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.rec_descriptor (
129 id BIGSERIAL PRIMARY KEY,
147 CREATE INDEX metabib_rec_descriptor_record_idx ON metabib.rec_descriptor (record);
148 CREATE INDEX metabib_rec_descriptor_item_type_idx ON metabib.rec_descriptor (item_type);
149 CREATE INDEX metabib_rec_descriptor_item_form_idx ON metabib.rec_descriptor (item_form);
150 CREATE INDEX metabib_rec_descriptor_bib_level_idx ON metabib.rec_descriptor (bib_level);
151 CREATE INDEX metabib_rec_descriptor_control_type_idx ON metabib.rec_descriptor (control_type);
152 CREATE INDEX metabib_rec_descriptor_char_encoding_idx ON metabib.rec_descriptor (char_encoding);
153 CREATE INDEX metabib_rec_descriptor_enc_level_idx ON metabib.rec_descriptor (enc_level);
154 CREATE INDEX metabib_rec_descriptor_audience_idx ON metabib.rec_descriptor (audience);
155 CREATE INDEX metabib_rec_descriptor_lit_form_idx ON metabib.rec_descriptor (lit_form);
156 CREATE INDEX metabib_rec_descriptor_cat_form_idx ON metabib.rec_descriptor (cat_form);
157 CREATE INDEX metabib_rec_descriptor_pub_status_idx ON metabib.rec_descriptor (pub_status);
158 CREATE INDEX metabib_rec_descriptor_item_lang_idx ON metabib.rec_descriptor (item_lang);
159 CREATE INDEX metabib_rec_descriptor_vr_format_idx ON metabib.rec_descriptor (vr_format);
160 CREATE INDEX metabib_rec_descriptor_date1_idx ON metabib.rec_descriptor (date1);
161 CREATE INDEX metabib_rec_descriptor_dates_idx ON metabib.rec_descriptor (date1,date2);
163 -- Use a sequence that matches previous version, for easier upgrading.
164 CREATE SEQUENCE metabib.full_rec_id_seq;
166 CREATE TABLE metabib.real_full_rec (
167 id BIGINT NOT NULL DEFAULT NEXTVAL('metabib.full_rec_id_seq'::REGCLASS),
168 record BIGINT NOT NULL,
169 tag CHAR(3) NOT NULL,
174 index_vector tsvector NOT NULL
176 ALTER TABLE metabib.real_full_rec ADD PRIMARY KEY (id);
178 CREATE INDEX metabib_full_rec_tag_subfield_idx ON metabib.real_full_rec (tag,subfield);
179 CREATE INDEX metabib_full_rec_value_idx ON metabib.real_full_rec (substring(value,1,1024));
180 /* Enable LIKE to use an index for database clusters with locales other than C or POSIX */
181 CREATE INDEX metabib_full_rec_value_tpo_index ON metabib.real_full_rec (substring(value,1,1024) text_pattern_ops);
182 CREATE INDEX metabib_full_rec_record_idx ON metabib.real_full_rec (record);
183 CREATE INDEX metabib_full_rec_index_vector_idx ON metabib.real_full_rec USING GIST (index_vector);
185 CREATE TRIGGER metabib_full_rec_fti_trigger
186 BEFORE UPDATE OR INSERT ON metabib.real_full_rec
187 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('default');
189 CREATE OR REPLACE VIEW metabib.full_rec AS
196 SUBSTRING(value,1,1024) AS value,
198 FROM metabib.real_full_rec;
200 CREATE OR REPLACE RULE metabib_full_rec_insert_rule
201 AS ON INSERT TO metabib.full_rec
203 INSERT INTO metabib.real_full_rec VALUES (
204 COALESCE(NEW.id, NEXTVAL('metabib.full_rec_id_seq'::REGCLASS)),
214 CREATE OR REPLACE RULE metabib_full_rec_update_rule
215 AS ON UPDATE TO metabib.full_rec
217 UPDATE metabib.real_full_rec SET
223 subfield = NEW.subfield,
225 index_vector = NEW.index_vector
228 CREATE OR REPLACE RULE metabib_full_rec_delete_rule
229 AS ON DELETE TO metabib.full_rec
231 DELETE FROM metabib.real_full_rec WHERE id = OLD.id;
233 CREATE TABLE metabib.metarecord_source_map (
234 id BIGSERIAL PRIMARY KEY,
235 metarecord BIGINT NOT NULL,
236 source BIGINT NOT NULL
238 CREATE INDEX metabib_metarecord_source_map_metarecord_idx ON metabib.metarecord_source_map (metarecord);
239 CREATE INDEX metabib_metarecord_source_map_source_record_idx ON metabib.metarecord_source_map (source);
241 CREATE TYPE metabib.field_entry_template AS (
248 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
250 bib biblio.record_entry%ROWTYPE;
251 idx config.metabib_field%ROWTYPE;
252 xfrm config.xml_transform%ROWTYPE;
254 transformed_xml TEXT;
256 xml_node_list TEXT[];
260 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
261 output_row metabib.field_entry_template%ROWTYPE;
265 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
267 -- Loop over the indexing entries
268 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
270 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
272 -- See if we can skip the XSLT ... it's expensive
273 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
274 -- Can't skip the transform
275 IF xfrm.xslt <> '---' THEN
276 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
278 transformed_xml := bib.marc;
