2 * Copyright (C) 2004-2008 Georgia Public Library Service
3 * Copyright (C) 2007-2008 Equinox Software, Inc.
4 * Mike Rylander <miker@esilibrary.com>
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public License
8 * as published by the Free Software Foundation; either version 2
9 * of the License, or (at your option) any later version.
11 * This program is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
18 DROP SCHEMA IF EXISTS metabib CASCADE;
21 CREATE SCHEMA metabib;
23 CREATE TABLE metabib.metarecord (
24 id BIGSERIAL PRIMARY KEY,
25 fingerprint TEXT NOT NULL,
29 CREATE INDEX metabib_metarecord_master_record_idx ON metabib.metarecord (master_record);
30 CREATE INDEX metabib_metarecord_fingerprint_idx ON metabib.metarecord (fingerprint);
32 CREATE TABLE metabib.identifier_field_entry (
33 id BIGSERIAL PRIMARY KEY,
34 source BIGINT NOT NULL,
37 index_vector tsvector NOT NULL
39 CREATE TRIGGER metabib_identifier_field_entry_fti_trigger
40 BEFORE UPDATE OR INSERT ON metabib.identifier_field_entry
41 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('identifier');
43 CREATE INDEX metabib_identifier_field_entry_index_vector_idx ON metabib.identifier_field_entry USING GIST (index_vector);
44 CREATE INDEX metabib_identifier_field_entry_value_idx ON metabib.identifier_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
45 CREATE INDEX metabib_identifier_field_entry_source_idx ON metabib.identifier_field_entry (source);
48 CREATE TABLE metabib.title_field_entry (
49 id BIGSERIAL PRIMARY KEY,
50 source BIGINT NOT NULL,
53 index_vector tsvector NOT NULL
55 CREATE TRIGGER metabib_title_field_entry_fti_trigger
56 BEFORE UPDATE OR INSERT ON metabib.title_field_entry
57 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('title');
59 CREATE INDEX metabib_title_field_entry_index_vector_idx ON metabib.title_field_entry USING GIST (index_vector);
60 CREATE INDEX metabib_title_field_entry_value_idx ON metabib.title_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
61 CREATE INDEX metabib_title_field_entry_source_idx ON metabib.title_field_entry (source);
64 CREATE TABLE metabib.author_field_entry (
65 id BIGSERIAL PRIMARY KEY,
66 source BIGINT NOT NULL,
69 index_vector tsvector NOT NULL
71 CREATE TRIGGER metabib_author_field_entry_fti_trigger
72 BEFORE UPDATE OR INSERT ON metabib.author_field_entry
73 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('author');
75 CREATE INDEX metabib_author_field_entry_index_vector_idx ON metabib.author_field_entry USING GIST (index_vector);
76 CREATE INDEX metabib_author_field_entry_value_idx ON metabib.author_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
77 CREATE INDEX metabib_author_field_entry_source_idx ON metabib.author_field_entry (source);
80 CREATE TABLE metabib.subject_field_entry (
81 id BIGSERIAL PRIMARY KEY,
82 source BIGINT NOT NULL,
85 index_vector tsvector NOT NULL
87 CREATE TRIGGER metabib_subject_field_entry_fti_trigger
88 BEFORE UPDATE OR INSERT ON metabib.subject_field_entry
89 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('subject');
91 CREATE INDEX metabib_subject_field_entry_index_vector_idx ON metabib.subject_field_entry USING GIST (index_vector);
92 CREATE INDEX metabib_subject_field_entry_value_idx ON metabib.subject_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
93 CREATE INDEX metabib_subject_field_entry_source_idx ON metabib.subject_field_entry (source);
96 CREATE TABLE metabib.keyword_field_entry (
97 id BIGSERIAL PRIMARY KEY,
98 source BIGINT NOT NULL,
101 index_vector tsvector NOT NULL
103 CREATE TRIGGER metabib_keyword_field_entry_fti_trigger
104 BEFORE UPDATE OR INSERT ON metabib.keyword_field_entry
105 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('keyword');
107 CREATE INDEX metabib_keyword_field_entry_index_vector_idx ON metabib.keyword_field_entry USING GIST (index_vector);
108 CREATE INDEX metabib_keyword_field_entry_value_idx ON metabib.keyword_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
109 CREATE INDEX metabib_keyword_field_entry_source_idx ON metabib.keyword_field_entry (source);
112 CREATE TABLE metabib.series_field_entry (
113 id BIGSERIAL PRIMARY KEY,
