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);
47 CREATE TABLE metabib.combined_identifier_field_entry (
48 record BIGINT NOT NULL,
49 metabib_field INT NULL,
50 index_vector tsvector NOT NULL
52 CREATE UNIQUE INDEX metabib_combined_identifier_field_entry_fakepk_idx ON metabib.combined_identifier_field_entry (record, COALESCE(metabib_field::TEXT,''));
53 CREATE INDEX metabib_combined_identifier_field_entry_index_vector_idx ON metabib.combined_identifier_field_entry USING GIST (index_vector);
54 CREATE INDEX metabib_combined_identifier_field_source_idx ON metabib.combined_identifier_field_entry (metabib_field);
56 CREATE TABLE metabib.title_field_entry (
57 id BIGSERIAL PRIMARY KEY,
58 source BIGINT NOT NULL,
61 index_vector tsvector NOT NULL
63 CREATE TRIGGER metabib_title_field_entry_fti_trigger
64 BEFORE UPDATE OR INSERT ON metabib.title_field_entry
65 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('title');
67 CREATE INDEX metabib_title_field_entry_index_vector_idx ON metabib.title_field_entry USING GIST (index_vector);
68 CREATE INDEX metabib_title_field_entry_value_idx ON metabib.title_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
69 CREATE INDEX metabib_title_field_entry_source_idx ON metabib.title_field_entry (source);
71 CREATE TABLE metabib.combined_title_field_entry (
72 record BIGINT NOT NULL,
73 metabib_field INT NULL,
74 index_vector tsvector NOT NULL
76 CREATE UNIQUE INDEX metabib_combined_title_field_entry_fakepk_idx ON metabib.combined_title_field_entry (record, COALESCE(metabib_field::TEXT,''));
77 CREATE INDEX metabib_combined_title_field_entry_index_vector_idx ON metabib.combined_title_field_entry USING GIST (index_vector);
78 CREATE INDEX metabib_combined_title_field_source_idx ON metabib.combined_title_field_entry (metabib_field);
80 CREATE TABLE metabib.author_field_entry (
81 id BIGSERIAL PRIMARY KEY,
82 source BIGINT NOT NULL,
85 index_vector tsvector NOT NULL
87 CREATE TRIGGER metabib_author_field_entry_fti_trigger
88 BEFORE UPDATE OR INSERT ON metabib.author_field_entry
89 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('author');
91 CREATE INDEX metabib_author_field_entry_index_vector_idx ON metabib.author_field_entry USING GIST (index_vector);
92 CREATE INDEX metabib_author_field_entry_value_idx ON metabib.author_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
93 CREATE INDEX metabib_author_field_entry_source_idx ON metabib.author_field_entry (source);
95 CREATE TABLE metabib.combined_author_field_entry (
96 record BIGINT NOT NULL,
97 metabib_field INT NULL,
98 index_vector tsvector NOT NULL
100 CREATE UNIQUE INDEX metabib_combined_author_field_entry_fakepk_idx ON metabib.combined_author_field_entry (record, COALESCE(metabib_field::TEXT,''));
101 CREATE INDEX metabib_combined_author_field_entry_index_vector_idx ON metabib.combined_author_field_entry USING GIST (index_vector);
102 CREATE INDEX metabib_combined_author_field_source_idx ON metabib.combined_author_field_entry (metabib_field);
104 CREATE TABLE metabib.subject_field_entry (
105 id BIGSERIAL PRIMARY KEY,
106 source BIGINT NOT NULL,
109 index_vector tsvector NOT NULL
111 CREATE TRIGGER metabib_subject_field_entry_fti_trigger
112 BEFORE UPDATE OR INSERT ON metabib.subject_field_entry
113 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('subject');
115 CREATE INDEX metabib_subject_field_entry_index_vector_idx ON metabib.subject_field_entry USING GIST (index_vector);
116 CREATE INDEX metabib_subject_field_entry_value_idx ON metabib.subject_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
117 CREATE INDEX metabib_subject_field_entry_source_idx ON metabib.subject_field_entry (source);
119 CREATE TABLE metabib.combined_subject_field_entry (
120 record BIGINT NOT NULL,
121 metabib_field INT NULL,
122 index_vector tsvector NOT NULL
124 CREATE UNIQUE INDEX metabib_combined_subject_field_entry_fakepk_idx ON metabib.combined_subject_field_entry (record, COALESCE(metabib_field::TEXT,''));
125 CREATE INDEX metabib_combined_subject_field_entry_index_vector_idx ON metabib.combined_subject_field_entry USING GIST (index_vector);
126 CREATE INDEX metabib_combined_subject_field_source_idx ON metabib.combined_subject_field_entry (metabib_field);
128 CREATE TABLE metabib.keyword_field_entry (
129 id BIGSERIAL PRIMARY KEY,
130 source BIGINT NOT NULL,
133 index_vector tsvector NOT NULL
135 CREATE TRIGGER metabib_keyword_field_entry_fti_trigger
136 BEFORE UPDATE OR INSERT ON metabib.keyword_field_entry
137 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('keyword');
139 CREATE INDEX metabib_keyword_field_entry_index_vector_idx ON metabib.keyword_field_entry USING GIST (index_vector);
140 CREATE INDEX metabib_keyword_field_entry_value_idx ON metabib.keyword_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
141 CREATE INDEX metabib_keyword_field_entry_source_idx ON metabib.keyword_field_entry (source);
143 CREATE TABLE metabib.combined_keyword_field_entry (
144 record BIGINT NOT NULL,
145 metabib_field INT NULL,
146 index_vector tsvector NOT NULL
148 CREATE UNIQUE INDEX metabib_combined_keyword_field_entry_fakepk_idx ON metabib.combined_keyword_field_entry (record, COALESCE(metabib_field::TEXT,''));
149 CREATE INDEX metabib_combined_keyword_field_entry_index_vector_idx ON metabib.combined_keyword_field_entry USING GIST (index_vector);
150 CREATE INDEX metabib_combined_keyword_field_source_idx ON metabib.combined_keyword_field_entry (metabib_field);
152 CREATE TABLE metabib.series_field_entry (
153 id BIGSERIAL PRIMARY KEY,
154 source BIGINT NOT NULL,
157 index_vector tsvector NOT NULL
