1 DROP SCHEMA IF EXISTS vandelay CASCADE;
5 CREATE SCHEMA vandelay;
7 CREATE TABLE vandelay.match_set (
10 owner INT NOT NULL REFERENCES actor.org_unit (id) ON DELETE CASCADE,
11 mtype TEXT NOT NULL DEFAULT 'biblio', -- 'biblio','authority','mfhd'?, others?
12 CONSTRAINT name_once_per_owner_mtype UNIQUE (name, owner, mtype)
15 -- Table to define match points, either FF via SVF or tag+subfield
16 CREATE TABLE vandelay.match_set_point (
17 id SERIAL PRIMARY KEY,
18 match_set INT REFERENCES vandelay.match_set (id) ON DELETE CASCADE,
19 parent INT REFERENCES vandelay.match_set_point (id),
20 bool_op TEXT CHECK (bool_op IS NULL OR (bool_op IN ('AND','OR','NOT'))),
21 svf TEXT REFERENCES config.record_attr_definition (name),
24 negate BOOL DEFAULT FALSE,
25 quality INT NOT NULL DEFAULT 1, -- higher is better
26 heading BOOLEAN NOT NULL DEFAULT FALSE, -- match on authority heading
27 CONSTRAINT vmsp_need_a_subfield_with_a_tag CHECK ((tag IS NOT NULL AND subfield IS NOT NULL) OR tag IS NULL),
28 CONSTRAINT vmsp_need_a_tag_or_a_ff_or_a_bo CHECK (
29 (tag IS NOT NULL AND svf IS NULL AND heading IS FALSE AND bool_op IS NULL) OR
30 (tag IS NULL AND svf IS NOT NULL AND heading IS FALSE AND bool_op IS NULL) OR
31 (tag IS NULL AND svf IS NULL AND heading IS TRUE AND bool_op IS NULL) OR
32 (tag IS NULL AND svf IS NULL AND heading IS FALSE AND bool_op IS NOT NULL)
36 CREATE TABLE vandelay.match_set_quality (
37 id SERIAL PRIMARY KEY,
38 match_set INT NOT NULL REFERENCES vandelay.match_set (id) ON DELETE CASCADE,
39 svf TEXT REFERENCES config.record_attr_definition,
43 quality INT NOT NULL DEFAULT 1, -- higher is better
44 CONSTRAINT vmsq_need_a_subfield_with_a_tag CHECK ((tag IS NOT NULL AND subfield IS NOT NULL) OR tag IS NULL),
45 CONSTRAINT vmsq_need_a_tag_or_a_ff CHECK ((tag IS NOT NULL AND svf IS NULL) OR (tag IS NULL AND svf IS NOT NULL))
47 CREATE UNIQUE INDEX vmsq_def_once_per_set ON vandelay.match_set_quality (match_set, COALESCE(tag,''), COALESCE(subfield,''), COALESCE(svf,''), value);
50 CREATE TABLE vandelay.queue (
51 id BIGSERIAL PRIMARY KEY,
52 owner INT NOT NULL REFERENCES actor.usr (id) DEFERRABLE INITIALLY DEFERRED,
54 complete BOOL NOT NULL DEFAULT FALSE,
55 match_set INT REFERENCES vandelay.match_set (id) ON UPDATE CASCADE ON DELETE SET NULL DEFERRABLE INITIALLY DEFERRED
58 CREATE TABLE vandelay.queued_record (
59 id BIGSERIAL PRIMARY KEY,
60 create_time TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT NOW(),
61 import_time TIMESTAMP WITH TIME ZONE,
62 purpose TEXT NOT NULL DEFAULT 'import' CHECK (purpose IN ('import','overlay')),
64 quality INT NOT NULL DEFAULT 0
69 /* Bib stuff at the top */
70 ----------------------------------------------------
72 CREATE TABLE vandelay.bib_attr_definition (
73 id SERIAL PRIMARY KEY,
74 code TEXT UNIQUE NOT NULL,
77 remove TEXT NOT NULL DEFAULT ''
80 -- Each TEXT field (other than 'name') should hold an XPath predicate for pulling the data needed
81 -- DROP TABLE vandelay.import_item_attr_definition CASCADE;
82 CREATE TABLE vandelay.import_item_attr_definition (
83 id BIGSERIAL PRIMARY KEY,
84 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
87 keep BOOL NOT NULL DEFAULT FALSE,
107 priv_note_title TEXT,
112 CONSTRAINT vand_import_item_attr_def_idx UNIQUE (owner,name)
115 CREATE TABLE vandelay.import_error (
116 code TEXT PRIMARY KEY,
117 description TEXT NOT NULL -- i18n
120 CREATE TYPE vandelay.bib_queue_queue_type AS ENUM ('bib', 'acq');
122 CREATE TABLE vandelay.bib_queue (
123 queue_type vandelay.bib_queue_queue_type NOT NULL DEFAULT 'bib',
124 item_attr_def BIGINT REFERENCES vandelay.import_item_attr_definition (id) ON DELETE SET NULL DEFERRABLE INITIALLY DEFERRED,
125 match_bucket INTEGER, -- REFERENCES container.biblio_record_entry_bucket(id);
126 CONSTRAINT vand_bib_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
127 ) INHERITS (vandelay.queue);
128 ALTER TABLE vandelay.bib_queue ADD PRIMARY KEY (id);
130 CREATE TABLE vandelay.queued_bib_record (
131 queue INT NOT NULL REFERENCES vandelay.bib_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
132 bib_source INT REFERENCES config.bib_source (id) DEFERRABLE INITIALLY DEFERRED,
133 imported_as BIGINT REFERENCES biblio.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
134 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
136 ) INHERITS (vandelay.queued_record);
137 ALTER TABLE vandelay.queued_bib_record ADD PRIMARY KEY (id);
138 CREATE INDEX queued_bib_record_queue_idx ON vandelay.queued_bib_record (queue);
140 CREATE TABLE vandelay.queued_bib_record_attr (
141 id BIGSERIAL PRIMARY KEY,
142 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) DEFERRABLE INITIALLY DEFERRED,
143 field INT NOT NULL REFERENCES vandelay.bib_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
144 attr_value TEXT NOT NULL
146 CREATE INDEX queued_bib_record_attr_record_idx ON vandelay.queued_bib_record_attr (record);
148 CREATE TABLE vandelay.bib_match (
149 id BIGSERIAL PRIMARY KEY,
150 queued_record BIGINT REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