114 source BIGINT NOT NULL,
117 index_vector tsvector NOT NULL
119 CREATE TRIGGER metabib_series_field_entry_fti_trigger
120 BEFORE UPDATE OR INSERT ON metabib.series_field_entry
121 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('series');
123 CREATE INDEX metabib_series_field_entry_index_vector_idx ON metabib.series_field_entry USING GIST (index_vector);
124 CREATE INDEX metabib_series_field_entry_value_idx ON metabib.series_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
125 CREATE INDEX metabib_series_field_entry_source_idx ON metabib.series_field_entry (source);
128 CREATE TABLE metabib.facet_entry (
129 id BIGSERIAL PRIMARY KEY,
130 source BIGINT NOT NULL,
134 CREATE INDEX metabib_facet_entry_field_idx ON metabib.facet_entry (field);
135 CREATE INDEX metabib_facet_entry_value_idx ON metabib.facet_entry (SUBSTRING(value,1,1024));
136 CREATE INDEX metabib_facet_entry_source_idx ON metabib.facet_entry (source);
138 CREATE TABLE metabib.record_attr (
139 id BIGINT PRIMARY KEY REFERENCES biblio.record_entry (id) ON DELETE CASCADE,
140 attrs HSTORE NOT NULL DEFAULT ''::HSTORE
142 CREATE INDEX metabib_svf_attrs_idx ON metabib.record_attr USING GIST (attrs);
143 CREATE INDEX metabib_svf_date1_idx ON metabib.record_attr ((attrs->'date1'));
144 CREATE INDEX metabib_svf_dates_idx ON metabib.record_attr ((attrs->'date1'),(attrs->'date2'));
146 -- Back-compat view ... we're moving to an HSTORE world
147 CREATE TYPE metabib.rec_desc_type AS (
165 CREATE VIEW metabib.rec_descriptor AS
168 (populate_record(NULL::metabib.rec_desc_type, attrs)).*
169 FROM metabib.record_attr;
171 -- Use a sequence that matches previous version, for easier upgrading.
172 CREATE SEQUENCE metabib.full_rec_id_seq;
174 CREATE TABLE metabib.real_full_rec (
175 id BIGINT NOT NULL DEFAULT NEXTVAL('metabib.full_rec_id_seq'::REGCLASS),
176 record BIGINT NOT NULL,
177 tag CHAR(3) NOT NULL,
182 index_vector tsvector NOT NULL
184 ALTER TABLE metabib.real_full_rec ADD PRIMARY KEY (id);
186 CREATE INDEX metabib_full_rec_tag_subfield_idx ON metabib.real_full_rec (tag,subfield);
187 CREATE INDEX metabib_full_rec_value_idx ON metabib.real_full_rec (substring(value,1,1024));
188 /* Enable LIKE to use an index for database clusters with locales other than C or POSIX */
189 CREATE INDEX metabib_full_rec_value_tpo_index ON metabib.real_full_rec (substring(value,1,1024) text_pattern_ops);
190 CREATE INDEX metabib_full_rec_record_idx ON metabib.real_full_rec (record);
191 CREATE INDEX metabib_full_rec_index_vector_idx ON metabib.real_full_rec USING GIST (index_vector);
193 CREATE TRIGGER metabib_full_rec_fti_trigger
194 BEFORE UPDATE OR INSERT ON metabib.real_full_rec
195 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('default');
197 CREATE OR REPLACE VIEW metabib.full_rec AS
204 SUBSTRING(value,1,1024) AS value,
206 FROM metabib.real_full_rec;
208 CREATE OR REPLACE RULE metabib_full_rec_insert_rule
209 AS ON INSERT TO metabib.full_rec
211 INSERT INTO metabib.real_full_rec VALUES (
212 COALESCE(NEW.id, NEXTVAL('metabib.full_rec_id_seq'::REGCLASS)),
222 CREATE OR REPLACE RULE metabib_full_rec_update_rule
223 AS ON UPDATE TO metabib.full_rec
225 UPDATE metabib.real_full_rec SET
231 subfield = NEW.subfield,
233 index_vector = NEW.index_vector
236 CREATE OR REPLACE RULE metabib_full_rec_delete_rule
237 AS ON DELETE TO metabib.full_rec
239 DELETE FROM metabib.real_full_rec WHERE id = OLD.id;
241 CREATE TABLE metabib.metarecord_source_map (
242 id BIGSERIAL PRIMARY KEY,
243 metarecord BIGINT NOT NULL,
244 source BIGINT NOT NULL
246 CREATE INDEX metabib_metarecord_source_map_metarecord_idx ON metabib.metarecord_source_map (metarecord);
247 CREATE INDEX metabib_metarecord_source_map_source_record_idx ON metabib.metarecord_source_map (source);
249 CREATE TYPE metabib.field_entry_template AS (
256 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
258 bib biblio.record_entry%ROWTYPE;
259 idx config.metabib_field%ROWTYPE;
260 xfrm config.xml_transform%ROWTYPE;
262 transformed_xml TEXT;
264 xml_node_list TEXT[];
268 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
269 output_row metabib.field_entry_template%ROWTYPE;
