2 * Copyright (C) 2004-2008 Georgia Public Library Service
3 * Copyright (C) 2007-2008 Equinox Software, Inc.
4 * Mike Rylander <miker@esilibrary.com>
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public License
8 * as published by the Free Software Foundation; either version 2
9 * of the License, or (at your option) any later version.
11 * This program is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
18 DROP SCHEMA IF EXISTS metabib CASCADE;
21 CREATE SCHEMA metabib;
23 CREATE TABLE metabib.metarecord (
24 id BIGSERIAL PRIMARY KEY,
25 fingerprint TEXT NOT NULL,
29 CREATE INDEX metabib_metarecord_master_record_idx ON metabib.metarecord (master_record);
30 CREATE INDEX metabib_metarecord_fingerprint_idx ON metabib.metarecord (fingerprint);
32 CREATE TABLE metabib.identifier_field_entry (
33 id BIGSERIAL PRIMARY KEY,
34 source BIGINT NOT NULL,
37 index_vector tsvector NOT NULL
39 CREATE TRIGGER metabib_identifier_field_entry_fti_trigger
40 BEFORE UPDATE OR INSERT ON metabib.identifier_field_entry
41 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('identifier');
43 CREATE INDEX metabib_identifier_field_entry_index_vector_idx ON metabib.identifier_field_entry USING GIST (index_vector);
44 CREATE INDEX metabib_identifier_field_entry_value_idx ON metabib.identifier_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
45 CREATE INDEX metabib_identifier_field_entry_source_idx ON metabib.identifier_field_entry (source);
48 CREATE TABLE metabib.title_field_entry (
49 id BIGSERIAL PRIMARY KEY,
50 source BIGINT NOT NULL,
53 index_vector tsvector NOT NULL
55 CREATE TRIGGER metabib_title_field_entry_fti_trigger
56 BEFORE UPDATE OR INSERT ON metabib.title_field_entry
57 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('title');
59 CREATE INDEX metabib_title_field_entry_index_vector_idx ON metabib.title_field_entry USING GIST (index_vector);
60 CREATE INDEX metabib_title_field_entry_value_idx ON metabib.title_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
61 CREATE INDEX metabib_title_field_entry_source_idx ON metabib.title_field_entry (source);
64 CREATE TABLE metabib.author_field_entry (
65 id BIGSERIAL PRIMARY KEY,
66 source BIGINT NOT NULL,
69 index_vector tsvector NOT NULL
71 CREATE TRIGGER metabib_author_field_entry_fti_trigger
72 BEFORE UPDATE OR INSERT ON metabib.author_field_entry
73 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('author');
75 CREATE INDEX metabib_author_field_entry_index_vector_idx ON metabib.author_field_entry USING GIST (index_vector);
76 CREATE INDEX metabib_author_field_entry_value_idx ON metabib.author_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
77 CREATE INDEX metabib_author_field_entry_source_idx ON metabib.author_field_entry (source);
80 CREATE TABLE metabib.subject_field_entry (
81 id BIGSERIAL PRIMARY KEY,
82 source BIGINT NOT NULL,
85 index_vector tsvector NOT NULL
87 CREATE TRIGGER metabib_subject_field_entry_fti_trigger
88 BEFORE UPDATE OR INSERT ON metabib.subject_field_entry
89 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('subject');
91 CREATE INDEX metabib_subject_field_entry_index_vector_idx ON metabib.subject_field_entry USING GIST (index_vector);
92 CREATE INDEX metabib_subject_field_entry_value_idx ON metabib.subject_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
93 CREATE INDEX metabib_subject_field_entry_source_idx ON metabib.subject_field_entry (source);
