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);
139 CREATE TABLE metabib.rec_descriptor (
140 id BIGSERIAL PRIMARY KEY,
158 CREATE INDEX metabib_rec_descriptor_record_idx ON metabib.rec_descriptor (record);
159 CREATE INDEX metabib_rec_descriptor_item_type_idx ON metabib.rec_descriptor (item_type);
160 CREATE INDEX metabib_rec_descriptor_item_form_idx ON metabib.rec_descriptor (item_form);
161 CREATE INDEX metabib_rec_descriptor_bib_level_idx ON metabib.rec_descriptor (bib_level);
162 CREATE INDEX metabib_rec_descriptor_control_type_idx ON metabib.rec_descriptor (control_type);
163 CREATE INDEX metabib_rec_descriptor_char_encoding_idx ON metabib.rec_descriptor (char_encoding);
164 CREATE INDEX metabib_rec_descriptor_enc_level_idx ON metabib.rec_descriptor (enc_level);
165 CREATE INDEX metabib_rec_descriptor_audience_idx ON metabib.rec_descriptor (audience);
166 CREATE INDEX metabib_rec_descriptor_lit_form_idx ON metabib.rec_descriptor (lit_form);
167 CREATE INDEX metabib_rec_descriptor_cat_form_idx ON metabib.rec_descriptor (cat_form);
168 CREATE INDEX metabib_rec_descriptor_pub_status_idx ON metabib.rec_descriptor (pub_status);
169 CREATE INDEX metabib_rec_descriptor_item_lang_idx ON metabib.rec_descriptor (item_lang);
170 CREATE INDEX metabib_rec_descriptor_vr_format_idx ON metabib.rec_descriptor (vr_format);
171 CREATE INDEX metabib_rec_descriptor_date1_idx ON metabib.rec_descriptor (date1);
172 CREATE INDEX metabib_rec_descriptor_dates_idx ON metabib.rec_descriptor (date1,date2);
174 -- Use a sequence that matches previous version, for easier upgrading.
175 CREATE SEQUENCE metabib.full_rec_id_seq;
177 CREATE TABLE metabib.real_full_rec (
178 id BIGINT NOT NULL DEFAULT NEXTVAL('metabib.full_rec_id_seq'::REGCLASS),
179 record BIGINT NOT NULL,
180 tag CHAR(3) NOT NULL,
185 index_vector tsvector NOT NULL
187 ALTER TABLE metabib.real_full_rec ADD PRIMARY KEY (id);
189 CREATE INDEX metabib_full_rec_tag_subfield_idx ON metabib.real_full_rec (tag,subfield);
190 CREATE INDEX metabib_full_rec_value_idx ON metabib.real_full_rec (substring(value,1,1024));
191 /* Enable LIKE to use an index for database clusters with locales other than C or POSIX */
192 CREATE INDEX metabib_full_rec_value_tpo_index ON metabib.real_full_rec (substring(value,1,1024) text_pattern_ops);
193 CREATE INDEX metabib_full_rec_record_idx ON metabib.real_full_rec (record);
194 CREATE INDEX metabib_full_rec_index_vector_idx ON metabib.real_full_rec USING GIST (index_vector);
196 CREATE TRIGGER metabib_full_rec_fti_trigger
197 BEFORE UPDATE OR INSERT ON metabib.real_full_rec
198 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('default');
200 CREATE OR REPLACE VIEW metabib.full_rec AS
207 SUBSTRING(value,1,1024) AS value,
209 FROM metabib.real_full_rec;
211 CREATE OR REPLACE RULE metabib_full_rec_insert_rule
212 AS ON INSERT TO metabib.full_rec
214 INSERT INTO metabib.real_full_rec VALUES (
215 COALESCE(NEW.id, NEXTVAL('metabib.full_rec_id_seq'::REGCLASS)),
225 CREATE OR REPLACE RULE metabib_full_rec_update_rule
226 AS ON UPDATE TO metabib.full_rec
228 UPDATE metabib.real_full_rec SET
234 subfield = NEW.subfield,
236 index_vector = NEW.index_vector
239 CREATE OR REPLACE RULE metabib_full_rec_delete_rule
240 AS ON DELETE TO metabib.full_rec
242 DELETE FROM metabib.real_full_rec WHERE id = OLD.id;
244 CREATE TABLE metabib.metarecord_source_map (
245 id BIGSERIAL PRIMARY KEY,
246 metarecord BIGINT NOT NULL,
247 source BIGINT NOT NULL
249 CREATE INDEX metabib_metarecord_source_map_metarecord_idx ON metabib.metarecord_source_map (metarecord);
250 CREATE INDEX metabib_metarecord_source_map_source_record_idx ON metabib.metarecord_source_map (source);
252 CREATE TYPE metabib.field_entry_template AS (
259 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
261 bib biblio.record_entry%ROWTYPE;
262 idx config.metabib_field%ROWTYPE;
263 xfrm config.xml_transform%ROWTYPE;
265 transformed_xml TEXT;
267 xml_node_list TEXT[];
271 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
272 output_row metabib.field_entry_template%ROWTYPE;
276 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
278 -- Loop over the indexing entries
279 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
281 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
283 -- See if we can skip the XSLT ... it's expensive
284 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
285 -- Can't skip the transform
286 IF xfrm.xslt <> '---' THEN
