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;
415 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
417 -- Loop over the indexing entries
418 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
420 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
422 -- See if we can skip the XSLT ... it's expensive
423 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
424 -- Can't skip the transform
425 IF xfrm.xslt <> '---' THEN
426 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
428 transformed_xml := bib.marc;
431 prev_xfrm := xfrm.name;
434 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
437 FOR xml_node IN SELECT x FROM unnest(xml_node_list) AS x LOOP
438 CONTINUE WHEN xml_node !~ E'^\\s*<';
440 curr_text := ARRAY_TO_STRING(
441 oils_xpath( '//text()',
442 REGEXP_REPLACE( -- This escapes all &s not followed by "amp;". Data ise returned from oils_xpath (above) in UTF-8, not entity encoded
443 REGEXP_REPLACE( -- This escapes embeded <s
445 $re$(>[^<]+)(<)([^>]+<)$re$,
457 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
459 IF raw_text IS NOT NULL THEN
460 raw_text := raw_text || joiner;
463 raw_text := COALESCE(raw_text,'') || curr_text;
465 -- autosuggest/metabib.browse_entry
466 IF idx.browse_field THEN
468 IF idx.browse_xpath IS NOT NULL AND idx.browse_xpath <> '' THEN
469 browse_text := oils_xpath_string( idx.browse_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
471 browse_text := curr_text;
474 output_row.field_class = idx.field_class;
475 output_row.field = idx.id;
476 output_row.source = rid;
477 output_row.value = BTRIM(REGEXP_REPLACE(browse_text, E'\\s+', ' ', 'g'));
479 output_row.browse_field = TRUE;
480 RETURN NEXT output_row;
481 output_row.browse_field = FALSE;
484 -- insert raw node text for faceting
485 IF idx.facet_field THEN
487 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
488 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
490 facet_text := curr_text;
493 output_row.field_class = idx.field_class;
494 output_row.field = -1 * idx.id;
495 output_row.source = rid;
496 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
498 output_row.facet_field = TRUE;
499 RETURN NEXT output_row;
500 output_row.facet_field = FALSE;
505 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
507 -- insert combined node text for searching
508 IF idx.search_field THEN
509 output_row.field_class = idx.field_class;
510 output_row.field = idx.id;
511 output_row.source = rid;
512 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
514 output_row.search_field = TRUE;
515 RETURN NEXT output_row;
522 $func$ LANGUAGE PLPGSQL;
524 CREATE OR REPLACE FUNCTION metabib.update_combined_index_vectors(bib_id BIGINT) RETURNS VOID AS $func$
526 DELETE FROM metabib.combined_keyword_field_entry WHERE record = bib_id;
527 INSERT INTO metabib.combined_keyword_field_entry(record, metabib_field, index_vector)
528 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
529 FROM metabib.keyword_field_entry WHERE source = bib_id GROUP BY field;
530 INSERT INTO metabib.combined_keyword_field_entry(record, metabib_field, index_vector)
531 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
532 FROM metabib.keyword_field_entry WHERE source = bib_id;
534 DELETE FROM metabib.combined_title_field_entry WHERE record = bib_id;
535 INSERT INTO metabib.combined_title_field_entry(record, metabib_field, index_vector)
536 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
537 FROM metabib.title_field_entry WHERE source = bib_id GROUP BY field;
538 INSERT INTO metabib.combined_title_field_entry(record, metabib_field, index_vector)
539 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
540 FROM metabib.title_field_entry WHERE source = bib_id;
542 DELETE FROM metabib.combined_author_field_entry WHERE record = bib_id;
543 INSERT INTO metabib.combined_author_field_entry(record, metabib_field, index_vector)
544 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
545 FROM metabib.author_field_entry WHERE source = bib_id GROUP BY field;
546 INSERT INTO metabib.combined_author_field_entry(record, metabib_field, index_vector)
547 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
548 FROM metabib.author_field_entry WHERE source = bib_id;
550 DELETE FROM metabib.combined_subject_field_entry WHERE record = bib_id;
551 INSERT INTO metabib.combined_subject_field_entry(record, metabib_field, index_vector)
552 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
553 FROM metabib.subject_field_entry WHERE source = bib_id GROUP BY field;
554 INSERT INTO metabib.combined_subject_field_entry(record, metabib_field, index_vector)
555 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
556 FROM metabib.subject_field_entry WHERE source = bib_id;
558 DELETE FROM metabib.combined_series_field_entry WHERE record = bib_id;
559 INSERT INTO metabib.combined_series_field_entry(record, metabib_field, index_vector)
560 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
561 FROM metabib.series_field_entry WHERE source = bib_id GROUP BY field;
562 INSERT INTO metabib.combined_series_field_entry(record, metabib_field, index_vector)