281 prev_xfrm := xfrm.name;
284 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
287 FOR xml_node IN SELECT x FROM explode_array(xml_node_list) AS x LOOP
288 CONTINUE WHEN xml_node !~ E'^\\s*<';
290 curr_text := ARRAY_TO_STRING(
291 oils_xpath( '//text()',
292 REGEXP_REPLACE( -- This escapes all &s not followed by "amp;". Data ise returned from oils_xpath (above) in UTF-8, not entity encoded
293 REGEXP_REPLACE( -- This escapes embeded <s
295 $re$(>[^<]+)(<)([^>]+<)$re$,
307 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
309 IF raw_text IS NOT NULL THEN
310 raw_text := raw_text || joiner;
313 raw_text := COALESCE(raw_text,'') || curr_text;
315 -- insert raw node text for faceting
316 IF idx.facet_field THEN
318 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
319 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
321 facet_text := curr_text;
324 output_row.field_class = idx.field_class;
325 output_row.field = -1 * idx.id;
326 output_row.source = rid;
327 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
329 RETURN NEXT output_row;
334 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
336 -- insert combined node text for searching
337 IF idx.search_field THEN
338 output_row.field_class = idx.field_class;
339 output_row.field = idx.id;
340 output_row.source = rid;
341 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
343 RETURN NEXT output_row;
349 $func$ LANGUAGE PLPGSQL;
351 -- default to a space joiner
352 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
353 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
356 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
358 bib biblio.record_entry%ROWTYPE;
359 output metabib.full_rec%ROWTYPE;
362 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
364 FOR field IN SELECT * FROM biblio.flatten_marc( bib.marc ) LOOP
365 output.record := rid;
366 output.ind1 := field.ind1;
367 output.ind2 := field.ind2;
368 output.tag := field.tag;
369 output.subfield := field.subfield;
370 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
371 output.value := naco_normalize(field.value, field.subfield);
373 output.value := field.value;
376 CONTINUE WHEN output.value IS NULL;
381 $func$ LANGUAGE PLPGSQL;
383 /* Old form of biblio.flatten_marc() relied on contrib/xml2 functions that got all crashy in PostgreSQL 8.4 */
384 -- CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT, BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
385 -- 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
387 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), NULL::TEXT, NULL::TEXT, NULL::TEXT, x.value, NULL::tsvector AS index_vector
388 -- FROM oils_xpath_table(
391 -- 'biblio.record_entry',
392 -- '//*[local-name()="controlfield"]/@tag|//*[local-name()="controlfield"]',
394 -- )x(record int, tag text, value text)
396 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), x.ind1, x.ind2, x.subfield, x.value, NULL::tsvector AS index_vector
397 -- FROM oils_xpath_table(
400 -- 'biblio.record_entry',
401 -- '//*[local-name()="datafield"]/@tag|' ||
402 -- '//*[local-name()="datafield"]/@ind1|' ||
403 -- '//*[local-name()="datafield"]/@ind2|' ||
404 -- '//*[local-name()="datafield"]/*/@code|' ||
405 -- '//*[local-name()="datafield"]/*[@code]',
407 -- )x(record int, tag text, ind1 text, ind2 text, subfield text, value text);
408 -- $func$ LANGUAGE SQL;
410 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT ) RETURNS SETOF metabib.full_rec AS $func$
413 use MARC::File::XML (BinaryEncoding => 'UTF-8');
416 my $r = MARC::Record->new_from_xml( $xml );
418 return_next( { tag => 'LDR', value => $r->leader } );
420 for my $f ( $r->fields ) {
421 if ($f->is_control_field) {
422 return_next({ tag => $f->tag, value => $f->data });
424 for my $s ($f->subfields) {
427 ind1 => $f->indicator(1),
428 ind2 => $f->indicator(2),
433 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
434 my $trim = $f->indicator(2) || 0;
437 ind1 => $f->indicator(1),
438 ind2 => $f->indicator(2),
440 value => substr( $s->[1], $trim )
449 $func$ LANGUAGE PLPERLU;
451 CREATE OR REPLACE FUNCTION biblio.marc21_record_type( rid BIGINT ) RETURNS config.marc21_rec_type_map AS $func$
458 retval config.marc21_rec_type_map%ROWTYPE;
460 SELECT * INTO ldr FROM metabib.full_rec WHERE record = rid AND tag = 'LDR' LIMIT 1;
462 IF ldr.id IS NULL THEN
463 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
467 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
468 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