287 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
289 transformed_xml := bib.marc;
292 prev_xfrm := xfrm.name;
295 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
298 FOR xml_node IN SELECT x FROM explode_array(xml_node_list) AS x LOOP
299 CONTINUE WHEN xml_node !~ E'^\\s*<';
301 curr_text := ARRAY_TO_STRING(
302 oils_xpath( '//text()',
303 REGEXP_REPLACE( -- This escapes all &s not followed by "amp;". Data ise returned from oils_xpath (above) in UTF-8, not entity encoded
304 REGEXP_REPLACE( -- This escapes embeded <s
306 $re$(>[^<]+)(<)([^>]+<)$re$,
318 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
320 IF raw_text IS NOT NULL THEN
321 raw_text := raw_text || joiner;
324 raw_text := COALESCE(raw_text,'') || curr_text;
326 -- insert raw node text for faceting
327 IF idx.facet_field THEN
329 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
330 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
332 facet_text := curr_text;
335 output_row.field_class = idx.field_class;
336 output_row.field = -1 * idx.id;
337 output_row.source = rid;
338 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
340 RETURN NEXT output_row;
345 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
347 -- insert combined node text for searching
348 IF idx.search_field THEN
349 output_row.field_class = idx.field_class;
350 output_row.field = idx.id;
351 output_row.source = rid;
352 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
354 RETURN NEXT output_row;
360 $func$ LANGUAGE PLPGSQL;
362 -- default to a space joiner
363 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
364 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
367 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
369 bib biblio.record_entry%ROWTYPE;
370 output metabib.full_rec%ROWTYPE;
373 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
375 FOR field IN SELECT * FROM biblio.flatten_marc( bib.marc ) LOOP
376 output.record := rid;
377 output.ind1 := field.ind1;
378 output.ind2 := field.ind2;
379 output.tag := field.tag;
380 output.subfield := field.subfield;
381 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
382 output.value := naco_normalize(field.value, field.subfield);
384 output.value := field.value;
387 CONTINUE WHEN output.value IS NULL;
392 $func$ LANGUAGE PLPGSQL;
394 /* Old form of biblio.flatten_marc() relied on contrib/xml2 functions that got all crashy in PostgreSQL 8.4 */
395 -- CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT, BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
396 -- 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
398 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), NULL::TEXT, NULL::TEXT, NULL::TEXT, x.value, NULL::tsvector AS index_vector
399 -- FROM oils_xpath_table(
402 -- 'biblio.record_entry',
403 -- '//*[local-name()="controlfield"]/@tag|//*[local-name()="controlfield"]',
405 -- )x(record int, tag text, value text)
407 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), x.ind1, x.ind2, x.subfield, x.value, NULL::tsvector AS index_vector
408 -- FROM oils_xpath_table(
411 -- 'biblio.record_entry',
412 -- '//*[local-name()="datafield"]/@tag|' ||
413 -- '//*[local-name()="datafield"]/@ind1|' ||
414 -- '//*[local-name()="datafield"]/@ind2|' ||
415 -- '//*[local-name()="datafield"]/*/@code|' ||
416 -- '//*[local-name()="datafield"]/*[@code]',
418 -- )x(record int, tag text, ind1 text, ind2 text, subfield text, value text);
419 -- $func$ LANGUAGE SQL;
421 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT ) RETURNS SETOF metabib.full_rec AS $func$
424 use MARC::File::XML (BinaryEncoding => 'UTF-8');
427 my $r = MARC::Record->new_from_xml( $xml );
429 return_next( { tag => 'LDR', value => $r->leader } );
431 for my $f ( $r->fields ) {
432 if ($f->is_control_field) {
433 return_next({ tag => $f->tag, value => $f->data });
435 for my $s ($f->subfields) {
438 ind1 => $f->indicator(1),
439 ind2 => $f->indicator(2),
444 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
445 my $trim = $f->indicator(2) || 0;
448 ind1 => $f->indicator(1),
449 ind2 => $f->indicator(2),
451 value => substr( $s->[1], $trim )
460 $func$ LANGUAGE PLPERLU;
462 CREATE OR REPLACE FUNCTION biblio.marc21_record_type( rid BIGINT ) RETURNS config.marc21_rec_type_map AS $func$
469 retval config.marc21_rec_type_map%ROWTYPE;
471 SELECT * INTO ldr FROM metabib.full_rec WHERE record = rid AND tag = 'LDR' LIMIT 1;
473 IF ldr.id IS NULL THEN