563 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
564 FROM metabib.series_field_entry WHERE source = bib_id;
566 DELETE FROM metabib.combined_identifier_field_entry WHERE record = bib_id;
567 INSERT INTO metabib.combined_identifier_field_entry(record, metabib_field, index_vector)
568 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
569 FROM metabib.identifier_field_entry WHERE source = bib_id GROUP BY field;
570 INSERT INTO metabib.combined_identifier_field_entry(record, metabib_field, index_vector)
571 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
572 FROM metabib.identifier_field_entry WHERE source = bib_id;
575 $func$ LANGUAGE PLPGSQL;
577 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 $func$
580 ind_data metabib.field_entry_template%ROWTYPE;
581 mbe_row metabib.browse_entry%ROWTYPE;
584 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
586 IF NOT skip_search THEN
587 FOR fclass IN SELECT * FROM config.metabib_class LOOP
588 -- RAISE NOTICE 'Emptying out %', fclass.name;
589 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
592 IF NOT skip_facet THEN
593 DELETE FROM metabib.facet_entry WHERE source = bib_id;
595 IF NOT skip_browse THEN
596 DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
600 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
601 IF ind_data.field < 0 THEN
602 ind_data.field = -1 * ind_data.field;
605 IF ind_data.facet_field AND NOT skip_facet THEN
606 INSERT INTO metabib.facet_entry (field, source, value)
607 VALUES (ind_data.field, ind_data.source, ind_data.value);
610 IF ind_data.browse_field AND NOT skip_browse THEN
611 -- A caveat about this SELECT: this should take care of replacing
612 -- old mbe rows when data changes, but not if normalization (by
613 -- which I mean specifically the output of
614 -- evergreen.oils_tsearch2()) changes. It may or may not be
615 -- expensive to add a comparison of index_vector to index_vector
616 -- to the WHERE clause below.
617 SELECT INTO mbe_row * FROM metabib.browse_entry WHERE value = ind_data.value;
619 mbe_id := mbe_row.id;
621 INSERT INTO metabib.browse_entry (value) VALUES
622 (metabib.browse_normalize(ind_data.value, ind_data.field));
623 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
626 INSERT INTO metabib.browse_entry_def_map (entry, def, source)
627 VALUES (mbe_id, ind_data.field, ind_data.source);
630 IF ind_data.search_field AND NOT skip_search THEN
632 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
634 quote_literal(ind_data.field) || $$, $$ ||
635 quote_literal(ind_data.source) || $$, $$ ||
636 quote_literal(ind_data.value) ||
642 IF NOT skip_search THEN
643 PERFORM metabib.update_combined_index_vectors(bib_id);
648 $func$ LANGUAGE PLPGSQL;
650 -- default to a space joiner
651 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
652 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
655 CREATE OR REPLACE FUNCTION authority.flatten_marc ( rid BIGINT ) RETURNS SETOF authority.full_rec AS $func$
657 auth authority.record_entry%ROWTYPE;
658 output authority.full_rec%ROWTYPE;
661 SELECT INTO auth * FROM authority.record_entry WHERE id = rid;
663 FOR field IN SELECT * FROM vandelay.flatten_marc( auth.marc ) LOOP
664 output.record := rid;
665 output.ind1 := field.ind1;
666 output.ind2 := field.ind2;
667 output.tag := field.tag;
668 output.subfield := field.subfield;
669 output.value := field.value;
674 $func$ LANGUAGE PLPGSQL;
676 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
678 bib biblio.record_entry%ROWTYPE;
679 output metabib.full_rec%ROWTYPE;
682 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
684 FOR field IN SELECT * FROM vandelay.flatten_marc( bib.marc ) LOOP
685 output.record := rid;
686 output.ind1 := field.ind1;
687 output.ind2 := field.ind2;
688 output.tag := field.tag;
689 output.subfield := field.subfield;
690 output.value := field.value;
695 $func$ LANGUAGE PLPGSQL;
697 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
704 retval config.marc21_rec_type_map%ROWTYPE;
706 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
708 IF ldr IS NULL OR ldr = '' THEN
709 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
713 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
714 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
717 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
718 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
720 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
722 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
725 IF retval.code IS NULL THEN
726 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
731 $func$ LANGUAGE PLPGSQL;
733 CREATE OR REPLACE FUNCTION biblio.marc21_record_type( rid BIGINT ) RETURNS config.marc21_rec_type_map AS $func$
734 SELECT * FROM vandelay.marc21_record_type( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
737 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
744 rtype := (vandelay.marc21_record_type( marc )).code;
745 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
746 IF ff_pos.tag = 'ldr' THEN
747 val := oils_xpath_string('//*[local-name()="leader"]', marc);
