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);
44 CREATE INDEX metabib_title_field_entry_value_idx ON metabib.title_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
45 CREATE INDEX metabib_title_field_entry_source_idx ON metabib.title_field_entry (source);
48 CREATE TABLE metabib.author_field_entry (
49 id BIGSERIAL PRIMARY KEY,
50 source BIGINT NOT NULL,
53 index_vector tsvector NOT NULL
55 CREATE TRIGGER metabib_author_field_entry_fti_trigger
56 BEFORE UPDATE OR INSERT ON metabib.author_field_entry
57 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('author');
59 CREATE INDEX metabib_author_field_entry_index_vector_idx ON metabib.author_field_entry USING GIST (index_vector);
60 CREATE INDEX metabib_author_field_entry_value_idx ON metabib.author_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
61 CREATE INDEX metabib_author_field_entry_source_idx ON metabib.author_field_entry (source);
64 CREATE TABLE metabib.subject_field_entry (
65 id BIGSERIAL PRIMARY KEY,
66 source BIGINT NOT NULL,
69 index_vector tsvector NOT NULL
71 CREATE TRIGGER metabib_subject_field_entry_fti_trigger
72 BEFORE UPDATE OR INSERT ON metabib.subject_field_entry
73 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('subject');
75 CREATE INDEX metabib_subject_field_entry_index_vector_idx ON metabib.subject_field_entry USING GIST (index_vector);
76 CREATE INDEX metabib_subject_field_entry_value_idx ON metabib.subject_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
77 CREATE INDEX metabib_subject_field_entry_source_idx ON metabib.subject_field_entry (source);
80 CREATE TABLE metabib.keyword_field_entry (
81 id BIGSERIAL PRIMARY KEY,
82 source BIGINT NOT NULL,
85 index_vector tsvector NOT NULL
87 CREATE TRIGGER metabib_keyword_field_entry_fti_trigger
88 BEFORE UPDATE OR INSERT ON metabib.keyword_field_entry
89 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('keyword');
91 CREATE INDEX metabib_keyword_field_entry_index_vector_idx ON metabib.keyword_field_entry USING GIST (index_vector);
92 CREATE INDEX metabib_keyword_field_entry_value_idx ON metabib.keyword_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
93 CREATE INDEX metabib_keyword_field_entry_source_idx ON metabib.keyword_field_entry (source);
96 CREATE TABLE metabib.series_field_entry (
97 id BIGSERIAL PRIMARY KEY,
98 source BIGINT NOT NULL,
101 index_vector tsvector NOT NULL
103 CREATE TRIGGER metabib_series_field_entry_fti_trigger
104 BEFORE UPDATE OR INSERT ON metabib.series_field_entry
105 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('series');
107 CREATE INDEX metabib_series_field_entry_index_vector_idx ON metabib.series_field_entry USING GIST (index_vector);
108 CREATE INDEX metabib_series_field_entry_value_idx ON metabib.series_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
109 CREATE INDEX metabib_series_field_entry_source_idx ON metabib.series_field_entry (source);
112 CREATE TABLE metabib.rec_descriptor (
113 id BIGSERIAL PRIMARY KEY,
131 CREATE INDEX metabib_rec_descriptor_record_idx ON metabib.rec_descriptor (record);
132 CREATE INDEX metabib_rec_descriptor_item_type_idx ON metabib.rec_descriptor (item_type);
133 CREATE INDEX metabib_rec_descriptor_item_form_idx ON metabib.rec_descriptor (item_form);
134 CREATE INDEX metabib_rec_descriptor_bib_level_idx ON metabib.rec_descriptor (bib_level);
135 CREATE INDEX metabib_rec_descriptor_control_type_idx ON metabib.rec_descriptor (control_type);
136 CREATE INDEX metabib_rec_descriptor_char_encoding_idx ON metabib.rec_descriptor (char_encoding);
137 CREATE INDEX metabib_rec_descriptor_enc_level_idx ON metabib.rec_descriptor (enc_level);
138 CREATE INDEX metabib_rec_descriptor_audience_idx ON metabib.rec_descriptor (audience);
139 CREATE INDEX metabib_rec_descriptor_lit_form_idx ON metabib.rec_descriptor (lit_form);
140 CREATE INDEX metabib_rec_descriptor_cat_form_idx ON metabib.rec_descriptor (cat_form);
141 CREATE INDEX metabib_rec_descriptor_pub_status_idx ON metabib.rec_descriptor (pub_status);
142 CREATE INDEX metabib_rec_descriptor_item_lang_idx ON metabib.rec_descriptor (item_lang);
143 CREATE INDEX metabib_rec_descriptor_vr_format_idx ON metabib.rec_descriptor (vr_format);
144 CREATE INDEX metabib_rec_descriptor_date1_idx ON metabib.rec_descriptor (date1);
145 CREATE INDEX metabib_rec_descriptor_dates_idx ON metabib.rec_descriptor (date1,date2);
147 -- Use a sequence that matches previous version, for easier upgrading.