471 tval := SUBSTRING( ldr.value, tval_rec.start_pos + 1, tval_rec.length );
472 bval := SUBSTRING( ldr.value, bval_rec.start_pos + 1, bval_rec.length );
474 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr.value;
476 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
479 IF retval.code IS NULL THEN
480 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
485 $func$ LANGUAGE PLPGSQL;
487 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
494 rtype := (biblio.marc21_record_type( rid )).code;
495 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
496 FOR tag_data IN SELECT * FROM metabib.full_rec WHERE tag = UPPER(ff_pos.tag) AND record = rid LOOP
497 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
500 val := REPEAT( ff_pos.default_val, ff_pos.length );
506 $func$ LANGUAGE PLPGSQL;
508 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
509 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
513 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
514 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
515 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
516 retval biblio.marc21_physical_characteristics%ROWTYPE;
519 SELECT * INTO _007 FROM metabib.full_rec WHERE record = rid AND tag = '007' LIMIT 1;
521 IF _007.id IS NOT NULL THEN
522 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007.value, 1, 1 );
524 IF ptype.ptype_key IS NOT NULL THEN
525 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
526 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 );
528 IF pval.id IS NOT NULL THEN
531 retval.record := rid;
532 retval.ptype := ptype.ptype_key;
533 retval.subfield := psf.id;
534 retval.value := pval.id;
544 $func$ LANGUAGE PLPGSQL;
546 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
559 IF marc IS NULL OR marc = '' THEN
563 -- First, the count of tags
564 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
566 -- now go through a bunch of pain to get the record type
567 IF best_type IS NOT NULL THEN
568 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
570 IF ldr IS NOT NULL THEN
571 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
572 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
575 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
576 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
578 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
580 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
582 IF type_map.code IS NOT NULL THEN
583 IF best_type = type_map.code THEN
584 qual := qual + qual / 2;
587 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
588 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
589 IF ff_tag_data = best_lang THEN
597 -- Now look for some quality metrics
599 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
604 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
611 $func$ LANGUAGE PLPGSQL;
613 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
615 idx config.biblio_fingerprint%ROWTYPE;
616 xfrm config.xml_transform%ROWTYPE;
618 transformed_xml TEXT;
620 xml_node_list TEXT[];
622 output_text TEXT := '';
625 IF marc IS NULL OR marc = '' THEN
629 -- Loop over the indexing entries
630 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
632 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
634 -- See if we can skip the XSLT ... it's expensive
635 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
636 -- Can't skip the transform
637 IF xfrm.xslt <> '---' THEN
638 transformed_xml := oils_xslt_process(marc,xfrm.xslt);
640 transformed_xml := marc;
643 prev_xfrm := xfrm.name;
646 raw_text := COALESCE(
654 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
663 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
664 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
666 IF idx.first_word IS TRUE THEN
667 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
670 output_text := output_text || REGEXP_REPLACE(raw_text, E'\\s+', '', 'g');
677 $func$ LANGUAGE PLPGSQL;
679 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
680 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
683 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
685 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
689 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
690 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
695 $func$ LANGUAGE PLPGSQL;
697 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_rec_descriptor( bib_id BIGINT ) RETURNS VOID AS $func$