748 IF val IS NOT NULL THEN
749 val := SUBSTRING( val, ff_pos.start_pos + 1, ff_pos.length );
753 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
754 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
758 val := REPEAT( ff_pos.default_val, ff_pos.length );
764 $func$ LANGUAGE PLPGSQL;
766 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
767 SELECT * FROM vandelay.marc21_extract_fixed_field( (SELECT marc FROM biblio.record_entry WHERE id = $1), $2 );
770 -- CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
771 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
776 output biblio.record_ff_map%ROWTYPE;
778 rtype := (vandelay.marc21_record_type( marc )).code;
780 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
781 output.ff_name := ff_pos.fixed_field;
782 output.ff_value := NULL;
784 IF ff_pos.tag = 'ldr' THEN
785 output.ff_value := oils_xpath_string('//*[local-name()="leader"]', marc);
786 IF output.ff_value IS NOT NULL THEN
787 output.ff_value := SUBSTRING( output.ff_value, ff_pos.start_pos + 1, ff_pos.length );
789 output.ff_value := NULL;
792 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
793 output.ff_value := SUBSTRING( tag_data, ff_pos.start_pos + 1, ff_pos.length );
794 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
796 output.ff_value := NULL;
804 $func$ LANGUAGE PLPGSQL;
806 CREATE OR REPLACE FUNCTION biblio.marc21_extract_all_fixed_fields( rid BIGINT ) RETURNS SETOF biblio.record_ff_map AS $func$
807 SELECT $1 AS record, ff_name, ff_value FROM vandelay.marc21_extract_all_fixed_fields( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
810 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
811 SELECT id, $1 AS record, ptype, subfield, value FROM vandelay.marc21_physical_characteristics( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
814 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
827 IF marc IS NULL OR marc = '' THEN
831 -- First, the count of tags
832 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
834 -- now go through a bunch of pain to get the record type
835 IF best_type IS NOT NULL THEN
836 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
838 IF ldr IS NOT NULL THEN
839 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
840 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
843 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
844 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
846 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
848 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
850 IF type_map.code IS NOT NULL THEN
851 IF best_type = type_map.code THEN
852 qual := qual + qual / 2;
855 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
856 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
857 IF ff_tag_data = best_lang THEN
865 -- Now look for some quality metrics
867 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
872 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
879 $func$ LANGUAGE PLPGSQL;
881 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
883 idx config.biblio_fingerprint%ROWTYPE;
884 xfrm config.xml_transform%ROWTYPE;
886 transformed_xml TEXT;
888 xml_node_list TEXT[];
890 output_text TEXT := '';
893 IF marc IS NULL OR marc = '' THEN
897 -- Loop over the indexing entries
898 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
900 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
902 -- See if we can skip the XSLT ... it's expensive
903 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
904 -- Can't skip the transform
905 IF xfrm.xslt <> '---' THEN
906 transformed_xml := oils_xslt_process(marc,xfrm.xslt);
908 transformed_xml := marc;
911 prev_xfrm := xfrm.name;
914 raw_text := COALESCE(
922 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
931 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
932 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
934 IF idx.first_word IS TRUE THEN
935 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
938 output_text := output_text || REGEXP_REPLACE(raw_text, E'\\s+', '', 'g');
945 $func$ LANGUAGE PLPGSQL;
947 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
948 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
951 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
953 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
957 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
958 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
963 $func$ LANGUAGE PLPGSQL;
965 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_full_rec( bib_id BIGINT ) RETURNS VOID AS $func$
967 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
969 DELETE FROM metabib.real_full_rec WHERE record = bib_id;
971 INSERT INTO metabib.real_full_rec (record, tag, ind1, ind2, subfield, value)
972 SELECT record, tag, ind1, ind2, subfield, value FROM biblio.flatten_marc( bib_id );
976 $func$ LANGUAGE PLPGSQL;
978 CREATE OR REPLACE FUNCTION biblio.extract_located_uris( bib_id BIGINT, marcxml TEXT, editor_id INT ) RETURNS VOID AS $func$
985 uri_owner_list TEXT[];
993 -- Clear any URI mappings and call numbers for this bib.
