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 /* We may not need these...
134 CREATE INDEX metabib_rec_descriptor_item_type_idx ON metabib.rec_descriptor (item_type);
135 CREATE INDEX metabib_rec_descriptor_item_form_idx ON metabib.rec_descriptor (item_form);
136 CREATE INDEX metabib_rec_descriptor_bib_level_idx ON metabib.rec_descriptor (bib_level);
137 CREATE INDEX metabib_rec_descriptor_control_type_idx ON metabib.rec_descriptor (control_type);
138 CREATE INDEX metabib_rec_descriptor_char_encoding_idx ON metabib.rec_descriptor (char_encoding);
139 CREATE INDEX metabib_rec_descriptor_enc_level_idx ON metabib.rec_descriptor (enc_level);
140 CREATE INDEX metabib_rec_descriptor_audience_idx ON metabib.rec_descriptor (audience);
141 CREATE INDEX metabib_rec_descriptor_lit_form_idx ON metabib.rec_descriptor (lit_form);
142 CREATE INDEX metabib_rec_descriptor_cat_form_idx ON metabib.rec_descriptor (cat_form);
143 CREATE INDEX metabib_rec_descriptor_pub_status_idx ON metabib.rec_descriptor (pub_status);
144 CREATE INDEX metabib_rec_descriptor_item_lang_idx ON metabib.rec_descriptor (item_lang);
145 CREATE INDEX metabib_rec_descriptor_vr_format_idx ON metabib.rec_descriptor (vr_format);
149 -- Use a sequence that matches previous version, for easier upgrading.
150 CREATE SEQUENCE metabib.full_rec_id_seq;
152 CREATE TABLE metabib.real_full_rec (
153 id BIGINT NOT NULL DEFAULT NEXTVAL('metabib.full_rec_id_seq'::REGCLASS),
154 record BIGINT NOT NULL,
155 tag CHAR(3) NOT NULL,
160 index_vector tsvector NOT NULL
162 ALTER TABLE metabib.real_full_rec ADD PRIMARY KEY (id);
164 CREATE INDEX metabib_full_rec_tag_subfield_idx ON metabib.real_full_rec (tag,subfield);
165 CREATE INDEX metabib_full_rec_value_idx ON metabib.real_full_rec (substring(value,1,1024));
166 /* Enable LIKE to use an index for database clusters with locales other than C or POSIX */
167 CREATE INDEX metabib_full_rec_value_tpo_index ON metabib.real_full_rec (substring(value,1,1024) text_pattern_ops);
168 CREATE INDEX metabib_full_rec_record_idx ON metabib.real_full_rec (record);
169 CREATE INDEX metabib_full_rec_index_vector_idx ON metabib.real_full_rec USING GIST (index_vector);
171 CREATE TRIGGER metabib_full_rec_fti_trigger
172 BEFORE UPDATE OR INSERT ON metabib.real_full_rec
173 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('default');
175 CREATE OR REPLACE VIEW metabib.full_rec AS
182 SUBSTRING(value,1,1024) AS value,
184 FROM metabib.real_full_rec;
186 CREATE OR REPLACE RULE metabib_full_rec_insert_rule
187 AS ON INSERT TO metabib.full_rec
189 INSERT INTO metabib.real_full_rec VALUES (
190 COALESCE(NEW.id, NEXTVAL('metabib.full_rec_id_seq'::REGCLASS)),
200 CREATE OR REPLACE RULE metabib_full_rec_update_rule
201 AS ON UPDATE TO metabib.full_rec
203 UPDATE metabib.real_full_rec SET
209 subfield = NEW.subfield,
211 index_vector = NEW.index_vector
214 CREATE OR REPLACE RULE metabib_full_rec_delete_rule
215 AS ON DELETE TO metabib.full_rec
217 DELETE FROM metabib.real_full_rec WHERE id = OLD.id;
219 CREATE TABLE metabib.metarecord_source_map (
220 id BIGSERIAL PRIMARY KEY,
221 metarecord BIGINT NOT NULL,
222 source BIGINT NOT NULL
224 CREATE INDEX metabib_metarecord_source_map_metarecord_idx ON metabib.metarecord_source_map (metarecord);
225 CREATE INDEX metabib_metarecord_source_map_source_record_idx ON metabib.metarecord_source_map (source);
227 CREATE TYPE metabib.field_entry_template AS (
234 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
236 bib biblio.record_entry%ROWTYPE;
237 idx config.metabib_field%ROWTYPE;
238 xfrm config.xml_transform%ROWTYPE;
240 transformed_xml TEXT;
242 xml_node_list TEXT[];
245 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
246 output_row metabib.field_entry_template%ROWTYPE;
250 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
252 -- Loop over the indexing entries
253 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
255 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
257 -- See if we can skip the XSLT ... it's expensive
258 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
259 -- Can't skip the transform
260 IF xfrm.xslt <> '---' THEN
261 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
263 transformed_xml := bib.marc;
266 prev_xfrm := xfrm.name;
269 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
272 FOR xml_node IN SELECT x FROM explode_array(xml_node_list) AS x LOOP