148 CREATE SEQUENCE metabib.full_rec_id_seq;
150 CREATE TABLE metabib.real_full_rec (
151 id BIGINT NOT NULL DEFAULT NEXTVAL('metabib.full_rec_id_seq'::REGCLASS),
152 record BIGINT NOT NULL,
153 tag CHAR(3) NOT NULL,
158 index_vector tsvector NOT NULL
160 ALTER TABLE metabib.real_full_rec ADD PRIMARY KEY (id);
162 CREATE INDEX metabib_full_rec_tag_subfield_idx ON metabib.real_full_rec (tag,subfield);
163 CREATE INDEX metabib_full_rec_value_idx ON metabib.real_full_rec (substring(value,1,1024));
164 /* Enable LIKE to use an index for database clusters with locales other than C or POSIX */
165 CREATE INDEX metabib_full_rec_value_tpo_index ON metabib.real_full_rec (substring(value,1,1024) text_pattern_ops);
166 CREATE INDEX metabib_full_rec_record_idx ON metabib.real_full_rec (record);
167 CREATE INDEX metabib_full_rec_index_vector_idx ON metabib.real_full_rec USING GIST (index_vector);
169 CREATE TRIGGER metabib_full_rec_fti_trigger
170 BEFORE UPDATE OR INSERT ON metabib.real_full_rec
171 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('default');
173 CREATE OR REPLACE VIEW metabib.full_rec AS
180 SUBSTRING(value,1,1024) AS value,
182 FROM metabib.real_full_rec;
184 CREATE OR REPLACE RULE metabib_full_rec_insert_rule
185 AS ON INSERT TO metabib.full_rec
187 INSERT INTO metabib.real_full_rec VALUES (
188 COALESCE(NEW.id, NEXTVAL('metabib.full_rec_id_seq'::REGCLASS)),
198 CREATE OR REPLACE RULE metabib_full_rec_update_rule
199 AS ON UPDATE TO metabib.full_rec
201 UPDATE metabib.real_full_rec SET
207 subfield = NEW.subfield,
209 index_vector = NEW.index_vector
212 CREATE OR REPLACE RULE metabib_full_rec_delete_rule
213 AS ON DELETE TO metabib.full_rec
215 DELETE FROM metabib.real_full_rec WHERE id = OLD.id;
217 CREATE TABLE metabib.metarecord_source_map (
218 id BIGSERIAL PRIMARY KEY,
219 metarecord BIGINT NOT NULL,
220 source BIGINT NOT NULL
222 CREATE INDEX metabib_metarecord_source_map_metarecord_idx ON metabib.metarecord_source_map (metarecord);
223 CREATE INDEX metabib_metarecord_source_map_source_record_idx ON metabib.metarecord_source_map (source);
225 CREATE TYPE metabib.field_entry_template AS (
232 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
234 bib biblio.record_entry%ROWTYPE;
235 idx config.metabib_field%ROWTYPE;
236 xfrm config.xml_transform%ROWTYPE;
238 transformed_xml TEXT;
240 xml_node_list TEXT[];
244 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
245 output_row metabib.field_entry_template%ROWTYPE;
249 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
251 -- Loop over the indexing entries
252 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
254 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
256 -- See if we can skip the XSLT ... it's expensive
257 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
258 -- Can't skip the transform
259 IF xfrm.xslt <> '---' THEN
260 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
262 transformed_xml := bib.marc;
265 prev_xfrm := xfrm.name;
268 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
271 FOR xml_node IN SELECT x FROM explode_array(xml_node_list) AS x LOOP
272 CONTINUE WHEN xml_node !~ E'^\\s*<';
274 curr_text := ARRAY_TO_STRING(
275 oils_xpath( '//text()',
276 REGEXP_REPLACE( -- This escapes all &s not followed by "amp;". Data ise returned from oils_xpath (above) in UTF-8, not entity encoded
277 REGEXP_REPLACE( -- This escapes embeded <s
279 $re$(>[^<]+)(<)([^>]+<)$re$,
291 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
293 IF raw_text IS NOT NULL THEN
294 raw_text := raw_text || joiner;
297 raw_text := COALESCE(raw_text,'') || curr_text;
299 -- insert raw node text for faceting
300 IF idx.facet_field THEN
302 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
303 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
305 facet_text := curr_text;
308 output_row.field_class = idx.field_class;
309 output_row.field = -1 * idx.id;
310 output_row.source = rid;
311 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
313 RETURN NEXT output_row;
318 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
320 -- insert combined node text for searching
321 IF idx.search_field THEN
322 output_row.field_class = idx.field_class;
323 output_row.field = idx.id;
324 output_row.source = rid;
325 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
327 RETURN NEXT output_row;
333 $func$ LANGUAGE PLPGSQL;
335 -- default to a space joiner
336 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
337 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
340 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
342 bib biblio.record_entry%ROWTYPE;
343 output metabib.full_rec%ROWTYPE;