994 -- This leads to acn / auricnm inflation, but also enables
995 -- old acn/auricnm's to go away and for bibs to be deleted.
996 FOR uri_cn_id IN SELECT id FROM asset.call_number WHERE record = bib_id AND label = '##URI##' AND NOT deleted LOOP
997 DELETE FROM asset.uri_call_number_map WHERE call_number = uri_cn_id;
998 DELETE FROM asset.call_number WHERE id = uri_cn_id;
1001 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',marcxml);
1002 IF ARRAY_UPPER(uris,1) > 0 THEN
1003 FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
1004 -- First we pull info out of the 856
1007 uri_href := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
1008 uri_label := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()',uri_xml))[1];
1009 uri_use := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()',uri_xml))[1];
1011 IF uri_label IS NULL THEN
1012 uri_label := uri_href;
1014 CONTINUE WHEN uri_href IS NULL;
1016 -- Get the distinct list of libraries wanting to use
1018 DISTINCT REGEXP_REPLACE(
1020 $re$^.*?\((\w+)\).*$$re$,
1023 ) INTO uri_owner_list
1026 '//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',
1031 IF ARRAY_UPPER(uri_owner_list,1) > 0 THEN
1033 -- look for a matching uri
1034 IF uri_use IS NULL THEN
1035 SELECT id INTO uri_id
1037 WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active
1038 ORDER BY id LIMIT 1;
1039 IF NOT FOUND THEN -- create one
1040 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
1041 SELECT id INTO uri_id
1043 WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active;
1046 SELECT id INTO uri_id
1048 WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active
1049 ORDER BY id LIMIT 1;
1050 IF NOT FOUND THEN -- create one
1051 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
1052 SELECT id INTO uri_id
1054 WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
1058 FOR j IN 1 .. ARRAY_UPPER(uri_owner_list, 1) LOOP
1059 uri_owner := uri_owner_list[j];
1061 SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = uri_owner;
1062 CONTINUE WHEN NOT FOUND;
1064 -- we need a call number to link through
1065 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;
1067 INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
1068 VALUES (uri_owner_id, bib_id, 'now', 'now', editor_id, editor_id, '##URI##');
1069 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;
1072 -- now, link them if they're not already
1073 SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
1075 INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
1087 $func$ LANGUAGE PLPGSQL;
1089 CREATE OR REPLACE FUNCTION metabib.remap_metarecord_for_bib( bib_id BIGINT, fp TEXT ) RETURNS BIGINT AS $func$
1093 tmp_mr metabib.metarecord%ROWTYPE;
1094 deleted_mrs BIGINT[];
1097 DELETE FROM metabib.metarecord_source_map WHERE source = bib_id; -- Rid ourselves of the search-estimate-killing linkage
1099 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
1101 IF old_mr IS NULL AND fp = tmp_mr.fingerprint THEN -- Find the first fingerprint-matching
1102 old_mr := tmp_mr.id;
1104 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
1105 IF source_count = 0 THEN -- No other records
1106 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
1107 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
1113 IF old_mr IS NULL THEN -- we found no suitable, preexisting MR based on old source maps
1114 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp; -- is there one for our current fingerprint?