159 CREATE TRIGGER metabib_series_field_entry_fti_trigger
160 BEFORE UPDATE OR INSERT ON metabib.series_field_entry
161 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('series');
163 CREATE INDEX metabib_series_field_entry_index_vector_idx ON metabib.series_field_entry USING GIST (index_vector);
164 CREATE INDEX metabib_series_field_entry_value_idx ON metabib.series_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
165 CREATE INDEX metabib_series_field_entry_source_idx ON metabib.series_field_entry (source);
167 CREATE TABLE metabib.combined_series_field_entry (
168 record BIGINT NOT NULL,
169 metabib_field INT NULL,
170 index_vector tsvector NOT NULL
172 CREATE UNIQUE INDEX metabib_combined_series_field_entry_fakepk_idx ON metabib.combined_series_field_entry (record, COALESCE(metabib_field::TEXT,''));
173 CREATE INDEX metabib_combined_series_field_entry_index_vector_idx ON metabib.combined_series_field_entry USING GIST (index_vector);
174 CREATE INDEX metabib_combined_series_field_source_idx ON metabib.combined_series_field_entry (metabib_field);
176 CREATE TABLE metabib.facet_entry (
177 id BIGSERIAL PRIMARY KEY,
178 source BIGINT NOT NULL,
182 CREATE INDEX metabib_facet_entry_field_idx ON metabib.facet_entry (field);
183 CREATE INDEX metabib_facet_entry_value_idx ON metabib.facet_entry (SUBSTRING(value,1,1024));
184 CREATE INDEX metabib_facet_entry_source_idx ON metabib.facet_entry (source);
186 CREATE TABLE metabib.browse_entry (
187 id BIGSERIAL PRIMARY KEY,
189 index_vector tsvector
191 CREATE INDEX metabib_browse_entry_index_vector_idx ON metabib.browse_entry USING GIN (index_vector);
192 CREATE TRIGGER metabib_browse_entry_fti_trigger
193 BEFORE INSERT OR UPDATE ON metabib.browse_entry
194 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('keyword');
197 CREATE TABLE metabib.browse_entry_def_map (
198 id BIGSERIAL PRIMARY KEY,
199 entry BIGINT REFERENCES metabib.browse_entry (id),
200 def INT REFERENCES config.metabib_field (id),
201 source BIGINT REFERENCES biblio.record_entry (id)
203 CREATE INDEX browse_entry_def_map_def_idx ON metabib.browse_entry_def_map (def);
204 CREATE INDEX browse_entry_def_map_entry_idx ON metabib.browse_entry_def_map (entry);
205 CREATE INDEX browse_entry_def_map_source_idx ON metabib.browse_entry_def_map (source);
209 CREATE OR REPLACE FUNCTION metabib.facet_normalize_trigger () RETURNS TRIGGER AS $$
214 facet_text := NEW.value;
217 SELECT n.func AS func,
218 n.param_count AS param_count,
220 FROM config.index_normalizer n
221 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
222 WHERE m.field = NEW.field AND m.pos < 0
225 EXECUTE 'SELECT ' || normalizer.func || '(' ||
226 quote_literal( facet_text ) ||
228 WHEN normalizer.param_count > 0
229 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
236 NEW.value = facet_text;
242 CREATE TRIGGER facet_normalize_tgr
243 BEFORE UPDATE OR INSERT ON metabib.facet_entry
244 FOR EACH ROW EXECUTE PROCEDURE metabib.facet_normalize_trigger();
246 CREATE OR REPLACE FUNCTION evergreen.facet_force_nfc() RETURNS TRIGGER AS $$
248 NEW.value := force_unicode_normal_form(NEW.value,'NFC');
253 CREATE TRIGGER facet_force_nfc_tgr
254 BEFORE UPDATE OR INSERT ON metabib.facet_entry
255 FOR EACH ROW EXECUTE PROCEDURE evergreen.facet_force_nfc();
257 CREATE TABLE metabib.record_attr (
258 id BIGINT PRIMARY KEY REFERENCES biblio.record_entry (id) ON DELETE CASCADE,
259 attrs HSTORE NOT NULL DEFAULT ''::HSTORE
261 CREATE INDEX metabib_svf_attrs_idx ON metabib.record_attr USING GIST (attrs);
262 CREATE INDEX metabib_svf_date1_idx ON metabib.record_attr ((attrs->'date1'));
263 CREATE INDEX metabib_svf_dates_idx ON metabib.record_attr ((attrs->'date1'),(attrs->'date2'));
265 -- Back-compat view ... we're moving to an HSTORE world
266 CREATE TYPE metabib.rec_desc_type AS (
284 CREATE VIEW metabib.rec_descriptor AS
287 (populate_record(NULL::metabib.rec_desc_type, attrs)).*
288 FROM metabib.record_attr;
290 -- Use a sequence that matches previous version, for easier upgrading.
291 CREATE SEQUENCE metabib.full_rec_id_seq;
293 CREATE TABLE metabib.real_full_rec (
294 id BIGINT NOT NULL DEFAULT NEXTVAL('metabib.full_rec_id_seq'::REGCLASS),
295 record BIGINT NOT NULL,
296 tag CHAR(3) NOT NULL,
301 index_vector tsvector NOT NULL
303 ALTER TABLE metabib.real_full_rec ADD PRIMARY KEY (id);
305 CREATE INDEX metabib_full_rec_tag_subfield_idx ON metabib.real_full_rec (tag,subfield);
306 CREATE INDEX metabib_full_rec_value_idx ON metabib.real_full_rec (substring(value,1,1024));
307 /* Enable LIKE to use an index for database clusters with locales other than C or POSIX */
308 CREATE INDEX metabib_full_rec_value_tpo_index ON metabib.real_full_rec (substring(value,1,1024) text_pattern_ops);
309 CREATE INDEX metabib_full_rec_record_idx ON metabib.real_full_rec (record);
310 CREATE INDEX metabib_full_rec_index_vector_idx ON metabib.real_full_rec USING GIST (index_vector);
311 CREATE INDEX metabib_full_rec_isxn_caseless_idx
312 ON metabib.real_full_rec (LOWER(value))
313 WHERE tag IN ('020', '022', '024');
314 -- This next index might fully supplant the one above, but leaving both for now.
315 -- (they are not too large)
316 -- The reason we need this index is to ensure that the query parser always
317 -- prefers this index over the simpler tag/subfield index, as this greatly
318 -- increases Vandelay overlay speed for these identifiers, especially when
319 -- a record has many of these fields (around > 4-6 seems like the cutoff
320 -- on at least one PG9.1 system)
321 -- A similar index could be added for other fields (e.g. 010), but one should
322 -- leave out the LOWER() in all other cases.
