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 (
398 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
400 bib biblio.record_entry%ROWTYPE;
401 idx config.metabib_field%ROWTYPE;
402 xfrm config.xml_transform%ROWTYPE;
404 transformed_xml TEXT;
406 xml_node_list TEXT[];
411 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
413 authority_link BIGINT;
414 output_row metabib.field_entry_template%ROWTYPE;
417 -- Start out with no field-use bools set
418 output_row.browse_field = FALSE;
419 output_row.facet_field = FALSE;
420 output_row.search_field = FALSE;
423 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
425 -- Loop over the indexing entries
426 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
428 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
430 -- See if we can skip the XSLT ... it's expensive
431 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
432 -- Can't skip the transform
433 IF xfrm.xslt <> '---' THEN
434 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
436 transformed_xml := bib.marc;
439 prev_xfrm := xfrm.name;
442 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
445 FOR xml_node IN SELECT x FROM unnest(xml_node_list) AS x LOOP
446 CONTINUE WHEN xml_node !~ E'^\\s*<';
448 curr_text := ARRAY_TO_STRING(
449 oils_xpath( '//text()',
450 REGEXP_REPLACE( -- This escapes all &s not followed by "amp;". Data ise returned from oils_xpath (above) in UTF-8, not entity encoded
451 REGEXP_REPLACE( -- This escapes embeded <s
453 $re$(>[^<]+)(<)([^>]+<)$re$,
465 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
467 IF raw_text IS NOT NULL THEN
468 raw_text := raw_text || joiner;
471 raw_text := COALESCE(raw_text,'') || curr_text;
473 -- autosuggest/metabib.browse_entry
474 IF idx.browse_field THEN
476 IF idx.browse_xpath IS NOT NULL AND idx.browse_xpath <> '' THEN
477 browse_text := oils_xpath_string( idx.browse_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
479 browse_text := curr_text;
482 output_row.field_class = idx.field_class;
483 output_row.field = idx.id;
484 output_row.source = rid;
485 output_row.value = BTRIM(REGEXP_REPLACE(browse_text, E'\\s+', ' ', 'g'));
487 IF idx.authority_xpath IS NOT NULL AND idx.authority_xpath <> '' THEN
488 authority_text := oils_xpath_string(
489 idx.authority_xpath, xml_node, joiner,
491 ARRAY[xfrm.prefix, xfrm.namespace_uri],
492 ARRAY['xlink','http://www.w3.org/1999/xlink']
496 IF authority_text ~ '^\d+$' THEN
497 authority_link := authority_text::BIGINT;
498 PERFORM * FROM authority.record_entry WHERE id = authority_link;
500 output_row.authority := authority_link;
506 output_row.browse_field = TRUE;
507 RETURN NEXT output_row;
508 output_row.browse_field = FALSE;
511 -- insert raw node text for faceting
512 IF idx.facet_field THEN
514 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
515 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
517 facet_text := curr_text;
520 output_row.field_class = idx.field_class;
521 output_row.field = -1 * idx.id;
522 output_row.source = rid;
523 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
525 output_row.facet_field = TRUE;
526 RETURN NEXT output_row;
527 output_row.facet_field = FALSE;
532 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
534 -- insert combined node text for searching
535 IF idx.search_field THEN
536 output_row.field_class = idx.field_class;
537 output_row.field = idx.id;
538 output_row.source = rid;
539 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
541 output_row.search_field = TRUE;
542 RETURN NEXT output_row;
543 output_row.search_field = FALSE;
550 $func$ LANGUAGE PLPGSQL;
552 CREATE OR REPLACE FUNCTION metabib.update_combined_index_vectors(bib_id BIGINT) RETURNS VOID AS $func$
554 DELETE FROM metabib.combined_keyword_field_entry WHERE record = bib_id;
555 INSERT INTO metabib.combined_keyword_field_entry(record, metabib_field, index_vector)
556 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
557 FROM metabib.keyword_field_entry WHERE source = bib_id GROUP BY field;
558 INSERT INTO metabib.combined_keyword_field_entry(record, metabib_field, index_vector)