96 CREATE TABLE metabib.keyword_field_entry (
97 id BIGSERIAL PRIMARY KEY,
98 source BIGINT NOT NULL,
101 index_vector tsvector NOT NULL
103 CREATE TRIGGER metabib_keyword_field_entry_fti_trigger
104 BEFORE UPDATE OR INSERT ON metabib.keyword_field_entry
105 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('keyword');
107 CREATE INDEX metabib_keyword_field_entry_index_vector_idx ON metabib.keyword_field_entry USING GIST (index_vector);
108 CREATE INDEX metabib_keyword_field_entry_value_idx ON metabib.keyword_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
109 CREATE INDEX metabib_keyword_field_entry_source_idx ON metabib.keyword_field_entry (source);
112 CREATE TABLE metabib.series_field_entry (
113 id BIGSERIAL PRIMARY KEY,
114 source BIGINT NOT NULL,
117 index_vector tsvector NOT NULL
119 CREATE TRIGGER metabib_series_field_entry_fti_trigger
120 BEFORE UPDATE OR INSERT ON metabib.series_field_entry
121 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('series');
123 CREATE INDEX metabib_series_field_entry_index_vector_idx ON metabib.series_field_entry USING GIST (index_vector);
124 CREATE INDEX metabib_series_field_entry_value_idx ON metabib.series_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
125 CREATE INDEX metabib_series_field_entry_source_idx ON metabib.series_field_entry (source);
128 CREATE TABLE metabib.facet_entry (
129 id BIGSERIAL PRIMARY KEY,
130 source BIGINT NOT NULL,
134 CREATE INDEX metabib_facet_entry_field_idx ON metabib.facet_entry (field);
135 CREATE INDEX metabib_facet_entry_value_idx ON metabib.facet_entry (SUBSTRING(value,1,1024));
136 CREATE INDEX metabib_facet_entry_source_idx ON metabib.facet_entry (source);
138 CREATE OR REPLACE FUNCTION metabib.facet_normalize_trigger () RETURNS TRIGGER AS $$
143 facet_text := NEW.value;
146 SELECT n.func AS func,
147 n.param_count AS param_count,
149 FROM config.index_normalizer n
150 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
151 WHERE m.field = NEW.field AND m.pos < 0
154 EXECUTE 'SELECT ' || normalizer.func || '(' ||
155 quote_literal( facet_text ) ||
157 WHEN normalizer.param_count > 0
158 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
165 NEW.value = facet_text;
171 CREATE TRIGGER facet_normalize_tgr
172 BEFORE UPDATE OR INSERT ON metabib.facet_entry
173 FOR EACH ROW EXECUTE PROCEDURE metabib.facet_normalize_trigger();
175 CREATE OR REPLACE FUNCTION evergreen.facet_force_nfc() RETURNS TRIGGER AS $$
177 NEW.value := force_unicode_normal_form(NEW.value,'NFC');
182 CREATE TRIGGER facet_force_nfc_tgr
183 BEFORE UPDATE OR INSERT ON metabib.facet_entry
184 FOR EACH ROW EXECUTE PROCEDURE evergreen.facet_force_nfc();
186 CREATE TABLE metabib.record_attr (
187 id BIGINT PRIMARY KEY REFERENCES biblio.record_entry (id) ON DELETE CASCADE,
188 attrs HSTORE NOT NULL DEFAULT ''::HSTORE
190 CREATE INDEX metabib_svf_attrs_idx ON metabib.record_attr USING GIST (attrs);
191 CREATE INDEX metabib_svf_date1_idx ON metabib.record_attr ((attrs->'date1'));
192 CREATE INDEX metabib_svf_dates_idx ON metabib.record_attr ((attrs->'date1'),(attrs->'date2'));
194 -- Back-compat view ... we're moving to an HSTORE world
195 CREATE TYPE metabib.rec_desc_type AS (
213 CREATE VIEW metabib.rec_descriptor AS
216 (populate_record(NULL::metabib.rec_desc_type, attrs)).*
217 FROM metabib.record_attr;
219 -- Use a sequence that matches previous version, for easier upgrading.