151 eg_record BIGINT REFERENCES biblio.record_entry (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
152 quality INT NOT NULL DEFAULT 1,
153 match_score INT NOT NULL DEFAULT 0
155 CREATE INDEX bib_match_queued_record_idx ON vandelay.bib_match (queued_record);
157 CREATE TABLE vandelay.import_item (
158 id BIGSERIAL PRIMARY KEY,
159 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
160 definition BIGINT NOT NULL REFERENCES vandelay.import_item_attr_definition (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
161 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
164 import_time TIMESTAMP WITH TIME ZONE,
173 deposit_amount NUMERIC(8,2),
186 internal_id BIGINT -- queue_type == 'acq' ? acq.lineitem_detail.id : asset.copy.id
188 CREATE INDEX import_item_record_idx ON vandelay.import_item (record);
190 CREATE TABLE vandelay.import_bib_trash_group(
191 id SERIAL PRIMARY KEY,
192 owner INTEGER NOT NULL REFERENCES actor.org_unit(id),
193 label TEXT NOT NULL, --i18n
194 always_apply BOOLEAN NOT NULL DEFAULT FALSE,
195 CONSTRAINT vand_import_bib_trash_grp_owner_label UNIQUE (owner, label)
198 CREATE TABLE vandelay.import_bib_trash_fields (
199 id BIGSERIAL PRIMARY KEY,
200 grp INTEGER NOT NULL REFERENCES vandelay.import_bib_trash_group,
202 CONSTRAINT vand_import_bib_trash_fields_once_per UNIQUE (grp, field)
205 CREATE TABLE vandelay.merge_profile (
206 id BIGSERIAL PRIMARY KEY,
207 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
213 update_bib_source BOOLEAN NOT NULL DEFAULT FALSE,
215 CONSTRAINT vand_merge_prof_owner_name_idx UNIQUE (owner,name),
216 CONSTRAINT add_replace_strip_or_preserve CHECK ((preserve_spec IS NOT NULL OR replace_spec IS NOT NULL) OR (preserve_spec IS NULL AND replace_spec IS NULL))
219 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
226 retval config.marc21_rec_type_map%ROWTYPE;
228 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
230 IF ldr IS NULL OR ldr = '' THEN
231 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
235 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
236 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
239 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
240 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
242 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
244 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
247 IF retval.code IS NULL THEN
248 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
253 $func$ LANGUAGE PLPGSQL;
255 CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
256 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT, use_default BOOL DEFAULT FALSE ) RETURNS SETOF biblio.record_ff_map AS $func$
261 output biblio.record_ff_map%ROWTYPE;
263 rtype := (vandelay.marc21_record_type( marc )).code;
265 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
266 output.ff_name := ff_pos.fixed_field;
267 output.ff_value := NULL;
269 IF ff_pos.tag = 'ldr' THEN
270 output.ff_value := oils_xpath_string('//*[local-name()="leader"]', marc);
271 IF output.ff_value IS NOT NULL THEN
272 output.ff_value := SUBSTRING( output.ff_value, ff_pos.start_pos + 1, ff_pos.length );
274 output.ff_value := NULL;
277 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
278 output.ff_value := SUBSTRING( tag_data, ff_pos.start_pos + 1, ff_pos.length );
279 CONTINUE WHEN output.ff_value IS NULL AND NOT use_default;
280 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
282 output.ff_value := NULL;
290 $func$ LANGUAGE PLPGSQL;
292 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field_list( marc TEXT, ff TEXT, use_default BOOL DEFAULT FALSE ) RETURNS TEXT[] AS $func$
298 collection TEXT[] := '{}'::TEXT[];
300 rtype := (vandelay.marc21_record_type( marc )).code;
301 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
302 IF ff_pos.tag = 'ldr' THEN
303 val := oils_xpath_string('//*[local-name()="leader"]', marc);
304 IF val IS NOT NULL THEN
305 val := SUBSTRING( val, ff_pos.start_pos + 1, ff_pos.length );
306 collection := collection || val;
309 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
310 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
311 collection := collection || val;
314 CONTINUE WHEN NOT use_default;
315 CONTINUE WHEN ARRAY_UPPER(collection, 1) > 0;
316 val := REPEAT( ff_pos.default_val, ff_pos.length );
317 collection := collection || val;
322 $func$ LANGUAGE PLPGSQL;
324 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT, use_default BOOL DEFAULT FALSE ) RETURNS TEXT AS $func$
331 rtype := (vandelay.marc21_record_type( marc )).code;
332 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
333 IF ff_pos.tag = 'ldr' THEN
334 val := oils_xpath_string('//*[local-name()="leader"]', marc);
335 IF val IS NOT NULL THEN
336 val := SUBSTRING( val, ff_pos.start_pos + 1, ff_pos.length );
340 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
341 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
345 CONTINUE WHEN NOT use_default;
346 val := REPEAT( ff_pos.default_val, ff_pos.length );
