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 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.title_field_entry (
33 id BIGSERIAL PRIMARY KEY,
34 source BIGINT NOT NULL,
37 index_vector tsvector NOT NULL
39 CREATE TRIGGER metabib_title_field_entry_fti_trigger
40 BEFORE UPDATE OR INSERT ON metabib.title_field_entry
41 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('title');
43 CREATE INDEX metabib_title_field_entry_index_vector_idx ON metabib.title_field_entry USING GIST (index_vector);
46 CREATE TABLE metabib.author_field_entry (
47 id BIGSERIAL PRIMARY KEY,
48 source BIGINT NOT NULL,
51 index_vector tsvector NOT NULL
53 CREATE TRIGGER metabib_author_field_entry_fti_trigger
54 BEFORE UPDATE OR INSERT ON metabib.author_field_entry
55 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('author');
57 CREATE INDEX metabib_author_field_entry_index_vector_idx ON metabib.author_field_entry USING GIST (index_vector);
60 CREATE TABLE metabib.subject_field_entry (
61 id BIGSERIAL PRIMARY KEY,
62 source BIGINT NOT NULL,
65 index_vector tsvector NOT NULL
67 CREATE TRIGGER metabib_subject_field_entry_fti_trigger
68 BEFORE UPDATE OR INSERT ON metabib.subject_field_entry
69 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('subject');
71 CREATE INDEX metabib_subject_field_entry_index_vector_idx ON metabib.subject_field_entry USING GIST (index_vector);
72 CREATE INDEX metabib_subject_field_entry_source_idx ON metabib.subject_field_entry (source);
75 CREATE TABLE metabib.keyword_field_entry (
76 id BIGSERIAL PRIMARY KEY,
77 source BIGINT NOT NULL,
80 index_vector tsvector NOT NULL
82 CREATE TRIGGER metabib_keyword_field_entry_fti_trigger
83 BEFORE UPDATE OR INSERT ON metabib.keyword_field_entry
84 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('keyword');
86 CREATE INDEX metabib_keyword_field_entry_index_vector_idx ON metabib.keyword_field_entry USING GIST (index_vector);
89 CREATE TABLE metabib.series_field_entry (
90 id BIGSERIAL PRIMARY KEY,
91 source BIGINT NOT NULL,
94 index_vector tsvector NOT NULL
96 CREATE TRIGGER metabib_series_field_entry_fti_trigger
97 BEFORE UPDATE OR INSERT ON metabib.series_field_entry
98 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('series');
100 CREATE INDEX metabib_series_field_entry_index_vector_idx ON metabib.series_field_entry USING GIST (index_vector);
103 CREATE TABLE metabib.rec_descriptor (
104 id BIGSERIAL PRIMARY KEY,
122 CREATE INDEX metabib_rec_descriptor_record_idx ON metabib.rec_descriptor (record);
123 /* We may not need these...
125 CREATE INDEX metabib_rec_descriptor_item_type_idx ON metabib.rec_descriptor (item_type);
126 CREATE INDEX metabib_rec_descriptor_item_form_idx ON metabib.rec_descriptor (item_form);
127 CREATE INDEX metabib_rec_descriptor_bib_level_idx ON metabib.rec_descriptor (bib_level);
128 CREATE INDEX metabib_rec_descriptor_control_type_idx ON metabib.rec_descriptor (control_type);
129 CREATE INDEX metabib_rec_descriptor_char_encoding_idx ON metabib.rec_descriptor (char_encoding);
130 CREATE INDEX metabib_rec_descriptor_enc_level_idx ON metabib.rec_descriptor (enc_level);
131 CREATE INDEX metabib_rec_descriptor_audience_idx ON metabib.rec_descriptor (audience);
132 CREATE INDEX metabib_rec_descriptor_lit_form_idx ON metabib.rec_descriptor (lit_form);
133 CREATE INDEX metabib_rec_descriptor_cat_form_idx ON metabib.rec_descriptor (cat_form);
134 CREATE INDEX metabib_rec_descriptor_pub_status_idx ON metabib.rec_descriptor (pub_status);
135 CREATE INDEX metabib_rec_descriptor_item_lang_idx ON metabib.rec_descriptor (item_lang);
136 CREATE INDEX metabib_rec_descriptor_vr_format_idx ON metabib.rec_descriptor (vr_format);
140 -- Use a sequence that matches previous version, for easier upgrading.