323 -- TODO: verify whether we can discard the non tag/subfield/substring version
324 -- above (metabib_full_rec_isxn_caseless_idx)
325 CREATE INDEX metabib_full_rec_02x_tag_subfield_lower_substring
326 ON metabib.real_full_rec (tag, subfield, LOWER(substring(value, 1, 1024)))
327 WHERE tag IN ('020', '022', '024');
330 CREATE TRIGGER metabib_full_rec_fti_trigger
331 BEFORE UPDATE OR INSERT ON metabib.real_full_rec
332 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('default');
334 CREATE OR REPLACE VIEW metabib.full_rec AS
341 SUBSTRING(value,1,1024) AS value,
343 FROM metabib.real_full_rec;
345 CREATE OR REPLACE RULE metabib_full_rec_insert_rule
346 AS ON INSERT TO metabib.full_rec
348 INSERT INTO metabib.real_full_rec VALUES (
349 COALESCE(NEW.id, NEXTVAL('metabib.full_rec_id_seq'::REGCLASS)),
359 CREATE OR REPLACE RULE metabib_full_rec_update_rule
360 AS ON UPDATE TO metabib.full_rec
362 UPDATE metabib.real_full_rec SET
368 subfield = NEW.subfield,
370 index_vector = NEW.index_vector
373 CREATE OR REPLACE RULE metabib_full_rec_delete_rule
374 AS ON DELETE TO metabib.full_rec
376 DELETE FROM metabib.real_full_rec WHERE id = OLD.id;
378 CREATE TABLE metabib.metarecord_source_map (
379 id BIGSERIAL PRIMARY KEY,
380 metarecord BIGINT NOT NULL,
381 source BIGINT NOT NULL
383 CREATE INDEX metabib_metarecord_source_map_metarecord_idx ON metabib.metarecord_source_map (metarecord);
384 CREATE INDEX metabib_metarecord_source_map_source_record_idx ON metabib.metarecord_source_map (source);
386 CREATE TYPE metabib.field_entry_template AS (
397 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
399 bib biblio.record_entry%ROWTYPE;
400 idx config.metabib_field%ROWTYPE;
401 xfrm config.xml_transform%ROWTYPE;
403 transformed_xml TEXT;
405 xml_node_list TEXT[];
410 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
411 output_row metabib.field_entry_template%ROWTYPE;
414 -- Start out with no field-use bools set
415 output_row.browse_field = FALSE;
416 output_row.facet_field = FALSE;
417 output_row.search_field = FALSE;
420 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
422 -- Loop over the indexing entries
423 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
425 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
427 -- See if we can skip the XSLT ... it's expensive
428 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
429 -- Can't skip the transform
430 IF xfrm.xslt <> '---' THEN
431 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
433 transformed_xml := bib.marc;
436 prev_xfrm := xfrm.name;
439 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
442 FOR xml_node IN SELECT x FROM unnest(xml_node_list) AS x LOOP
443 CONTINUE WHEN xml_node !~ E'^\\s*<';
445 curr_text := ARRAY_TO_STRING(
446 oils_xpath( '//text()',
447 REGEXP_REPLACE( -- This escapes all &s not followed by "amp;". Data ise returned from oils_xpath (above) in UTF-8, not entity encoded
448 REGEXP_REPLACE( -- This escapes embeded <s
450 $re$(>[^<]+)(<)([^>]+<)$re$,
462 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
464 IF raw_text IS NOT NULL THEN
465 raw_text := raw_text || joiner;
468 raw_text := COALESCE(raw_text,'') || curr_text;
470 -- autosuggest/metabib.browse_entry
471 IF idx.browse_field THEN
473 IF idx.browse_xpath IS NOT NULL AND idx.browse_xpath <> '' THEN
474 browse_text := oils_xpath_string( idx.browse_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
476 browse_text := curr_text;
479 output_row.field_class = idx.field_class;
480 output_row.field = idx.id;
481 output_row.source = rid;
482 output_row.value = BTRIM(REGEXP_REPLACE(browse_text, E'\\s+', ' ', 'g'));
484 output_row.browse_field = TRUE;
485 RETURN NEXT output_row;
486 output_row.browse_field = FALSE;
489 -- insert raw node text for faceting
490 IF idx.facet_field THEN
492 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
493 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
495 facet_text := curr_text;
498 output_row.field_class = idx.field_class;
499 output_row.field = -1 * idx.id;
500 output_row.source = rid;
501 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
503 output_row.facet_field = TRUE;
504 RETURN NEXT output_row;
505 output_row.facet_field = FALSE;
510 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
512 -- insert combined node text for searching
513 IF idx.search_field THEN
514 output_row.field_class = idx.field_class;
515 output_row.field = idx.id;
516 output_row.source = rid;
517 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
519 output_row.search_field = TRUE;
520 RETURN NEXT output_row;
521 output_row.search_field = FALSE;
528 $func$ LANGUAGE PLPGSQL;
530 CREATE OR REPLACE FUNCTION metabib.update_combined_index_vectors(bib_id BIGINT) RETURNS VOID AS $func$
532 DELETE FROM metabib.combined_keyword_field_entry WHERE record = bib_id;
533 INSERT INTO metabib.combined_keyword_field_entry(record, metabib_field, index_vector)
534 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
535 FROM metabib.keyword_field_entry WHERE source = bib_id GROUP BY field;
536 INSERT INTO metabib.combined_keyword_field_entry(record, metabib_field, index_vector)
537 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
538 FROM metabib.keyword_field_entry WHERE source = bib_id;
540 DELETE FROM metabib.combined_title_field_entry WHERE record = bib_id;
541 INSERT INTO metabib.combined_title_field_entry(record, metabib_field, index_vector)
542 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
543 FROM metabib.title_field_entry WHERE source = bib_id GROUP BY field;
544 INSERT INTO metabib.combined_title_field_entry(record, metabib_field, index_vector)
545 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
546 FROM metabib.title_field_entry WHERE source = bib_id;
548 DELETE FROM metabib.combined_author_field_entry WHERE record = bib_id;
549 INSERT INTO metabib.combined_author_field_entry(record, metabib_field, index_vector)
550 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
551 FROM metabib.author_field_entry WHERE source = bib_id GROUP BY field;
552 INSERT INTO metabib.combined_author_field_entry(record, metabib_field, index_vector)
553 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