699 DELETE FROM metabib.rec_descriptor WHERE record = bib_id;
700 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)
702 biblio.marc21_extract_fixed_field( bib_id, 'Type' ),
703 biblio.marc21_extract_fixed_field( bib_id, 'Form' ),
704 biblio.marc21_extract_fixed_field( bib_id, 'BLvl' ),
705 biblio.marc21_extract_fixed_field( bib_id, 'Ctrl' ),
706 biblio.marc21_extract_fixed_field( bib_id, 'ELvl' ),
707 biblio.marc21_extract_fixed_field( bib_id, 'Audn' ),
708 biblio.marc21_extract_fixed_field( bib_id, 'LitF' ),
709 biblio.marc21_extract_fixed_field( bib_id, 'TMat' ),
710 biblio.marc21_extract_fixed_field( bib_id, 'Desc' ),
711 biblio.marc21_extract_fixed_field( bib_id, 'DtSt' ),
712 biblio.marc21_extract_fixed_field( bib_id, 'Lang' ),
714 FROM biblio.marc21_physical_characteristics( bib_id) p
715 JOIN config.marc21_physical_characteristic_subfield_map s ON (s.id = p.subfield)
716 JOIN config.marc21_physical_characteristic_value_map v ON (v.id = p.value)
717 WHERE p.ptype = 'v' AND s.subfield = 'e' ),
718 biblio.marc21_extract_fixed_field( bib_id, 'Date1'),
719 biblio.marc21_extract_fixed_field( bib_id, 'Date2');
723 $func$ LANGUAGE PLPGSQL;
725 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_full_rec( bib_id BIGINT ) RETURNS VOID AS $func$
727 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;
741 FOR fclass IN SELECT * FROM config.metabib_class LOOP
742 -- RAISE NOTICE 'Emptying out %', fclass.name;
743 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
746 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
747 IF ind_data.field < 0 THEN
749 ind_data.field = -1 * ind_data.field;
755 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value, index_vector)
757 quote_literal(ind_data.field) || $$, $$ ||
758 quote_literal(ind_data.source) || $$, $$ ||
759 quote_literal(ind_data.value) || $$, $$ ||
760 COALESCE(quote_literal(ind_vector),'NULL'::TEXT) ||
766 $func$ LANGUAGE PLPGSQL;
768 CREATE OR REPLACE FUNCTION biblio.extract_located_uris( bib_id BIGINT, marcxml TEXT, editor_id INT ) RETURNS VOID AS $func$
782 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',marcxml);
783 IF ARRAY_UPPER(uris,1) > 0 THEN
784 FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
785 -- First we pull info out of the 856
788 uri_href := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
789 CONTINUE WHEN uri_href IS NULL;
791 uri_label := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()|//*[@code="u"]/text()',uri_xml))[1];
792 CONTINUE WHEN uri_label IS NULL;
794 uri_owner := (oils_xpath('//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',uri_xml))[1];
795 CONTINUE WHEN uri_owner IS NULL;
797 uri_use := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()',uri_xml))[1];
799 uri_owner := REGEXP_REPLACE(uri_owner, $re$^.*?\((\w+)\).*$$re$, E'\\1');
801 SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = uri_owner;
802 CONTINUE WHEN NOT FOUND;
804 -- now we look for a matching uri
805 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
806 IF NOT FOUND THEN -- create one
807 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
808 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
811 -- we need a call number to link through
812 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;
814 INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
815 VALUES (uri_owner_id, bib_id, 'now', 'now', editor_id, editor_id, '##URI##');
816 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;
819 -- now, link them if they're not already
820 SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
822 INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
830 $func$ LANGUAGE PLPGSQL;
833 CREATE OR REPLACE FUNCTION metabib.remap_metarecord_for_bib( bib_id BIGINT, fp TEXT ) RETURNS BIGINT AS $func$
837 tmp_mr metabib.metarecord%ROWTYPE;
838 deleted_mrs BIGINT[];
841 DELETE FROM metabib.metarecord_source_map WHERE source = bib_id; -- Rid ourselves of the search-estimate-killing linkage
843 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
845 IF old_mr IS NULL AND fp = tmp_mr.fingerprint THEN -- Find the first fingerprint-matching
848 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
849 IF source_count = 0 THEN -- No other records
850 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
851 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
857 IF old_mr IS NULL THEN -- we found no suitable, preexisting MR based on old source maps
858 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp; -- is there one for our current fingerprint?