141 CREATE SEQUENCE metabib.full_rec_id_seq;
143 CREATE TABLE metabib.real_full_rec (
144 id BIGINT NOT NULL DEFAULT NEXTVAL('metabib.full_rec_id_seq'::REGCLASS),
145 record BIGINT NOT NULL,
146 tag CHAR(3) NOT NULL,
151 index_vector tsvector NOT NULL
153 ALTER TABLE metabib.real_full_rec ADD PRIMARY KEY (id);
155 CREATE INDEX metabib_full_rec_tag_subfield_idx ON metabib.real_full_rec (tag,subfield);
156 CREATE INDEX metabib_full_rec_value_idx ON metabib.real_full_rec (substring(value,1,1024));
157 /* Enable LIKE to use an index for database clusters with locales other than C or POSIX */
158 CREATE INDEX metabib_full_rec_value_tpo_index ON metabib.real_full_rec (substring(value,1,1024) text_pattern_ops);
159 CREATE INDEX metabib_full_rec_record_idx ON metabib.real_full_rec (record);
160 CREATE INDEX metabib_full_rec_index_vector_idx ON metabib.real_full_rec USING GIST (index_vector);
162 CREATE TRIGGER metabib_full_rec_fti_trigger
163 BEFORE UPDATE OR INSERT ON metabib.real_full_rec
164 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('default');
166 CREATE OR REPLACE VIEW metabib.full_rec AS
173 SUBSTRING(value,1,1024) AS value,
175 FROM metabib.real_full_rec;
177 CREATE OR REPLACE RULE metabib_full_rec_insert_rule
178 AS ON INSERT TO metabib.full_rec
180 INSERT INTO metabib.real_full_rec VALUES (
181 COALESCE(NEW.id, NEXTVAL('metabib.full_rec_id_seq'::REGCLASS)),
191 CREATE OR REPLACE RULE metabib_full_rec_update_rule
192 AS ON UPDATE TO metabib.full_rec
194 UPDATE metabib.real_full_rec SET
200 subfield = NEW.subfield,
202 index_vector = NEW.index_vector
205 CREATE OR REPLACE RULE metabib_full_rec_delete_rule
206 AS ON DELETE TO metabib.full_rec
208 DELETE FROM metabib.real_full_rec WHERE id = OLD.id;
210 CREATE TABLE metabib.metarecord_source_map (
211 id BIGSERIAL PRIMARY KEY,
212 metarecord BIGINT NOT NULL,
213 source BIGINT NOT NULL
215 CREATE INDEX metabib_metarecord_source_map_metarecord_idx ON metabib.metarecord_source_map (metarecord);
216 CREATE INDEX metabib_metarecord_source_map_source_record_idx ON metabib.metarecord_source_map (source);
218 CREATE FUNCTION version_specific_xpath () RETURNS TEXT AS $wrapper_function$
223 IF REGEXP_REPLACE(VERSION(),E'^.+?(\\d+\\.\\d+).*?$',E'\\1')::FLOAT < 8.3 THEN
224 out_text := 'Creating XPath functions that work like the native XPATH function in 8.3+';
226 EXECUTE $create_82_funcs$
228 CREATE OR REPLACE FUNCTION oils_xpath ( xpath TEXT, xml TEXT, ns ANYARRAY ) RETURNS TEXT[] AS $func$
235 munged_xpath := xpath;
237 IF ns IS NOT NULL THEN
238 FOR namespace IN 1 .. array_upper(ns, 1) LOOP
239 munged_xpath := REGEXP_REPLACE(
241 E'(' || ns[namespace][1] || E'):(\\w+)',
242 E'*[local-name() = "\\2" and namespace-uri() = "' || ns[namespace][2] || E'"]',
247 munged_xpath := REGEXP_REPLACE( munged_xpath, E'\\]\\[(\\D)',E' and \\1', 'g');
250 node_text := xpath_nodeset(xml, munged_xpath, 'XXX_OILS_NODESET');
251 node_text := REGEXP_REPLACE(node_text,'^<XXX_OILS_NODESET>', '');
252 node_text := REGEXP_REPLACE(node_text,'</XXX_OILS_NODESET>$', '');
254 RETURN STRING_TO_ARRAY(node_text, '</XXX_OILS_NODESET><XXX_OILS_NODESET>');
256 $func$ LANGUAGE PLPGSQL;
258 CREATE OR REPLACE FUNCTION oils_xpath ( TEXT, TEXT ) RETURNS TEXT[] AS 'SELECT oils_xpath( $1, $2, NULL::TEXT[] );' LANGUAGE SQL;
262 out_text := 'Creating XPath wrapper functions around the native XPATH function in 8.3+';
264 EXECUTE $create_83_funcs$
266 CREATE OR REPLACE FUNCTION oils_xpath ( TEXT, TEXT, ANYARRAY ) RETURNS TEXT[] AS 'SELECT XPATH( $1, $2::XML, $3 )::TEXT[];' LANGUAGE SQL;
267 CREATE OR REPLACE FUNCTION oils_xpath ( TEXT, TEXT ) RETURNS TEXT[] AS 'SELECT XPATH( $1, $2::XML )::TEXT[];' LANGUAGE SQL;
275 $wrapper_function$ LANGUAGE PLPGSQL;
277 SELECT version_specific_xpath();
278 DROP FUNCTION version_specific_xpath();
280 CREATE TYPE metabib.field_entry_template AS (
287 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
289 bib biblio.record_entry%ROWTYPE;
290 idx config.metabib_field%ROWTYPE;
291 xfrm config.xml_transform%ROWTYPE;
293 transformed_xml TEXT;
295 xml_node_list TEXT[];
297 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
298 output_row metabib.field_entry_template%ROWTYPE;
302 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
304 -- Loop over the indexing entries
305 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
307 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
309 -- See if we can skip the XSLT ... it's expensive
