%ROWTYPE;
533 output authority.full_rec%ROWTYPE;
536 SELECT INTO auth * FROM authority.record_entry WHERE id = rid;
538 FOR field IN SELECT * FROM vandelay.flatten_marc( auth.marc ) LOOP
539 output.record := rid;
540 output.ind1 := field.ind1;
541 output.ind2 := field.ind2;
542 output.tag := field.tag;
543 output.subfield := field.subfield;
544 output.value := field.value;
549 $func$ LANGUAGE PLPGSQL;
551 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
553 bib biblio.record_entry%ROWTYPE;
554 output metabib.full_rec%ROWTYPE;
557 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
559 FOR field IN SELECT * FROM vandelay.flatten_marc( bib.marc ) LOOP
560 output.record := rid;
561 output.ind1 := field.ind1;
562 output.ind2 := field.ind2;
563 output.tag := field.tag;
564 output.subfield := field.subfield;
565 output.value := field.value;
570 $func$ LANGUAGE PLPGSQL;
572 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
579 retval config.marc21_rec_type_map%ROWTYPE;
581 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
583 IF ldr IS NULL OR ldr = '' THEN
584 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
588 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
589 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
592 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
593 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
595 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
597 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
600 IF retval.code IS NULL THEN
601 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
606 $func$ LANGUAGE PLPGSQL;
608 CREATE OR REPLACE FUNCTION biblio.marc21_record_type( rid BIGINT ) RETURNS config.marc21_rec_type_map AS $func$
609 SELECT * FROM vandelay.marc21_record_type( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
612 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
619 rtype := (vandelay.marc21_record_type( marc )).code;
620 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
621 IF ff_pos.tag = 'ldr' THEN
622 val := oils_xpath_string('//*[local-name()="leader"]', marc);
623 IF val IS NOT NULL THEN
624 val := SUBSTRING( val, ff_pos.start_pos + 1, ff_pos.length );
628 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
629 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
633 val := REPEAT( ff_pos.default_val, ff_pos.length );
639 $func$ LANGUAGE PLPGSQL;
641 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
642 SELECT * FROM vandelay.marc21_extract_fixed_field( (SELECT marc FROM biblio.record_entry WHERE id = $1), $2 );
645 -- CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
646 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
651 output biblio.record_ff_map%ROWTYPE;
653 rtype := (vandelay.marc21_record_type( marc )).code;
655 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
656 output.ff_name := ff_pos.fixed_field;
657 output.ff_value := NULL;
659 IF ff_pos.tag = 'ldr' THEN
660 output.ff_value := oils_xpath_string('//*[local-name()="leader"]', marc);
661 IF output.ff_value IS NOT NULL THEN
662 output.ff_value := SUBSTRING( output.ff_value, ff_pos.start_pos + 1, ff_pos.length );
664 output.ff_value := NULL;
667 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
668 output.ff_value := SUBSTRING( tag_data, ff_pos.start_pos + 1, ff_pos.length );
669 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
671 output.ff_value := NULL;
679 $func$ LANGUAGE PLPGSQL;
681 CREATE OR REPLACE FUNCTION biblio.marc21_extract_all_fixed_fields( rid BIGINT ) RETURNS SETOF biblio.record_ff_map AS $func$
682 SELECT $1 AS record, ff_name, ff_value FROM vandelay.marc21_extract_all_fixed_fields( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
685 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
686 SELECT id, $1 AS record, ptype, subfield, value FROM vandelay.marc21_physical_characteristics( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
689 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
702 IF marc IS NULL OR marc = '' THEN
706 -- First, the count of tags
707 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
709 -- now go through a bunch of pain to get the record type
710 IF best_type IS NOT NULL THEN