220 CREATE SEQUENCE metabib.full_rec_id_seq;
222 CREATE TABLE metabib.real_full_rec (
223 id BIGINT NOT NULL DEFAULT NEXTVAL('metabib.full_rec_id_seq'::REGCLASS),
224 record BIGINT NOT NULL,
225 tag CHAR(3) NOT NULL,
230 index_vector tsvector NOT NULL
232 ALTER TABLE metabib.real_full_rec ADD PRIMARY KEY (id);
234 CREATE INDEX metabib_full_rec_tag_subfield_idx ON metabib.real_full_rec (tag,subfield);
235 CREATE INDEX metabib_full_rec_value_idx ON metabib.real_full_rec (substring(value,1,1024));
236 /* Enable LIKE to use an index for database clusters with locales other than C or POSIX */
237 CREATE INDEX metabib_full_rec_value_tpo_index ON metabib.real_full_rec (substring(value,1,1024) text_pattern_ops);
238 CREATE INDEX metabib_full_rec_record_idx ON metabib.real_full_rec (record);
239 CREATE INDEX metabib_full_rec_index_vector_idx ON metabib.real_full_rec USING GIST (index_vector);
241 CREATE TRIGGER metabib_full_rec_fti_trigger
242 BEFORE UPDATE OR INSERT ON metabib.real_full_rec
243 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('default');
245 CREATE OR REPLACE VIEW metabib.full_rec AS
252 SUBSTRING(value,1,1024) AS value,
254 FROM metabib.real_full_rec;
256 CREATE OR REPLACE RULE metabib_full_rec_insert_rule
257 AS ON INSERT TO metabib.full_rec
259 INSERT INTO metabib.real_full_rec VALUES (
260 COALESCE(NEW.id, NEXTVAL('metabib.full_rec_id_seq'::REGCLASS)),
270 CREATE OR REPLACE RULE metabib_full_rec_update_rule
271 AS ON UPDATE TO metabib.full_rec
273 UPDATE metabib.real_full_rec SET
279 subfield = NEW.subfield,
281 index_vector = NEW.index_vector
284 CREATE OR REPLACE RULE metabib_full_rec_delete_rule
285 AS ON DELETE TO metabib.full_rec
287 DELETE FROM metabib.real_full_rec WHERE id = OLD.id;
289 CREATE TABLE metabib.metarecord_source_map (
290 id BIGSERIAL PRIMARY KEY,
291 metarecord BIGINT NOT NULL,
292 source BIGINT NOT NULL
294 CREATE INDEX metabib_metarecord_source_map_metarecord_idx ON metabib.metarecord_source_map (metarecord);
295 CREATE INDEX metabib_metarecord_source_map_source_record_idx ON metabib.metarecord_source_map (source);
297 CREATE TYPE metabib.field_entry_template AS (
304 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
306 bib biblio.record_entry%ROWTYPE;
307 idx config.metabib_field%ROWTYPE;
308 xfrm config.xml_transform%ROWTYPE;
310 transformed_xml TEXT;
312 xml_node_list TEXT[];
316 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
317 output_row metabib.field_entry_template%ROWTYPE;
321 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
323 -- Loop over the indexing entries
324 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
326 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
328 -- See if we can skip the XSLT ... it's expensive
329 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
330 -- Can't skip the transform
331 IF xfrm.xslt <> '---' THEN
332 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
334 transformed_xml := bib.marc;
337 prev_xfrm := xfrm.name;
340 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
343 FOR xml_node IN SELECT x FROM unnest(xml_node_list) AS x LOOP
344 CONTINUE WHEN xml_node !~ E'^\\s*<';
346 curr_text := ARRAY_TO_STRING(
347 oils_xpath( '//text()',
348 REGEXP_REPLACE( -- This escapes all &s not followed by "amp;". Data ise returned from oils_xpath (above) in UTF-8, not entity encoded
349 REGEXP_REPLACE( -- This escapes embeded <s
351 $re$(>[^<]+)(<)([^>]+<)$re$,
363 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
365 IF raw_text IS NOT NULL THEN
366 raw_text := raw_text || joiner;
369 raw_text := COALESCE(raw_text,'') || curr_text;
371 -- insert raw node text for faceting
372 IF idx.facet_field THEN
374 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
375 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
377 facet_text := curr_text;
380 output_row.field_class = idx.field_class;
381 output_row.field = -1 * idx.id;
382 output_row.source = rid;
383 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
385 RETURN NEXT output_row;
390 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
392 -- insert combined node text for searching
393 IF idx.search_field THEN
394 output_row.field_class = idx.field_class;
395 output_row.field = idx.id;
396 output_row.source = rid;
397 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