352 $func$ LANGUAGE PLPGSQL;
354 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
355 CREATE OR REPLACE FUNCTION vandelay.marc21_physical_characteristics( marc TEXT) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
359 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
360 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
361 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
362 retval biblio.marc21_physical_characteristics%ROWTYPE;
365 FOR _007 IN SELECT oils_xpath_string('//*', value) FROM UNNEST(oils_xpath('//*[@tag="007"]', marc)) x(value) LOOP
366 IF _007 IS NOT NULL AND _007 <> '' THEN
367 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007, 1, 1 );
369 IF ptype.ptype_key IS NOT NULL THEN
370 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
371 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 );
373 IF pval.id IS NOT NULL THEN
376 retval.ptype := ptype.ptype_key;
377 retval.subfield := psf.id;
378 retval.value := pval.id;
389 $func$ LANGUAGE PLPGSQL;
391 CREATE TYPE vandelay.flat_marc AS ( tag CHAR(3), ind1 TEXT, ind2 TEXT, subfield TEXT, value TEXT );
392 CREATE OR REPLACE FUNCTION vandelay.flay_marc ( TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
395 use MARC::File::XML (BinaryEncoding => 'UTF-8');
399 MARC::Charset->assume_unicode(1);
402 my $r = MARC::Record->new_from_xml( $xml );
404 return_next( { tag => 'LDR', value => $r->leader } );
406 for my $f ( $r->fields ) {
407 if ($f->is_control_field) {
408 return_next({ tag => $f->tag, value => $f->data });
410 for my $s ($f->subfields) {
413 ind1 => $f->indicator(1),
414 ind2 => $f->indicator(2),
419 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
420 my $trim = $f->indicator(2) || 0;
423 ind1 => $f->indicator(1),
424 ind2 => $f->indicator(2),
426 value => substr( $s->[1], $trim )
435 $func$ LANGUAGE PLPERLU;
437 CREATE OR REPLACE FUNCTION vandelay.flatten_marc ( marc TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
439 output vandelay.flat_marc%ROWTYPE;
442 FOR field IN SELECT * FROM vandelay.flay_marc( marc ) LOOP
443 output.ind1 := field.ind1;
444 output.ind2 := field.ind2;
445 output.tag := field.tag;
446 output.subfield := field.subfield;
447 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
448 output.value := naco_normalize(field.value, field.subfield);
450 output.value := field.value;
453 CONTINUE WHEN output.value IS NULL;
458 $func$ LANGUAGE PLPGSQL;
460 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT, attr_defs TEXT[]) RETURNS hstore AS $_$
462 transformed_xml TEXT;
465 xfrm config.xml_transform%ROWTYPE;
467 new_attrs HSTORE := ''::HSTORE;
468 attr_def config.record_attr_definition%ROWTYPE;
471 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE name IN (SELECT * FROM UNNEST(attr_defs)) ORDER BY format LOOP
473 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
474 SELECT STRING_AGG(x.value, COALESCE(attr_def.joiner,' ')) INTO attr_value
475 FROM vandelay.flatten_marc(xml) AS x
476 WHERE x.tag LIKE attr_def.tag
478 WHEN attr_def.sf_list IS NOT NULL
479 THEN POSITION(x.subfield IN attr_def.sf_list) > 0
486 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
487 attr_value := vandelay.marc21_extract_fixed_field(xml, attr_def.fixed_field);
489 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
491 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
493 -- See if we can skip the XSLT ... it's expensive
494 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
495 -- Can't skip the transform
496 IF xfrm.xslt <> '---' THEN
497 transformed_xml := oils_xslt_process(xml,xfrm.xslt);
499 transformed_xml := xml;
502 prev_xfrm := xfrm.name;
505 IF xfrm.name IS NULL THEN
506 -- just grab the marcxml (empty) transform
507 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
508 prev_xfrm := xfrm.name;
511 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
513 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
514 SELECT m.value::TEXT INTO attr_value
515 FROM vandelay.marc21_physical_characteristics(xml) v
516 JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
517 WHERE v.subfield = attr_def.phys_char_sf
518 LIMIT 1; -- Just in case ...
522 -- apply index normalizers to attr_value
524 SELECT n.func AS func,
525 n.param_count AS param_count,
527 FROM config.index_normalizer n
528 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
529 WHERE attr = attr_def.name
531 EXECUTE 'SELECT ' || normalizer.func || '(' ||
532 quote_nullable( attr_value ) ||
534 WHEN normalizer.param_count > 0
535 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
542 -- Add the new value to the hstore
543 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
549 $_$ LANGUAGE PLPGSQL;
551 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT ) RETURNS hstore AS $_$
552 SELECT vandelay.extract_rec_attrs( $1, (SELECT ARRAY_AGG(name) FROM config.record_attr_definition));
555 -- Everything between this comment and the beginning of the definition of
556 -- vandelay.match_bib_record() is strictly in service of that function.