474 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
478 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
479 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
482 tval := SUBSTRING( ldr.value, tval_rec.start_pos + 1, tval_rec.length );
483 bval := SUBSTRING( ldr.value, bval_rec.start_pos + 1, bval_rec.length );
485 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr.value;
487 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
490 IF retval.code IS NULL THEN
491 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
496 $func$ LANGUAGE PLPGSQL;
498 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
505 rtype := (biblio.marc21_record_type( rid )).code;
506 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
507 FOR tag_data IN SELECT * FROM metabib.full_rec WHERE tag = UPPER(ff_pos.tag) AND record = rid LOOP
508 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
511 val := REPEAT( ff_pos.default_val, ff_pos.length );
517 $func$ LANGUAGE PLPGSQL;
519 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
520 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
524 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
525 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
526 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
527 retval biblio.marc21_physical_characteristics%ROWTYPE;
530 SELECT * INTO _007 FROM metabib.full_rec WHERE record = rid AND tag = '007' LIMIT 1;
532 IF _007.id IS NOT NULL THEN
533 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007.value, 1, 1 );
535 IF ptype.ptype_key IS NOT NULL THEN
536 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
537 SELECT * INTO pval FROM config.marc21_physical_characteristic_value_map WHERE ptype_subfield = psf.id AND value = SUBSTRING( _007.value, psf.start_pos + 1, psf.length );
539 IF pval.id IS NOT NULL THEN
542 retval.record := rid;
543 retval.ptype := ptype.ptype_key;
544 retval.subfield := psf.id;
545 retval.value := pval.id;
555 $func$ LANGUAGE PLPGSQL;
557 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
570 IF marc IS NULL OR marc = '' THEN
574 -- First, the count of tags
575 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
577 -- now go through a bunch of pain to get the record type
578 IF best_type IS NOT NULL THEN
579 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
581 IF ldr IS NOT NULL THEN
582 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
583 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
586 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
587 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
589 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
591 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
593 IF type_map.code IS NOT NULL THEN
594 IF best_type = type_map.code THEN
595 qual := qual + qual / 2;
598 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
599 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
600 IF ff_tag_data = best_lang THEN
608 -- Now look for some quality metrics
610 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
615 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
622 $func$ LANGUAGE PLPGSQL;
624 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
626 idx config.biblio_fingerprint%ROWTYPE;
627 xfrm config.xml_transform%ROWTYPE;
629 transformed_xml TEXT;
631 xml_node_list TEXT[];
633 output_text TEXT := '';
636 IF marc IS NULL OR marc = '' THEN
640 -- Loop over the indexing entries
641 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
643 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
645 -- See if we can skip the XSLT ... it's expensive
646 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
647 -- Can't skip the transform
648 IF xfrm.xslt <> '---' THEN
649 transformed_xml := oils_xslt_process(marc,xfrm.xslt);
651 transformed_xml := marc;
654 prev_xfrm := xfrm.name;
657 raw_text := COALESCE(
665 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
674 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
675 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
677 IF idx.first_word IS TRUE THEN
678 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
681 output_text := output_text || REGEXP_REPLACE(raw_text, E'\\s+', '', 'g');
688 $func$ LANGUAGE PLPGSQL;
690 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
691 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
694 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
696 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