346 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
348 FOR field IN SELECT * FROM biblio.flatten_marc( bib.marc ) LOOP
349 output.record := rid;
350 output.ind1 := field.ind1;
351 output.ind2 := field.ind2;
352 output.tag := field.tag;
353 output.subfield := field.subfield;
354 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
355 output.value := naco_normalize(field.value, field.subfield);
357 output.value := field.value;
360 CONTINUE WHEN output.value IS NULL;
365 $func$ LANGUAGE PLPGSQL;
367 /* Old form of biblio.flatten_marc() relied on contrib/xml2 functions that got all crashy in PostgreSQL 8.4 */
368 -- CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT, BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
369 -- 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
371 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), NULL::TEXT, NULL::TEXT, NULL::TEXT, x.value, NULL::tsvector AS index_vector
372 -- FROM oils_xpath_table(
375 -- 'biblio.record_entry',
376 -- '//*[local-name()="controlfield"]/@tag|//*[local-name()="controlfield"]',
378 -- )x(record int, tag text, value text)
380 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), x.ind1, x.ind2, x.subfield, x.value, NULL::tsvector AS index_vector
381 -- FROM oils_xpath_table(
384 -- 'biblio.record_entry',
385 -- '//*[local-name()="datafield"]/@tag|' ||
386 -- '//*[local-name()="datafield"]/@ind1|' ||
387 -- '//*[local-name()="datafield"]/@ind2|' ||
388 -- '//*[local-name()="datafield"]/*/@code|' ||
389 -- '//*[local-name()="datafield"]/*[@code]',
391 -- )x(record int, tag text, ind1 text, ind2 text, subfield text, value text);
392 -- $func$ LANGUAGE SQL;
394 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT ) RETURNS SETOF metabib.full_rec AS $func$
397 use MARC::File::XML (BinaryEncoding => 'UTF-8');
400 my $r = MARC::Record->new_from_xml( $xml );
402 return_next( { tag => 'LDR', value => $r->leader } );
404 for my $f ( $r->fields ) {
405 if ($f->is_control_field) {
406 return_next({ tag => $f->tag, value => $f->data });
408 for my $s ($f->subfields) {
411 ind1 => $f->indicator(1),
412 ind2 => $f->indicator(2),
417 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
418 my $trim = $f->indicator(2) || 0;
421 ind1 => $f->indicator(1),
422 ind2 => $f->indicator(2),
424 value => substr( $s->[1], $trim )
433 $func$ LANGUAGE PLPERLU;
435 CREATE OR REPLACE FUNCTION biblio.marc21_record_type( rid BIGINT ) RETURNS config.marc21_rec_type_map AS $func$
442 retval config.marc21_rec_type_map%ROWTYPE;
444 SELECT * INTO ldr FROM metabib.full_rec WHERE record = rid AND tag = 'LDR' LIMIT 1;
446 IF ldr.id IS NULL THEN
447 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
451 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
452 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
455 tval := SUBSTRING( ldr.value, tval_rec.start_pos + 1, tval_rec.length );
456 bval := SUBSTRING( ldr.value, bval_rec.start_pos + 1, bval_rec.length );
458 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr.value;
460 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
463 IF retval.code IS NULL THEN
464 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
469 $func$ LANGUAGE PLPGSQL;
471 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
478 rtype := (biblio.marc21_record_type( rid )).code;
479 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
480 FOR tag_data IN SELECT * FROM metabib.full_rec WHERE tag = UPPER(ff_pos.tag) AND record = rid LOOP
481 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
484 val := REPEAT( ff_pos.default_val, ff_pos.length );
490 $func$ LANGUAGE PLPGSQL;
492 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
493 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
497 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
498 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
499 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
500 retval biblio.marc21_physical_characteristics%ROWTYPE;
503 SELECT * INTO _007 FROM metabib.full_rec WHERE record = rid AND tag = '007' LIMIT 1;
505 IF _007.id IS NOT NULL THEN
506 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007.value, 1, 1 );
508 IF ptype.ptype_key IS NOT NULL THEN
509 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
510 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 );
512 IF pval.id IS NOT NULL THEN
515 retval.record := rid;
516 retval.ptype := ptype.ptype_key;
517 retval.subfield := psf.id;
518 retval.value := pval.id;
528 $func$ LANGUAGE PLPGSQL;
530 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
543 IF marc IS NULL OR marc = '' THEN
547 -- First, the count of tags