554 FROM metabib.author_field_entry WHERE source = bib_id;
556 DELETE FROM metabib.combined_subject_field_entry WHERE record = bib_id;
557 INSERT INTO metabib.combined_subject_field_entry(record, metabib_field, index_vector)
558 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
559 FROM metabib.subject_field_entry WHERE source = bib_id GROUP BY field;
560 INSERT INTO metabib.combined_subject_field_entry(record, metabib_field, index_vector)
561 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
562 FROM metabib.subject_field_entry WHERE source = bib_id;
564 DELETE FROM metabib.combined_series_field_entry WHERE record = bib_id;
565 INSERT INTO metabib.combined_series_field_entry(record, metabib_field, index_vector)
566 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
567 FROM metabib.series_field_entry WHERE source = bib_id GROUP BY field;
568 INSERT INTO metabib.combined_series_field_entry(record, metabib_field, index_vector)
569 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
570 FROM metabib.series_field_entry WHERE source = bib_id;
572 DELETE FROM metabib.combined_identifier_field_entry WHERE record = bib_id;
573 INSERT INTO metabib.combined_identifier_field_entry(record, metabib_field, index_vector)
574 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
575 FROM metabib.identifier_field_entry WHERE source = bib_id GROUP BY field;
576 INSERT INTO metabib.combined_identifier_field_entry(record, metabib_field, index_vector)
577 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
578 FROM metabib.identifier_field_entry WHERE source = bib_id;
581 $func$ LANGUAGE PLPGSQL;
583 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries( bib_id BIGINT, skip_facet BOOL DEFAULT FALSE, skip_browse BOOL DEFAULT FALSE, skip_search BOOL DEFAULT FALSE ) RETURNS VOID AS $func$
586 ind_data metabib.field_entry_template%ROWTYPE;
587 mbe_row metabib.browse_entry%ROWTYPE;
594 SELECT COALESCE(NULLIF(skip_facet, FALSE), EXISTS (SELECT enabled FROM config.internal_flag WHERE name = 'ingest.skip_facet_indexing' AND enabled)) INTO b_skip_facet;
595 SELECT COALESCE(NULLIF(skip_browse, FALSE), EXISTS (SELECT enabled FROM config.internal_flag WHERE name = 'ingest.skip_browse_indexing' AND enabled)) INTO b_skip_browse;
596 SELECT COALESCE(NULLIF(skip_search, FALSE), EXISTS (SELECT enabled FROM config.internal_flag WHERE name = 'ingest.skip_search_indexing' AND enabled)) INTO b_skip_search;
598 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
600 IF NOT b_skip_search THEN
601 FOR fclass IN SELECT * FROM config.metabib_class LOOP
602 -- RAISE NOTICE 'Emptying out %', fclass.name;
603 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
606 IF NOT b_skip_facet THEN
607 DELETE FROM metabib.facet_entry WHERE source = bib_id;
609 IF NOT b_skip_browse THEN
610 DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
614 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
615 IF ind_data.field < 0 THEN
616 ind_data.field = -1 * ind_data.field;
619 IF ind_data.facet_field AND NOT b_skip_facet THEN
620 INSERT INTO metabib.facet_entry (field, source, value)
621 VALUES (ind_data.field, ind_data.source, ind_data.value);
624 IF ind_data.browse_field AND NOT b_skip_browse THEN
625 -- A caveat about this SELECT: this should take care of replacing
626 -- old mbe rows when data changes, but not if normalization (by
627 -- which I mean specifically the output of
628 -- evergreen.oils_tsearch2()) changes. It may or may not be
629 -- expensive to add a comparison of index_vector to index_vector
630 -- to the WHERE clause below.
631 SELECT INTO mbe_row * FROM metabib.browse_entry WHERE value = ind_data.value;
633 mbe_id := mbe_row.id;
635 INSERT INTO metabib.browse_entry (value) VALUES
636 (metabib.browse_normalize(ind_data.value, ind_data.field));
637 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
640 INSERT INTO metabib.browse_entry_def_map (entry, def, source)
641 VALUES (mbe_id, ind_data.field, ind_data.source);
644 -- Avoid inserting duplicate rows, but retain granularity of being
645 -- able to search browse fields with "starts with" type operators
646 -- (for example, for titles of songs in music albums)
647 IF (ind_data.search_field OR ind_data.browse_field) AND NOT b_skip_search THEN
648 EXECUTE 'SELECT 1 FROM metabib.' || ind_data.field_class ||
649 '_field_entry WHERE field = $1 AND source = $2 AND value = $3'
650 INTO mbe_id USING ind_data.field, ind_data.source, ind_data.value;
651 -- RAISE NOTICE 'Search for an already matching row returned %', mbe_id;
652 IF mbe_id IS NULL THEN
654 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
656 quote_literal(ind_data.field) || $$, $$ ||
657 quote_literal(ind_data.source) || $$, $$ ||
658 quote_literal(ind_data.value) ||
665 IF NOT b_skip_search THEN
666 PERFORM metabib.update_combined_index_vectors(bib_id);
671 $func$ LANGUAGE PLPGSQL;
673 -- default to a space joiner
674 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
675 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
678 CREATE OR REPLACE FUNCTION authority.flatten_marc ( rid BIGINT ) RETURNS SETOF authority.full_rec AS $func$
680 auth authority.record_entry%ROWTYPE;
681 output authority.full_rec%ROWTYPE;
684 SELECT INTO auth * FROM authority.record_entry WHERE id = rid;
686 FOR field IN SELECT * FROM vandelay.flatten_marc( auth.marc ) LOOP
687 output.record := rid;
688 output.ind1 := field.ind1;
689 output.ind2 := field.ind2;
690 output.tag := field.tag;
691 output.subfield := field.subfield;
692 output.value := field.value;
697 $func$ LANGUAGE PLPGSQL;
699 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
701 bib biblio.record_entry%ROWTYPE;
702 output metabib.full_rec%ROWTYPE;
705 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
707 FOR field IN SELECT * FROM vandelay.flatten_marc( bib.marc ) LOOP
708 output.record := rid;
709 output.ind1 := field.ind1;
710 output.ind2 := field.ind2;
711 output.tag := field.tag;
712 output.subfield := field.subfield;
713 output.value := field.value;
718 $func$ LANGUAGE PLPGSQL;
720 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
727 retval config.marc21_rec_type_map%ROWTYPE;
729 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
731 IF ldr IS NULL OR ldr = '' THEN