557 CREATE TYPE vandelay.match_set_test_result AS (record BIGINT, quality INTEGER);
559 CREATE OR REPLACE FUNCTION vandelay.match_set_test_marcxml(
560 match_set_id INTEGER, record_xml TEXT, bucket_id INTEGER
561 ) RETURNS SETOF vandelay.match_set_test_result AS $$
572 tags_rstore := vandelay.flatten_marc_hstore(record_xml);
573 svf_rstore := vandelay.extract_rec_attrs(record_xml);
575 CREATE TEMPORARY TABLE _vandelay_tmp_qrows (q INTEGER);
576 CREATE TEMPORARY TABLE _vandelay_tmp_jrows (j TEXT);
578 -- generate the where clause and return that directly (into wq), and as
579 -- a side-effect, populate the _vandelay_tmp_[qj]rows tables.
580 wq := vandelay.get_expr_from_match_set(match_set_id, tags_rstore);
582 query_ := 'SELECT DISTINCT(record), ';
584 -- qrows table is for the quality bits we add to the SELECT clause
586 'COALESCE(n' || q::TEXT || '.quality, 0)', ' + '
587 ) INTO coal FROM _vandelay_tmp_qrows;
589 -- our query string so far is the SELECT clause and the inital FROM.
590 -- no JOINs yet nor the WHERE clause
591 query_ := query_ || coal || ' AS quality ' || E'\n';
593 -- jrows table is for the joins we must make (and the real text conditions)
594 SELECT STRING_AGG(j, E'\n') INTO joins
595 FROM _vandelay_tmp_jrows;
597 -- add those joins and the where clause to our query.
598 query_ := query_ || joins || E'\n';
600 -- join the record bucket
601 IF bucket_id IS NOT NULL THEN
602 query_ := query_ || 'JOIN container.biblio_record_entry_bucket_item ' ||
603 'brebi ON (brebi.target_biblio_record_entry = record ' ||
604 'AND brebi.bucket = ' || bucket_id || E')\n';
607 query_ := query_ || 'JOIN biblio.record_entry bre ON (bre.id = record) ' || 'WHERE ' || wq || ' AND not bre.deleted';
609 -- this will return rows of record,quality
610 FOR rec IN EXECUTE query_ USING tags_rstore, svf_rstore LOOP
614 DROP TABLE _vandelay_tmp_qrows;
615 DROP TABLE _vandelay_tmp_jrows;
621 CREATE OR REPLACE FUNCTION vandelay.flatten_marc_hstore(
623 ) RETURNS HSTORE AS $func$
627 ARRAY_AGG(tag || (COALESCE(subfield, ''))),
631 SELECT tag, subfield, ARRAY_AGG(value)::TEXT AS value
634 CASE WHEN tag = '020' THEN -- caseless -- isbn
635 LOWER((REGEXP_MATCHES(value,$$^(\S{10,17})$$))[1] || '%')
636 WHEN tag = '022' THEN -- caseless -- issn
637 LOWER((REGEXP_MATCHES(value,$$^(\S{4}[- ]?\S{4})$$))[1] || '%')
638 WHEN tag = '024' THEN -- caseless -- upc (other)
643 FROM vandelay.flatten_marc(record_xml)) x
644 GROUP BY tag, subfield ORDER BY tag, subfield
648 $func$ LANGUAGE PLPGSQL;
650 -- backwards compat version so we don't have
651 -- to modify vandelay.match_set_test_marcxml()
652 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set(
653 match_set_id INTEGER,
657 RETURN vandelay.get_expr_from_match_set(
658 match_set_id, tags_rstore, NULL);
662 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set(
663 match_set_id INTEGER,
668 root vandelay.match_set_point;
670 SELECT * INTO root FROM vandelay.match_set_point
671 WHERE parent IS NULL AND match_set = match_set_id;
673 RETURN vandelay.get_expr_from_match_set_point(
674 root, tags_rstore, auth_heading);
678 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set_point(
679 node vandelay.match_set_point,
688 child vandelay.match_set_point;
690 SELECT ARRAY_AGG(id) INTO children FROM vandelay.match_set_point
691 WHERE parent = node.id;
693 IF ARRAY_LENGTH(children, 1) > 0 THEN
694 this_op := vandelay._get_expr_render_one(node);
697 WHILE children[i] IS NOT NULL LOOP