399 RETURN NEXT output_row;
405 $func$ LANGUAGE PLPGSQL;
407 -- default to a space joiner
408 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
409 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
412 CREATE OR REPLACE FUNCTION authority.flatten_marc ( rid BIGINT ) RETURNS SETOF authority.full_rec AS $func$
414 auth authority.record_entry%ROWTYPE;
415 output authority.full_rec%ROWTYPE;
418 SELECT INTO auth * FROM authority.record_entry WHERE id = rid;
420 FOR field IN SELECT * FROM vandelay.flatten_marc( auth.marc ) LOOP
421 output.record := rid;
422 output.ind1 := field.ind1;
423 output.ind2 := field.ind2;
424 output.tag := field.tag;
425 output.subfield := field.subfield;
426 output.value := field.value;
431 $func$ LANGUAGE PLPGSQL;
433 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
435 bib biblio.record_entry%ROWTYPE;
436 output metabib.full_rec%ROWTYPE;
439 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
441 FOR field IN SELECT * FROM vandelay.flatten_marc( bib.marc ) LOOP
442 output.record := rid;
443 output.ind1 := field.ind1;
444 output.ind2 := field.ind2;
445 output.tag := field.tag;
446 output.subfield := field.subfield;
447 output.value := field.value;
452 $func$ LANGUAGE PLPGSQL;
454 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
461 retval config.marc21_rec_type_map%ROWTYPE;
463 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
465 IF ldr IS NULL OR ldr = '' THEN
466 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
470 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
471 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
474 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
475 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
477 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
479 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
482 IF retval.code IS NULL THEN
483 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
488 $func$ LANGUAGE PLPGSQL;
490 CREATE OR REPLACE FUNCTION biblio.marc21_record_type( rid BIGINT ) RETURNS config.marc21_rec_type_map AS $func$
491 SELECT * FROM vandelay.marc21_record_type( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
494 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
501 rtype := (vandelay.marc21_record_type( marc )).code;
502 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
503 IF ff_pos.tag = 'ldr' THEN
504 val := oils_xpath_string('//*[local-name()="leader"]', marc);
505 IF val IS NOT NULL THEN
506 val := SUBSTRING( val, ff_pos.start_pos + 1, ff_pos.length );
510 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
511 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
515 val := REPEAT( ff_pos.default_val, ff_pos.length );
521 $func$ LANGUAGE PLPGSQL;
523 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
524 SELECT * FROM vandelay.marc21_extract_fixed_field( (SELECT marc FROM biblio.record_entry WHERE id = $1), $2 );
527 -- CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
528 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
533 output biblio.record_ff_map%ROWTYPE;
535 rtype := (vandelay.marc21_record_type( marc )).code;
537 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
538 output.ff_name := ff_pos.fixed_field;
539 output.ff_value := NULL;
541 IF ff_pos.tag = 'ldr' THEN
542 output.ff_value := oils_xpath_string('//*[local-name()="leader"]', marc);
543 IF output.ff_value IS NOT NULL THEN
544 output.ff_value := SUBSTRING( output.ff_value, ff_pos.start_pos + 1, ff_pos.length );
546 output.ff_value := NULL;
549 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
550 output.ff_value := SUBSTRING( tag_data, ff_pos.start_pos + 1, ff_pos.length );
551 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
553 output.ff_value := NULL;
561 $func$ LANGUAGE PLPGSQL;
563 CREATE OR REPLACE FUNCTION biblio.marc21_extract_all_fixed_fields( rid BIGINT ) RETURNS SETOF biblio.record_ff_map AS $func$
564 SELECT $1 AS record, ff_name, ff_value FROM vandelay.marc21_extract_all_fixed_fields( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
567 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
568 SELECT id, $1 AS record, ptype, subfield, value FROM vandelay.marc21_physical_characteristics( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