559 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
560 FROM metabib.keyword_field_entry WHERE source = bib_id;
562 DELETE FROM metabib.combined_title_field_entry WHERE record = bib_id;
563 INSERT INTO metabib.combined_title_field_entry(record, metabib_field, index_vector)
564 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
565 FROM metabib.title_field_entry WHERE source = bib_id GROUP BY field;
566 INSERT INTO metabib.combined_title_field_entry(record, metabib_field, index_vector)
567 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
568 FROM metabib.title_field_entry WHERE source = bib_id;
570 DELETE FROM metabib.combined_author_field_entry WHERE record = bib_id;
571 INSERT INTO metabib.combined_author_field_entry(record, metabib_field, index_vector)
572 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
573 FROM metabib.author_field_entry WHERE source = bib_id GROUP BY field;
574 INSERT INTO metabib.combined_author_field_entry(record, metabib_field, index_vector)
575 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
576 FROM metabib.author_field_entry WHERE source = bib_id;
578 DELETE FROM metabib.combined_subject_field_entry WHERE record = bib_id;
579 INSERT INTO metabib.combined_subject_field_entry(record, metabib_field, index_vector)
580 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
581 FROM metabib.subject_field_entry WHERE source = bib_id GROUP BY field;
582 INSERT INTO metabib.combined_subject_field_entry(record, metabib_field, index_vector)
583 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
584 FROM metabib.subject_field_entry WHERE source = bib_id;
586 DELETE FROM metabib.combined_series_field_entry WHERE record = bib_id;
587 INSERT INTO metabib.combined_series_field_entry(record, metabib_field, index_vector)
588 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
589 FROM metabib.series_field_entry WHERE source = bib_id GROUP BY field;
590 INSERT INTO metabib.combined_series_field_entry(record, metabib_field, index_vector)
591 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
592 FROM metabib.series_field_entry WHERE source = bib_id;
594 DELETE FROM metabib.combined_identifier_field_entry WHERE record = bib_id;
595 INSERT INTO metabib.combined_identifier_field_entry(record, metabib_field, index_vector)
596 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
597 FROM metabib.identifier_field_entry WHERE source = bib_id GROUP BY field;
598 INSERT INTO metabib.combined_identifier_field_entry(record, metabib_field, index_vector)
599 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
600 FROM metabib.identifier_field_entry WHERE source = bib_id;
603 $func$ LANGUAGE PLPGSQL;
605 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 $func$
608 ind_data metabib.field_entry_template%ROWTYPE;
609 mbe_row metabib.browse_entry%ROWTYPE;
616 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;
617 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;
618 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;
620 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
622 IF NOT b_skip_search THEN
623 FOR fclass IN SELECT * FROM config.metabib_class LOOP
624 -- RAISE NOTICE 'Emptying out %', fclass.name;
625 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
628 IF NOT b_skip_facet THEN
629 DELETE FROM metabib.facet_entry WHERE source = bib_id;
631 IF NOT b_skip_browse THEN
632 DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
636 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
637 IF ind_data.field < 0 THEN
638 ind_data.field = -1 * ind_data.field;
641 IF ind_data.facet_field AND NOT b_skip_facet THEN
642 INSERT INTO metabib.facet_entry (field, source, value)
643 VALUES (ind_data.field, ind_data.source, ind_data.value);
646 IF ind_data.browse_field AND NOT b_skip_browse THEN
647 -- A caveat about this SELECT: this should take care of replacing
648 -- old mbe rows when data changes, but not if normalization (by
649 -- which I mean specifically the output of
650 -- evergreen.oils_tsearch2()) changes. It may or may not be
651 -- expensive to add a comparison of index_vector to index_vector
652 -- to the WHERE clause below.