698 SELECT * INTO child FROM vandelay.match_set_point
699 WHERE id = children[i];
701 q := q || ' ' || this_op || ' ';
704 q := q || vandelay.get_expr_from_match_set_point(
705 child, tags_rstore, auth_heading);
709 ELSIF node.bool_op IS NULL THEN
710 PERFORM vandelay._get_expr_push_qrow(node);
711 PERFORM vandelay._get_expr_push_jrow(node, tags_rstore, auth_heading);
712 RETURN vandelay._get_expr_render_one(node);
719 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_qrow(
720 node vandelay.match_set_point
724 INSERT INTO _vandelay_tmp_qrows (q) VALUES (node.id);
728 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_jrow(
729 node vandelay.match_set_point,
744 -- remember $1 is tags_rstore, and $2 is svf_rstore
745 -- a non-NULL auth_heading means we're matching authority records
747 IF auth_heading IS NOT NULL THEN
748 rec_table := 'authority.full_rec';
750 rec_table := 'metabib.full_rec';
754 SELECT COUNT(*) INTO jrow_count FROM _vandelay_tmp_jrows;
755 IF jrow_count > 0 THEN
756 my_using := ' USING (record)';
757 my_join := 'FULL OUTER JOIN';
763 IF node.tag IS NOT NULL THEN
764 caseless := (node.tag IN ('020', '022', '024'));
766 IF node.subfield IS NOT NULL THEN
767 tagkey := tagkey || node.subfield;
785 my_alias := 'n' || node.id::TEXT;
787 jrow := my_join || ' (SELECT *, ';
788 IF node.tag IS NOT NULL THEN
789 jrow := jrow || node.quality ||
790 ' AS quality FROM ' || rec_table || ' mfr WHERE mfr.tag = ''' ||
792 IF node.subfield IS NOT NULL THEN
793 jrow := jrow || ' AND mfr.subfield = ''' ||
794 node.subfield || '''';
796 jrow := jrow || ' AND (';
797 jrow := jrow || vandelay._node_tag_comparisons(caseless, op, tags_rstore, tagkey);
798 jrow := jrow || ')) ' || my_alias || my_using || E'\n';
800 IF auth_heading IS NOT NULL THEN -- authority record
801 IF node.heading AND auth_heading <> '' THEN
802 jrow := jrow || 'id AS record, ' || node.quality ||
803 ' AS quality FROM authority.record_entry are ' ||
804 ' WHERE are.heading = ''' || auth_heading || '''';
805 jrow := jrow || ') ' || my_alias || my_using || E'\n';
808 jrow := jrow || 'id AS record, ' || node.quality ||
809 ' AS quality FROM metabib.record_attr_flat mraf WHERE mraf.attr = ''' ||
810 node.svf || ''' AND mraf.value ' || op || ' $2->''' || node.svf || ''') ' ||
811 my_alias || my_using || E'\n';
814 INSERT INTO _vandelay_tmp_jrows (j) VALUES (jrow);
818 CREATE OR REPLACE FUNCTION vandelay._node_tag_comparisons(
830 vals := tags_rstore->tagkey;
835 IF vals[i] IS NULL THEN
838 result := result || ' OR ';
843 result := result || 'LOWER(mfr.value) ' || op;
845 result := result || 'mfr.value ' || op;
848 result := result || ' ' || COALESCE('''' || vals[i] || '''', 'NULL');
850 IF vals[i] IS NULL THEN
861 CREATE OR REPLACE FUNCTION vandelay._get_expr_render_one(
862 node vandelay.match_set_point
867 IF node.bool_op IS NOT NULL THEN
870 RETURN '(n' || node.id::TEXT || '.id IS NOT NULL)';
875 CREATE OR REPLACE FUNCTION vandelay.match_bib_record() RETURNS TRIGGER AS $func$
877 incoming_existing_id TEXT;
878 test_result vandelay.match_set_test_result%ROWTYPE;
883 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
887 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
889 SELECT q.match_set INTO match_set FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
891 IF match_set IS NOT NULL THEN
892 NEW.quality := vandelay.measure_record_quality( NEW.marc, match_set );
895 -- Perfect matches on 901$c exit early with a match with high quality.
