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 TYPE metabib.field_entry_template AS (
225 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
227 bib biblio.record_entry%ROWTYPE;
228 idx config.metabib_field%ROWTYPE;
229 xfrm config.xml_transform%ROWTYPE;
231 transformed_xml TEXT;
233 xml_node_list TEXT[];
235 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
236 output_row metabib.field_entry_template%ROWTYPE;
240 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
242 -- Loop over the indexing entries
243 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
245 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
247 -- See if we can skip the XSLT ... it's expensive
248 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
249 -- Can't skip the transform
250 IF xfrm.xslt <> '---' THEN
251 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
253 transformed_xml := bib.marc;
256 prev_xfrm := xfrm.name;
259 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
262 FOR xml_node IN SELECT x FROM explode_array(xml_node_list) AS x LOOP
263 IF raw_text IS NOT NULL THEN
264 raw_text := raw_text || joiner;
266 raw_text := COALESCE(raw_text,'') || ARRAY_TO_STRING(oils_xpath( '//text()', REGEXP_REPLACE(xml_node,'&(?!amp;)','&','g')), ' ');
269 CONTINUE WHEN raw_text IS NULL;
271 output_row.field_class = idx.field_class;
272 output_row.field = idx.id;
273 output_row.source = rid;
274 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
276 RETURN NEXT output_row;
281 $func$ LANGUAGE PLPGSQL;
283 -- default to a space joiner
284 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
285 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
288 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
290 bib biblio.record_entry%ROWTYPE;
291 output metabib.full_rec%ROWTYPE;
294 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
296 FOR field IN SELECT * FROM biblio.flatten_marc( bib.marc ) LOOP
297 output.record := rid;
298 output.ind1 := field.ind1;
299 output.ind2 := field.ind2;
300 output.tag := field.tag;
301 output.subfield := field.subfield;
302 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
303 output.value := naco_normalize(field.value, field.subfield);
305 output.value := field.value;
308 CONTINUE WHEN output.value IS NULL;
313 $func$ LANGUAGE PLPGSQL;
315 /* Old form of biblio.flatten_marc() relied on contrib/xml2 functions that got all crashy in PostgreSQL 8.4 */
316 -- CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT, BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
317 -- 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
319 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), NULL::TEXT, NULL::TEXT, NULL::TEXT, x.value, NULL::tsvector AS index_vector
320 -- FROM oils_xpath_table(
323 -- 'biblio.record_entry',
324 -- '//*[local-name()="controlfield"]/@tag|//*[local-name()="controlfield"]',
326 -- )x(record int, tag text, value text)
328 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), x.ind1, x.ind2, x.subfield, x.value, NULL::tsvector AS index_vector
329 -- FROM oils_xpath_table(
332 -- 'biblio.record_entry',
333 -- '//*[local-name()="datafield"]/@tag|' ||
334 -- '//*[local-name()="datafield"]/@ind1|' ||
335 -- '//*[local-name()="datafield"]/@ind2|' ||
336 -- '//*[local-name()="datafield"]/*/@code|' ||
337 -- '//*[local-name()="datafield"]/*[@code]',
339 -- )x(record int, tag text, ind1 text, ind2 text, subfield text, value text);
340 -- $func$ LANGUAGE SQL;
342 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT ) RETURNS SETOF metabib.full_rec AS $func$
345 use MARC::File::XML (BinaryEncoding => 'UTF-8');
348 my $r = MARC::Record->new_from_xml( $xml );
350 return_next( { tag => 'LDR', value => $r->leader } );
352 for my $f ( $r->fields ) {
353 if ($f->is_control_field) {
354 return_next({ tag => $f->tag, value => $f->data });
356 for my $s ($f->subfields) {
359 ind1 => $f->indicator(1),
360 ind2 => $f->indicator(2),
365 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
366 my $trim = $f->indicator(2) || 0;
369 ind1 => $f->indicator(1),
370 ind2 => $f->indicator(2),
372 value => substr( $s->[1], $trim )
381 $func$ LANGUAGE PLPERLU;
383 CREATE OR REPLACE FUNCTION biblio.marc21_record_type( rid BIGINT ) RETURNS config.marc21_rec_type_map AS $func$
390 retval config.marc21_rec_type_map%ROWTYPE;
392 SELECT * INTO ldr FROM metabib.full_rec WHERE record = rid AND tag = 'LDR' LIMIT 1;
394 IF ldr.id IS NULL THEN
395 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
399 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
400 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
403 tval := SUBSTRING( ldr.value, tval_rec.start_pos + 1, tval_rec.length );
404 bval := SUBSTRING( ldr.value, bval_rec.start_pos + 1, bval_rec.length );
406 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr.value;
408 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
411 IF retval.code IS NULL THEN
412 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
417 $func$ LANGUAGE PLPGSQL;
419 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