273 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
275 -- Loop over the indexing entries
276 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
278 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
280 -- See if we can skip the XSLT ... it's expensive
281 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
282 -- Can't skip the transform
283 IF xfrm.xslt <> '---' THEN
284 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
286 transformed_xml := bib.marc;
289 prev_xfrm := xfrm.name;
292 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
295 FOR xml_node IN SELECT x FROM explode_array(xml_node_list) AS x LOOP
296 CONTINUE WHEN xml_node !~ E'^\\s*<';
298 curr_text := ARRAY_TO_STRING(
299 oils_xpath( '//text()',
300 REGEXP_REPLACE( -- This escapes all &s not followed by "amp;". Data ise returned from oils_xpath (above) in UTF-8, not entity encoded
301 REGEXP_REPLACE( -- This escapes embeded <s
303 $re$(>[^<]+)(<)([^>]+<)$re$,
315 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
317 IF raw_text IS NOT NULL THEN
318 raw_text := raw_text || joiner;
321 raw_text := COALESCE(raw_text,'') || curr_text;
323 -- insert raw node text for faceting
324 IF idx.facet_field THEN
326 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
327 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
329 facet_text := curr_text;
332 output_row.field_class = idx.field_class;
333 output_row.field = -1 * idx.id;
334 output_row.source = rid;
335 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
337 RETURN NEXT output_row;
342 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
344 -- insert combined node text for searching
345 IF idx.search_field THEN
346 output_row.field_class = idx.field_class;
347 output_row.field = idx.id;
348 output_row.source = rid;
349 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
351 RETURN NEXT output_row;
357 $func$ LANGUAGE PLPGSQL;
359 -- default to a space joiner
360 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
361 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
364 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
366 bib biblio.record_entry%ROWTYPE;
367 output metabib.full_rec%ROWTYPE;
370 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
372 FOR field IN SELECT * FROM biblio.flatten_marc( bib.marc ) LOOP
373 output.record := rid;
374 output.ind1 := field.ind1;
375 output.ind2 := field.ind2;
376 output.tag := field.tag;
377 output.subfield := field.subfield;
378 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
379 output.value := naco_normalize(field.value, field.subfield);
381 output.value := field.value;
384 CONTINUE WHEN output.value IS NULL;
389 $func$ LANGUAGE PLPGSQL;
391 /* Old form of biblio.flatten_marc() relied on contrib/xml2 functions that got all crashy in PostgreSQL 8.4 */
392 -- CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT, BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
393 -- 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
395 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), NULL::TEXT, NULL::TEXT, NULL::TEXT, x.value, NULL::tsvector AS index_vector
396 -- FROM oils_xpath_table(
399 -- 'biblio.record_entry',
400 -- '//*[local-name()="controlfield"]/@tag|//*[local-name()="controlfield"]',
402 -- )x(record int, tag text, value text)
404 -- SELECT NULL::bigint AS id, NULL::bigint, x.tag::char(3), x.ind1, x.ind2, x.subfield, x.value, NULL::tsvector AS index_vector
405 -- FROM oils_xpath_table(
408 -- 'biblio.record_entry',
409 -- '//*[local-name()="datafield"]/@tag|' ||
410 -- '//*[local-name()="datafield"]/@ind1|' ||
411 -- '//*[local-name()="datafield"]/@ind2|' ||
412 -- '//*[local-name()="datafield"]/*/@code|' ||
413 -- '//*[local-name()="datafield"]/*[@code]',
415 -- )x(record int, tag text, ind1 text, ind2 text, subfield text, value text);
416 -- $func$ LANGUAGE SQL;
418 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( TEXT ) RETURNS SETOF metabib.full_rec AS $func$
421 use MARC::File::XML (BinaryEncoding => 'UTF-8');
424 my $r = MARC::Record->new_from_xml( $xml );
426 return_next( { tag => 'LDR', value => $r->leader } );
428 for my $f ( $r->fields ) {
429 if ($f->is_control_field) {