1115 IF old_mr IS NULL THEN -- nope, create one and grab its id
1116 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( fp, bib_id );
1117 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp;
1118 ELSE -- indeed there is. update it with a null cache and recalcualated master record
1119 UPDATE metabib.metarecord
1121 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
1124 ELSE -- there was one we already attached to, update its mods cache and master_record
1125 UPDATE metabib.metarecord
1127 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
1131 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, bib_id); -- new source mapping
1133 IF ARRAY_UPPER(deleted_mrs,1) > 0 THEN
1134 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
1140 $func$ LANGUAGE PLPGSQL;
1142 CREATE OR REPLACE FUNCTION biblio.map_authority_linking (bibid BIGINT, marc TEXT) RETURNS BIGINT AS $func$
1143 DELETE FROM authority.bib_linking WHERE bib = $1;
1144 INSERT INTO authority.bib_linking (bib, authority)
1147 FROM ( SELECT DISTINCT $1 AS bib,
1148 BTRIM(remove_paren_substring(txt))::BIGINT AS authority
1149 FROM unnest(oils_xpath('//*[@code="0"]/text()',$2)) x(txt)
1150 WHERE BTRIM(remove_paren_substring(txt)) ~ $re$^\d+$$re$
1151 ) y JOIN authority.record_entry r ON r.id = y.authority;
1153 $func$ LANGUAGE SQL;
1155 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
1156 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
1158 transformed_xml TEXT;
1161 xfrm config.xml_transform%ROWTYPE;
1163 new_attrs HSTORE := ''::HSTORE;
1164 attr_def config.record_attr_definition%ROWTYPE;
1167 IF NEW.deleted IS TRUE THEN -- If this bib is deleted
1168 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage
1169 DELETE FROM metabib.record_attr WHERE id = NEW.id; -- Kill the attrs hash, useless on deleted records
1170 DELETE FROM authority.bib_linking WHERE bib = NEW.id; -- Avoid updating fields in bibs that are no longer visible
1171 DELETE FROM biblio.peer_bib_copy_map WHERE peer_record = NEW.id; -- Separate any multi-homed items
1172 DELETE FROM metabib.browse_entry_def_map WHERE source = NEW.id; -- Don't auto-suggest deleted bibs
1173 RETURN NEW; -- and we're done
1176 IF TG_OP = 'UPDATE' THEN -- re-ingest?
1177 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
1179 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
1184 -- Record authority linking
1185 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_linking' AND enabled;
1187 PERFORM biblio.map_authority_linking( NEW.id, NEW.marc );
1190 -- Flatten and insert the mfr data
1191 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_full_rec' AND enabled;
1193 PERFORM metabib.reingest_metabib_full_rec(NEW.id);
1195 -- Now we pull out attribute data, which is dependent on the mfr for all but XPath-based fields
1196 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_rec_descriptor' AND enabled;
1198 FOR attr_def IN SELECT * FROM config.record_attr_definition ORDER BY format LOOP
1200 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
1201 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(value), COALESCE(attr_def.joiner,' ')) INTO attr_value
1202 FROM (SELECT * FROM metabib.full_rec ORDER BY tag, subfield) AS x
1203 WHERE record = NEW.id
1204 AND tag LIKE attr_def.tag
1206 WHEN attr_def.sf_list IS NOT NULL
1207 THEN POSITION(subfield IN attr_def.sf_list) > 0
1214 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
1215 attr_value := biblio.marc21_extract_fixed_field(NEW.id, attr_def.fixed_field);
1217 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
1219 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
1221 -- See if we can skip the XSLT ... it's expensive
1222 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
1223 -- Can't skip the transform
1224 IF xfrm.xslt <> '---' THEN
1225 transformed_xml := oils_xslt_process(NEW.marc,xfrm.xslt);
1227 transformed_xml := NEW.marc;
1230 prev_xfrm := xfrm.name;
1233 IF xfrm.name IS NULL THEN
1234 -- just grab the marcxml (empty) transform
1235 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
1236 prev_xfrm := xfrm.name;