896 incoming_existing_id :=
897 oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]', NEW.marc);
899 IF incoming_existing_id IS NOT NULL AND incoming_existing_id != '' THEN
900 SELECT id INTO tmp_rec FROM biblio.record_entry WHERE id = incoming_existing_id::bigint;
901 IF tmp_rec IS NOT NULL THEN
902 INSERT INTO vandelay.bib_match (queued_record, eg_record, match_score, quality)
907 -- note: no match_set means quality==0
908 vandelay.measure_record_quality( b.marc, match_set )
909 FROM biblio.record_entry b
910 WHERE id = incoming_existing_id::bigint;
914 IF match_set IS NULL THEN
918 SELECT q.match_bucket INTO match_bucket FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
920 FOR test_result IN SELECT * FROM
921 vandelay.match_set_test_marcxml(match_set, NEW.marc, match_bucket) LOOP
923 INSERT INTO vandelay.bib_match ( queued_record, eg_record, match_score, quality )
928 vandelay.measure_record_quality( b.marc, match_set )
929 FROM biblio.record_entry b
930 WHERE id = test_result.record;
936 $func$ LANGUAGE PLPGSQL;
938 CREATE OR REPLACE FUNCTION vandelay.measure_record_quality ( xml TEXT, match_set_id INT ) RETURNS INT AS $_$
942 test vandelay.match_set_quality%ROWTYPE;
945 FOR test IN SELECT * FROM vandelay.match_set_quality WHERE match_set = match_set_id LOOP
946 IF test.tag IS NOT NULL THEN
947 FOR rvalue IN SELECT value FROM vandelay.flatten_marc( xml ) WHERE tag = test.tag AND subfield = test.subfield LOOP
948 IF test.value = rvalue THEN
949 out_q := out_q + test.quality;
953 IF test.value = vandelay.extract_rec_attrs(xml, ARRAY[test.svf]) -> test.svf THEN
954 out_q := out_q + test.quality;
961 $_$ LANGUAGE PLPGSQL;
963 CREATE TYPE vandelay.tcn_data AS (tcn TEXT, tcn_source TEXT, used BOOL);
964 CREATE OR REPLACE FUNCTION vandelay.find_bib_tcn_data ( xml TEXT ) RETURNS SETOF vandelay.tcn_data AS $_$
968 output vandelay.tcn_data%ROWTYPE;
972 eg_tcn := BTRIM((oils_xpath('//*[@tag="001"]/text()',xml))[1]);
973 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
975 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="003"]/text()',xml))[1]);
976 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
977 eg_tcn_source := 'System Local';
980 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
983 output.used := FALSE;
988 output.tcn := eg_tcn;
989 output.tcn_source := eg_tcn_source;
995 eg_tcn := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="a"]/text()',xml))[1]);
996 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
998 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="b"]/text()',xml))[1]);
999 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
1000 eg_tcn_source := 'System Local';
1003 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1006 output.used := FALSE;
1008 output.used := TRUE;
1011 output.tcn := eg_tcn;
1012 output.tcn_source := eg_tcn_source;
1018 eg_tcn := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="a"]/text()',xml))[1]);
1019 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
1021 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="b"]/text()',xml))[1]);
1022 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
1023 eg_tcn_source := 'System Local';
1026 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1029 output.used := FALSE;
1031 output.used := TRUE;
1034 output.tcn := eg_tcn;
1035 output.tcn_source := eg_tcn_source;
1041 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="020"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
1042 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
1044 eg_tcn_source := 'ISBN';
1046 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1049 output.used := FALSE;
1051 output.used := TRUE;
1054 output.tcn := eg_tcn;
1055 output.tcn_source := eg_tcn_source;
1061 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="022"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
1062 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
1064 eg_tcn_source := 'ISSN';
1066 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1069 output.used := FALSE;
1071 output.used := TRUE;
1074 output.tcn := eg_tcn;
1075 output.tcn_source := eg_tcn_source;
1081 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="010"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
1082 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
1084 eg_tcn_source := 'LCCN';
1086 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1089 output.used := FALSE;
1091 output.used := TRUE;
1094 output.tcn := eg_tcn;
1095 output.tcn_source := eg_tcn_source;
1101 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="035"]/*[@code="a"]/text()',xml))[1], $re$^.*?(\w+)$$re$, $re$\1$re$);
1102 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
1104 eg_tcn_source := 'System Legacy';
1106 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1109 output.used := FALSE;
1111 output.used := TRUE;
1114 output.tcn := eg_tcn;
1115 output.tcn_source := eg_tcn_source;
1122 $_$ LANGUAGE PLPGSQL;
1124 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT, force_add INT ) RETURNS TEXT AS $_$
1127 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1131 MARC::Charset->assume_unicode(1);
1133 my $target_xml = shift;
1134 my $source_xml = shift;
1135 my $field_spec = shift;
1136 my $force_add = shift || 0;
1138 my $target_r = MARC::Record->new_from_xml( $target_xml );
1139 my $source_r = MARC::Record->new_from_xml( $source_xml );
1141 return $target_xml unless ($target_r && $source_r);
1143 my @field_list = split(',', $field_spec);
1146 for my $f (@field_list) {
1147 $f =~ s/^\s*//; $f =~ s/\s*$//;
1148 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1154 $match =~ s/^\s*//; $match =~ s/\s*$//;
1155 $fields{$field} = { sf => [ split('', $sf) ] };
1157 my ($msf,$mre) = split('~', $match);
1158 if (length($msf) > 0 and length($mre) > 0) {
1159 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1160 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1161 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1167 for my $f ( keys %fields) {
1168 if ( @{$fields{$f}{sf}} ) {
1169 for my $from_field ($source_r->field( $f )) {
1170 my @tos = $target_r->field( $f );
1172 next if (exists($fields{$f}{match}) and !$force_add);
1173 my @new_fields = map { $_->clone } $source_r->field( $f );
1174 $target_r->insert_fields_ordered( @new_fields );
1176 for my $to_field (@tos) {
1177 if (exists($fields{$f}{match})) {
1178 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1180 for my $old_sf ($from_field->subfields) {
1181 $to_field->add_subfields( @$old_sf ) if grep(/$$old_sf[0]/,@{$fields{$f}{sf}});
1187 my @new_fields = map { $_->clone } $source_r->field( $f );
1188 $target_r->insert_fields_ordered( @new_fields );