430 return_next({ tag => $f->tag, value => $f->data });
432 for my $s ($f->subfields) {
435 ind1 => $f->indicator(1),
436 ind2 => $f->indicator(2),
441 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
442 my $trim = $f->indicator(2) || 0;
445 ind1 => $f->indicator(1),
446 ind2 => $f->indicator(2),
448 value => substr( $s->[1], $trim )
457 $func$ LANGUAGE PLPERLU;
459 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
466 retval config.marc21_rec_type_map%ROWTYPE;
468 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
470 IF ldr IS NULL OR ldr = '' THEN
471 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
475 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
476 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
479 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
480 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
482 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
484 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
487 IF retval.code IS NULL THEN
488 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
493 $func$ LANGUAGE PLPGSQL;
495 CREATE OR REPLACE FUNCTION biblio.marc21_record_type( rid BIGINT ) RETURNS config.marc21_rec_type_map AS $func$
496 SELECT * FROM vandelay.marc21_record_type( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
499 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
506 rtype := (vandelay.marc21_record_type( marc )).code;
507 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
508 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(tag) || '"]/text()', marc ) ) x(value) LOOP
509 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
512 val := REPEAT( ff_pos.default_val, ff_pos.length );
518 $func$ LANGUAGE PLPGSQL;
520 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
521 SELECT * FROM vandelay.marc21_extract_fixed_field( (SELECT marc FROM biblio.record_entry WHERE id = $1), $2 );
524 CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
525 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
530 output biblio.record_ff_map%ROWTYPE;
532 rtype := (vandelay.marc21_record_type( marc )).code;
534 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
535 output.ff_name := ff_pos.fixed_field;
536 output.ff_value := NULL;
538 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(tag) || '"]/text()', marc ) ) x(value) LOOP
539 output.ff_value := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
540 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
542 output.ff_value := NULL;
549 $func$ LANGUAGE PLPGSQL;
551 CREATE OR REPLACE FUNCTION biblio.marc21_extract_all_fixed_fields( rid BIGINT ) RETURNS SETOF biblio.record_ff_map AS $func$
552 SELECT $1 AS record, ff_name, ff_value FROM vandelay.marc21_extract_all_fixed_fields( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
555 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
556 CREATE OR REPLACE FUNCTION vandelay.marc21_physical_characteristics( marc TEXT) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
560 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
561 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
562 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
563 retval biblio.marc21_physical_characteristics%ROWTYPE;
566 _007 := oils_xpath_string( '//*[@tag="007"]', marc );
568 IF _007 IS NOT NULL AND _007 <> '' THEN
569 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007, 1, 1 );
571 IF ptype.ptype_key IS NOT NULL THEN
572 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
573 SELECT * INTO pval FROM config.marc21_physical_characteristic_value_map WHERE ptype_subfield = psf.id AND value = SUBSTRING( _007, psf.start_pos + 1, psf.length );
575 IF pval.id IS NOT NULL THEN
578 retval.ptype := ptype.ptype_key;
579 retval.subfield := psf.id;
580 retval.value := pval.id;
590 $func$ LANGUAGE PLPGSQL;
592 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
593 SELECT id, $1 AS record, ptype, subfield, value FROM vandelay.marc21_physical_characteristics( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
596 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
609 IF marc IS NULL OR marc = '' THEN
613 -- First, the count of tags