732 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
736 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
737 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
740 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
741 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
743 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
745 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
748 IF retval.code IS NULL THEN
749 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
754 $func$ LANGUAGE PLPGSQL;
756 CREATE OR REPLACE FUNCTION biblio.marc21_record_type( rid BIGINT ) RETURNS config.marc21_rec_type_map AS $func$
757 SELECT * FROM vandelay.marc21_record_type( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
760 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
767 rtype := (vandelay.marc21_record_type( marc )).code;
768 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
769 IF ff_pos.tag = 'ldr' THEN
770 val := oils_xpath_string('//*[local-name()="leader"]', marc);
771 IF val IS NOT NULL THEN
772 val := SUBSTRING( val, ff_pos.start_pos + 1, ff_pos.length );
776 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
777 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
781 val := REPEAT( ff_pos.default_val, ff_pos.length );
787 $func$ LANGUAGE PLPGSQL;
789 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
790 SELECT * FROM vandelay.marc21_extract_fixed_field( (SELECT marc FROM biblio.record_entry WHERE id = $1), $2 );
793 -- CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
794 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
799 output biblio.record_ff_map%ROWTYPE;
801 rtype := (vandelay.marc21_record_type( marc )).code;
803 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
804 output.ff_name := ff_pos.fixed_field;
805 output.ff_value := NULL;
807 IF ff_pos.tag = 'ldr' THEN
808 output.ff_value := oils_xpath_string('//*[local-name()="leader"]', marc);
809 IF output.ff_value IS NOT NULL THEN
810 output.ff_value := SUBSTRING( output.ff_value, ff_pos.start_pos + 1, ff_pos.length );
812 output.ff_value := NULL;
815 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
816 output.ff_value := SUBSTRING( tag_data, ff_pos.start_pos + 1, ff_pos.length );
817 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
819 output.ff_value := NULL;
827 $func$ LANGUAGE PLPGSQL;
829 CREATE OR REPLACE FUNCTION biblio.marc21_extract_all_fixed_fields( rid BIGINT ) RETURNS SETOF biblio.record_ff_map AS $func$
830 SELECT $1 AS record, ff_name, ff_value FROM vandelay.marc21_extract_all_fixed_fields( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
833 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
834 SELECT id, $1 AS record, ptype, subfield, value FROM vandelay.marc21_physical_characteristics( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
837 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
850 IF marc IS NULL OR marc = '' THEN
854 -- First, the count of tags
855 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
857 -- now go through a bunch of pain to get the record type
858 IF best_type IS NOT NULL THEN
859 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
861 IF ldr IS NOT NULL THEN
862 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
863 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
866 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
867 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
869 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
871 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
873 IF type_map.code IS NOT NULL THEN
874 IF best_type = type_map.code THEN
875 qual := qual + qual / 2;
878 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
879 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
880 IF ff_tag_data = best_lang THEN
888 -- Now look for some quality metrics
890 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
895 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
902 $func$ LANGUAGE PLPGSQL;
904 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
906 idx config.biblio_fingerprint%ROWTYPE;
907 xfrm config.xml_transform%ROWTYPE;
909 transformed_xml TEXT;
911 xml_node_list TEXT[];
913 output_text TEXT := '';
916 IF marc IS NULL OR marc = '' THEN
920 -- Loop over the indexing entries
921 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
923 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
925 -- See if we can skip the XSLT ... it's expensive
926 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
927 -- Can't skip the transform
928 IF xfrm.xslt <> '---' THEN
929 transformed_xml := oils_xslt_process(marc,xfrm.xslt);
931 transformed_xml := marc;
934 prev_xfrm := xfrm.name;
937 raw_text := COALESCE(
945 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
954 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
955 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
957 IF idx.first_word IS TRUE THEN
958 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
961 output_text := output_text || REGEXP_REPLACE(raw_text, E'\\s+', '', 'g');
968 $func$ LANGUAGE PLPGSQL;
970 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
971 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
974 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
976 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
980 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
981 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
986 $func$ LANGUAGE PLPGSQL;
988 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_full_rec( bib_id BIGINT ) RETURNS VOID AS $func$
990 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
992 DELETE FROM metabib.real_full_rec WHERE record = bib_id;
994 INSERT INTO metabib.real_full_rec (record, tag, ind1, ind2, subfield, value)
995 SELECT record, tag, ind1, ind2, subfield, value FROM biblio.flatten_marc( bib_id );
999 $func$ LANGUAGE PLPGSQL;
1001 CREATE OR REPLACE FUNCTION biblio.extract_located_uris( bib_id BIGINT, marcxml TEXT, editor_id INT ) RETURNS VOID AS $func$
1008 uri_owner_list TEXT[];
1016 -- Clear any URI mappings and call numbers for this bib.
1017 -- This leads to acn / auricnm inflation, but also enables
1018 -- old acn/auricnm's to go away and for bibs to be deleted.
