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 evergreen.facet_force_nfc() RETURNS TRIGGER AS $$
140 NEW.value := force_unicode_normal_form(NEW.value,'NFC');
145 CREATE TRIGGER facet_force_nfc_tgr
146 BEFORE UPDATE OR INSERT ON metabib.facet_entry
147 FOR EACH ROW EXECUTE PROCEDURE evergreen.facet_force_nfc();
149 CREATE TABLE metabib.record_attr (
150 id BIGINT PRIMARY KEY REFERENCES biblio.record_entry (id) ON DELETE CASCADE,
151 attrs HSTORE NOT NULL DEFAULT ''::HSTORE
153 CREATE INDEX metabib_svf_attrs_idx ON metabib.record_attr USING GIST (attrs);
154 CREATE INDEX metabib_svf_date1_idx ON metabib.record_attr ((attrs->'date1'));
155 CREATE INDEX metabib_svf_dates_idx ON metabib.record_attr ((attrs->'date1'),(attrs->'date2'));
157 -- Back-compat view ... we're moving to an HSTORE world
158 CREATE TYPE metabib.rec_desc_type AS (
176 CREATE VIEW metabib.rec_descriptor AS
179 (populate_record(NULL::metabib.rec_desc_type, attrs)).*
180 FROM metabib.record_attr;
182 -- Use a sequence that matches previous version, for easier upgrading.
183 CREATE SEQUENCE metabib.full_rec_id_seq;
185 CREATE TABLE metabib.real_full_rec (
186 id BIGINT NOT NULL DEFAULT NEXTVAL('metabib.full_rec_id_seq'::REGCLASS),
187 record BIGINT NOT NULL,
188 tag CHAR(3) NOT NULL,
193 index_vector tsvector NOT NULL
195 ALTER TABLE metabib.real_full_rec ADD PRIMARY KEY (id);
197 CREATE INDEX metabib_full_rec_tag_subfield_idx ON metabib.real_full_rec (tag,subfield);
198 CREATE INDEX metabib_full_rec_value_idx ON metabib.real_full_rec (substring(value,1,1024));
199 /* Enable LIKE to use an index for database clusters with locales other than C or POSIX */
200 CREATE INDEX metabib_full_rec_value_tpo_index ON metabib.real_full_rec (substring(value,1,1024) text_pattern_ops);
201 CREATE INDEX metabib_full_rec_record_idx ON metabib.real_full_rec (record);
202 CREATE INDEX metabib_full_rec_index_vector_idx ON metabib.real_full_rec USING GIST (index_vector);
204 CREATE TRIGGER metabib_full_rec_fti_trigger
205 BEFORE UPDATE OR INSERT ON metabib.real_full_rec
206 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('default');
208 CREATE OR REPLACE VIEW metabib.full_rec AS
215 SUBSTRING(value,1,1024) AS value,
217 FROM metabib.real_full_rec;
219 CREATE OR REPLACE RULE metabib_full_rec_insert_rule
220 AS ON INSERT TO metabib.full_rec
222 INSERT INTO metabib.real_full_rec VALUES (
223 COALESCE(NEW.id, NEXTVAL('metabib.full_rec_id_seq'::REGCLASS)),
233 CREATE OR REPLACE RULE metabib_full_rec_update_rule
234 AS ON UPDATE TO metabib.full_rec
236 UPDATE metabib.real_full_rec SET
242 subfield = NEW.subfield,
244 index_vector = NEW.index_vector
247 CREATE OR REPLACE RULE metabib_full_rec_delete_rule
248 AS ON DELETE TO metabib.full_rec
250 DELETE FROM metabib.real_full_rec WHERE id = OLD.id;
252 CREATE TABLE metabib.metarecord_source_map (
253 id BIGSERIAL PRIMARY KEY,
254 metarecord BIGINT NOT NULL,
255 source BIGINT NOT NULL
257 CREATE INDEX metabib_metarecord_source_map_metarecord_idx ON metabib.metarecord_source_map (metarecord);
258 CREATE INDEX metabib_metarecord_source_map_source_record_idx ON metabib.metarecord_source_map (source);
260 CREATE TYPE metabib.field_entry_template AS (
267 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
269 bib biblio.record_entry%ROWTYPE;
270 idx config.metabib_field%ROWTYPE;
271 xfrm config.xml_transform%ROWTYPE;
273 transformed_xml TEXT;
275 xml_node_list TEXT[];
279 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
280 output_row metabib.field_entry_template%ROWTYPE;
284 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
286 -- Loop over the indexing entries
287 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
289 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
291 -- See if we can skip the XSLT ... it's expensive
292 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
293 -- Can't skip the transform
294 IF xfrm.xslt <> '---' THEN
295 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
297 transformed_xml := bib.marc;
300 prev_xfrm := xfrm.name;
303 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
306 FOR xml_node IN SELECT x FROM explode_array(xml_node_list) AS x LOOP
307 CONTINUE WHEN xml_node !~ E'^\\s*<';
309 curr_text := ARRAY_TO_STRING(
310 oils_xpath( '//text()',
311 REGEXP_REPLACE( -- This escapes all &s not followed by "amp;". Data ise returned from oils_xpath (above) in UTF-8, not entity encoded
312 REGEXP_REPLACE( -- This escapes embeded <s
314 $re$(>[^<]+)(<)([^>]+<)$re$,
326 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
328 IF raw_text IS NOT NULL THEN