614 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
616 -- now go through a bunch of pain to get the record type
617 IF best_type IS NOT NULL THEN
618 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
620 IF ldr IS NOT NULL THEN
621 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
622 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
625 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
626 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
628 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
630 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
632 IF type_map.code IS NOT NULL THEN
633 IF best_type = type_map.code THEN
634 qual := qual + qual / 2;
637 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
638 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
639 IF ff_tag_data = best_lang THEN
647 -- Now look for some quality metrics
649 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
654 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
661 $func$ LANGUAGE PLPGSQL;
663 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
665 idx config.biblio_fingerprint%ROWTYPE;
666 xfrm config.xml_transform%ROWTYPE;
668 transformed_xml TEXT;
670 xml_node_list TEXT[];
672 output_text TEXT := '';
675 IF marc IS NULL OR marc = '' THEN
679 -- Loop over the indexing entries
680 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
682 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
684 -- See if we can skip the XSLT ... it's expensive
685 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
686 -- Can't skip the transform
687 IF xfrm.xslt <> '---' THEN
688 transformed_xml := oils_xslt_process(marc,xfrm.xslt);
690 transformed_xml := marc;
693 prev_xfrm := xfrm.name;
696 raw_text := COALESCE(
704 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
713 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
714 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
716 IF idx.first_word IS TRUE THEN
717 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
720 output_text := output_text || REGEXP_REPLACE(raw_text, E'\\s+', '', 'g');
727 $func$ LANGUAGE PLPGSQL;
729 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
730 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
733 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
735 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
739 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
740 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
745 $func$ LANGUAGE PLPGSQL;
747 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_full_rec( bib_id BIGINT ) RETURNS VOID AS $func$
749 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
751 DELETE FROM metabib.real_full_rec WHERE record = bib_id;
753 INSERT INTO metabib.real_full_rec (record, tag, ind1, ind2, subfield, value)
754 SELECT record, tag, ind1, ind2, subfield, value FROM biblio.flatten_marc( bib_id );
758 $func$ LANGUAGE PLPGSQL;
760 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries( bib_id BIGINT ) RETURNS VOID AS $func$
763 ind_data metabib.field_entry_template%ROWTYPE;
765 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
767 FOR fclass IN SELECT * FROM config.metabib_class LOOP
768 -- RAISE NOTICE 'Emptying out %', fclass.name;
769 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
771 DELETE FROM metabib.facet_entry WHERE source = bib_id;
774 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
775 IF ind_data.field < 0 THEN
776 ind_data.field = -1 * ind_data.field;
777 INSERT INTO metabib.facet_entry (field, source, value)
778 VALUES (ind_data.field, ind_data.source, ind_data.value);
781 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
783 quote_literal(ind_data.field) || $$, $$ ||
784 quote_literal(ind_data.source) || $$, $$ ||
785 quote_literal(ind_data.value) ||
793 $func$ LANGUAGE PLPGSQL;
795 CREATE OR REPLACE FUNCTION biblio.extract_located_uris( bib_id BIGINT, marcxml TEXT, editor_id INT ) RETURNS VOID AS $func$
809 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',marcxml);
810 IF ARRAY_UPPER(uris,1) > 0 THEN
811 FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
812 -- First we pull info out of the 856