1239 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
1241 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
1242 SELECT m.value INTO attr_value
1243 FROM biblio.marc21_physical_characteristics(NEW.id) v
1244 JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
1245 WHERE v.subfield = attr_def.phys_char_sf
1246 LIMIT 1; -- Just in case ...
1250 -- apply index normalizers to attr_value
1252 SELECT n.func AS func,
1253 n.param_count AS param_count,
1255 FROM config.index_normalizer n
1256 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
1257 WHERE attr = attr_def.name
1259 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1260 COALESCE( quote_literal( attr_value ), 'NULL' ) ||
1262 WHEN normalizer.param_count > 0
1263 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1266 ')' INTO attr_value;
1270 -- Add the new value to the hstore
1271 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
1275 IF TG_OP = 'INSERT' OR OLD.deleted THEN -- initial insert OR revivication
1276 INSERT INTO metabib.record_attr (id, attrs) VALUES (NEW.id, new_attrs);
1278 UPDATE metabib.record_attr SET attrs = new_attrs WHERE id = NEW.id;
1284 -- Gather and insert the field entry data
1285 PERFORM metabib.reingest_metabib_field_entries(NEW.id);
1287 -- Located URI magic
1288 IF TG_OP = 'INSERT' THEN
1289 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1291 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1294 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1296 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1300 -- (re)map metarecord-bib linking
1301 IF TG_OP = 'INSERT' THEN -- if not deleted and performing an insert, check for the flag
1302 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
1304 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1306 ELSE -- we're doing an update, and we're not deleted, remap
1307 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_update' AND enabled;
1309 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1315 $func$ LANGUAGE PLPGSQL;
1317 CREATE OR REPLACE FUNCTION metabib.browse_normalize(facet_text TEXT, mapped_field INT) RETURNS TEXT AS $$
1323 SELECT n.func AS func,
1324 n.param_count AS param_count,
1326 FROM config.index_normalizer n
1327 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
1328 WHERE m.field = mapped_field AND m.pos < 0
1331 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1332 quote_literal( facet_text ) ||
1334 WHEN normalizer.param_count > 0
1335 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1338 ')' INTO facet_text;
1345 $$ LANGUAGE PLPGSQL;
1348 -- This mimics a specific part of QueryParser, turning the first part of a
1349 -- classed search (search_class) into a set of classes and possibly fields.
1350 -- search_class might look like "author" or "title|proper" or "ti|uniform"
1351 -- or "au" or "au|corporate|personal" or anything like that, where the first
1352 -- element of the list you get by separating on the "|" character is either
1353 -- a registered class (config.metabib_class) or an alias
1354 -- (config.metabib_search_alias), and the rest of any such elements are
1355 -- fields (config.metabib_field).
1357 FUNCTION metabib.search_class_to_registered_components(search_class TEXT)
1358 RETURNS SETOF RECORD AS $func$
1360 search_parts TEXT[];
1362 search_part_count INTEGER;
1364 registered_class config.metabib_class%ROWTYPE;
1365 registered_alias config.metabib_search_alias%ROWTYPE;
1366 registered_field config.metabib_field%ROWTYPE;
1368 search_parts := REGEXP_SPLIT_TO_ARRAY(search_class, E'\\|');
1370 search_part_count := ARRAY_LENGTH(search_parts, 1);
1371 IF search_part_count = 0 THEN
1374 SELECT INTO registered_class
1375 * FROM config.metabib_class WHERE name = search_parts[1];
1377 IF search_part_count < 2 THEN -- all fields
1378 rec := (registered_class.name, NULL::INTEGER);
1382 FOR field_name IN SELECT *
1383 FROM UNNEST(search_parts[2:search_part_count]) LOOP
1384 SELECT INTO registered_field
1385 * FROM config.metabib_field
1386 WHERE name = field_name AND
1387 field_class = registered_class.name;
1389 rec := (registered_class.name, registered_field.id);
1394 -- maybe we have an alias?