548 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
550 -- now go through a bunch of pain to get the record type
551 IF best_type IS NOT NULL THEN
552 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
554 IF ldr IS NOT NULL THEN
555 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
556 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
559 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
560 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
562 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
564 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
566 IF type_map.code IS NOT NULL THEN
567 IF best_type = type_map.code THEN
568 qual := qual + qual / 2;
571 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
572 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
573 IF ff_tag_data = best_lang THEN
581 -- Now look for some quality metrics
583 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
588 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
595 $func$ LANGUAGE PLPGSQL;
597 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
599 idx config.biblio_fingerprint%ROWTYPE;
600 xfrm config.xml_transform%ROWTYPE;
602 transformed_xml TEXT;
604 xml_node_list TEXT[];
606 output_text TEXT := '';
609 IF marc IS NULL OR marc = '' THEN
613 -- Loop over the indexing entries
614 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
616 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
618 -- See if we can skip the XSLT ... it's expensive
619 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
620 -- Can't skip the transform
621 IF xfrm.xslt <> '---' THEN
622 transformed_xml := oils_xslt_process(marc,xfrm.xslt);
624 transformed_xml := marc;
627 prev_xfrm := xfrm.name;
630 raw_text := COALESCE(
638 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
647 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
648 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
650 IF idx.first_word IS TRUE THEN
651 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
654 output_text := output_text || REGEXP_REPLACE(raw_text, E'\\s+', '', 'g');
661 $func$ LANGUAGE PLPGSQL;
663 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
664 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
667 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
669 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
673 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
674 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
679 $func$ LANGUAGE PLPGSQL;
681 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
682 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
684 ind_data metabib.field_entry_template%ROWTYPE;
686 tmp_mr metabib.metarecord%ROWTYPE;
703 IF NEW.deleted IS TRUE THEN
704 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage
705 RETURN NEW; -- and we're done
708 IF TG_OP = 'UPDATE' THEN -- re-ingest?
709 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
711 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
715 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage
719 IF TG_OP = 'UPDATE' THEN -- Clean out the cruft
720 DELETE FROM metabib.title_field_entry WHERE source = NEW.id;
721 DELETE FROM metabib.author_field_entry WHERE source = NEW.id;
722 DELETE FROM metabib.subject_field_entry WHERE source = NEW.id;
723 DELETE FROM metabib.keyword_field_entry WHERE source = NEW.id;
724 DELETE FROM metabib.series_field_entry WHERE source = NEW.id;
725 DELETE FROM metabib.full_rec WHERE record = NEW.id;
726 DELETE FROM metabib.rec_descriptor WHERE record = NEW.id;
730 -- Shove the flattened MARC in
731 INSERT INTO metabib.full_rec (record, tag, ind1, ind2, subfield, value)
732 SELECT record, tag, ind1, ind2, subfield, value FROM biblio.flatten_marc( NEW.id );
734 -- And now the indexing data
735 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( NEW.id ) LOOP
736 IF ind_data.field < 0 THEN
738 ind_data.field = -1 * ind_data.field;
743 IF ind_data.field_class = 'title' THEN
744 INSERT INTO metabib.title_field_entry (field, source, value, index_vector)
745 VALUES (ind_data.field, ind_data.source, ind_data.value, ind_vector);
746 ELSIF ind_data.field_class = 'author' THEN
747 INSERT INTO metabib.author_field_entry (field, source, value, index_vector)
748 VALUES (ind_data.field, ind_data.source, ind_data.value, ind_vector);
749 ELSIF ind_data.field_class = 'subject' THEN
750 INSERT INTO metabib.subject_field_entry (field, source, value, index_vector)
751 VALUES (ind_data.field, ind_data.source, ind_data.value, ind_vector);
752 ELSIF ind_data.field_class = 'keyword' THEN
753 INSERT INTO metabib.keyword_field_entry (field, source, value, index_vector)