1019 FOR uri_cn_id IN SELECT id FROM asset.call_number WHERE record = bib_id AND label = '##URI##' AND NOT deleted LOOP
1020 DELETE FROM asset.uri_call_number_map WHERE call_number = uri_cn_id;
1021 DELETE FROM asset.call_number WHERE id = uri_cn_id;
1024 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',marcxml);
1025 IF ARRAY_UPPER(uris,1) > 0 THEN
1026 FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
1027 -- First we pull info out of the 856
1030 uri_href := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
1031 uri_label := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()',uri_xml))[1];
1032 uri_use := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()',uri_xml))[1];
1034 IF uri_label IS NULL THEN
1035 uri_label := uri_href;
1037 CONTINUE WHEN uri_href IS NULL;
1039 -- Get the distinct list of libraries wanting to use
1041 DISTINCT REGEXP_REPLACE(
1043 $re$^.*?\((\w+)\).*$$re$,
1046 ) INTO uri_owner_list
1049 '//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',
1054 IF ARRAY_UPPER(uri_owner_list,1) > 0 THEN
1056 -- look for a matching uri
1057 IF uri_use IS NULL THEN
1058 SELECT id INTO uri_id
1060 WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active
1061 ORDER BY id LIMIT 1;
1062 IF NOT FOUND THEN -- create one
1063 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
1064 SELECT id INTO uri_id
1066 WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active;
1069 SELECT id INTO uri_id
1071 WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active
1072 ORDER BY id LIMIT 1;
1073 IF NOT FOUND THEN -- create one
1074 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
1075 SELECT id INTO uri_id
1077 WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
1081 FOR j IN 1 .. ARRAY_UPPER(uri_owner_list, 1) LOOP
1082 uri_owner := uri_owner_list[j];
1084 SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = uri_owner;
1085 CONTINUE WHEN NOT FOUND;
1087 -- we need a call number to link through
1088 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;
1090 INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
1091 VALUES (uri_owner_id, bib_id, 'now', 'now', editor_id, editor_id, '##URI##');
1092 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;
1095 -- now, link them if they're not already
1096 SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
1098 INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
1110 $func$ LANGUAGE PLPGSQL;
1112 CREATE OR REPLACE FUNCTION metabib.remap_metarecord_for_bib( bib_id BIGINT, fp TEXT ) RETURNS BIGINT AS $func$
1116 tmp_mr metabib.metarecord%ROWTYPE;
1117 deleted_mrs BIGINT[];
1120 DELETE FROM metabib.metarecord_source_map WHERE source = bib_id; -- Rid ourselves of the search-estimate-killing linkage
1122 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
1124 IF old_mr IS NULL AND fp = tmp_mr.fingerprint THEN -- Find the first fingerprint-matching
1125 old_mr := tmp_mr.id;
1127 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
1128 IF source_count = 0 THEN -- No other records
1129 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
1130 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
1136 IF old_mr IS NULL THEN -- we found no suitable, preexisting MR based on old source maps
1137 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp; -- is there one for our current fingerprint?
1138 IF old_mr IS NULL THEN -- nope, create one and grab its id
1139 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( fp, bib_id );
1140 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp;
1141 ELSE -- indeed there is. update it with a null cache and recalcualated master record
1142 UPDATE metabib.metarecord
1144 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
1147 ELSE -- there was one we already attached to, update its mods cache and master_record
1148 UPDATE metabib.metarecord
1150 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
1154 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, bib_id); -- new source mapping
1156 IF ARRAY_UPPER(deleted_mrs,1) > 0 THEN
1157 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
1163 $func$ LANGUAGE PLPGSQL;
1165 CREATE OR REPLACE FUNCTION biblio.map_authority_linking (bibid BIGINT, marc TEXT) RETURNS BIGINT AS $func$
1166 DELETE FROM authority.bib_linking WHERE bib = $1;
1167 INSERT INTO authority.bib_linking (bib, authority)
1170 FROM ( SELECT DISTINCT $1 AS bib,
1171 BTRIM(remove_paren_substring(txt))::BIGINT AS authority
1172 FROM unnest(oils_xpath('//*[@code="0"]/text()',$2)) x(txt)
1173 WHERE BTRIM(remove_paren_substring(txt)) ~ $re$^\d+$$re$
1174 ) y JOIN authority.record_entry r ON r.id = y.authority;
1176 $func$ LANGUAGE SQL;
1178 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
1179 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
1181 transformed_xml TEXT;
1184 xfrm config.xml_transform%ROWTYPE;
1186 new_attrs HSTORE := ''::HSTORE;
1187 attr_def config.record_attr_definition%ROWTYPE;
1190 IF NEW.deleted IS TRUE THEN -- If this bib is deleted
1191 PERFORM * FROM config.internal_flag WHERE
1192 name = 'ingest.metarecord_mapping.preserve_on_delete' AND enabled;
1194 -- One needs to keep these around to support searches
1195 -- with the #deleted modifier, so one should turn on the named
1196 -- internal flag for that functionality.
1197 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id;
1198 DELETE FROM metabib.record_attr WHERE id = NEW.id;
1201 DELETE FROM authority.bib_linking WHERE bib = NEW.id; -- Avoid updating fields in bibs that are no longer visible
1202 DELETE FROM biblio.peer_bib_copy_map WHERE peer_record = NEW.id; -- Separate any multi-homed items
1203 DELETE FROM metabib.browse_entry_def_map WHERE source = NEW.id; -- Don't auto-suggest deleted bibs
1204 RETURN NEW; -- and we're done
1207 IF TG_OP = 'UPDATE' THEN -- re-ingest?
1208 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
1210 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
1215 -- Record authority linking
1216 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_linking' AND enabled;
1218 PERFORM biblio.map_authority_linking( NEW.id, NEW.marc );
1221 -- Flatten and insert the mfr data
1222 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_full_rec' AND enabled;
1224 PERFORM metabib.reingest_metabib_full_rec(NEW.id);
1226 -- Now we pull out attribute data, which is dependent on the mfr for all but XPath-based fields
1227 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_rec_descriptor' AND enabled;
1229 FOR attr_def IN SELECT * FROM config.record_attr_definition ORDER BY format LOOP
1231 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
1232 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(value), COALESCE(attr_def.joiner,' ')) INTO attr_value
1233 FROM (SELECT * FROM metabib.full_rec ORDER BY tag, subfield) AS x
1234 WHERE record = NEW.id
1235 AND tag LIKE attr_def.tag
1237 WHEN attr_def.sf_list IS NOT NULL
1238 THEN POSITION(subfield IN attr_def.sf_list) > 0
1245 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
1246 attr_value := biblio.marc21_extract_fixed_field(NEW.id, attr_def.fixed_field);
1248 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
1250 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
1252 -- See if we can skip the XSLT ... it's expensive
1253 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
1254 -- Can't skip the transform
1255 IF xfrm.xslt <> '---' THEN
1256 transformed_xml := oils_xslt_process(NEW.marc,xfrm.xslt);
1258 transformed_xml := NEW.marc;
1261 prev_xfrm := xfrm.name;
1264 IF xfrm.name IS NULL THEN
1265 -- just grab the marcxml (empty) transform
1266 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
1267 prev_xfrm := xfrm.name;