653 SELECT INTO mbe_row * FROM metabib.browse_entry WHERE value = ind_data.value;
655 mbe_id := mbe_row.id;
657 INSERT INTO metabib.browse_entry (value) VALUES
658 (metabib.browse_normalize(ind_data.value, ind_data.field));
659 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
662 INSERT INTO metabib.browse_entry_def_map (entry, def, source, authority)
663 VALUES (mbe_id, ind_data.field, ind_data.source, ind_data.authority);
666 -- Avoid inserting duplicate rows, but retain granularity of being
667 -- able to search browse fields with "starts with" type operators
668 -- (for example, for titles of songs in music albums)
669 IF (ind_data.search_field OR ind_data.browse_field) AND NOT b_skip_search THEN
670 EXECUTE 'SELECT 1 FROM metabib.' || ind_data.field_class ||
671 '_field_entry WHERE field = $1 AND source = $2 AND value = $3'
672 INTO mbe_id USING ind_data.field, ind_data.source, ind_data.value;
673 -- RAISE NOTICE 'Search for an already matching row returned %', mbe_id;
674 IF mbe_id IS NULL THEN
676 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
678 quote_literal(ind_data.field) || $$, $$ ||
679 quote_literal(ind_data.source) || $$, $$ ||
680 quote_literal(ind_data.value) ||
687 IF NOT b_skip_search THEN
688 PERFORM metabib.update_combined_index_vectors(bib_id);
693 $func$ LANGUAGE PLPGSQL;
695 -- default to a space joiner
696 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
697 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
700 CREATE OR REPLACE FUNCTION authority.flatten_marc ( rid BIGINT ) RETURNS SETOF authority.full_rec AS $func$
702 auth authority.record_entry%ROWTYPE;
703 output authority.full_rec%ROWTYPE;
706 SELECT INTO auth * FROM authority.record_entry WHERE id = rid;
708 FOR field IN SELECT * FROM vandelay.flatten_marc( auth.marc ) LOOP
709 output.record := rid;
710 output.ind1 := field.ind1;
711 output.ind2 := field.ind2;
712 output.tag := field.tag;
713 output.subfield := field.subfield;
714 output.value := field.value;
719 $func$ LANGUAGE PLPGSQL;
721 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
723 bib biblio.record_entry%ROWTYPE;
724 output metabib.full_rec%ROWTYPE;
727 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
729 FOR field IN SELECT * FROM vandelay.flatten_marc( bib.marc ) LOOP
730 output.record := rid;
731 output.ind1 := field.ind1;
732 output.ind2 := field.ind2;
733 output.tag := field.tag;
734 output.subfield := field.subfield;
735 output.value := field.value;
740 $func$ LANGUAGE PLPGSQL;
742 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
749 retval config.marc21_rec_type_map%ROWTYPE;
751 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
753 IF ldr IS NULL OR ldr = '' THEN
754 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
758 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
759 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
762 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
763 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
765 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
767 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
770 IF retval.code IS NULL THEN
771 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
776 $func$ LANGUAGE PLPGSQL;
778 CREATE OR REPLACE FUNCTION biblio.marc21_record_type( rid BIGINT ) RETURNS config.marc21_rec_type_map AS $func$
779 SELECT * FROM vandelay.marc21_record_type( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
782 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
789 rtype := (vandelay.marc21_record_type( marc )).code;
790 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
791 IF ff_pos.tag = 'ldr' THEN
792 val := oils_xpath_string('//*[local-name()="leader"]', marc);
793 IF val IS NOT NULL THEN
794 val := SUBSTRING( val, ff_pos.start_pos + 1, ff_pos.length );
798 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
799 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
803 val := REPEAT( ff_pos.default_val, ff_pos.length );
809 $func$ LANGUAGE PLPGSQL;
811 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
812 SELECT * FROM vandelay.marc21_extract_fixed_field( (SELECT marc FROM biblio.record_entry WHERE id = $1), $2 );
815 -- CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
816 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
821 output biblio.record_ff_map%ROWTYPE;
823 rtype := (vandelay.marc21_record_type( marc )).code;
825 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
826 output.ff_name := ff_pos.fixed_field;
827 output.ff_value := NULL;
829 IF ff_pos.tag = 'ldr' THEN
830 output.ff_value := oils_xpath_string('//*[local-name()="leader"]', marc);
831 IF output.ff_value IS NOT NULL THEN
832 output.ff_value := SUBSTRING( output.ff_value, ff_pos.start_pos + 1, ff_pos.length );
834 output.ff_value := NULL;
837 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
838 output.ff_value := SUBSTRING( tag_data, ff_pos.start_pos + 1, ff_pos.length );
839 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
841 output.ff_value := NULL;
849 $func$ LANGUAGE PLPGSQL;
851 CREATE OR REPLACE FUNCTION biblio.marc21_extract_all_fixed_fields( rid BIGINT ) RETURNS SETOF biblio.record_ff_map AS $func$
852 SELECT $1 AS record, ff_name, ff_value FROM vandelay.marc21_extract_all_fixed_fields( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
855 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
856 SELECT id, $1 AS record, ptype, subfield, value FROM vandelay.marc21_physical_characteristics( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
859 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
872 IF marc IS NULL OR marc = '' THEN
876 -- First, the count of tags
877 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
879 -- now go through a bunch of pain to get the record type
880 IF best_type IS NOT NULL THEN
881 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
883 IF ldr IS NOT NULL THEN
884 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
885 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
888 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
889 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