571 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
584 IF marc IS NULL OR marc = '' THEN
588 -- First, the count of tags
589 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
591 -- now go through a bunch of pain to get the record type
592 IF best_type IS NOT NULL THEN
593 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
595 IF ldr IS NOT NULL THEN
596 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
597 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
600 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
601 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
603 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
605 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
607 IF type_map.code IS NOT NULL THEN
608 IF best_type = type_map.code THEN
609 qual := qual + qual / 2;
612 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
613 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
614 IF ff_tag_data = best_lang THEN
622 -- Now look for some quality metrics
624 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
629 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
636 $func$ LANGUAGE PLPGSQL;
638 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
640 idx config.biblio_fingerprint%ROWTYPE;
641 xfrm config.xml_transform%ROWTYPE;
643 transformed_xml TEXT;
645 xml_node_list TEXT[];
647 output_text TEXT := '';
650 IF marc IS NULL OR marc = '' THEN
654 -- Loop over the indexing entries
655 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
657 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
659 -- See if we can skip the XSLT ... it's expensive
660 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
661 -- Can't skip the transform
662 IF xfrm.xslt <> '---' THEN
663 transformed_xml := oils_xslt_process(marc,xfrm.xslt);
665 transformed_xml := marc;
668 prev_xfrm := xfrm.name;
671 raw_text := COALESCE(
679 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
688 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
689 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
691 IF idx.first_word IS TRUE THEN
692 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
695 output_text := output_text || REGEXP_REPLACE(raw_text, E'\\s+', '', 'g');
702 $func$ LANGUAGE PLPGSQL;
704 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
705 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
708 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
710 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
714 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
715 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
720 $func$ LANGUAGE PLPGSQL;
722 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_full_rec( bib_id BIGINT ) RETURNS VOID AS $func$
724 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
726 DELETE FROM metabib.real_full_rec WHERE record = bib_id;
728 INSERT INTO metabib.real_full_rec (record, tag, ind1, ind2, subfield, value)
729 SELECT record, tag, ind1, ind2, subfield, value FROM biblio.flatten_marc( bib_id );
733 $func$ LANGUAGE PLPGSQL;
735 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries( bib_id BIGINT ) RETURNS VOID AS $func$
738 ind_data metabib.field_entry_template%ROWTYPE;
740 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
742 FOR fclass IN SELECT * FROM config.metabib_class LOOP
743 -- RAISE NOTICE 'Emptying out %', fclass.name;
744 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
746 DELETE FROM metabib.facet_entry WHERE source = bib_id;
749 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
750 IF ind_data.field < 0 THEN
751 ind_data.field = -1 * ind_data.field;
752 INSERT INTO metabib.facet_entry (field, source, value)
753 VALUES (ind_data.field, ind_data.source, ind_data.value);
756 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
758 quote_literal(ind_data.field) || $$, $$ ||
759 quote_literal(ind_data.source) || $$, $$ ||
760 quote_literal(ind_data.value) ||
768 $func$ LANGUAGE PLPGSQL;
770 CREATE OR REPLACE FUNCTION biblio.extract_located_uris( bib_id BIGINT, marcxml TEXT, editor_id INT ) RETURNS VOID AS $func$
777 uri_owner_list TEXT[];
785 -- Clear any URI mappings and call numbers for this bib.
786 -- This leads to acn / auricnm inflation, but also enables
787 -- old acn/auricnm's to go away and for bibs to be deleted.