1395 SELECT INTO registered_alias
1396 * FROM config.metabib_search_alias WHERE alias=search_parts[1];
1400 IF search_part_count < 2 THEN -- return w/e the alias says
1402 registered_alias.field_class, registered_alias.field
1407 FOR field_name IN SELECT *
1408 FROM UNNEST(search_parts[2:search_part_count]) LOOP
1409 SELECT INTO registered_field
1410 * FROM config.metabib_field
1411 WHERE name = field_name AND
1412 field_class = registered_alias.field_class;
1415 registered_alias.field_class,
1426 $func$ LANGUAGE PLPGSQL ROWS 1;
1429 -- Given a string such as a user might type into a search box, prepare
1430 -- two changed variants for TO_TSQUERY(). See
1431 -- http://www.postgresql.org/docs/9.0/static/textsearch-controls.html
1432 -- The first variant is normalized to match indexed documents regardless
1433 -- of diacritics. The second variant keeps its diacritics for proper
1434 -- highlighting via TS_HEADLINE().
1436 FUNCTION metabib.autosuggest_prepare_tsquery(orig TEXT) RETURNS TEXT[] AS
1439 orig_ended_in_space BOOLEAN;
1444 orig_ended_in_space := orig ~ E'\\s$';
1446 orig := ARRAY_TO_STRING(
1447 evergreen.regexp_split_to_array(orig, E'\\W+'), ' '
1450 normalized := public.search_normalize(orig); -- also trim()s
1451 plain := trim(orig);
1453 IF NOT orig_ended_in_space THEN
1454 plain := plain || ':*';
1455 normalized := normalized || ':*';
1458 plain := ARRAY_TO_STRING(
1459 evergreen.regexp_split_to_array(plain, E'\\s+'), ' & '
1461 normalized := ARRAY_TO_STRING(
1462 evergreen.regexp_split_to_array(normalized, E'\\s+'), ' & '
1465 RETURN ARRAY[normalized, plain];
1467 $$ LANGUAGE PLPGSQL;
1471 FUNCTION metabib.suggest_browse_entries(
1472 raw_query_text TEXT, -- actually typed by humans at the UI level
1473 search_class TEXT, -- 'alias' or 'class' or 'class|field..', etc
1474 headline_opts TEXT, -- markup options for ts_headline()
1475 visibility_org INTEGER,-- null if you don't want opac visibility test
1476 query_limit INTEGER,-- use in LIMIT clause of interal query
1477 normalization INTEGER -- argument to TS_RANK_CD()
1479 value TEXT, -- plain
1481 buoyant_and_class_match BOOL,
1483 field_weight INTEGER,
1486 match TEXT -- marked up
1489 prepared_query_texts TEXT[];
1491 plain_query TSQUERY;
1492 opac_visibility_join TEXT;
1493 search_class_join TEXT;
1496 prepared_query_texts := metabib.autosuggest_prepare_tsquery(raw_query_text);
1498 query := TO_TSQUERY('keyword', prepared_query_texts[1]);
1499 plain_query := TO_TSQUERY('keyword', prepared_query_texts[2]);
1501 visibility_org := NULLIF(visibility_org,-1);
1502 IF visibility_org IS NOT NULL THEN
1503 opac_visibility_join := '
1504 JOIN asset.opac_visible_copies aovc ON (
1505 aovc.record = x.source AND
1506 aovc.circ_lib IN (SELECT id FROM actor.org_unit_descendants($4))
1509 opac_visibility_join := '';
1512 -- The following determines whether we only provide suggestsons matching
1513 -- the user's selected search_class, or whether we show other suggestions
1514 -- too. The reason for MIN() is that for search_classes like
1515 -- 'title|proper|uniform' you would otherwise get multiple rows. The
1516 -- implication is that if title as a class doesn't have restrict,
1517 -- nor does the proper field, but the uniform field does, you're going
1518 -- to get 'false' for your overall evaluation of 'should we restrict?'
1519 -- To invert that, change from MIN() to MAX().
1523 MIN(cmc.restrict::INT) AS restrict_class,
1524 MIN(cmf.restrict::INT) AS restrict_field
1525 FROM metabib.search_class_to_registered_components(search_class)
1526 AS _registered (field_class TEXT, field INT)
1528 config.metabib_class cmc ON (cmc.name = _registered.field_class)
1530 config.metabib_field cmf ON (cmf.id = _registered.field);
1532 -- evaluate 'should we restrict?'