891 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
893 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
895 IF type_map.code IS NOT NULL THEN
896 IF best_type = type_map.code THEN
897 qual := qual + qual / 2;
900 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
901 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
902 IF ff_tag_data = best_lang THEN
910 -- Now look for some quality metrics
912 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
917 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
924 $func$ LANGUAGE PLPGSQL;
926 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
928 idx config.biblio_fingerprint%ROWTYPE;
929 xfrm config.xml_transform%ROWTYPE;
931 transformed_xml TEXT;
933 xml_node_list TEXT[];
935 output_text TEXT := '';
938 IF marc IS NULL OR marc = '' THEN
942 -- Loop over the indexing entries
943 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
945 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
947 -- See if we can skip the XSLT ... it's expensive
948 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
949 -- Can't skip the transform
950 IF xfrm.xslt <> '---' THEN
951 transformed_xml := oils_xslt_process(marc,xfrm.xslt);
953 transformed_xml := marc;
956 prev_xfrm := xfrm.name;
959 raw_text := COALESCE(
967 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
976 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
977 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
979 IF idx.first_word IS TRUE THEN
980 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
983 output_text := output_text || REGEXP_REPLACE(raw_text, E'\\s+', '', 'g');
990 $func$ LANGUAGE PLPGSQL;
992 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
993 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
996 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
998 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
1002 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
1003 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
1008 $func$ LANGUAGE PLPGSQL;
1010 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_full_rec( bib_id BIGINT ) RETURNS VOID AS $func$
1012 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
1014 DELETE FROM metabib.real_full_rec WHERE record = bib_id;
1016 INSERT INTO metabib.real_full_rec (record, tag, ind1, ind2, subfield, value)
1017 SELECT record, tag, ind1, ind2, subfield, value FROM biblio.flatten_marc( bib_id );
1021 $func$ LANGUAGE PLPGSQL;
1023 CREATE OR REPLACE FUNCTION biblio.extract_located_uris( bib_id BIGINT, marcxml TEXT, editor_id INT ) RETURNS VOID AS $func$
1030 uri_owner_list TEXT[];
1038 -- Clear any URI mappings and call numbers for this bib.
1039 -- This leads to acn / auricnm inflation, but also enables
1040 -- old acn/auricnm's to go away and for bibs to be deleted.
1041 FOR uri_cn_id IN SELECT id FROM asset.call_number WHERE record = bib_id AND label = '##URI##' AND NOT deleted LOOP
1042 DELETE FROM asset.uri_call_number_map WHERE call_number = uri_cn_id;
1043 DELETE FROM asset.call_number WHERE id = uri_cn_id;
1046 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',marcxml);
1047 IF ARRAY_UPPER(uris,1) > 0 THEN
1048 FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
1049 -- First we pull info out of the 856
1052 uri_href := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
1053 uri_label := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()',uri_xml))[1];
1054 uri_use := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()',uri_xml))[1];
1056 IF uri_label IS NULL THEN
1057 uri_label := uri_href;
1059 CONTINUE WHEN uri_href IS NULL;
1061 -- Get the distinct list of libraries wanting to use
1063 DISTINCT REGEXP_REPLACE(
1065 $re$^.*?\((\w+)\).*$$re$,
1068 ) INTO uri_owner_list
1071 '//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',
1076 IF ARRAY_UPPER(uri_owner_list,1) > 0 THEN
1078 -- look for a matching uri
1079 IF uri_use IS NULL THEN
1080 SELECT id INTO uri_id
1082 WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active
1083 ORDER BY id LIMIT 1;
1084 IF NOT FOUND THEN -- create one
1085 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
1086 SELECT id INTO uri_id
1088 WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active;
1091 SELECT id INTO uri_id
1093 WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active
1094 ORDER BY id LIMIT 1;
1095 IF NOT FOUND THEN -- create one
1096 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
1097 SELECT id INTO uri_id
1099 WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
1103 FOR j IN 1 .. ARRAY_UPPER(uri_owner_list, 1) LOOP
1104 uri_owner := uri_owner_list[j];
1106 SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = uri_owner;
1107 CONTINUE WHEN NOT FOUND;
1109 -- we need a call number to link through
1110 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;
1112 INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
1113 VALUES (uri_owner_id, bib_id, 'now', 'now', editor_id, editor_id, '##URI##');
1114 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;
1117 -- now, link them if they're not already
1118 SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
1120 INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
1132 $func$ LANGUAGE PLPGSQL;
1134 CREATE OR REPLACE FUNCTION metabib.remap_metarecord_for_bib( bib_id BIGINT, fp TEXT ) RETURNS BIGINT AS $func$
1138 tmp_mr metabib.metarecord%ROWTYPE;
1139 deleted_mrs BIGINT[];
1142 DELETE FROM metabib.metarecord_source_map WHERE source = bib_id; -- Rid ourselves of the search-estimate-killing linkage
1144 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
1146 IF old_mr IS NULL AND fp = tmp_mr.fingerprint THEN -- Find the first fingerprint-matching
1147 old_mr := tmp_mr.id;
1149 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
1150 IF source_count = 0 THEN -- No other records
1151 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
1152 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
1158 IF old_mr IS NULL THEN -- we found no suitable, preexisting MR based on old source maps
1159 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp; -- is there one for our current fingerprint?