273 CONTINUE WHEN xml_node !~ E'^\\s*<';
275 curr_text := ARRAY_TO_STRING(
276 oils_xpath( '//text()',
277 REGEXP_REPLACE( -- This escapes all &s not followed by "amp;". Data ise returned from oils_xpath (above) in UTF-8, not entity encoded
278 REGEXP_REPLACE( -- This escapes embeded <s
280 $re$(>[^<]+)(<)([^>]+<)$re$,
292 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
294 IF raw_text IS NOT NULL THEN
295 raw_text := raw_text || joiner;
298 raw_text := COALESCE(raw_text,'') || curr_text;
300 -- insert raw node text for faceting
301 IF idx.facet_field THEN
303 output_row.field_class = idx.field_class;
304 output_row.field = -1 *idx.id;
305 output_row.source = rid;
306 output_row.value = BTRIM(REGEXP_REPLACE(curr_text, E'\\s+', ' ', 'g'));
308 RETURN NEXT output_row;
313 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
315 -- insert combined node text for searching
316 IF idx.search_field THEN
317 output_row.field_class = idx.field_class;
318 output_row.field = idx.id;
319 output_row.source = rid;
320 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
322 RETURN NEXT output_row;
328 $func$ LANGUAGE PLPGSQL;
330 -- default to a space joiner
331 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
332 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
335 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
337 bib biblio.record_entry%ROWTYPE;
338 output metabib.full_rec%ROWTYPE;
341 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
343 FOR field IN SELECT * FROM biblio.flatten_marc( bib.marc ) LOOP
344 output.record := rid;
345 output.ind1 := field.ind1;
346 output.ind2 := field.ind2;
347 output.tag := field.tag;
348 output.subfield := field.subfield;
349 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
350 output.value := naco_normalize(field.value, field.subfield);
352 output.value := field.value;
355 CONTINUE WHEN output.value IS NULL;
360 $func$ LANGUAGE PLPGSQL;
362 /* Old form of biblio.flatten_marc() relied on contrib/xml2 functions that got all crashy in PostgreSQL 8.4 */
363 -- CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT, BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
364 -- 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
366 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), NULL::TEXT, NULL::TEXT, NULL::TEXT, x.value, NULL::tsvector AS index_vector
367 -- FROM oils_xpath_table(
370 -- 'biblio.record_entry',
371 -- '//*[local-name()="controlfield"]/@tag|//*[local-name()="controlfield"]',
373 -- )x(record int, tag text, value text)
375 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), x.ind1, x.ind2, x.subfield, x.value, NULL::tsvector AS index_vector
376 -- FROM oils_xpath_table(
379 -- 'biblio.record_entry',
380 -- '//*[local-name()="datafield"]/@tag|' ||
381 -- '//*[local-name()="datafield"]/@ind1|' ||
382 -- '//*[local-name()="datafield"]/@ind2|' ||
383 -- '//*[local-name()="datafield"]/*/@code|' ||
384 -- '//*[local-name()="datafield"]/*[@code]',
386 -- )x(record int, tag text, ind1 text, ind2 text, subfield text, value text);
387 -- $func$ LANGUAGE SQL;
389 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT ) RETURNS SETOF metabib.full_rec AS $func$
392 use MARC::File::XML (BinaryEncoding => 'UTF-8');
395 my $r = MARC::Record->new_from_xml( $xml );
397 return_next( { tag => 'LDR', value => $r->leader } );
399 for my $f ( $r->fields ) {
400 if ($f->is_control_field) {
401 return_next({ tag => $f->tag, value => $f->data });
403 for my $s ($f->subfields) {
406 ind1 => $f->indicator(1),
407 ind2 => $f->indicator(2),
412 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
413 my $trim = $f->indicator(2) || 0;
416 ind1 => $f->indicator(1),
417 ind2 => $f->indicator(2),
419 value => substr( $s->[1], $trim )
428 $func$ LANGUAGE PLPERLU;
430 CREATE OR REPLACE FUNCTION biblio.marc21_record_type( rid BIGINT ) RETURNS config.marc21_rec_type_map AS $func$
437 retval config.marc21_rec_type_map%ROWTYPE;
439 SELECT * INTO ldr FROM metabib.full_rec WHERE record = rid AND tag = 'LDR' LIMIT 1;
441 IF ldr.id IS NULL THEN
442 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
446 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
447 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
450 tval := SUBSTRING( ldr.value, tval_rec.start_pos + 1, tval_rec.length );
451 bval := SUBSTRING( ldr.value, bval_rec.start_pos + 1, bval_rec.length );