329 raw_text := raw_text || joiner;
332 raw_text := COALESCE(raw_text,'') || curr_text;
334 -- insert raw node text for faceting
335 IF idx.facet_field THEN
337 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
338 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
340 facet_text := curr_text;
343 output_row.field_class = idx.field_class;
344 output_row.field = -1 * idx.id;
345 output_row.source = rid;
346 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
348 RETURN NEXT output_row;
353 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
355 -- insert combined node text for searching
356 IF idx.search_field THEN
357 output_row.field_class = idx.field_class;
358 output_row.field = idx.id;
359 output_row.source = rid;
360 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
362 RETURN NEXT output_row;
368 $func$ LANGUAGE PLPGSQL;
370 -- default to a space joiner
371 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
372 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
375 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
377 bib biblio.record_entry%ROWTYPE;
378 output metabib.full_rec%ROWTYPE;
381 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
383 FOR field IN SELECT * FROM biblio.flatten_marc( bib.marc ) LOOP
384 output.record := rid;
385 output.ind1 := field.ind1;
386 output.ind2 := field.ind2;
387 output.tag := field.tag;
388 output.subfield := field.subfield;
389 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
390 output.value := naco_normalize(field.value, field.subfield);
392 output.value := field.value;
395 CONTINUE WHEN output.value IS NULL;
400 $func$ LANGUAGE PLPGSQL;
402 /* Old form of biblio.flatten_marc() relied on contrib/xml2 functions that got all crashy in PostgreSQL 8.4 */
403 -- CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT, BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
404 -- SELECT NULL::bigint AS id, NULL::bigint, 'LDR'::char(3), NULL::TEXT, NULL::TEXT, NULL::TEXT, oils_xpath_string( '//*[local-name()="leader"]', $1 ), NULL::tsvector AS index_vector
406 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), NULL::TEXT, NULL::TEXT, NULL::TEXT, x.value, NULL::tsvector AS index_vector
407 -- FROM oils_xpath_table(
410 -- 'biblio.record_entry',
411 -- '//*[local-name()="controlfield"]/@tag|//*[local-name()="controlfield"]',
413 -- )x(record int, tag text, value text)
415 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), x.ind1, x.ind2, x.subfield, x.value, NULL::tsvector AS index_vector
416 -- FROM oils_xpath_table(
419 -- 'biblio.record_entry',
420 -- '//*[local-name()="datafield"]/@tag|' ||
421 -- '//*[local-name()="datafield"]/@ind1|' ||
422 -- '//*[local-name()="datafield"]/@ind2|' ||
423 -- '//*[local-name()="datafield"]/*/@code|' ||
424 -- '//*[local-name()="datafield"]/*[@code]',
426 -- )x(record int, tag text, ind1 text, ind2 text, subfield text, value text);
427 -- $func$ LANGUAGE SQL;
429 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT ) RETURNS SETOF metabib.full_rec AS $func$
432 use MARC::File::XML (BinaryEncoding => 'UTF-8');
435 my $r = MARC::Record->new_from_xml( $xml );
437 return_next( { tag => 'LDR', value => $r->leader } );
439 for my $f ( $r->fields ) {
440 if ($f->is_control_field) {
441 return_next({ tag => $f->tag, value => $f->data });
443 for my $s ($f->subfields) {
446 ind1 => $f->indicator(1),
447 ind2 => $f->indicator(2),
452 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
453 my $trim = $f->indicator(2) || 0;
456 ind1 => $f->indicator(1),
457 ind2 => $f->indicator(2),
459 value => substr( $s->[1], $trim )
468 $func$ LANGUAGE PLPERLU;
470 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
477 retval config.marc21_rec_type_map%ROWTYPE;
479 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
481 IF ldr IS NULL OR ldr = '' THEN
482 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
486 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
487 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
490 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
491 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
493 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
495 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
498 IF retval.code IS NULL THEN
499 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
504 $func$ LANGUAGE PLPGSQL;
506 CREATE OR REPLACE FUNCTION biblio.marc21_record_type( rid BIGINT ) RETURNS config.marc21_rec_type_map AS $func$
507 SELECT * FROM vandelay.marc21_record_type( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
510 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