1160 IF old_mr IS NULL THEN -- nope, create one and grab its id
1161 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( fp, bib_id );
1162 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp;
1163 ELSE -- indeed there is. update it with a null cache and recalcualated master record
1164 UPDATE metabib.metarecord
1166 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
1169 ELSE -- there was one we already attached to, update its mods cache and master_record
1170 UPDATE metabib.metarecord
1172 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
1176 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, bib_id); -- new source mapping
1178 IF ARRAY_UPPER(deleted_mrs,1) > 0 THEN
1179 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
1185 $func$ LANGUAGE PLPGSQL;
1187 CREATE OR REPLACE FUNCTION biblio.map_authority_linking (bibid BIGINT, marc TEXT) RETURNS BIGINT AS $func$
1188 DELETE FROM authority.bib_linking WHERE bib = $1;
1189 INSERT INTO authority.bib_linking (bib, authority)
1192 FROM ( SELECT DISTINCT $1 AS bib,
1193 BTRIM(remove_paren_substring(txt))::BIGINT AS authority
1194 FROM unnest(oils_xpath('//*[@code="0"]/text()',$2)) x(txt)
1195 WHERE BTRIM(remove_paren_substring(txt)) ~ $re$^\d+$$re$
1196 ) y JOIN authority.record_entry r ON r.id = y.authority;
1198 $func$ LANGUAGE SQL;
1200 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
1201 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
1203 transformed_xml TEXT;
1206 xfrm config.xml_transform%ROWTYPE;
1208 new_attrs HSTORE := ''::HSTORE;
1209 attr_def config.record_attr_definition%ROWTYPE;
1212 IF NEW.deleted IS TRUE THEN -- If this bib is deleted
1213 PERFORM * FROM config.internal_flag WHERE
1214 name = 'ingest.metarecord_mapping.preserve_on_delete' AND enabled;
1216 -- One needs to keep these around to support searches
1217 -- with the #deleted modifier, so one should turn on the named
1218 -- internal flag for that functionality.
1219 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id;
1220 DELETE FROM metabib.record_attr WHERE id = NEW.id;
1223 DELETE FROM authority.bib_linking WHERE bib = NEW.id; -- Avoid updating fields in bibs that are no longer visible
1224 DELETE FROM biblio.peer_bib_copy_map WHERE peer_record = NEW.id; -- Separate any multi-homed items
1225 DELETE FROM metabib.browse_entry_def_map WHERE source = NEW.id; -- Don't auto-suggest deleted bibs
1226 RETURN NEW; -- and we're done
1229 IF TG_OP = 'UPDATE' THEN -- re-ingest?
1230 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
1232 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
1237 -- Record authority linking
1238 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_linking' AND enabled;
1240 PERFORM biblio.map_authority_linking( NEW.id, NEW.marc );
1243 -- Flatten and insert the mfr data
1244 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_full_rec' AND enabled;
1246 PERFORM metabib.reingest_metabib_full_rec(NEW.id);
1248 -- Now we pull out attribute data, which is dependent on the mfr for all but XPath-based fields
1249 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_rec_descriptor' AND enabled;
1251 FOR attr_def IN SELECT * FROM config.record_attr_definition ORDER BY format LOOP
1253 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
1254 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(value), COALESCE(attr_def.joiner,' ')) INTO attr_value
1255 FROM (SELECT * FROM metabib.full_rec ORDER BY tag, subfield) AS x
1256 WHERE record = NEW.id
1257 AND tag LIKE attr_def.tag
1259 WHEN attr_def.sf_list IS NOT NULL
1260 THEN POSITION(subfield IN attr_def.sf_list) > 0
1267 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
1268 attr_value := biblio.marc21_extract_fixed_field(NEW.id, attr_def.fixed_field);
1270 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
1272 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
1274 -- See if we can skip the XSLT ... it's expensive
1275 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
1276 -- Can't skip the transform
1277 IF xfrm.xslt <> '---' THEN
1278 transformed_xml := oils_xslt_process(NEW.marc,xfrm.xslt);
1280 transformed_xml := NEW.marc;
1283 prev_xfrm := xfrm.name;
1286 IF xfrm.name IS NULL THEN
1287 -- just grab the marcxml (empty) transform
1288 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
1289 prev_xfrm := xfrm.name;