700 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
701 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
706 $func$ LANGUAGE PLPGSQL;
708 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_rec_descriptor( bib_id BIGINT ) RETURNS VOID AS $func$
710 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
712 DELETE FROM metabib.rec_descriptor WHERE record = bib_id;
714 INSERT INTO metabib.rec_descriptor (record, item_type, item_form, bib_level, control_type, enc_level, audience, lit_form, type_mat, cat_form, pub_status, item_lang, vr_format, date1, date2)
716 biblio.marc21_extract_fixed_field( bib_id, 'Type' ),
717 biblio.marc21_extract_fixed_field( bib_id, 'Form' ),
718 biblio.marc21_extract_fixed_field( bib_id, 'BLvl' ),
719 biblio.marc21_extract_fixed_field( bib_id, 'Ctrl' ),
720 biblio.marc21_extract_fixed_field( bib_id, 'ELvl' ),
721 biblio.marc21_extract_fixed_field( bib_id, 'Audn' ),
722 biblio.marc21_extract_fixed_field( bib_id, 'LitF' ),
723 biblio.marc21_extract_fixed_field( bib_id, 'TMat' ),
724 biblio.marc21_extract_fixed_field( bib_id, 'Desc' ),
725 biblio.marc21_extract_fixed_field( bib_id, 'DtSt' ),
726 biblio.marc21_extract_fixed_field( bib_id, 'Lang' ),
728 FROM biblio.marc21_physical_characteristics( bib_id) p
729 JOIN config.marc21_physical_characteristic_subfield_map s ON (s.id = p.subfield)
730 JOIN config.marc21_physical_characteristic_value_map v ON (v.id = p.value)
731 WHERE p.ptype = 'v' AND s.subfield = 'e' ),
732 LPAD(NULLIF(REGEXP_REPLACE(NULLIF(biblio.marc21_extract_fixed_field( bib_id, 'Date1'), ''), E'\\D', '0', 'g')::INT,0)::TEXT,4,'0'),
733 LPAD(NULLIF(REGEXP_REPLACE(NULLIF(biblio.marc21_extract_fixed_field( bib_id, 'Date2'), ''), E'\\D', '9', 'g')::INT,9999)::TEXT,4,'0');
737 $func$ LANGUAGE PLPGSQL;
739 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_full_rec( bib_id BIGINT ) RETURNS VOID AS $func$
741 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
743 DELETE FROM metabib.real_full_rec WHERE record = bib_id;
745 INSERT INTO metabib.real_full_rec (record, tag, ind1, ind2, subfield, value)
746 SELECT record, tag, ind1, ind2, subfield, value FROM biblio.flatten_marc( bib_id );
750 $func$ LANGUAGE PLPGSQL;
752 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries( bib_id BIGINT ) RETURNS VOID AS $func$
755 ind_data metabib.field_entry_template%ROWTYPE;
757 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
759 FOR fclass IN SELECT * FROM config.metabib_class LOOP
760 -- RAISE NOTICE 'Emptying out %', fclass.name;
761 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
763 DELETE FROM metabib.facet_entry WHERE source = bib_id;
766 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
767 IF ind_data.field < 0 THEN
768 ind_data.field = -1 * ind_data.field;
769 INSERT INTO metabib.facet_entry (field, source, value)
770 VALUES (ind_data.field, ind_data.source, ind_data.value);
773 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
775 quote_literal(ind_data.field) || $$, $$ ||
776 quote_literal(ind_data.source) || $$, $$ ||
777 quote_literal(ind_data.value) ||
785 $func$ LANGUAGE PLPGSQL;
787 CREATE OR REPLACE FUNCTION biblio.extract_located_uris( bib_id BIGINT, marcxml TEXT, editor_id INT ) RETURNS VOID AS $func$
801 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',marcxml);
802 IF ARRAY_UPPER(uris,1) > 0 THEN
803 FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
804 -- First we pull info out of the 856
807 uri_href := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
808 CONTINUE WHEN uri_href IS NULL;
810 uri_label := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()|//*[@code="u"]/text()',uri_xml))[1];
811 CONTINUE WHEN uri_label IS NULL;
813 uri_owner := (oils_xpath('//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',uri_xml))[1];
814 CONTINUE WHEN uri_owner IS NULL;
816 uri_use := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()|//*[@code="u"]/text()',uri_xml))[1];
818 uri_owner := REGEXP_REPLACE(uri_owner, $re$^.*?\((\w+)\).*$$re$, E'\\1');
820 SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = uri_owner;
821 CONTINUE WHEN NOT FOUND;
823 -- now we look for a matching uri
824 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
825 IF NOT FOUND THEN -- create one
826 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
827 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
830 -- we need a call number to link through
831 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;