1270 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
1272 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
1273 SELECT m.value INTO attr_value
1274 FROM biblio.marc21_physical_characteristics(NEW.id) v
1275 JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
1276 WHERE v.subfield = attr_def.phys_char_sf
1277 LIMIT 1; -- Just in case ...
1281 -- apply index normalizers to attr_value
1283 SELECT n.func AS func,
1284 n.param_count AS param_count,
1286 FROM config.index_normalizer n
1287 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
1288 WHERE attr = attr_def.name
1290 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1291 COALESCE( quote_literal( attr_value ), 'NULL' ) ||
1293 WHEN normalizer.param_count > 0
1294 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1297 ')' INTO attr_value;
1301 -- Add the new value to the hstore
1302 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
1306 IF TG_OP = 'INSERT' OR OLD.deleted THEN -- initial insert OR revivication
1307 DELETE FROM metabib.record_attr WHERE id = NEW.id;
1308 INSERT INTO metabib.record_attr (id, attrs) VALUES (NEW.id, new_attrs);
1310 UPDATE metabib.record_attr SET attrs = new_attrs WHERE id = NEW.id;
1316 -- Gather and insert the field entry data
1317 PERFORM metabib.reingest_metabib_field_entries(NEW.id);
1319 -- Located URI magic
1320 IF TG_OP = 'INSERT' THEN
1321 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1323 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1326 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1328 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1332 -- (re)map metarecord-bib linking
1333 IF TG_OP = 'INSERT' THEN -- if not deleted and performing an insert, check for the flag
1334 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
1336 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1338 ELSE -- we're doing an update, and we're not deleted, remap
1339 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_update' AND enabled;
1341 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1347 $func$ LANGUAGE PLPGSQL;
1349 CREATE OR REPLACE FUNCTION metabib.browse_normalize(facet_text TEXT, mapped_field INT) RETURNS TEXT AS $$
1355 SELECT n.func AS func,
1356 n.param_count AS param_count,
1358 FROM config.index_normalizer n
1359 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
1360 WHERE m.field = mapped_field AND m.pos < 0
1363 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1364 quote_literal( facet_text ) ||
1366 WHEN normalizer.param_count > 0
1367 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1370 ')' INTO facet_text;
1377 $$ LANGUAGE PLPGSQL;
1380 -- This mimics a specific part of QueryParser, turning the first part of a
1381 -- classed search (search_class) into a set of classes and possibly fields.
1382 -- search_class might look like "author" or "title|proper" or "ti|uniform"
1383 -- or "au" or "au|corporate|personal" or anything like that, where the first
1384 -- element of the list you get by separating on the "|" character is either
1385 -- a registered class (config.metabib_class) or an alias
1386 -- (config.metabib_search_alias), and the rest of any such elements are
1387 -- fields (config.metabib_field).
1389 FUNCTION metabib.search_class_to_registered_components(search_class TEXT)
1390 RETURNS SETOF RECORD AS $func$
1392 search_parts TEXT[];
1394 search_part_count INTEGER;
1396 registered_class config.metabib_class%ROWTYPE;
1397 registered_alias config.metabib_search_alias%ROWTYPE;
1398 registered_field config.metabib_field%ROWTYPE;
1400 search_parts := REGEXP_SPLIT_TO_ARRAY(search_class, E'\\|');
1402 search_part_count := ARRAY_LENGTH(search_parts, 1);
1403 IF search_part_count = 0 THEN
1406 SELECT INTO registered_class
1407 * FROM config.metabib_class WHERE name = search_parts[1];
1409 IF search_part_count < 2 THEN -- all fields
1410 rec := (registered_class.name, NULL::INTEGER);
1414 FOR field_name IN SELECT *
1415 FROM UNNEST(search_parts[2:search_part_count]) LOOP
1416 SELECT INTO registered_field
1417 * FROM config.metabib_field
1418 WHERE name = field_name AND
1419 field_class = registered_class.name;
1421 rec := (registered_class.name, registered_field.id);
1426 -- maybe we have an alias?
1427 SELECT INTO registered_alias
1428 * FROM config.metabib_search_alias WHERE alias=search_parts[1];
1432 IF search_part_count < 2 THEN -- return w/e the alias says
1434 registered_alias.field_class, registered_alias.field
1439 FOR field_name IN SELECT *
1440 FROM UNNEST(search_parts[2:search_part_count]) LOOP
1441 SELECT INTO registered_field
1442 * FROM config.metabib_field
1443 WHERE name = field_name AND
1444 field_class = registered_alias.field_class;
1447 registered_alias.field_class,
1458 $func$ LANGUAGE PLPGSQL ROWS 1;
1461 -- Given a string such as a user might type into a search box, prepare
1462 -- two changed variants for TO_TSQUERY(). See
1463 -- http://www.postgresql.org/docs/9.0/static/textsearch-controls.html
1464 -- The first variant is normalized to match indexed documents regardless
1465 -- of diacritics. The second variant keeps its diacritics for proper
1466 -- highlighting via TS_HEADLINE().
1468 FUNCTION metabib.autosuggest_prepare_tsquery(orig TEXT) RETURNS TEXT[] AS
1471 orig_ended_in_space BOOLEAN;
1476 orig_ended_in_space := orig ~ E'\\s$';
1478 orig := ARRAY_TO_STRING(
1479 evergreen.regexp_split_to_array(orig, E'\\W+'), ' '
1482 normalized := public.search_normalize(orig); -- also trim()s
1483 plain := trim(orig);
1485 IF NOT orig_ended_in_space THEN
1486 plain := plain || ':*';
1487 normalized := normalized || ':*';
1490 plain := ARRAY_TO_STRING(
1491 evergreen.regexp_split_to_array(plain, E'\\s+'), ' & '
1493 normalized := ARRAY_TO_STRING(
1494 evergreen.regexp_split_to_array(normalized, E'\\s+'), ' & '
1497 RETURN ARRAY[normalized, plain];
1499 $$ LANGUAGE PLPGSQL;
1503 FUNCTION metabib.suggest_browse_entries(
1504 raw_query_text TEXT, -- actually typed by humans at the UI level
1505 search_class TEXT, -- 'alias' or 'class' or 'class|field..', etc
1506 headline_opts TEXT, -- markup options for ts_headline()
1507 visibility_org INTEGER,-- null if you don't want opac visibility test
1508 query_limit INTEGER,-- use in LIMIT clause of interal query
1509 normalization INTEGER -- argument to TS_RANK_CD()
1511 value TEXT, -- plain
1513 buoyant_and_class_match BOOL,
1515 field_weight INTEGER,
1518 match TEXT -- marked up
1521 prepared_query_texts TEXT[];
1523 plain_query TSQUERY;
1524 opac_visibility_join TEXT;
1525 search_class_join TEXT;
1528 prepared_query_texts := metabib.autosuggest_prepare_tsquery(raw_query_text);
1530 query := TO_TSQUERY('keyword', prepared_query_texts[1]);
1531 plain_query := TO_TSQUERY('keyword', prepared_query_texts[2]);
1533 visibility_org := NULLIF(visibility_org,-1);
1534 IF visibility_org IS NOT NULL THEN
1535 opac_visibility_join := '
1536 JOIN asset.opac_visible_copies aovc ON (
1537 aovc.record = x.source AND
1538 aovc.circ_lib IN (SELECT id FROM actor.org_unit_descendants($4))
1541 opac_visibility_join := '';
1544 -- The following determines whether we only provide suggestsons matching
1545 -- the user's selected search_class, or whether we show other suggestions
1546 -- too. The reason for MIN() is that for search_classes like
1547 -- 'title|proper|uniform' you would otherwise get multiple rows. The
1548 -- implication is that if title as a class doesn't have restrict,
1549 -- nor does the proper field, but the uniform field does, you're going
1550 -- to get 'false' for your overall evaluation of 'should we restrict?'