517 rtype := (vandelay.marc21_record_type( marc )).code;
518 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
519 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
520 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
523 val := REPEAT( ff_pos.default_val, ff_pos.length );
529 $func$ LANGUAGE PLPGSQL;
531 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
532 SELECT * FROM vandelay.marc21_extract_fixed_field( (SELECT marc FROM biblio.record_entry WHERE id = $1), $2 );
535 CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
536 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
541 output biblio.record_ff_map%ROWTYPE;
543 rtype := (vandelay.marc21_record_type( marc )).code;
545 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
546 output.ff_name := ff_pos.fixed_field;
547 output.ff_value := NULL;
549 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(tag) || '"]/text()', marc ) ) x(value) LOOP
550 output.ff_value := SUBSTRING( tag_data.value, 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;
560 $func$ LANGUAGE PLPGSQL;
562 CREATE OR REPLACE FUNCTION biblio.marc21_extract_all_fixed_fields( rid BIGINT ) RETURNS SETOF biblio.record_ff_map AS $func$
563 SELECT $1 AS record, ff_name, ff_value FROM vandelay.marc21_extract_all_fixed_fields( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
566 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
567 CREATE OR REPLACE FUNCTION vandelay.marc21_physical_characteristics( marc TEXT) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
571 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
572 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
573 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
574 retval biblio.marc21_physical_characteristics%ROWTYPE;
577 _007 := oils_xpath_string( '//*[@tag="007"]', marc );
579 IF _007 IS NOT NULL AND _007 <> '' THEN
580 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007, 1, 1 );
582 IF ptype.ptype_key IS NOT NULL THEN
583 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
584 SELECT * INTO pval FROM config.marc21_physical_characteristic_value_map WHERE ptype_subfield = psf.id AND value = SUBSTRING( _007, psf.start_pos + 1, psf.length );
586 IF pval.id IS NOT NULL THEN
589 retval.ptype := ptype.ptype_key;
590 retval.subfield := psf.id;
591 retval.value := pval.id;
601 $func$ LANGUAGE PLPGSQL;
603 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
604 SELECT id, $1 AS record, ptype, subfield, value FROM vandelay.marc21_physical_characteristics( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
607 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
620 IF marc IS NULL OR marc = '' THEN
624 -- First, the count of tags
625 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
627 -- now go through a bunch of pain to get the record type
628 IF best_type IS NOT NULL THEN
629 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
631 IF ldr IS NOT NULL THEN
632 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
633 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
636 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
637 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
639 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
641 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
643 IF type_map.code IS NOT NULL THEN
644 IF best_type = type_map.code THEN
645 qual := qual + qual / 2;
648 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
649 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
650 IF ff_tag_data = best_lang THEN
658 -- Now look for some quality metrics
660 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
665 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
672 $func$ LANGUAGE PLPGSQL;
674 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
676 idx config.biblio_fingerprint%ROWTYPE;
677 xfrm config.xml_transform%ROWTYPE;
679 transformed_xml TEXT;
681 xml_node_list TEXT[];
683 output_text TEXT := '';
686 IF marc IS NULL OR marc = '' THEN
690 -- Loop over the indexing entries
691 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
693 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
695 -- See if we can skip the XSLT ... it's expensive
696 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
697 -- Can't skip the transform
698 IF xfrm.xslt <> '---' THEN
699 transformed_xml := oils_xslt_process(marc,xfrm.xslt);
701 transformed_xml := marc;
704 prev_xfrm := xfrm.name;
707 raw_text := COALESCE(
715 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
724 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