859 IF old_mr IS NULL THEN -- nope, create one and grab its id
860 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( fp, bib_id );
861 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp;
862 ELSE -- indeed there is. update it with a null cache and recalcualated master record
863 UPDATE metabib.metarecord
865 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
868 ELSE -- there was one we already attached to, update its mods cache and master_record
869 UPDATE metabib.metarecord
871 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
875 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, bib_id); -- new source mapping
877 IF ARRAY_UPPER(deleted_mrs,1) > 0 THEN
878 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
884 $func$ LANGUAGE PLPGSQL;
886 CREATE OR REPLACE FUNCTION authority.propagate_changes (aid BIGINT, bid BIGINT) RETURNS BIGINT AS $func$
887 UPDATE biblio.record_entry
888 SET marc = vandelay.merge_record_xml( marc, authority.generate_overlay_template( $1 ) )
893 CREATE OR REPLACE FUNCTION authority.propagate_changes (aid BIGINT) RETURNS SETOF BIGINT AS $func$
894 SELECT authority.propagate_changes( authority, bib ) FROM authority.bib_linking WHERE authority = $1;
897 CREATE OR REPLACE FUNCTION authority.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
900 IF NEW.deleted IS TRUE THEN -- If this authority is deleted
901 RETURN NEW; -- and ... we're done
904 IF TG_OP = 'UPDATE' THEN -- re-ingest?
905 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
907 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
913 -- authority change propogation
914 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_auto_update' AND enabled;
916 PERFORM authority.propagate_changes( NEW.id );
921 $func$ 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(x))::BIGINT AS authority
931 FROM explode_array(oils_xpath('//*[@code="0"]/text()',$2)) x
932 ) y JOIN authority.record_entry r ON r.id = y.authority;
936 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
937 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
940 IF NEW.deleted IS TRUE THEN -- If this bib is deleted
941 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage
942 RETURN NEW; -- and we're done
945 IF TG_OP = 'UPDATE' THEN -- re-ingest?
946 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
948 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
953 -- Record authority linking
954 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_linking' AND enabled;
956 PERFORM biblio.map_authority_linking( NEW.id, NEW.marc );
959 -- Flatten and insert the mfr data
960 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_full_rec' AND enabled;
962 PERFORM metabib.reingest_metabib_full_rec(NEW.id);
963 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_rec_descriptor' AND enabled;
965 PERFORM metabib.reingest_metabib_rec_descriptor(NEW.id);
969 -- Gather and insert the field entry data
970 PERFORM metabib.reingest_metabib_field_entries(NEW.id);
973 IF TG_OP = 'INSERT' THEN
974 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
976 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
979 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
981 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
985 -- (re)map metarecord-bib linking
986 IF TG_OP = 'INSERT' THEN -- if not deleted and performing an insert, check for the flag
987 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
989 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
991 ELSE -- we're doing an update, and we're not deleted, remap
992 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
997 $func$ LANGUAGE PLPGSQL;