1551 -- To invert that, change from MIN() to MAX().
1555 MIN(cmc.restrict::INT) AS restrict_class,
1556 MIN(cmf.restrict::INT) AS restrict_field
1557 FROM metabib.search_class_to_registered_components(search_class)
1558 AS _registered (field_class TEXT, field INT)
1560 config.metabib_class cmc ON (cmc.name = _registered.field_class)
1562 config.metabib_field cmf ON (cmf.id = _registered.field);
1564 -- evaluate 'should we restrict?'
1565 IF r_fields.restrict_field::BOOL OR r_fields.restrict_class::BOOL THEN
1566 search_class_join := '
1568 metabib.search_class_to_registered_components($2)
1569 AS _registered (field_class TEXT, field INT) ON (
1570 (_registered.field IS NULL AND
1571 _registered.field_class = cmf.field_class) OR
1572 (_registered.field = cmf.id)
1576 search_class_join := '
1578 metabib.search_class_to_registered_components($2)
1579 AS _registered (field_class TEXT, field INT) ON (
1580 _registered.field_class = cmc.name
1585 RETURN QUERY EXECUTE '
1594 TS_HEADLINE(value, $7, $3)
1595 FROM (SELECT DISTINCT
1598 cmc.buoyant AND _registered.field_class IS NOT NULL AS push,
1599 _registered.field = cmf.id AS restrict,
1601 TS_RANK_CD(mbe.index_vector, $1, $6),
1604 FROM metabib.browse_entry_def_map mbedm
1605 JOIN (SELECT * FROM metabib.browse_entry WHERE index_vector @@ $1 LIMIT 10000) mbe ON (mbe.id = mbedm.entry)
1606 JOIN config.metabib_field cmf ON (cmf.id = mbedm.def)
1607 JOIN config.metabib_class cmc ON (cmf.field_class = cmc.name)
1608 ' || search_class_join || '
1609 ORDER BY 3 DESC, 4 DESC NULLS LAST, 5 DESC, 6 DESC, 7 DESC, 1 ASC
1611 ' || opac_visibility_join || '
1612 ORDER BY 3 DESC, 4 DESC NULLS LAST, 5 DESC, 6 DESC, 7 DESC, 1 ASC
1614 ' -- sic, repeat the order by clause in the outer select too
1616 query, search_class, headline_opts,
1617 visibility_org, query_limit, normalization, plain_query
1621 -- buoyant AND chosen class = match class
1622 -- chosen field = match field
1629 $func$ LANGUAGE PLPGSQL;
1631 CREATE OR REPLACE FUNCTION public.oils_tsearch2 () RETURNS TRIGGER AS $$
1635 temp_vector TEXT := '';
1641 NEW.index_vector = ''::tsvector;
1643 IF TG_TABLE_NAME::TEXT ~ 'field_entry$' THEN
1645 SELECT n.func AS func,
1646 n.param_count AS param_count,
1648 FROM config.index_normalizer n
1649 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
1650 WHERE field = NEW.field AND m.pos < 0
1652 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1653 quote_literal( value ) ||
1655 WHEN normalizer.param_count > 0
1656 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1666 SELECT n.func AS func,
1667 n.param_count AS param_count,
1669 FROM config.index_normalizer n
1670 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
1671 WHERE field = NEW.field AND m.pos >= 0
1673 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1674 quote_literal( value ) ||
1676 WHEN normalizer.param_count > 0
1677 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1685 IF TG_TABLE_NAME::TEXT ~ 'browse_entry$' THEN
1686 value := ARRAY_TO_STRING(
1687 evergreen.regexp_split_to_array(value, E'\\W+'), ' '
1689 value := public.search_normalize(value);
1690 NEW.index_vector = to_tsvector(TG_ARGV[0]::regconfig, value);
1691 ELSIF TG_TABLE_NAME::TEXT ~ 'field_entry$' THEN
1693 SELECT ts_config, index_weight
1694 FROM config.metabib_class_ts_map
1695 WHERE field_class = TG_ARGV[0]
1696 AND index_lang IS NULL OR EXISTS (SELECT 1 FROM metabib.record_attr WHERE id = NEW.source AND index_lang IN(attrs->'item_lang',attrs->'language'))
1697 AND always OR NOT EXISTS (SELECT 1 FROM config.metabib_field_ts_map WHERE metabib_field = NEW.field)
1699 SELECT ts_config, index_weight
1700 FROM config.metabib_field_ts_map
1701 WHERE metabib_field = NEW.field
1702 AND index_lang IS NULL OR EXISTS (SELECT 1 FROM metabib.record_attr WHERE id = NEW.source AND index_lang IN(attrs->'item_lang',attrs->'language'))
1703 ORDER BY index_weight ASC
1705 IF cur_weight IS NOT NULL AND cur_weight != ts_rec.index_weight THEN
1706 NEW.index_vector = NEW.index_vector || setweight(temp_vector::tsvector,cur_weight);
1709 cur_weight = ts_rec.index_weight;
1710 SELECT INTO temp_vector temp_vector || ' ' || to_tsvector(ts_rec.ts_config::regconfig, value)::TEXT;
1712 NEW.index_vector = NEW.index_vector || setweight(temp_vector::tsvector,cur_weight);
1714 NEW.index_vector = to_tsvector(TG_ARGV[0]::regconfig, value);
1719 $$ LANGUAGE PLPGSQL;