711 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
713 IF ldr IS NOT NULL THEN
714 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
715 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
718 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
719 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
721 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
723 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
725 IF type_map.code IS NOT NULL THEN
726 IF best_type = type_map.code THEN
727 qual := qual + qual / 2;
730 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = 'Lang' AND rec_type = type_map.code ORDER BY tag DESC LOOP
731 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
732 IF ff_tag_data = best_lang THEN
740 -- Now look for some quality metrics
742 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
747 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
754 $func$ LANGUAGE PLPGSQL;
756 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
758 idx config.biblio_fingerprint%ROWTYPE;
759 xfrm config.xml_transform%ROWTYPE;
761 transformed_xml TEXT;
763 xml_node_list TEXT[];
765 output_text TEXT := '';
768 IF marc IS NULL OR marc = '' THEN
772 -- Loop over the indexing entries
773 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
775 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
777 -- See if we can skip the XSLT ... it's expensive
778 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
779 -- Can't skip the transform
780 IF xfrm.xslt <> '---' THEN
781 transformed_xml := oils_xslt_process(marc,xfrm.xslt);
783 transformed_xml := marc;
786 prev_xfrm := xfrm.name;
789 raw_text := COALESCE(
797 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
806 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
807 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
809 IF idx.first_word IS TRUE THEN
810 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
813 output_text := output_text || REGEXP_REPLACE(raw_text, E'\\s+', '', 'g');
820 $func$ LANGUAGE PLPGSQL;
822 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
823 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
826 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
828 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
832 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
833 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
838 $func$ LANGUAGE PLPGSQL;
840 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_full_rec( bib_id BIGINT ) RETURNS VOID AS $func$
842 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
844 DELETE FROM metabib.real_full_rec WHERE record = bib_id;
846 INSERT INTO metabib.real_full_rec (record, tag, ind1, ind2, subfield, value)
847 SELECT record, tag, ind1, ind2, subfield, value FROM biblio.flatten_marc( bib_id );
851 $func$ LANGUAGE PLPGSQL;
853 CREATE OR REPLACE FUNCTION biblio.extract_located_uris( bib_id BIGINT, marcxml TEXT, editor_id INT ) RETURNS VOID AS $func$
860 uri_owner_list TEXT[];
868 -- Clear any URI mappings and call numbers for this bib.
869 -- This leads to acn / auricnm inflation, but also enables
870 -- old acn/auricnm's to go away and for bibs to be deleted.
871 FOR uri_cn_id IN SELECT id FROM asset.call_number WHERE record = bib_id AND label = '##URI##' AND NOT deleted LOOP
872 DELETE FROM asset.uri_call_number_map WHERE call_number = uri_cn_id;
873 DELETE FROM asset.call_number WHERE id = uri_cn_id;
876 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',marcxml);
877 IF ARRAY_UPPER(uris,1) > 0 THEN
878 FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
879 -- First we pull info out of the 856
882 uri_href := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
883 uri_label := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()',uri_xml))[1];
884 uri_use := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()',uri_xml))[1];
886 IF uri_label IS NULL THEN
887 uri_label := uri_href;
889 CONTINUE WHEN uri_href IS NULL;
891 -- Get the distinct list of libraries wanting to use
893 DISTINCT REGEXP_REPLACE(
895 $re$^.*?\((\w+)\).*$$re$,
898 ) INTO uri_owner_list
901 '//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',