815 uri_href := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
816 CONTINUE WHEN uri_href IS NULL;
818 uri_label := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()|//*[@code="u"]/text()',uri_xml))[1];
819 CONTINUE WHEN uri_label IS NULL;
821 uri_owner := (oils_xpath('//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',uri_xml))[1];
822 CONTINUE WHEN uri_owner IS NULL;
824 uri_use := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()|//*[@code="u"]/text()',uri_xml))[1];
826 uri_owner := REGEXP_REPLACE(uri_owner, $re$^.*?\((\w+)\).*$$re$, E'\\1');
828 SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = uri_owner;
829 CONTINUE WHEN NOT FOUND;
831 -- now we look for a matching uri
832 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
833 IF NOT FOUND THEN -- create one
834 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
835 SELECT id INTO uri_id FROM asset.uri WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
838 -- we need a call number to link through
839 SELECT id INTO uri_cn_id FROM asset.call_number WHERE owning_lib = uri_owner_id AND record = bib_id AND label = '##URI##' AND NOT deleted;
841 INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
842 VALUES (uri_owner_id, bib_id, 'now', 'now', editor_id, editor_id, '##URI##');
843 SELECT id INTO uri_cn_id FROM asset.call_number WHERE owning_lib = uri_owner_id AND record = bib_id AND label = '##URI##' AND NOT deleted;
846 -- now, link them if they're not already
847 SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
849 INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
857 $func$ LANGUAGE PLPGSQL;
860 CREATE OR REPLACE FUNCTION metabib.remap_metarecord_for_bib( bib_id BIGINT, fp TEXT ) RETURNS BIGINT AS $func$
864 tmp_mr metabib.metarecord%ROWTYPE;
865 deleted_mrs BIGINT[];
868 DELETE FROM metabib.metarecord_source_map WHERE source = bib_id; -- Rid ourselves of the search-estimate-killing linkage
870 FOR tmp_mr IN SELECT m.* FROM metabib.metarecord m JOIN metabib.metarecord_source_map s ON (s.metarecord = m.id) WHERE s.source = bib_id LOOP
872 IF old_mr IS NULL AND fp = tmp_mr.fingerprint THEN -- Find the first fingerprint-matching
875 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
876 IF source_count = 0 THEN -- No other records
877 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
878 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
884 IF old_mr IS NULL THEN -- we found no suitable, preexisting MR based on old source maps
885 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp; -- is there one for our current fingerprint?
886 IF old_mr IS NULL THEN -- nope, create one and grab its id
887 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( fp, bib_id );
888 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp;
889 ELSE -- indeed there is. update it with a null cache and recalcualated master record
890 UPDATE metabib.metarecord
892 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
895 ELSE -- there was one we already attached to, update its mods cache and master_record
896 UPDATE metabib.metarecord
898 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp ORDER BY quality DESC LIMIT 1)
902 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, bib_id); -- new source mapping
904 IF ARRAY_UPPER(deleted_mrs,1) > 0 THEN
905 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
911 $func$ LANGUAGE PLPGSQL;
913 CREATE OR REPLACE FUNCTION biblio.map_authority_linking (bibid BIGINT, marc TEXT) RETURNS BIGINT AS $func$
914 DELETE FROM authority.bib_linking WHERE bib = $1;
915 INSERT INTO authority.bib_linking (bib, authority)
918 FROM ( SELECT DISTINCT $1 AS bib,
919 BTRIM(remove_paren_substring(txt))::BIGINT AS authority
920 FROM explode_array(oils_xpath('//*[@code="0"]/text()',$2)) x(txt)
921 WHERE BTRIM(remove_paren_substring(txt)) ~ $re$^\d+$$re$
922 ) y JOIN authority.record_entry r ON r.id = y.authority;
926 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
927 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
929 transformed_xml TEXT;
932 xfrm config.xml_transform%ROWTYPE;
934 new_attrs HSTORE := ''::HSTORE;
935 attr_def config.record_attr_definition%ROWTYPE;