1292 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
1294 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
1295 SELECT m.value INTO attr_value
1296 FROM biblio.marc21_physical_characteristics(NEW.id) v
1297 JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
1298 WHERE v.subfield = attr_def.phys_char_sf
1299 LIMIT 1; -- Just in case ...
1303 -- apply index normalizers to attr_value
1305 SELECT n.func AS func,
1306 n.param_count AS param_count,
1308 FROM config.index_normalizer n
1309 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
1310 WHERE attr = attr_def.name
1312 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1313 COALESCE( quote_literal( attr_value ), 'NULL' ) ||
1315 WHEN normalizer.param_count > 0
1316 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1319 ')' INTO attr_value;
1323 -- Add the new value to the hstore
1324 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
1328 IF TG_OP = 'INSERT' OR OLD.deleted THEN -- initial insert OR revivication
1329 DELETE FROM metabib.record_attr WHERE id = NEW.id;
1330 INSERT INTO metabib.record_attr (id, attrs) VALUES (NEW.id, new_attrs);
1332 UPDATE metabib.record_attr SET attrs = new_attrs WHERE id = NEW.id;
1338 -- Gather and insert the field entry data
1339 PERFORM metabib.reingest_metabib_field_entries(NEW.id);
1341 -- Located URI magic
1342 IF TG_OP = 'INSERT' THEN
1343 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1345 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1348 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1350 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1354 -- (re)map metarecord-bib linking
1355 IF TG_OP = 'INSERT' THEN -- if not deleted and performing an insert, check for the flag
1356 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
1358 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1360 ELSE -- we're doing an update, and we're not deleted, remap
1361 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_update' AND enabled;
1363 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1369 $func$ LANGUAGE PLPGSQL;
1371 CREATE OR REPLACE FUNCTION metabib.browse_normalize(facet_text TEXT, mapped_field INT) RETURNS TEXT AS $$
1377 SELECT n.func AS func,
1378 n.param_count AS param_count,
1380 FROM config.index_normalizer n
1381 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
1382 WHERE m.field = mapped_field AND m.pos < 0
1385 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1386 quote_literal( facet_text ) ||
1388 WHEN normalizer.param_count > 0
1389 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1392 ')' INTO facet_text;
1399 $$ LANGUAGE PLPGSQL;
1402 -- This mimics a specific part of QueryParser, turning the first part of a
1403 -- classed search (search_class) into a set of classes and possibly fields.
1404 -- search_class might look like "author" or "title|proper" or "ti|uniform"
1405 -- or "au" or "au|corporate|personal" or anything like that, where the first
1406 -- element of the list you get by separating on the "|" character is either
1407 -- a registered class (config.metabib_class) or an alias
1408 -- (config.metabib_search_alias), and the rest of any such elements are
1409 -- fields (config.metabib_field).
1411 FUNCTION metabib.search_class_to_registered_components(search_class TEXT)
1412 RETURNS SETOF RECORD AS $func$
1414 search_parts TEXT[];
1416 search_part_count INTEGER;
1418 registered_class config.metabib_class%ROWTYPE;
1419 registered_alias config.metabib_search_alias%ROWTYPE;
1420 registered_field config.metabib_field%ROWTYPE;
1422 search_parts := REGEXP_SPLIT_TO_ARRAY(search_class, E'\\|');
1424 search_part_count := ARRAY_LENGTH(search_parts, 1);
1425 IF search_part_count = 0 THEN
1428 SELECT INTO registered_class
1429 * FROM config.metabib_class WHERE name = search_parts[1];
1431 IF search_part_count < 2 THEN -- all fields
1432 rec := (registered_class.name, NULL::INTEGER);
1436 FOR field_name IN SELECT *
1437 FROM UNNEST(search_parts[2:search_part_count]) LOOP
1438 SELECT INTO registered_field
1439 * FROM config.metabib_field
1440 WHERE name = field_name AND
1441 field_class = registered_class.name;
1443 rec := (registered_class.name, registered_field.id);
1448 -- maybe we have an alias?