754 VALUES (ind_data.field, ind_data.source, ind_data.value, ind_vector);
755 ELSIF ind_data.field_class = 'series' THEN
756 INSERT INTO metabib.series_field_entry (field, source, value, index_vector)
757 VALUES (ind_data.field, ind_data.source, ind_data.value, ind_vector);
761 -- Then, the rec_descriptor
762 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)
764 biblio.marc21_extract_fixed_field( NEW.id, 'Type' ),
765 biblio.marc21_extract_fixed_field( NEW.id, 'Form' ),
766 biblio.marc21_extract_fixed_field( NEW.id, 'BLvl' ),
767 biblio.marc21_extract_fixed_field( NEW.id, 'Ctrl' ),
768 biblio.marc21_extract_fixed_field( NEW.id, 'ELvl' ),
769 biblio.marc21_extract_fixed_field( NEW.id, 'Audn' ),
770 biblio.marc21_extract_fixed_field( NEW.id, 'LitF' ),
771 biblio.marc21_extract_fixed_field( NEW.id, 'TMat' ),
772 biblio.marc21_extract_fixed_field( NEW.id, 'Desc' ),
773 biblio.marc21_extract_fixed_field( NEW.id, 'DtSt' ),
774 biblio.marc21_extract_fixed_field( NEW.id, 'Lang' ),
776 FROM biblio.marc21_physical_characteristics( NEW.id) p
777 JOIN config.marc21_physical_characteristic_subfield_map s ON (s.id = p.subfield)
778 JOIN config.marc21_physical_characteristic_value_map v ON (v.id = p.value)
779 WHERE p.ptype = 'v' AND s.subfield = 'e' ),
780 biblio.marc21_extract_fixed_field( NEW.id, 'Date1'),
781 biblio.marc21_extract_fixed_field( NEW.id, 'Date2');
784 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.skip_located_uri' AND enabled;
786 IF NOT FOUND OR TG_OP = 'INSERT' THEN
787 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',NEW.marc);
788 IF ARRAY_UPPER(uris,1) > 0 THEN
789 FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
790 -- First we pull infot out of the 856
793 uri_href := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
794 CONTINUE WHEN uri_href IS NULL;
796 uri_label := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()|//*[@code="u"]/text()',uri_xml))[1];
797 CONTINUE WHEN uri_label IS NULL;
799 uri_owner := (oils_xpath('//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',uri_xml))[1];
800 CONTINUE WHEN uri_owner IS NULL;
802 uri_use := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()',uri_xml))[1];
804 uri_owner := REGEXP_REPLACE(uri_owner, $re$^.*?\((\w+)\).*$$re$, E'\\1');
806 SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = uri_owner;
807 CONTINUE WHEN NOT FOUND;
809 -- now we look for a matching uri
810 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
811 IF NOT FOUND THEN -- create one
812 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
813 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
816 -- we need a call number to link through
817 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;
819 INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
820 VALUES (uri_owner_id, NEW.id, 'now', 'now', NEW.editor, NEW.editor, '##URI##');
821 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;
824 -- now, link them if they're not already
825 SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
827 INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
834 -- And, finally, metarecord mapping!
836 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
838 IF NOT FOUND OR TG_OP = 'UPDATE' THEN
839 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
841 IF old_mr IS NULL AND NEW.fingerprint = tmp_mr.fingerprint THEN -- Find the first fingerprint-matching
844 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
845 IF source_count = 0 THEN -- No other records
846 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
847 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
853 IF old_mr IS NULL THEN -- we found no suitable, preexisting MR based on old source maps
854 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = NEW.fingerprint; -- is there one for our current fingerprint?
855 IF old_mr IS NULL THEN -- nope, create one and grab its id
856 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( NEW.fingerprint, NEW.id );
857 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = NEW.fingerprint;
858 ELSE -- indeed there is. update it with a null cache and recalcualated master record
859 UPDATE metabib.metarecord
861 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = NEW.fingerprint ORDER BY quality DESC LIMIT 1)
864 ELSE -- there was one we already attached to, update its mods cache and master_record
865 UPDATE metabib.metarecord
867 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = NEW.fingerprint ORDER BY quality DESC LIMIT 1)
871 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, NEW.id); -- new source mapping
873 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
879 $func$ LANGUAGE PLPGSQL;