938 IF NEW.deleted IS TRUE THEN -- If this bib is deleted
939 DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage
940 DELETE FROM metabib.record_attr WHERE id = NEW.id; -- Kill the attrs hash, useless on deleted records
941 DELETE FROM authority.bib_linking WHERE bib = NEW.id; -- Avoid updating fields in bibs that are no longer visible
942 RETURN NEW; -- and we're done
945 IF TG_OP = 'UPDATE' THEN -- re-ingest?
946 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
948 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
953 -- Record authority linking
954 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_linking' AND enabled;
956 PERFORM biblio.map_authority_linking( NEW.id, NEW.marc );
959 -- Flatten and insert the mfr data
960 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_full_rec' AND enabled;
962 PERFORM metabib.reingest_metabib_full_rec(NEW.id);
964 -- Now we pull out attribute data, which is dependent on the mfr for all but XPath-based fields
965 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_rec_descriptor' AND enabled;
967 FOR attr_def IN SELECT * FROM config.record_attr_definition ORDER BY format LOOP
969 IF attr_def.tag THEN -- tag (and optional subfield list) selection
970 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(value), COALESCE(attr_def.joiner,' ')) INTO attr_value
971 FROM (SELECT * FROM metabib.full_rec ORDER BY tag, subfield) AS x
972 WHERE record = NEW.id
973 AND tag LIKE attr_def.tag
975 WHEN attr_def.sf_list IS NOT NULL
976 THEN POSITION(subfield IN attr_def.sf_list) > 0
983 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
984 attr_value := biblio.marc21_extract_fixed_field(NEW.id, attr_def.fixed_field);
986 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
988 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
990 -- See if we can skip the XSLT ... it's expensive
991 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
992 -- Can't skip the transform
993 IF xfrm.xslt <> '---' THEN
994 transformed_xml := oils_xslt_process(NEW.marc,xfrm.xslt);
996 transformed_xml := NEW.marc;
999 prev_xfrm := xfrm.name;
1002 IF xfrm.name IS NULL THEN
1003 -- just grab the marcxml (empty) transform
1004 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
1005 prev_xfrm := xfrm.name;
1008 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
1010 ELSIF attr_def.phys_char IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
1011 SELECT value::TEXT INTO attr_value
1012 FROM biblio.marc21_physical_characteristics(NEW.id)
1013 WHERE subfield = attr_def.phys_char_sf
1014 LIMIT 1; -- Just in case ...
1018 -- apply index normalizers to attr_value
1020 SELECT n.func AS func,
1021 n.param_count AS param_count,
1023 FROM config.index_normalizer n
1024 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
1025 WHERE attr = attr_def.name
1027 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1028 quote_literal( attr_value ) ||
1030 WHEN normalizer.param_count > 0
1031 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1034 ')' INTO attr_value;
1038 -- Add the new value to the hstore
1039 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
1043 IF TG_OP = 'INSERT' OR OLD.deleted THEN -- initial insert OR revivication
1044 INSERT INTO metabib.record_attr (id, attrs) VALUES (NEW.id, new_attrs);
1046 UPDATE metabib.record_attr SET attrs = attrs || new_attrs WHERE id = NEW.id;
1052 -- Gather and insert the field entry data
1053 PERFORM metabib.reingest_metabib_field_entries(NEW.id);
1055 -- Located URI magic
1056 IF TG_OP = 'INSERT' THEN
1057 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1059 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1062 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1064 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1068 -- (re)map metarecord-bib linking
1069 IF TG_OP = 'INSERT' THEN -- if not deleted and performing an insert, check for the flag
1070 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
1072 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1074 ELSE -- we're doing an update, and we're not deleted, remap
1075 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_update' AND enabled;
1077 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1083 $func$ LANGUAGE PLPGSQL;