310 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
311 -- Can't skip the transform
312 IF xfrm.xslt <> '---' THEN
313 transformed_xml := xslt_process(bib.marc,xfrm.xslt);
315 transformed_xml := bib.marc;
318 prev_xfrm := xfrm.name;
321 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
324 FOR xml_node IN SELECT x FROM explode_array(xml_node_list) AS x LOOP
325 IF raw_text IS NOT NULL THEN
326 raw_text := raw_text || joiner;
328 raw_text := COALESCE(raw_text,'') || ARRAY_TO_STRING(oils_xpath( '//text()', xml_node ), ' ');
331 CONTINUE WHEN raw_text IS NULL;
333 output_row.field_class = idx.field_class;
334 output_row.field = idx.id;
335 output_row.source = rid;
336 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
338 RETURN NEXT output_row;
343 $func$ LANGUAGE PLPGSQL;
345 -- default to a space joiner
346 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
347 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
350 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
352 bib biblio.record_entry%ROWTYPE;
353 output metabib.full_rec%ROWTYPE;
356 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
358 FOR field IN SELECT * FROM biblio.flatten_marc( bib.marc ) LOOP
359 output.record := rid;
360 output.ind1 := field.ind1;
361 output.ind2 := field.ind2;
362 output.tag := field.tag;
363 output.subfield := field.subfield;
364 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
365 output.value := naco_normalize(field.value, field.subfield);
367 output.value := field.value;
373 $func$ LANGUAGE PLPGSQL;
375 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT ) RETURNS SETOF metabib.full_rec AS $func$
378 use MARC::File::XML (BinaryEncoding => 'UTF-8');
381 my $r = MARC::Record->new_from_xml( $xml );
383 return_next( { tag => 'LDR', value => $r->leader } );
385 for my $f ( $r->fields ) {
386 if ($f->is_control_field) {
387 return_next({ tag => $f->tag, value => $f->data });
389 for my $s ($f->subfields) {
392 ind1 => $f->indicator(1),
393 ind2 => $f->indicator(2),
398 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
399 my $trim = $f->indicator(2) || 0;
402 ind1 => $f->indicator(1),
403 ind2 => $f->indicator(2),
405 value => substr( $s->[1], $trim )
414 $func$ LANGUAGE PLPERLU;
416 CREATE OR REPLACE FUNCTION biblio.marc21_record_type( rid BIGINT ) RETURNS config.marc21_rec_type_map AS $func$
423 retval config.marc21_rec_type_map%ROWTYPE;
425 SELECT * INTO ldr FROM metabib.full_rec WHERE record = rid AND tag = 'LDR' LIMIT 1;
427 IF ldr.id IS NULL THEN
428 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
432 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
433 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
436 tval := SUBSTRING( ldr.value, tval_rec.start_pos + 1, tval_rec.length );
437 bval := SUBSTRING( ldr.value, bval_rec.start_pos + 1, bval_rec.length );
439 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr.value;
441 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
444 IF retval.code IS NULL THEN
445 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
450 $func$ LANGUAGE PLPGSQL;
452 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
459 rtype := (biblio.marc21_record_type( rid )).code;
460 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
461 FOR tag_data IN SELECT * FROM metabib.full_rec WHERE tag = UPPER(ff_pos.tag) AND record = rid LOOP
462 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
465 val := REPEAT( ff_pos.default_val, ff_pos.length );
471 $func$ LANGUAGE PLPGSQL;
473 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
474 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
478 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
479 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
480 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
481 retval biblio.marc21_physical_characteristics%ROWTYPE;
484 SELECT * INTO _007 FROM metabib.full_rec WHERE record = rid AND tag = '007' LIMIT 1;
486 IF _007.id IS NOT NULL THEN
487 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007.value, 1, 1 );
489 IF ptype.ptype_key IS NOT NULL THEN
490 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
491 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 );
493 IF pval.id IS NOT NULL THEN
496 retval.record := rid;
497 retval.ptype := ptype.ptype_key;
498 retval.subfield := psf.id;
499 retval.value := pval.id;
509 $func$ LANGUAGE PLPGSQL;
511 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
524 IF marc IS NULL OR marc = '' THEN