426 rtype := (biblio.marc21_record_type( rid )).code;
427 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
428 FOR tag_data IN SELECT * FROM metabib.full_rec WHERE tag = UPPER(ff_pos.tag) AND record = rid LOOP
429 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
432 val := REPEAT( ff_pos.default_val, ff_pos.length );
438 $func$ LANGUAGE PLPGSQL;
440 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
441 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
445 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
446 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
447 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
448 retval biblio.marc21_physical_characteristics%ROWTYPE;
451 SELECT * INTO _007 FROM metabib.full_rec WHERE record = rid AND tag = '007' LIMIT 1;
453 IF _007.id IS NOT NULL THEN
454 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007.value, 1, 1 );
456 IF ptype.ptype_key IS NOT NULL THEN
457 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
458 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 );
460 IF pval.id IS NOT NULL THEN
463 retval.record := rid;
464 retval.ptype := ptype.ptype_key;
465 retval.subfield := psf.id;
466 retval.value := pval.id;
476 $func$ LANGUAGE PLPGSQL;
478 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
491 IF marc IS NULL OR marc = '' THEN
495 -- First, the count of tags
496 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
498 -- now go through a bunch of pain to get the record type
499 IF best_type IS NOT NULL THEN
500 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
502 IF ldr IS NOT NULL THEN
503 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
504 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
507 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
508 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
510 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
512 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
514 IF type_map.code IS NOT NULL THEN
515 IF best_type = type_map.code THEN
516 qual := qual + qual / 2;
519 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
520 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
521 IF ff_tag_data = best_lang THEN
529 -- Now look for some quality metrics
531 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
536 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
543 $func$ LANGUAGE PLPGSQL;
545 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
547 idx config.biblio_fingerprint%ROWTYPE;
548 xfrm config.xml_transform%ROWTYPE;
550 transformed_xml TEXT;
552 xml_node_list TEXT[];
554 output_text TEXT := '';
557 IF marc IS NULL OR marc = '' THEN
561 -- Loop over the indexing entries
562 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
564 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
566 -- See if we can skip the XSLT ... it's expensive
567 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
568 -- Can't skip the transform
569 IF xfrm.xslt <> '---' THEN
570 transformed_xml := oils_xslt_process(marc,xfrm.xslt);
572 transformed_xml := marc;
575 prev_xfrm := xfrm.name;
578 raw_text := COALESCE(
586 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
595 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
596 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
598 IF idx.first_word IS TRUE THEN
599 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
602 output_text := output_text || REGEXP_REPLACE(raw_text, E'\\s+', '', 'g');
609 $func$ LANGUAGE PLPGSQL;
611 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
612 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
615 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
617 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
621 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
622 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
627 $func$ LANGUAGE PLPGSQL;
629 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
630 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
632 ind_data metabib.field_entry_template%ROWTYPE;
634 tmp_mr metabib.metarecord%ROWTYPE;
649 IF TG_OP = 'UPDATE' THEN -- Clean out the cruft
650 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage
653 IF NEW.deleted IS TRUE THEN
654 RETURN NEW; -- and we're done
657 IF TG_OP = 'UPDATE' THEN -- Clean out the cruft
658 DELETE FROM metabib.title_field_entry WHERE source = NEW.id;
659 DELETE FROM metabib.author_field_entry WHERE source = NEW.id;
660 DELETE FROM metabib.subject_field_entry WHERE source = NEW.id;
661 DELETE FROM metabib.keyword_field_entry WHERE source = NEW.id;
662 DELETE FROM metabib.series_field_entry WHERE source = NEW.id;
663 DELETE FROM metabib.full_rec WHERE record = NEW.id;
664 DELETE FROM metabib.rec_descriptor WHERE record = NEW.id;
668 -- Shove the flattened MARC in
669 INSERT INTO metabib.full_rec (record, tag, ind1, ind2, subfield, value)
670 SELECT record, tag, ind1, ind2, subfield, value FROM biblio.flatten_marc( NEW.id );
672 -- And now the indexing data
673 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( NEW.id ) LOOP
674 IF ind_data.field_class = 'title' THEN