1449 SELECT INTO registered_alias
1450 * FROM config.metabib_search_alias WHERE alias=search_parts[1];
1454 IF search_part_count < 2 THEN -- return w/e the alias says
1456 registered_alias.field_class, registered_alias.field
1461 FOR field_name IN SELECT *
1462 FROM UNNEST(search_parts[2:search_part_count]) LOOP
1463 SELECT INTO registered_field
1464 * FROM config.metabib_field
1465 WHERE name = field_name AND
1466 field_class = registered_alias.field_class;
1469 registered_alias.field_class,
1480 $func$ LANGUAGE PLPGSQL ROWS 1;
1483 -- Given a string such as a user might type into a search box, prepare
1484 -- two changed variants for TO_TSQUERY(). See
1485 -- http://www.postgresql.org/docs/9.0/static/textsearch-controls.html
1486 -- The first variant is normalized to match indexed documents regardless
1487 -- of diacritics. The second variant keeps its diacritics for proper
1488 -- highlighting via TS_HEADLINE().
1490 FUNCTION metabib.autosuggest_prepare_tsquery(orig TEXT) RETURNS TEXT[] AS
1493 orig_ended_in_space BOOLEAN;
1498 orig_ended_in_space := orig ~ E'\\s$';
1500 orig := ARRAY_TO_STRING(
1501 evergreen.regexp_split_to_array(orig, E'\\W+'), ' '
1504 normalized := public.search_normalize(orig); -- also trim()s
1505 plain := trim(orig);
1507 IF NOT orig_ended_in_space THEN
1508 plain := plain || ':*';
1509 normalized := normalized || ':*';
1512 plain := ARRAY_TO_STRING(
1513 evergreen.regexp_split_to_array(plain, E'\\s+'), ' & '
1515 normalized := ARRAY_TO_STRING(
1516 evergreen.regexp_split_to_array(normalized, E'\\s+'), ' & '
1519 RETURN ARRAY[normalized, plain];
1521 $$ LANGUAGE PLPGSQL;
1525 FUNCTION metabib.suggest_browse_entries(
1526 raw_query_text TEXT, -- actually typed by humans at the UI level
1527 search_class TEXT, -- 'alias' or 'class' or 'class|field..', etc
1528 headline_opts TEXT, -- markup options for ts_headline()
1529 visibility_org INTEGER,-- null if you don't want opac visibility test
1530 query_limit INTEGER,-- use in LIMIT clause of interal query
1531 normalization INTEGER -- argument to TS_RANK_CD()
1533 value TEXT, -- plain
1535 buoyant_and_class_match BOOL,
1537 field_weight INTEGER,
1540 match TEXT -- marked up
1543 prepared_query_texts TEXT[];
1545 plain_query TSQUERY;
1546 opac_visibility_join TEXT;
1547 search_class_join TEXT;
1550 prepared_query_texts := metabib.autosuggest_prepare_tsquery(raw_query_text);
1552 query := TO_TSQUERY('keyword', prepared_query_texts[1]);
1553 plain_query := TO_TSQUERY('keyword', prepared_query_texts[2]);
1555 visibility_org := NULLIF(visibility_org,-1);
1556 IF visibility_org IS NOT NULL THEN
1557 opac_visibility_join := '
1558 JOIN asset.opac_visible_copies aovc ON (
1559 aovc.record = x.source AND
1560 aovc.circ_lib IN (SELECT id FROM actor.org_unit_descendants($4))
1563 opac_visibility_join := '';
1566 -- The following determines whether we only provide suggestsons matching
1567 -- the user's selected search_class, or whether we show other suggestions
1568 -- too. The reason for MIN() is that for search_classes like
1569 -- 'title|proper|uniform' you would otherwise get multiple rows. The
1570 -- implication is that if title as a class doesn't have restrict,
1571 -- nor does the proper field, but the uniform field does, you're going
1572 -- to get 'false' for your overall evaluation of 'should we restrict?'
1573 -- To invert that, change from MIN() to MAX().
1577 MIN(cmc.restrict::INT) AS restrict_class,
1578 MIN(cmf.restrict::INT) AS restrict_field
1579 FROM metabib.search_class_to_registered_components(search_class)
1580 AS _registered (field_class TEXT, field INT)
1582 config.metabib_class cmc ON (cmc.name = _registered.field_class)
1584 config.metabib_field cmf ON (cmf.id = _registered.field);
1586 -- evaluate 'should we restrict?'