906 IF ARRAY_UPPER(uri_owner_list,1) > 0 THEN
908 -- look for a matching uri
909 IF uri_use IS NULL THEN
910 SELECT id INTO uri_id
912 WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active
914 IF NOT FOUND THEN -- create one
915 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
916 SELECT id INTO uri_id
918 WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active;
921 SELECT id INTO uri_id
923 WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active
925 IF NOT FOUND THEN -- create one
926 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
927 SELECT id INTO uri_id
929 WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
933 FOR j IN 1 .. ARRAY_UPPER(uri_owner_list, 1) LOOP
934 uri_owner := uri_owner_list[j];
936 SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = uri_owner;
937 CONTINUE WHEN NOT FOUND;
939 -- we need a call number to link through
940 SELECT id INTO uri_cn_id FROM asset.call_number WHERE owning_lib = uri_owner_id AND record = bib_id AND label = '##URI##' AND NOT deleted;
942 INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
943 VALUES (uri_owner_id, bib_id, 'now', 'now', editor_id, editor_id, '##URI##');
944 SELECT id INTO uri_cn_id FROM asset.call_number WHERE owning_lib = uri_owner_id AND record = bib_id AND label = '##URI##' AND NOT deleted;
947 -- now, link them if they're not already
948 SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
950 INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
962 $func$ LANGUAGE PLPGSQL;
964 CREATE OR REPLACE FUNCTION metabib.remap_metarecord_for_bib( bib_id BIGINT, fp TEXT ) RETURNS BIGINT AS $func$
968 tmp_mr metabib.metarecord%ROWTYPE;
969 deleted_mrs BIGINT[];
972 DELETE FROM metabib.metarecord_source_map WHERE source = bib_id; -- Rid ourselves of the search-estimate-killing linkage
974 FOR tmp_mr IN SELECT m.* FROM metabib.metarecord m JOIN metabib.metarecord_source_map s ON (s.metarecord = m.id) WHERE s.source = bib_id LOOP
976 IF old_mr IS NULL AND fp = tmp_mr.fingerprint THEN -- Find the first fingerprint-matching
979 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
980 IF source_count = 0 THEN -- No other records
981 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
982 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
988 IF old_mr IS NULL THEN -- we found no suitable, preexisting MR based on old source maps
989 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp; -- is there one for our current fingerprint?
990 IF old_mr IS NULL THEN -- nope, create one and grab its id
991 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( fp, bib_id );
992 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp;
993 ELSE -- indeed there is. update it with a null cache and recalcualated master record
994 UPDATE metabib.metarecord
996 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
999 ELSE -- there was one we already attached to, update its mods cache and master_record
1000 UPDATE metabib.metarecord
1002 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
1006 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, bib_id); -- new source mapping
1008 IF ARRAY_UPPER(deleted_mrs,1) > 0 THEN
1009 UPDATE action.hold_request SET target = old_mr WHERE target IN ( SELECT unnest(deleted_mrs) ) AND hold_type = 'M'; -- if we had to delete any MRs above, make sure their holds are moved
1015 $func$ LANGUAGE PLPGSQL;
1017 CREATE OR REPLACE FUNCTION biblio.map_authority_linking (bibid BIGINT, marc TEXT) RETURNS BIGINT AS $func$
1018 DELETE FROM authority.bib_linking WHERE bib = $1;
1019 INSERT INTO authority.bib_linking (bib, authority)
1022 FROM ( SELECT DISTINCT $1 AS bib,
1023 BTRIM(remove_paren_substring(txt))::BIGINT AS authority
1024 FROM unnest(oils_xpath('//*[@code="0"]/text()',$2)) x(txt)
1025 WHERE BTRIM(remove_paren_substring(txt)) ~ $re$^\d+$$re$
1026 ) y JOIN authority.record_entry r ON r.id = y.authority;
1028 $func$ LANGUAGE SQL;
1030 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
1031 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
1033 transformed_xml TEXT;
1036 xfrm config.xml_transform%ROWTYPE;
1038 new_attrs HSTORE := ''::HSTORE;
1039 attr_def config.record_attr_definition%ROWTYPE;