675 INSERT INTO metabib.title_field_entry (field, source, value)
676 VALUES (ind_data.field, ind_data.source, ind_data.value);
677 ELSIF ind_data.field_class = 'author' THEN
678 INSERT INTO metabib.author_field_entry (field, source, value)
679 VALUES (ind_data.field, ind_data.source, ind_data.value);
680 ELSIF ind_data.field_class = 'subject' THEN
681 INSERT INTO metabib.subject_field_entry (field, source, value)
682 VALUES (ind_data.field, ind_data.source, ind_data.value);
683 ELSIF ind_data.field_class = 'keyword' THEN
684 INSERT INTO metabib.keyword_field_entry (field, source, value)
685 VALUES (ind_data.field, ind_data.source, ind_data.value);
686 ELSIF ind_data.field_class = 'series' THEN
687 INSERT INTO metabib.series_field_entry (field, source, value)
688 VALUES (ind_data.field, ind_data.source, ind_data.value);
692 -- Then, the rec_descriptor
693 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)
695 biblio.marc21_extract_fixed_field( NEW.id, 'Type' ),
696 biblio.marc21_extract_fixed_field( NEW.id, 'Form' ),
697 biblio.marc21_extract_fixed_field( NEW.id, 'BLvl' ),
698 biblio.marc21_extract_fixed_field( NEW.id, 'Ctrl' ),
699 biblio.marc21_extract_fixed_field( NEW.id, 'ELvl' ),
700 biblio.marc21_extract_fixed_field( NEW.id, 'Audn' ),
701 biblio.marc21_extract_fixed_field( NEW.id, 'LitF' ),
702 biblio.marc21_extract_fixed_field( NEW.id, 'TMat' ),
703 biblio.marc21_extract_fixed_field( NEW.id, 'Desc' ),
704 biblio.marc21_extract_fixed_field( NEW.id, 'DtSt' ),
705 biblio.marc21_extract_fixed_field( NEW.id, 'Lang' ),
707 FROM biblio.marc21_physical_characteristics( NEW.id) p
708 JOIN config.marc21_physical_characteristic_subfield_map s ON (s.id = p.subfield)
709 JOIN config.marc21_physical_characteristic_value_map v ON (v.id = p.value)
710 WHERE p.ptype = 'v' AND s.subfield = 'e' ),
711 biblio.marc21_extract_fixed_field( NEW.id, 'Date1'),
712 biblio.marc21_extract_fixed_field( NEW.id, 'Date2');
715 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',NEW.marc);
716 IF ARRAY_UPPER(uris,1) > 0 THEN
717 FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
718 -- First we pull infot out of the 856
721 uri_href := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
722 CONTINUE WHEN uri_href IS NULL;
724 uri_label := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()|//*[@code="u"]/text()',uri_xml))[1];
725 CONTINUE WHEN uri_label IS NULL;
727 uri_owner := (oils_xpath('//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',uri_xml))[1];
728 CONTINUE WHEN uri_owner IS NULL;
730 uri_use := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()',uri_xml))[1];
732 uri_owner := REGEXP_REPLACE(uri_owner, $re$^.*?\((\w+)\).*$$re$, E'\\1');
734 SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = uri_owner;
735 CONTINUE WHEN NOT FOUND;
737 -- now we look for a matching uri
738 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
739 IF NOT FOUND THEN -- create one
740 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
741 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
744 -- we need a call number to link through
745 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;
747 INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
748 VALUES (uri_owner_id, NEW.id, 'now', 'now', NEW.editor, NEW.editor, '##URI##');
749 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;
752 -- now, link them if they're not already
753 SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
755 INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
761 -- And, finally, metarecord mapping!
763 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
765 IF NOT FOUND OR TG_OP = 'UPDATE' THEN
766 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
768 IF old_mr IS NULL AND NEW.fingerprint = tmp_mr.fingerprint THEN -- Find the first fingerprint-matching
771 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
772 IF source_count = 0 THEN -- No other records
773 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
774 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
780 IF old_mr IS NULL THEN -- we found no suitable, preexisting MR based on old source maps
781 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = NEW.fingerprint; -- is there one for our current fingerprint?
782 IF old_mr IS NULL THEN -- nope, create one and grab its id
783 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( NEW.fingerprint, NEW.id );
784 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = NEW.fingerprint;
785 ELSE -- indeed there is. update it with a null cache and recalcualated master record
786 UPDATE metabib.metarecord
788 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = NEW.fingerprint ORDER BY quality DESC LIMIT 1)
791 ELSE -- there was one we already attached to, update its mods cache and master_record
792 UPDATE metabib.metarecord
794 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = NEW.fingerprint ORDER BY quality DESC LIMIT 1)
798 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, NEW.id); -- new source mapping
800 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
806 $func$ LANGUAGE PLPGSQL;