1533 IF r_fields.restrict_field::BOOL OR r_fields.restrict_class::BOOL THEN
1534 search_class_join := '
1536 metabib.search_class_to_registered_components($2)
1537 AS _registered (field_class TEXT, field INT) ON (
1538 (_registered.field IS NULL AND
1539 _registered.field_class = cmf.field_class) OR
1540 (_registered.field = cmf.id)
1544 search_class_join := '
1546 metabib.search_class_to_registered_components($2)
1547 AS _registered (field_class TEXT, field INT) ON (
1548 _registered.field_class = cmc.name
1553 RETURN QUERY EXECUTE '
1562 TS_HEADLINE(value, $7, $3)
1563 FROM (SELECT DISTINCT
1566 cmc.buoyant AND _registered.field_class IS NOT NULL AS push,
1567 _registered.field = cmf.id AS restrict,
1569 TS_RANK_CD(mbe.index_vector, $1, $6),
1572 FROM metabib.browse_entry_def_map mbedm
1573 JOIN (SELECT * FROM metabib.browse_entry WHERE index_vector @@ $1 LIMIT 10000) mbe ON (mbe.id = mbedm.entry)
1574 JOIN config.metabib_field cmf ON (cmf.id = mbedm.def)
1575 JOIN config.metabib_class cmc ON (cmf.field_class = cmc.name)
1576 ' || search_class_join || '
1577 ORDER BY 3 DESC, 4 DESC NULLS LAST, 5 DESC, 6 DESC, 7 DESC, 1 ASC
1579 ' || opac_visibility_join || '
1580 ORDER BY 3 DESC, 4 DESC NULLS LAST, 5 DESC, 6 DESC, 7 DESC, 1 ASC
1582 ' -- sic, repeat the order by clause in the outer select too
1584 query, search_class, headline_opts,
1585 visibility_org, query_limit, normalization, plain_query
1589 -- buoyant AND chosen class = match class
1590 -- chosen field = match field
1597 $func$ LANGUAGE PLPGSQL;
1599 CREATE OR REPLACE FUNCTION public.oils_tsearch2 () RETURNS TRIGGER AS $$
1603 temp_vector TEXT := '';
1609 NEW.index_vector = ''::tsvector;
1611 IF TG_TABLE_NAME::TEXT ~ 'field_entry$' THEN
1613 SELECT n.func AS func,
1614 n.param_count AS param_count,
1616 FROM config.index_normalizer n
1617 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
1618 WHERE field = NEW.field
1620 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1621 quote_literal( value ) ||
1623 WHEN normalizer.param_count > 0
1624 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1633 IF TG_TABLE_NAME::TEXT ~ 'browse_entry$' THEN
1634 value := ARRAY_TO_STRING(
1635 evergreen.regexp_split_to_array(value, E'\\W+'), ' '
1637 value := public.search_normalize(value);
1638 NEW.index_vector = to_tsvector(TG_ARGV[0]::regconfig, value);
1639 ELSIF TG_TABLE_NAME::TEXT ~ 'field_entry$' THEN
1641 SELECT ts_config, index_weight
1642 FROM config.metabib_class_ts_map
1643 WHERE field_class = TG_ARGV[0]
1644 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'))
1645 AND always OR NOT EXISTS (SELECT 1 FROM config.metabib_field_ts_map WHERE metabib_field = NEW.field)
1647 SELECT ts_config, index_weight
1648 FROM config.metabib_field_ts_map
1649 WHERE metabib_field = NEW.field
1650 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'))
1651 ORDER BY index_weight ASC
1653 IF cur_weight IS NOT NULL AND cur_weight != ts_rec.index_weight THEN
1654 NEW.index_vector = NEW.index_vector || setweight(temp_vector::tsvector,cur_weight);
1657 cur_weight = ts_rec.index_weight;
1658 SELECT INTO temp_vector temp_vector || ' ' || to_tsvector(ts_rec.ts_config::regconfig, value)::TEXT;
1660 NEW.index_vector = NEW.index_vector || setweight(temp_vector::tsvector,cur_weight);
1662 NEW.index_vector = to_tsvector(TG_ARGV[0]::regconfig, value);
1667 $$ LANGUAGE PLPGSQL;