788 FOR uri_cn_id IN SELECT id FROM asset.call_number WHERE record = bib_id AND label = '##URI##' AND NOT deleted LOOP
789 DELETE FROM asset.uri_call_number_map WHERE call_number = uri_cn_id;
790 DELETE FROM asset.call_number WHERE id = uri_cn_id;
793 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',marcxml);
794 IF ARRAY_UPPER(uris,1) > 0 THEN
795 FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
796 -- First we pull info out of the 856
799 uri_href := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
800 uri_label := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()',uri_xml))[1];
801 uri_use := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()',uri_xml))[1];
803 IF uri_label IS NULL THEN
804 uri_label := uri_href;
806 CONTINUE WHEN uri_href IS NULL;
808 -- Get the distinct list of libraries wanting to use
810 DISTINCT REGEXP_REPLACE(
812 $re$^.*?\((\w+)\).*$$re$,
815 ) INTO uri_owner_list
818 '//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',
823 IF ARRAY_UPPER(uri_owner_list,1) > 0 THEN
825 -- look for a matching uri
826 IF uri_use IS NULL THEN
827 SELECT id INTO uri_id
829 WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active
831 IF NOT FOUND THEN -- create one
832 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
833 SELECT id INTO uri_id
835 WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active;
838 SELECT id INTO uri_id
840 WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active
842 IF NOT FOUND THEN -- create one
843 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
844 SELECT id INTO uri_id
846 WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
850 FOR j IN 1 .. ARRAY_UPPER(uri_owner_list, 1) LOOP
851 uri_owner := uri_owner_list[j];
853 SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = uri_owner;
854 CONTINUE WHEN NOT FOUND;
856 -- we need a call number to link through
857 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;
859 INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
860 VALUES (uri_owner_id, bib_id, 'now', 'now', editor_id, editor_id, '##URI##');
861 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;
864 -- now, link them if they're not already
865 SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
867 INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
879 $func$ LANGUAGE PLPGSQL;
881 CREATE OR REPLACE FUNCTION metabib.remap_metarecord_for_bib( bib_id BIGINT, fp TEXT ) RETURNS BIGINT AS $func$
885 tmp_mr metabib.metarecord%ROWTYPE;
886 deleted_mrs BIGINT[];
889 DELETE FROM metabib.metarecord_source_map WHERE source = bib_id; -- Rid ourselves of the search-estimate-killing linkage
891 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
893 IF old_mr IS NULL AND fp = tmp_mr.fingerprint THEN -- Find the first fingerprint-matching
896 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
897 IF source_count = 0 THEN -- No other records
898 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
899 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
905 IF old_mr IS NULL THEN -- we found no suitable, preexisting MR based on old source maps
906 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp; -- is there one for our current fingerprint?
907 IF old_mr IS NULL THEN -- nope, create one and grab its id
908 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( fp, bib_id );
909 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp;
910 ELSE -- indeed there is. update it with a null cache and recalcualated master record
911 UPDATE metabib.metarecord
913 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
916 ELSE -- there was one we already attached to, update its mods cache and master_record
917 UPDATE metabib.metarecord
919 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
923 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, bib_id); -- new source mapping
925 IF ARRAY_UPPER(deleted_mrs,1) > 0 THEN
926 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
932 $func$ LANGUAGE PLPGSQL;
934 CREATE OR REPLACE FUNCTION biblio.map_authority_linking (bibid BIGINT, marc TEXT) RETURNS BIGINT AS $func$
935 DELETE FROM authority.bib_linking WHERE bib = $1;
936 INSERT INTO authority.bib_linking (bib, authority)
939 FROM ( SELECT DISTINCT $1 AS bib,
940 BTRIM(remove_paren_substring(txt))::BIGINT AS authority
941 FROM unnest(oils_xpath('//*[@code="0"]/text()',$2)) x(txt)
942 WHERE BTRIM(remove_paren_substring(txt)) ~ $re$^\d+$$re$
943 ) y JOIN authority.record_entry r ON r.id = y.authority;
947 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
948 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
950 transformed_xml TEXT;
953 xfrm config.xml_transform%ROWTYPE;
955 new_attrs HSTORE := ''::HSTORE;
956 attr_def config.record_attr_definition%ROWTYPE;