1042 IF NEW.deleted IS TRUE THEN -- If this bib is deleted
1043 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage
1044 DELETE FROM metabib.record_attr WHERE id = NEW.id; -- Kill the attrs hash, useless on deleted records
1045 DELETE FROM authority.bib_linking WHERE bib = NEW.id; -- Avoid updating fields in bibs that are no longer visible
1046 DELETE FROM biblio.peer_bib_copy_map WHERE peer_record = NEW.id; -- Separate any multi-homed items
1047 DELETE FROM metabib.browse_entry_def_map WHERE source = NEW.id; -- Don't auto-suggest deleted bibs
1048 RETURN NEW; -- and we're done
1051 IF TG_OP = 'UPDATE' THEN -- re-ingest?
1052 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
1054 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
1059 -- Record authority linking
1060 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_linking' AND enabled;
1062 PERFORM biblio.map_authority_linking( NEW.id, NEW.marc );
1065 -- Flatten and insert the mfr data
1066 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_full_rec' AND enabled;
1068 PERFORM metabib.reingest_metabib_full_rec(NEW.id);
1070 -- Now we pull out attribute data, which is dependent on the mfr for all but XPath-based fields
1071 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_rec_descriptor' AND enabled;
1073 FOR attr_def IN SELECT * FROM config.record_attr_definition ORDER BY format LOOP
1075 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
1076 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(value), COALESCE(attr_def.joiner,' ')) INTO attr_value
1077 FROM (SELECT * FROM metabib.full_rec ORDER BY tag, subfield) AS x
1078 WHERE record = NEW.id
1079 AND tag LIKE attr_def.tag
1081 WHEN attr_def.sf_list IS NOT NULL
1082 THEN POSITION(subfield IN attr_def.sf_list) > 0
1089 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
1090 attr_value := biblio.marc21_extract_fixed_field(NEW.id, attr_def.fixed_field);
1092 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
1094 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
1096 -- See if we can skip the XSLT ... it's expensive
1097 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
1098 -- Can't skip the transform
1099 IF xfrm.xslt <> '---' THEN
1100 transformed_xml := oils_xslt_process(NEW.marc,xfrm.xslt);
1102 transformed_xml := NEW.marc;
1105 prev_xfrm := xfrm.name;
1108 IF xfrm.name IS NULL THEN
1109 -- just grab the marcxml (empty) transform
1110 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
1111 prev_xfrm := xfrm.name;
1114 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
1116 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
1117 SELECT m.value INTO attr_value
1118 FROM biblio.marc21_physical_characteristics(NEW.id) v
1119 JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
1120 WHERE v.subfield = attr_def.phys_char_sf
1121 LIMIT 1; -- Just in case ...
1125 -- apply index normalizers to attr_value
1127 SELECT n.func AS func,
1128 n.param_count AS param_count,
1130 FROM config.index_normalizer n
1131 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
1132 WHERE attr = attr_def.name
1134 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1135 COALESCE( quote_literal( attr_value ), 'NULL' ) ||
1137 WHEN normalizer.param_count > 0
1138 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1141 ')' INTO attr_value;
1145 -- Add the new value to the hstore
1146 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
1150 IF TG_OP = 'INSERT' OR OLD.deleted THEN -- initial insert OR revivication
1151 INSERT INTO metabib.record_attr (id, attrs) VALUES (NEW.id, new_attrs);
1153 UPDATE metabib.record_attr SET attrs = new_attrs WHERE id = NEW.id;
1159 -- Gather and insert the field entry data
1160 PERFORM metabib.reingest_metabib_field_entries(NEW.id);
1162 -- Located URI magic
1163 IF TG_OP = 'INSERT' THEN
1164 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1166 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1169 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1171 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1175 -- (re)map metarecord-bib linking
1176 IF TG_OP = 'INSERT' THEN -- if not deleted and performing an insert, check for the flag
1177 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
1179 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1181 ELSE -- we're doing an update, and we're not deleted, remap
1182 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_update' AND enabled;
1184 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1190 $func$ LANGUAGE PLPGSQL;
1192 CREATE OR REPLACE FUNCTION metabib.browse_normalize(facet_text TEXT, mapped_field INT) RETURNS TEXT AS $$
1198 SELECT n.func AS func,
1199 n.param_count AS param_count,
1201 FROM config.index_normalizer n
1202 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
1203 WHERE m.field = mapped_field AND m.pos < 0
1206 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1207 quote_literal( facet_text ) ||
1209 WHEN normalizer.param_count > 0
1210 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1213 ')' INTO facet_text;
1220 $$ LANGUAGE PLPGSQL;
1223 -- This mimics a specific part of QueryParser, turning the first part of a
1224 -- classed search (search_class) into a set of classes and possibly fields.