1587 IF r_fields.restrict_field::BOOL OR r_fields.restrict_class::BOOL THEN
1588 search_class_join := '
1590 metabib.search_class_to_registered_components($2)
1591 AS _registered (field_class TEXT, field INT) ON (
1592 (_registered.field IS NULL AND
1593 _registered.field_class = cmf.field_class) OR
1594 (_registered.field = cmf.id)
1598 search_class_join := '
1600 metabib.search_class_to_registered_components($2)
1601 AS _registered (field_class TEXT, field INT) ON (
1602 _registered.field_class = cmc.name
1607 RETURN QUERY EXECUTE '
1616 TS_HEADLINE(value, $7, $3)
1617 FROM (SELECT DISTINCT
1620 cmc.buoyant AND _registered.field_class IS NOT NULL AS push,
1621 _registered.field = cmf.id AS restrict,
1623 TS_RANK_CD(mbe.index_vector, $1, $6),
1626 FROM metabib.browse_entry_def_map mbedm
1627 JOIN (SELECT * FROM metabib.browse_entry WHERE index_vector @@ $1 LIMIT 10000) mbe ON (mbe.id = mbedm.entry)
1628 JOIN config.metabib_field cmf ON (cmf.id = mbedm.def)
1629 JOIN config.metabib_class cmc ON (cmf.field_class = cmc.name)
1630 ' || search_class_join || '
1631 ORDER BY 3 DESC, 4 DESC NULLS LAST, 5 DESC, 6 DESC, 7 DESC, 1 ASC
1633 ' || opac_visibility_join || '
1634 ORDER BY 3 DESC, 4 DESC NULLS LAST, 5 DESC, 6 DESC, 7 DESC, 1 ASC
1636 ' -- sic, repeat the order by clause in the outer select too
1638 query, search_class, headline_opts,
1639 visibility_org, query_limit, normalization, plain_query
1643 -- buoyant AND chosen class = match class
1644 -- chosen field = match field
1651 $func$ LANGUAGE PLPGSQL;
1653 CREATE OR REPLACE FUNCTION public.oils_tsearch2 () RETURNS TRIGGER AS $$
1657 temp_vector TEXT := '';
1663 NEW.index_vector = ''::tsvector;
1665 IF TG_TABLE_NAME::TEXT ~ 'field_entry$' THEN
1667 SELECT n.func AS func,
1668 n.param_count AS param_count,
1670 FROM config.index_normalizer n
1671 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
1672 WHERE field = NEW.field AND m.pos < 0
1674 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1675 quote_literal( value ) ||
1677 WHEN normalizer.param_count > 0
1678 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1688 SELECT n.func AS func,
1689 n.param_count AS param_count,
1691 FROM config.index_normalizer n
1692 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
1693 WHERE field = NEW.field AND m.pos >= 0
1695 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1696 quote_literal( value ) ||
1698 WHEN normalizer.param_count > 0
1699 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1707 IF TG_TABLE_NAME::TEXT ~ 'browse_entry$' THEN
1708 value := ARRAY_TO_STRING(
1709 evergreen.regexp_split_to_array(value, E'\\W+'), ' '
1711 value := public.search_normalize(value);
1712 NEW.index_vector = to_tsvector(TG_ARGV[0]::regconfig, value);
1713 ELSIF TG_TABLE_NAME::TEXT ~ 'field_entry$' THEN
1715 SELECT ts_config, index_weight
1716 FROM config.metabib_class_ts_map
1717 WHERE field_class = TG_ARGV[0]
1718 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'))
1719 AND always OR NOT EXISTS (SELECT 1 FROM config.metabib_field_ts_map WHERE metabib_field = NEW.field)
1721 SELECT ts_config, index_weight
1722 FROM config.metabib_field_ts_map
1723 WHERE metabib_field = NEW.field
1724 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'))
1725 ORDER BY index_weight ASC
1727 IF cur_weight IS NOT NULL AND cur_weight != ts_rec.index_weight THEN
1728 NEW.index_vector = NEW.index_vector || setweight(temp_vector::tsvector,cur_weight);
1731 cur_weight = ts_rec.index_weight;
1732 SELECT INTO temp_vector temp_vector || ' ' || to_tsvector(ts_rec.ts_config::regconfig, value)::TEXT;
1734 NEW.index_vector = NEW.index_vector || setweight(temp_vector::tsvector,cur_weight);
1736 NEW.index_vector = to_tsvector(TG_ARGV[0]::regconfig, value);
1741 $$ LANGUAGE PLPGSQL;