528 -- First, the count of tags
529 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
531 -- now go through a bunch of pain to get the record type
532 IF best_type IS NOT NULL THEN
533 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
535 IF ldr IS NOT NULL THEN
536 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
537 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
540 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
541 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
543 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
545 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
547 IF type_map.code IS NOT NULL THEN
548 IF best_type = type_map.code THEN
549 qual := qual + qual / 2;
552 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
553 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
554 IF ff_tag_data = best_lang THEN
562 -- Now look for some quality metrics
564 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
569 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
576 $func$ LANGUAGE PLPGSQL;
578 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
580 idx config.biblio_fingerprint%ROWTYPE;
581 xfrm config.xml_transform%ROWTYPE;
583 transformed_xml TEXT;
585 xml_node_list TEXT[];
587 output_text TEXT := '';
590 IF marc IS NULL OR marc = '' THEN
594 -- Loop over the indexing entries
595 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
597 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
599 -- See if we can skip the XSLT ... it's expensive
600 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
601 -- Can't skip the transform
602 IF xfrm.xslt <> '---' THEN
603 transformed_xml := xslt_process(marc,xfrm.xslt);
605 transformed_xml := marc;
608 prev_xfrm := xfrm.name;
611 raw_text := COALESCE(
619 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
628 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
629 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
631 IF idx.first_word IS TRUE THEN
632 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
635 output_text := output_text || REGEXP_REPLACE(raw_text, E'\\s+', '', 'g');
642 $func$ LANGUAGE PLPGSQL;
644 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
645 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
648 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
650 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
654 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
655 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
660 $func$ LANGUAGE PLPGSQL;
662 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
663 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
665 ind_data metabib.field_entry_template%ROWTYPE;
667 tmp_mr metabib.metarecord%ROWTYPE;
682 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage
684 IF NEW.deleted IS TRUE THEN
685 RETURN NEW; -- and we're done
688 IF TG_OP = 'UPDATE' THEN -- Clean out the cruft
689 DELETE FROM metabib.title_field_entry WHERE source = NEW.id;
690 DELETE FROM metabib.author_field_entry WHERE source = NEW.id;
691 DELETE FROM metabib.subject_field_entry WHERE source = NEW.id;
692 DELETE FROM metabib.keyword_field_entry WHERE source = NEW.id;
693 DELETE FROM metabib.series_field_entry WHERE source = NEW.id;
694 DELETE FROM metabib.full_rec WHERE record = NEW.id;
695 DELETE FROM metabib.rec_descriptor WHERE record = NEW.id;
699 -- Shove the flattened MARC in
700 INSERT INTO metabib.full_rec (record, tag, ind1, ind2, subfield, value)
701 SELECT record, tag, ind1, ind2, subfield, value FROM biblio.flatten_marc( NEW.id );
703 -- And now the indexing data
704 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( NEW.id ) LOOP
705 IF ind_data.field_class = 'title' THEN
706 INSERT INTO metabib.title_field_entry (field, source, value)
707 VALUES (ind_data.field, ind_data.source, ind_data.value);
708 ELSIF ind_data.field_class = 'author' THEN
709 INSERT INTO metabib.author_field_entry (field, source, value)
710 VALUES (ind_data.field, ind_data.source, ind_data.value);
711 ELSIF ind_data.field_class = 'subject' THEN
712 INSERT INTO metabib.subject_field_entry (field, source, value)
713 VALUES (ind_data.field, ind_data.source, ind_data.value);
714 ELSIF ind_data.field_class = 'keyword' THEN
715 INSERT INTO metabib.keyword_field_entry (field, source, value)
716 VALUES (ind_data.field, ind_data.source, ind_data.value);
717 ELSIF ind_data.field_class = 'series' THEN
718 INSERT INTO metabib.series_field_entry (field, source, value)