1192 $target_xml = $target_r->as_xml_record;
1193 $target_xml =~ s/^<\?.+?\?>$//mo;
1194 $target_xml =~ s/\n//sgo;
1195 $target_xml =~ s/>\s+</></sgo;
1199 $_$ LANGUAGE PLPERLU;
1201 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1202 SELECT vandelay.add_field( $1, $2, $3, 0 );
1205 CREATE OR REPLACE FUNCTION vandelay.strip_field ( xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1208 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1212 MARC::Charset->assume_unicode(1);
1215 my $r = MARC::Record->new_from_xml( $xml );
1217 return $xml unless ($r);
1219 my $field_spec = shift;
1220 my @field_list = split(',', $field_spec);
1223 for my $f (@field_list) {
1224 $f =~ s/^\s*//; $f =~ s/\s*$//;
1225 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1231 $match =~ s/^\s*//; $match =~ s/\s*$//;
1232 $fields{$field} = { sf => [ split('', $sf) ] };
1234 my ($msf,$mre) = split('~', $match);
1235 if (length($msf) > 0 and length($mre) > 0) {
1236 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1237 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1238 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1244 for my $f ( keys %fields) {
1245 for my $to_field ($r->field( $f )) {
1246 if (exists($fields{$f}{match})) {
1247 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1250 if ( @{$fields{$f}{sf}} ) {
1251 $to_field->delete_subfield(code => $fields{$f}{sf});
1253 $r->delete_field( $to_field );
1258 $xml = $r->as_xml_record;
1259 $xml =~ s/^<\?.+?\?>$//mo;
1261 $xml =~ s/>\s+</></sgo;
1265 $_$ LANGUAGE PLPERLU;
1267 CREATE OR REPLACE FUNCTION vandelay.replace_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1274 parsed_target := vandelay.strip_field( target_xml, ''); -- this dance normalizes the format of the xml for the IF below
1275 xml_output := parsed_target; -- if there are no replace rules, just return the input
1277 FOR curr_field IN SELECT UNNEST( STRING_TO_ARRAY(field, ',') ) LOOP -- naive split, but it's the same we use in the perl
1279 xml_output := vandelay.strip_field( parsed_target, curr_field);
1281 IF xml_output <> parsed_target AND curr_field ~ E'~' THEN
1282 -- we removed something, and there was a regexp restriction in the curr_field definition, so proceed
1283 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 1 );
1284 ELSIF curr_field !~ E'~' THEN
1285 -- No regexp restriction, add the curr_field
1286 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 0 );
1289 parsed_target := xml_output; -- in prep for any following loop iterations
1295 $_$ LANGUAGE PLPGSQL;
1297 CREATE OR REPLACE FUNCTION vandelay.merge_record_xml ( target_xml TEXT, source_xml TEXT, add_rule TEXT, replace_preserve_rule TEXT, strip_rule TEXT ) RETURNS TEXT AS $_$
1298 SELECT vandelay.replace_field( vandelay.add_field( vandelay.strip_field( $1, $5) , $2, $3 ), $2, $4);
1301 CREATE TYPE vandelay.compile_profile AS (add_rule TEXT, replace_rule TEXT, preserve_rule TEXT, strip_rule TEXT);
1302 CREATE OR REPLACE FUNCTION vandelay.compile_profile ( incoming_xml TEXT ) RETURNS vandelay.compile_profile AS $_$
1304 output vandelay.compile_profile%ROWTYPE;
1305 profile vandelay.merge_profile%ROWTYPE;
1307 profile_tmpl_owner TEXT;
1308 add_rule TEXT := '';
1309 strip_rule TEXT := '';
1310 replace_rule TEXT := '';
1311 preserve_rule TEXT := '';
1315 profile_tmpl := (oils_xpath('//*[@tag="905"]/*[@code="t"]/text()',incoming_xml))[1];
1316 profile_tmpl_owner := (oils_xpath('//*[@tag="905"]/*[@code="o"]/text()',incoming_xml))[1];
1318 IF profile_tmpl IS NOT NULL AND profile_tmpl <> '' AND profile_tmpl_owner IS NOT NULL AND profile_tmpl_owner <> '' THEN
1319 SELECT p.* INTO profile
1320 FROM vandelay.merge_profile p
1321 JOIN actor.org_unit u ON (u.id = p.owner)
1322 WHERE p.name = profile_tmpl
1323 AND u.shortname = profile_tmpl_owner;
1325 IF profile.id IS NOT NULL THEN
1326 add_rule := COALESCE(profile.add_spec,'');
1327 strip_rule := COALESCE(profile.strip_spec,'');
1328 replace_rule := COALESCE(profile.replace_spec,'');
1329 preserve_rule := COALESCE(profile.preserve_spec,'');
1333 add_rule := add_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="a"]/text()',incoming_xml),','),'');
1334 strip_rule := strip_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="d"]/text()',incoming_xml),','),'');
1335 replace_rule := replace_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="r"]/text()',incoming_xml),','),'');
1336 preserve_rule := preserve_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="p"]/text()',incoming_xml),','),'');
1338 output.add_rule := BTRIM(add_rule,',');
1339 output.replace_rule := BTRIM(replace_rule,',');
1340 output.strip_rule := BTRIM(strip_rule,',');
1341 output.preserve_rule := BTRIM(preserve_rule,',');
1345 $_$ LANGUAGE PLPGSQL;
1347 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1349 merge_profile vandelay.merge_profile%ROWTYPE;
1350 dyn_profile vandelay.compile_profile%ROWTYPE;
1360 SELECT b.marc INTO eg_marc
1361 FROM biblio.record_entry b
1365 IF eg_marc IS NULL OR v_marc IS NULL THEN
1366 -- RAISE NOTICE 'no marc for template or bib record';
1370 dyn_profile := vandelay.compile_profile( v_marc );
1372 IF merge_profile_id IS NOT NULL THEN
1373 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1375 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1376 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1377 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1378 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1382 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1383 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1387 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1388 --Since we have nothing to do, just return a NOOP "we did it"
1390 ELSIF dyn_profile.replace_rule <> '' THEN
1391 source_marc = v_marc;
1392 target_marc = eg_marc;
1393 replace_rule = dyn_profile.replace_rule;
1395 source_marc = eg_marc;
1396 target_marc = v_marc;
1397 replace_rule = dyn_profile.preserve_rule;
1400 UPDATE biblio.record_entry
1401 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1405 -- RAISE NOTICE 'update of biblio.record_entry failed';
1412 $$ LANGUAGE PLPGSQL;
1414 CREATE OR REPLACE FUNCTION vandelay.merge_record_xml ( target_marc TEXT, template_marc TEXT ) RETURNS TEXT AS $$
1416 dyn_profile vandelay.compile_profile%ROWTYPE;
1424 IF target_marc IS NULL OR template_marc IS NULL THEN
1425 -- RAISE NOTICE 'no marc for target or template record';
1429 dyn_profile := vandelay.compile_profile( template_marc );
1431 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1432 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1436 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1437 --Since we have nothing to do, just return what we were given.