1225 -- search_class might look like "author" or "title|proper" or "ti|uniform"
1226 -- or "au" or "au|corporate|personal" or anything like that, where the first
1227 -- element of the list you get by separating on the "|" character is either
1228 -- a registered class (config.metabib_class) or an alias
1229 -- (config.metabib_search_alias), and the rest of any such elements are
1230 -- fields (config.metabib_field).
1232 FUNCTION metabib.search_class_to_registered_components(search_class TEXT)
1233 RETURNS SETOF RECORD AS $func$
1235 search_parts TEXT[];
1237 search_part_count INTEGER;
1239 registered_class config.metabib_class%ROWTYPE;
1240 registered_alias config.metabib_search_alias%ROWTYPE;
1241 registered_field config.metabib_field%ROWTYPE;
1243 search_parts := REGEXP_SPLIT_TO_ARRAY(search_class, E'\\|');
1245 search_part_count := ARRAY_LENGTH(search_parts, 1);
1246 IF search_part_count = 0 THEN
1249 SELECT INTO registered_class
1250 * FROM config.metabib_class WHERE name = search_parts[1];
1252 IF search_part_count < 2 THEN -- all fields
1253 rec := (registered_class.name, NULL::INTEGER);
1257 FOR field_name IN SELECT *
1258 FROM UNNEST(search_parts[2:search_part_count]) LOOP
1259 SELECT INTO registered_field
1260 * FROM config.metabib_field
1261 WHERE name = field_name AND
1262 field_class = registered_class.name;
1264 rec := (registered_class.name, registered_field.id);
1269 -- maybe we have an alias?
1270 SELECT INTO registered_alias
1271 * FROM config.metabib_search_alias WHERE alias=search_parts[1];
1275 IF search_part_count < 2 THEN -- return w/e the alias says
1277 registered_alias.field_class, registered_alias.field
1282 FOR field_name IN SELECT *
1283 FROM UNNEST(search_parts[2:search_part_count]) LOOP
1284 SELECT INTO registered_field
1285 * FROM config.metabib_field
1286 WHERE name = field_name AND
1287 field_class = registered_alias.field_class;
1290 registered_alias.field_class,
1301 $func$ LANGUAGE PLPGSQL ROWS 1;
1304 -- Given a string such as a user might type into a search box, prepare
1305 -- two changed variants for TO_TSQUERY(). See
1306 -- http://www.postgresql.org/docs/9.0/static/textsearch-controls.html
1307 -- The first variant is normalized to match indexed documents regardless
1308 -- of diacritics. The second variant keeps its diacritics for proper
1309 -- highlighting via TS_HEADLINE().
1311 FUNCTION metabib.autosuggest_prepare_tsquery(orig TEXT) RETURNS TEXT[] AS
1314 orig_ended_in_space BOOLEAN;
1319 orig_ended_in_space := orig ~ E'\\s$';
1321 orig := ARRAY_TO_STRING(
1322 evergreen.regexp_split_to_array(orig, E'\\W+'), ' '
1325 normalized := public.search_normalize(orig); -- also trim()s
1326 plain := trim(orig);
1328 IF NOT orig_ended_in_space THEN
1329 plain := plain || ':*';
1330 normalized := normalized || ':*';
1333 plain := ARRAY_TO_STRING(
1334 evergreen.regexp_split_to_array(plain, E'\\s+'), ' & '
1336 normalized := ARRAY_TO_STRING(
1337 evergreen.regexp_split_to_array(normalized, E'\\s+'), ' & '
1340 RETURN ARRAY[normalized, plain];
1342 $$ LANGUAGE PLPGSQL;
1346 FUNCTION metabib.suggest_browse_entries(
1347 raw_query_text TEXT, -- actually typed by humans at the UI level
1348 search_class TEXT, -- 'alias' or 'class' or 'class|field..', etc
1349 headline_opts TEXT, -- markup options for ts_headline()
1350 visibility_org INTEGER,-- null if you don't want opac visibility test