725 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
727 IF idx.first_word IS TRUE THEN
728 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
731 output_text := output_text || REGEXP_REPLACE(raw_text, E'\\s+', '', 'g');
738 $func$ LANGUAGE PLPGSQL;
740 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
741 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
744 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
746 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
750 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
751 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
756 $func$ LANGUAGE PLPGSQL;
758 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_full_rec( bib_id BIGINT ) RETURNS VOID AS $func$
760 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
762 DELETE FROM metabib.real_full_rec WHERE record = bib_id;
764 INSERT INTO metabib.real_full_rec (record, tag, ind1, ind2, subfield, value)
765 SELECT record, tag, ind1, ind2, subfield, value FROM biblio.flatten_marc( bib_id );
769 $func$ LANGUAGE PLPGSQL;
771 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries( bib_id BIGINT ) RETURNS VOID AS $func$
774 ind_data metabib.field_entry_template%ROWTYPE;
776 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
778 FOR fclass IN SELECT * FROM config.metabib_class LOOP
779 -- RAISE NOTICE 'Emptying out %', fclass.name;
780 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
782 DELETE FROM metabib.facet_entry WHERE source = bib_id;
785 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
786 IF ind_data.field < 0 THEN
787 ind_data.field = -1 * ind_data.field;
788 INSERT INTO metabib.facet_entry (field, source, value)
789 VALUES (ind_data.field, ind_data.source, ind_data.value);
792 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
794 quote_literal(ind_data.field) || $$, $$ ||
795 quote_literal(ind_data.source) || $$, $$ ||
796 quote_literal(ind_data.value) ||
804 $func$ LANGUAGE PLPGSQL;
806 CREATE OR REPLACE FUNCTION biblio.extract_located_uris( bib_id BIGINT, marcxml TEXT, editor_id INT ) RETURNS VOID AS $func$
820 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',marcxml);
821 IF ARRAY_UPPER(uris,1) > 0 THEN
822 FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
823 -- First we pull info out of the 856
826 uri_href := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
827 CONTINUE WHEN uri_href IS NULL;
829 uri_label := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()|//*[@code="u"]/text()',uri_xml))[1];
830 CONTINUE WHEN uri_label IS NULL;
832 uri_owner := (oils_xpath('//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',uri_xml))[1];
833 CONTINUE WHEN uri_owner IS NULL;
835 uri_use := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()|//*[@code="u"]/text()',uri_xml))[1];
837 uri_owner := REGEXP_REPLACE(uri_owner, $re$^.*?\((\w+)\).*$$re$, E'\\1');
839 SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = uri_owner;
840 CONTINUE WHEN NOT FOUND;
842 -- now we look for a matching uri
843 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
844 IF NOT FOUND THEN -- create one
845 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
846 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
849 -- we need a call number to link through
850 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;
852 INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
853 VALUES (uri_owner_id, bib_id, 'now', 'now', editor_id, editor_id, '##URI##');
854 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;
857 -- now, link them if they're not already
858 SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
860 INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
868 $func$ LANGUAGE PLPGSQL;
871 CREATE OR REPLACE FUNCTION metabib.remap_metarecord_for_bib( bib_id BIGINT, fp TEXT ) RETURNS BIGINT AS $func$
875 tmp_mr metabib.metarecord%ROWTYPE;
876 deleted_mrs BIGINT[];
879 DELETE FROM metabib.metarecord_source_map WHERE source = bib_id; -- Rid ourselves of the search-estimate-killing linkage
881 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
883 IF old_mr IS NULL AND fp = tmp_mr.fingerprint THEN -- Find the first fingerprint-matching
886 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
887 IF source_count = 0 THEN -- No other records
888 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
889 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
895 IF old_mr IS NULL THEN -- we found no suitable, preexisting MR based on old source maps
896 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp; -- is there one for our current fingerprint?