453 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr.value;
455 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
458 IF retval.code IS NULL THEN
459 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
464 $func$ LANGUAGE PLPGSQL;
466 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
473 rtype := (biblio.marc21_record_type( rid )).code;
474 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
475 FOR tag_data IN SELECT * FROM metabib.full_rec WHERE tag = UPPER(ff_pos.tag) AND record = rid LOOP
476 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
479 val := REPEAT( ff_pos.default_val, ff_pos.length );
485 $func$ LANGUAGE PLPGSQL;
487 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
488 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
492 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
493 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
494 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
495 retval biblio.marc21_physical_characteristics%ROWTYPE;
498 SELECT * INTO _007 FROM metabib.full_rec WHERE record = rid AND tag = '007' LIMIT 1;
500 IF _007.id IS NOT NULL THEN
501 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007.value, 1, 1 );
503 IF ptype.ptype_key IS NOT NULL THEN
504 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
505 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 );
507 IF pval.id IS NOT NULL THEN
510 retval.record := rid;
511 retval.ptype := ptype.ptype_key;
512 retval.subfield := psf.id;
513 retval.value := pval.id;
523 $func$ LANGUAGE PLPGSQL;
525 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
538 IF marc IS NULL OR marc = '' THEN
542 -- First, the count of tags
543 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
545 -- now go through a bunch of pain to get the record type
546 IF best_type IS NOT NULL THEN
547 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
549 IF ldr IS NOT NULL THEN
550 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
551 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
554 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
555 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
557 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
559 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
561 IF type_map.code IS NOT NULL THEN
562 IF best_type = type_map.code THEN
563 qual := qual + qual / 2;
566 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
567 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
568 IF ff_tag_data = best_lang THEN
576 -- Now look for some quality metrics
578 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
583 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
590 $func$ LANGUAGE PLPGSQL;
592 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
594 idx config.biblio_fingerprint%ROWTYPE;
595 xfrm config.xml_transform%ROWTYPE;
597 transformed_xml TEXT;
599 xml_node_list TEXT[];
601 output_text TEXT := '';
604 IF marc IS NULL OR marc = '' THEN
608 -- Loop over the indexing entries
609 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
611 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
613 -- See if we can skip the XSLT ... it's expensive
614 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
615 -- Can't skip the transform
616 IF xfrm.xslt <> '---' THEN
617 transformed_xml := oils_xslt_process(marc,xfrm.xslt);
619 transformed_xml := marc;
622 prev_xfrm := xfrm.name;
625 raw_text := COALESCE(
633 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
642 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
643 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
645 IF idx.first_word IS TRUE THEN
646 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
649 output_text := output_text || REGEXP_REPLACE(raw_text, E'\\s+', '', 'g');
656 $func$ LANGUAGE PLPGSQL;
658 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
659 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
662 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
664 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
668 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
669 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
674 $func$ LANGUAGE PLPGSQL;
676 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
677 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
679 ind_data metabib.field_entry_template%ROWTYPE;
681 tmp_mr metabib.metarecord%ROWTYPE;