833 INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
834 VALUES (uri_owner_id, bib_id, 'now', 'now', editor_id, editor_id, '##URI##');
835 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;
838 -- now, link them if they're not already
839 SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
841 INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
849 $func$ LANGUAGE PLPGSQL;
852 CREATE OR REPLACE FUNCTION metabib.remap_metarecord_for_bib( bib_id BIGINT, fp TEXT ) RETURNS BIGINT AS $func$
856 tmp_mr metabib.metarecord%ROWTYPE;
857 deleted_mrs BIGINT[];
860 DELETE FROM metabib.metarecord_source_map WHERE source = bib_id; -- Rid ourselves of the search-estimate-killing linkage
862 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
864 IF old_mr IS NULL AND fp = tmp_mr.fingerprint THEN -- Find the first fingerprint-matching
867 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
868 IF source_count = 0 THEN -- No other records
869 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
870 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
876 IF old_mr IS NULL THEN -- we found no suitable, preexisting MR based on old source maps
877 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp; -- is there one for our current fingerprint?
878 IF old_mr IS NULL THEN -- nope, create one and grab its id
879 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( fp, bib_id );
880 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp;
881 ELSE -- indeed there is. update it with a null cache and recalcualated master record
882 UPDATE metabib.metarecord
884 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
887 ELSE -- there was one we already attached to, update its mods cache and master_record
888 UPDATE metabib.metarecord
890 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
894 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, bib_id); -- new source mapping
896 IF ARRAY_UPPER(deleted_mrs,1) > 0 THEN
897 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
903 $func$ LANGUAGE PLPGSQL;
905 CREATE OR REPLACE FUNCTION biblio.map_authority_linking (bibid BIGINT, marc TEXT) RETURNS BIGINT AS $func$
906 DELETE FROM authority.bib_linking WHERE bib = $1;
907 INSERT INTO authority.bib_linking (bib, authority)
910 FROM ( SELECT DISTINCT $1 AS bib,
911 BTRIM(remove_paren_substring(txt))::BIGINT AS authority
912 FROM explode_array(oils_xpath('//*[@code="0"]/text()',$2)) x(txt)
913 WHERE BTRIM(remove_paren_substring(txt)) ~ $re$^\d+$$re$
914 ) y JOIN authority.record_entry r ON r.id = y.authority;
918 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
919 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
922 IF NEW.deleted IS TRUE THEN -- If this bib is deleted
923 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage
924 DELETE FROM authority.bib_linking WHERE bib = NEW.id; -- Avoid updating fields in bibs that are no longer visible
925 RETURN NEW; -- and we're done
928 IF TG_OP = 'UPDATE' THEN -- re-ingest?
929 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
931 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
936 -- Record authority linking
937 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_linking' AND enabled;
939 PERFORM biblio.map_authority_linking( NEW.id, NEW.marc );
942 -- Flatten and insert the mfr data
943 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_full_rec' AND enabled;
945 PERFORM metabib.reingest_metabib_full_rec(NEW.id);
946 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_rec_descriptor' AND enabled;
948 PERFORM metabib.reingest_metabib_rec_descriptor(NEW.id);
952 -- Gather and insert the field entry data
953 PERFORM metabib.reingest_metabib_field_entries(NEW.id);
956 IF TG_OP = 'INSERT' THEN
957 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
959 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
962 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
964 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
968 -- (re)map metarecord-bib linking
969 IF TG_OP = 'INSERT' THEN -- if not deleted and performing an insert, check for the flag
970 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
972 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
974 ELSE -- we're doing an update, and we're not deleted, remap
975 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_update' AND enabled;
977 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
983 $func$ LANGUAGE PLPGSQL;