897 IF old_mr IS NULL THEN -- nope, create one and grab its id
898 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( fp, bib_id );
899 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp;
900 ELSE -- indeed there is. update it with a null cache and recalcualated master record
901 UPDATE metabib.metarecord
903 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
906 ELSE -- there was one we already attached to, update its mods cache and master_record
907 UPDATE metabib.metarecord
909 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
913 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, bib_id); -- new source mapping
915 IF ARRAY_UPPER(deleted_mrs,1) > 0 THEN
916 UPDATE action.hold_request SET target = old_mr WHERE target IN ( SELECT explode_array(deleted_mrs) ) AND hold_type = 'M'; -- if we had to delete any MRs above, make sure their holds are moved
922 $func$ LANGUAGE PLPGSQL;
924 CREATE OR REPLACE FUNCTION biblio.map_authority_linking (bibid BIGINT, marc TEXT) RETURNS BIGINT AS $func$
925 DELETE FROM authority.bib_linking WHERE bib = $1;
926 INSERT INTO authority.bib_linking (bib, authority)
929 FROM ( SELECT DISTINCT $1 AS bib,
930 BTRIM(remove_paren_substring(txt))::BIGINT AS authority
931 FROM explode_array(oils_xpath('//*[@code="0"]/text()',$2)) x(txt)
932 WHERE BTRIM(remove_paren_substring(txt)) ~ $re$^\d+$$re$
933 ) y JOIN authority.record_entry r ON r.id = y.authority;
937 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
938 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
940 transformed_xml TEXT;
943 xfrm config.xml_transform%ROWTYPE;
945 new_attrs HSTORE := ''::HSTORE;
946 attr_def config.record_attr_definition%ROWTYPE;
949 IF NEW.deleted IS TRUE THEN -- If this bib is deleted
950 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage
951 DELETE FROM metabib.record_attr WHERE id = NEW.id; -- Kill the attrs hash, useless on deleted records
952 DELETE FROM authority.bib_linking WHERE bib = NEW.id; -- Avoid updating fields in bibs that are no longer visible
953 DELETE FROM biblio.peer_bib_copy_map WHERE peer_record = NEW.id; -- Separate any multi-homed items
954 RETURN NEW; -- and we're done
957 IF TG_OP = 'UPDATE' THEN -- re-ingest?
958 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
960 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
965 -- Record authority linking
966 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_linking' AND enabled;
968 PERFORM biblio.map_authority_linking( NEW.id, NEW.marc );
971 -- Flatten and insert the mfr data
972 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_full_rec' AND enabled;
974 PERFORM metabib.reingest_metabib_full_rec(NEW.id);
976 -- Now we pull out attribute data, which is dependent on the mfr for all but XPath-based fields
977 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_rec_descriptor' AND enabled;
979 FOR attr_def IN SELECT * FROM config.record_attr_definition ORDER BY format LOOP
981 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
982 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(value), COALESCE(attr_def.joiner,' ')) INTO attr_value
983 FROM (SELECT * FROM metabib.full_rec ORDER BY tag, subfield) AS x
984 WHERE record = NEW.id
985 AND tag LIKE attr_def.tag
987 WHEN attr_def.sf_list IS NOT NULL
988 THEN POSITION(subfield IN attr_def.sf_list) > 0
995 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
996 attr_value := biblio.marc21_extract_fixed_field(NEW.id, attr_def.fixed_field);
998 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
1000 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
1002 -- See if we can skip the XSLT ... it's expensive
1003 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
1004 -- Can't skip the transform
1005 IF xfrm.xslt <> '---' THEN
1006 transformed_xml := oils_xslt_process(NEW.marc,xfrm.xslt);
1008 transformed_xml := NEW.marc;
1011 prev_xfrm := xfrm.name;
1014 IF xfrm.name IS NULL THEN
1015 -- just grab the marcxml (empty) transform
1016 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
1017 prev_xfrm := xfrm.name;
1020 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
1022 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
1023 SELECT value::TEXT INTO attr_value
1024 FROM biblio.marc21_physical_characteristics(NEW.id)
1025 WHERE subfield = attr_def.phys_char_sf
1026 LIMIT 1; -- Just in case ...
1030 -- apply index normalizers to attr_value
1032 SELECT n.func AS func,
1033 n.param_count AS param_count,
1035 FROM config.index_normalizer n
1036 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
1037 WHERE attr = attr_def.name
1039 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1040 quote_literal( attr_value ) ||
1042 WHEN normalizer.param_count > 0
1043 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1046 ')' INTO attr_value;
1050 -- Add the new value to the hstore
1051 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
1055 IF TG_OP = 'INSERT' OR OLD.deleted THEN -- initial insert OR revivication
1056 INSERT INTO metabib.record_attr (id, attrs) VALUES (NEW.id, new_attrs);
1058 UPDATE metabib.record_attr SET attrs = attrs || new_attrs WHERE id = NEW.id;
1064 -- Gather and insert the field entry data
1065 PERFORM metabib.reingest_metabib_field_entries(NEW.id);
1067 -- Located URI magic
1068 IF TG_OP = 'INSERT' THEN
1069 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1071 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1074 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1076 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1080 -- (re)map metarecord-bib linking
1081 IF TG_OP = 'INSERT' THEN -- if not deleted and performing an insert, check for the flag
1082 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
1084 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1086 ELSE -- we're doing an update, and we're not deleted, remap
1087 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_update' AND enabled;
1089 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1095 $func$ LANGUAGE PLPGSQL;