719 VALUES (ind_data.field, ind_data.source, ind_data.value);
723 -- Then, the rec_descriptor
724 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)
726 biblio.marc21_extract_fixed_field( NEW.id, 'Type' ),
727 biblio.marc21_extract_fixed_field( NEW.id, 'Form' ),
728 biblio.marc21_extract_fixed_field( NEW.id, 'BLvl' ),
729 biblio.marc21_extract_fixed_field( NEW.id, 'Ctrl' ),
730 biblio.marc21_extract_fixed_field( NEW.id, 'ELvl' ),
731 biblio.marc21_extract_fixed_field( NEW.id, 'Audn' ),
732 biblio.marc21_extract_fixed_field( NEW.id, 'LitF' ),
733 biblio.marc21_extract_fixed_field( NEW.id, 'TMat' ),
734 biblio.marc21_extract_fixed_field( NEW.id, 'Desc' ),
735 biblio.marc21_extract_fixed_field( NEW.id, 'DtSt' ),
736 biblio.marc21_extract_fixed_field( NEW.id, 'Lang' ),
738 FROM biblio.marc21_physical_characteristics( NEW.id) p
739 JOIN config.marc21_physical_characteristic_subfield_map s ON (s.id = p.subfield)
740 JOIN config.marc21_physical_characteristic_value_map v ON (v.id = p.value)
741 WHERE p.ptype = 'v' AND s.subfield = 'e' ),
742 biblio.marc21_extract_fixed_field( NEW.id, 'Date1'),
743 biblio.marc21_extract_fixed_field( NEW.id, 'Date2');
746 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',NEW.marc);
747 IF ARRAY_UPPER(uris,1) > 0 THEN
748 FOR i IN 1 .. ARRAY_UPPER( uris ) LOOP
749 -- First we pull infot out of the 856
752 uri_href := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
753 CONTINUE WHEN uri_href IS NULL;
755 uri_label := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()|//*[@code="u"]/text()',uri_xml))[1];
756 CONTINUE WHEN uri_label IS NULL;
758 uri_owner := (oils_xpath('//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',uri_xml))[1];
759 CONTINUE WHEN uri_owner IS NULL;
761 uri_use := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()',uri_xml))[1];
763 uri_owner := REGEXP_REPLACE(uri_owner, $re$^.*?\((\w+)\).*$$re$, E'\\1');
765 SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = uri_owner;
766 CONTINUE WHEN NOT FOUND;
768 -- now we look for a matching uri
769 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
770 IF NOT FOUND THEN -- create one
771 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
772 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
775 -- we need a call number to link through
776 SELECT id INTO uri_cn_id FROM asset.call_number WHERE owning_lib = uri_owner_id AND record = NEW.id AND label = '##URI##' AND NOT deleted;
778 INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
779 VALUES (uri_owner_id, NEW.id, 'now', 'now', NEW.editor, NEW.editor, '##URI##');
780 SELECT id INTO uri_cn_id FROM asset.call_number WHERE owning_lib = uri_owner_id AND record = NEW.id AND label = '##URI##' AND NOT deleted;
783 -- now, link them if they're not already
784 SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
786 INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
792 -- And, finally, metarecord mapping!
794 FOR tmp_mr IN SELECT m.* FROM metabib.metarecord m JOIN metabib.metarecord_source_map s ON (s.metarecord = m.id) WHERE s.source = NEW.id LOOP
796 IF old_mr IS NULL AND NEW.fingerprint = tmp_mr.fingerprint THEN -- Find the first fingerprint-matching
799 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
800 IF source_count = 0 THEN -- No other records
801 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
802 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
808 IF old_mr IS NULL THEN -- we found no suitable, preexisting MR based on old source maps
809 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = NEW.fingerprint; -- is there one for our current fingerprint?
810 IF old_mr IS NULL THEN -- nope, create one and grab its id
811 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( NEW.fingerprint, NEW.id );
812 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = NEW.fingerprint;
813 ELSE -- indeed there is. update it with a null cache and recalcualated master record
814 UPDATE metabib.metarecord
816 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = NEW.fingerprint ORDER BY quality DESC LIMIT 1)
819 ELSE -- there was one we already attached to, update its mods cache and master_record
820 UPDATE metabib.metarecord
822 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = NEW.fingerprint ORDER BY quality DESC LIMIT 1)
826 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, NEW.id); -- new source mapping
828 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
833 $func$ LANGUAGE PLPGSQL;