698 IF NEW.deleted IS TRUE THEN
699 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage
700 RETURN NEW; -- and we're done
703 IF TG_OP = 'UPDATE' THEN -- re-ingest?
704 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
706 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
710 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage
714 IF TG_OP = 'UPDATE' THEN -- Clean out the cruft
715 DELETE FROM metabib.title_field_entry WHERE source = NEW.id;
716 DELETE FROM metabib.author_field_entry WHERE source = NEW.id;
717 DELETE FROM metabib.subject_field_entry WHERE source = NEW.id;
718 DELETE FROM metabib.keyword_field_entry WHERE source = NEW.id;
719 DELETE FROM metabib.series_field_entry WHERE source = NEW.id;
720 DELETE FROM metabib.full_rec WHERE record = NEW.id;
721 DELETE FROM metabib.rec_descriptor WHERE record = NEW.id;
725 -- Shove the flattened MARC in
726 INSERT INTO metabib.full_rec (record, tag, ind1, ind2, subfield, value)
727 SELECT record, tag, ind1, ind2, subfield, value FROM biblio.flatten_marc( NEW.id );
729 -- And now the indexing data
730 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( NEW.id ) LOOP
731 IF ind_data.field < 0 THEN
733 ind_data.field = -1 * ind_data.field;
738 IF ind_data.field_class = 'title' THEN
739 INSERT INTO metabib.title_field_entry (field, source, value, index_vector)
740 VALUES (ind_data.field, ind_data.source, ind_data.value, ind_vector);
741 ELSIF ind_data.field_class = 'author' THEN
742 INSERT INTO metabib.author_field_entry (field, source, value, index_vector)
743 VALUES (ind_data.field, ind_data.source, ind_data.value, ind_vector);
744 ELSIF ind_data.field_class = 'subject' THEN
745 INSERT INTO metabib.subject_field_entry (field, source, value, index_vector)
746 VALUES (ind_data.field, ind_data.source, ind_data.value, ind_vector);
747 ELSIF ind_data.field_class = 'keyword' THEN
748 INSERT INTO metabib.keyword_field_entry (field, source, value, index_vector)
749 VALUES (ind_data.field, ind_data.source, ind_data.value, ind_vector);
750 ELSIF ind_data.field_class = 'series' THEN
751 INSERT INTO metabib.series_field_entry (field, source, value, index_vector)
752 VALUES (ind_data.field, ind_data.source, ind_data.value, ind_vector);
756 -- Then, the rec_descriptor
757 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)
759 biblio.marc21_extract_fixed_field( NEW.id, 'Type' ),
760 biblio.marc21_extract_fixed_field( NEW.id, 'Form' ),
761 biblio.marc21_extract_fixed_field( NEW.id, 'BLvl' ),
762 biblio.marc21_extract_fixed_field( NEW.id, 'Ctrl' ),
763 biblio.marc21_extract_fixed_field( NEW.id, 'ELvl' ),
764 biblio.marc21_extract_fixed_field( NEW.id, 'Audn' ),
765 biblio.marc21_extract_fixed_field( NEW.id, 'LitF' ),
766 biblio.marc21_extract_fixed_field( NEW.id, 'TMat' ),
767 biblio.marc21_extract_fixed_field( NEW.id, 'Desc' ),
768 biblio.marc21_extract_fixed_field( NEW.id, 'DtSt' ),
769 biblio.marc21_extract_fixed_field( NEW.id, 'Lang' ),
771 FROM biblio.marc21_physical_characteristics( NEW.id) p
772 JOIN config.marc21_physical_characteristic_subfield_map s ON (s.id = p.subfield)
773 JOIN config.marc21_physical_characteristic_value_map v ON (v.id = p.value)
774 WHERE p.ptype = 'v' AND s.subfield = 'e' ),
775 biblio.marc21_extract_fixed_field( NEW.id, 'Date1'),
776 biblio.marc21_extract_fixed_field( NEW.id, 'Date2');
779 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.skip_located_uri' AND enabled;
781 IF NOT FOUND OR TG_OP = 'INSERT' THEN
782 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',NEW.marc);
783 IF ARRAY_UPPER(uris,1) > 0 THEN
784 FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
785 -- First we pull infot 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 = NEW.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, NEW.id, 'now', 'now', NEW.editor, NEW.editor, '##URI##');
816 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 -- 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);
829 -- And, finally, metarecord mapping!
831 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
833 IF NOT FOUND OR TG_OP = 'UPDATE' THEN
834 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
836 IF old_mr IS NULL AND NEW.fingerprint = tmp_mr.fingerprint THEN -- Find the first fingerprint-matching
839 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
840 IF source_count = 0 THEN -- No other records
841 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
842 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
848 IF old_mr IS NULL THEN -- we found no suitable, preexisting MR based on old source maps
849 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = NEW.fingerprint; -- is there one for our current fingerprint?
850 IF old_mr IS NULL THEN -- nope, create one and grab its id
851 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( NEW.fingerprint, NEW.id );
852 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = NEW.fingerprint;
853 ELSE -- indeed there is. update it with a null cache and recalcualated master record
854 UPDATE metabib.metarecord
856 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = NEW.fingerprint ORDER BY quality DESC LIMIT 1)
859 ELSE -- there was one we already attached to, update its mods cache and master_record
860 UPDATE metabib.metarecord
862 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = NEW.fingerprint ORDER BY quality DESC LIMIT 1)
866 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, NEW.id); -- new source mapping
868 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
874 $func$ LANGUAGE PLPGSQL;