1351 query_limit INTEGER,-- use in LIMIT clause of interal query
1352 normalization INTEGER -- argument to TS_RANK_CD()
1354 value TEXT, -- plain
1356 buoyant_and_class_match BOOL,
1358 field_weight INTEGER,
1361 match TEXT -- marked up
1364 prepared_query_texts TEXT[];
1366 plain_query TSQUERY;
1367 opac_visibility_join TEXT;
1368 search_class_join TEXT;
1371 prepared_query_texts := metabib.autosuggest_prepare_tsquery(raw_query_text);
1373 query := TO_TSQUERY('keyword', prepared_query_texts[1]);
1374 plain_query := TO_TSQUERY('keyword', prepared_query_texts[2]);
1376 IF visibility_org IS NOT NULL THEN
1377 opac_visibility_join := '
1378 JOIN asset.opac_visible_copies aovc ON (
1379 aovc.record = x.source AND
1380 aovc.circ_lib IN (SELECT id FROM actor.org_unit_descendants($4))
1383 opac_visibility_join := '';
1386 -- The following determines whether we only provide suggestsons matching
1387 -- the user's selected search_class, or whether we show other suggestions
1388 -- too. The reason for MIN() is that for search_classes like
1389 -- 'title|proper|uniform' you would otherwise get multiple rows. The
1390 -- implication is that if title as a class doesn't have restrict,
1391 -- nor does the proper field, but the uniform field does, you're going
1392 -- to get 'false' for your overall evaluation of 'should we restrict?'
1393 -- To invert that, change from MIN() to MAX().
1397 MIN(cmc.restrict::INT) AS restrict_class,
1398 MIN(cmf.restrict::INT) AS restrict_field
1399 FROM metabib.search_class_to_registered_components(search_class)
1400 AS _registered (field_class TEXT, field INT)
1402 config.metabib_class cmc ON (cmc.name = _registered.field_class)
1404 config.metabib_field cmf ON (cmf.id = _registered.field);
1406 -- evaluate 'should we restrict?'
1407 IF r_fields.restrict_field::BOOL OR r_fields.restrict_class::BOOL THEN
1408 search_class_join := '
1410 metabib.search_class_to_registered_components($2)
1411 AS _registered (field_class TEXT, field INT) ON (
1412 (_registered.field IS NULL AND
1413 _registered.field_class = cmf.field_class) OR
1414 (_registered.field = cmf.id)
1418 search_class_join := '
1420 metabib.search_class_to_registered_components($2)
1421 AS _registered (field_class TEXT, field INT) ON (
1422 _registered.field_class = cmc.name
1427 RETURN QUERY EXECUTE '
1436 TS_HEADLINE(value, $7, $3)
1437 FROM (SELECT DISTINCT
1440 cmc.buoyant AND _registered.field_class IS NOT NULL AS push,
1441 _registered.field = cmf.id AS restrict,
1443 TS_RANK_CD(mbe.index_vector, $1, $6),
1446 FROM metabib.browse_entry_def_map mbedm
1447 JOIN (SELECT * FROM metabib.browse_entry WHERE index_vector @@ $1 LIMIT 10000) mbe ON (mbe.id = mbedm.entry)
1448 JOIN config.metabib_field cmf ON (cmf.id = mbedm.def)
1449 JOIN config.metabib_class cmc ON (cmf.field_class = cmc.name)
1450 ' || search_class_join || '
1451 ORDER BY 3 DESC, 4 DESC NULLS LAST, 5 DESC, 6 DESC, 7 DESC, 1 ASC
1453 ' || opac_visibility_join || '
1454 ORDER BY 3 DESC, 4 DESC NULLS LAST, 5 DESC, 6 DESC, 7 DESC, 1 ASC
1456 ' -- sic, repeat the order by clause in the outer select too
1458 query, search_class, headline_opts,
1459 visibility_org, query_limit, normalization, plain_query
1463 -- buoyant AND chosen class = match class
1464 -- chosen field = match field
1471 $func$ LANGUAGE PLPGSQL;