959 IF NEW.deleted IS TRUE THEN -- If this bib is deleted
960 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage
961 DELETE FROM metabib.record_attr WHERE id = NEW.id; -- Kill the attrs hash, useless on deleted records
962 DELETE FROM authority.bib_linking WHERE bib = NEW.id; -- Avoid updating fields in bibs that are no longer visible
963 DELETE FROM biblio.peer_bib_copy_map WHERE peer_record = NEW.id; -- Separate any multi-homed items
964 RETURN NEW; -- and we're done
967 IF TG_OP = 'UPDATE' THEN -- re-ingest?
968 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
970 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
975 -- Record authority linking
976 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_linking' AND enabled;
978 PERFORM biblio.map_authority_linking( NEW.id, NEW.marc );
981 -- Flatten and insert the mfr data
982 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_full_rec' AND enabled;
984 PERFORM metabib.reingest_metabib_full_rec(NEW.id);
986 -- Now we pull out attribute data, which is dependent on the mfr for all but XPath-based fields
987 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_rec_descriptor' AND enabled;
989 FOR attr_def IN SELECT * FROM config.record_attr_definition ORDER BY format LOOP
991 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
992 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(value), COALESCE(attr_def.joiner,' ')) INTO attr_value
993 FROM (SELECT * FROM metabib.full_rec ORDER BY tag, subfield) AS x
994 WHERE record = NEW.id
995 AND tag LIKE attr_def.tag
997 WHEN attr_def.sf_list IS NOT NULL
998 THEN POSITION(subfield IN attr_def.sf_list) > 0
1005 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
1006 attr_value := biblio.marc21_extract_fixed_field(NEW.id, attr_def.fixed_field);
1008 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
1010 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
1012 -- See if we can skip the XSLT ... it's expensive
1013 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
1014 -- Can't skip the transform
1015 IF xfrm.xslt <> '---' THEN
1016 transformed_xml := oils_xslt_process(NEW.marc,xfrm.xslt);
1018 transformed_xml := NEW.marc;
1021 prev_xfrm := xfrm.name;
1024 IF xfrm.name IS NULL THEN
1025 -- just grab the marcxml (empty) transform
1026 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
1027 prev_xfrm := xfrm.name;
1030 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
1032 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
1033 SELECT m.value INTO attr_value
1034 FROM biblio.marc21_physical_characteristics(NEW.id) v
1035 JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
1036 WHERE v.subfield = attr_def.phys_char_sf
1037 LIMIT 1; -- Just in case ...
1041 -- apply index normalizers to attr_value
1043 SELECT n.func AS func,
1044 n.param_count AS param_count,
1046 FROM config.index_normalizer n
1047 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
1048 WHERE attr = attr_def.name
1050 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1051 quote_literal( attr_value ) ||
1053 WHEN normalizer.param_count > 0
1054 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1057 ')' INTO attr_value;
1061 -- Add the new value to the hstore
1062 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
1066 IF TG_OP = 'INSERT' OR OLD.deleted THEN -- initial insert OR revivication
1067 INSERT INTO metabib.record_attr (id, attrs) VALUES (NEW.id, new_attrs);
1069 UPDATE metabib.record_attr SET attrs = new_attrs WHERE id = NEW.id;
1075 -- Gather and insert the field entry data
1076 PERFORM metabib.reingest_metabib_field_entries(NEW.id);
1078 -- Located URI magic
1079 IF TG_OP = 'INSERT' THEN
1080 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1082 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1085 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1087 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1091 -- (re)map metarecord-bib linking
1092 IF TG_OP = 'INSERT' THEN -- if not deleted and performing an insert, check for the flag
1093 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
1095 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1097 ELSE -- we're doing an update, and we're not deleted, remap
1098 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_update' AND enabled;
1100 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1106 $func$ LANGUAGE PLPGSQL;