1439 ELSIF dyn_profile.replace_rule <> '' THEN
1440 trgt_marc = target_marc;
1441 tmpl_marc = template_marc;
1442 replace_rule = dyn_profile.replace_rule;
1444 tmp_marc = target_marc;
1445 trgt_marc = template_marc;
1446 tmpl_marc = tmp_marc;
1447 replace_rule = dyn_profile.preserve_rule;
1450 RETURN vandelay.merge_record_xml( trgt_marc, tmpl_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule );
1453 $$ LANGUAGE PLPGSQL;
1455 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT) RETURNS BOOL AS $$
1456 SELECT vandelay.template_overlay_bib_record( $1, $2, NULL);
1459 CREATE OR REPLACE FUNCTION vandelay.overlay_bib_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1465 update_fields TEXT[];
1470 SELECT q.marc, q.bib_source INTO v_marc, v_bib_source
1471 FROM vandelay.queued_bib_record q
1472 JOIN vandelay.bib_match m ON (m.queued_record = q.id AND q.id = import_id)
1475 IF v_marc IS NULL THEN
1476 -- RAISE NOTICE 'no marc for vandelay or bib record';
1480 IF vandelay.template_overlay_bib_record( v_marc, eg_id, merge_profile_id) THEN
1481 UPDATE vandelay.queued_bib_record
1482 SET imported_as = eg_id,
1484 WHERE id = import_id;
1486 SELECT q.update_bib_source INTO update_bib FROM vandelay.merge_profile q where q.id = merge_profile_Id;
1489 editor_string := (oils_xpath('//*[@tag="905"]/*[@code="u"]/text()',v_marc))[1];
1491 IF editor_string IS NOT NULL AND editor_string <> '' THEN
1492 SELECT usr INTO editor_id FROM actor.card WHERE barcode = editor_string;
1494 IF editor_id IS NULL THEN
1495 SELECT id INTO editor_id FROM actor.usr WHERE usrname = editor_string;
1498 IF editor_id IS NOT NULL THEN
1499 --only update the edit date if we have a valid editor
1500 update_fields := ARRAY_APPEND(update_fields, 'editor = ' || editor_id || ', edit_date = NOW()');
1504 IF v_bib_source IS NOT NULL THEN
1505 update_fields := ARRAY_APPEND(update_fields, 'source = ' || v_bib_source);
1508 IF ARRAY_LENGTH(update_fields, 1) > 0 THEN
1509 update_query := 'UPDATE biblio.record_entry SET ' || ARRAY_TO_STRING(update_fields, ',') || ' WHERE id = ' || eg_id || ';';
1510 --RAISE NOTICE 'query: %', update_query;
1511 EXECUTE update_query;
1518 -- RAISE NOTICE 'update of biblio.record_entry failed';
1523 $$ LANGUAGE PLPGSQL;
1525 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
1528 lwm_ratio_value NUMERIC;
1531 lwm_ratio_value := COALESCE(lwm_ratio_value_p, 0.0);
1533 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1536 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1540 SELECT m.eg_record INTO eg_id
1541 FROM vandelay.bib_match m
1542 JOIN vandelay.queued_bib_record qr ON (m.queued_record = qr.id)
1543 JOIN vandelay.bib_queue q ON (qr.queue = q.id)
1544 JOIN biblio.record_entry r ON (r.id = m.eg_record)
1545 WHERE m.queued_record = import_id
1546 AND qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC >= lwm_ratio_value
1547 ORDER BY m.match_score DESC, -- required match score
1548 qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC DESC, -- quality tie breaker
1549 m.id -- when in doubt, use the first match
1552 IF eg_id IS NULL THEN
1553 -- RAISE NOTICE 'incoming record is not of high enough quality';
1557 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1559 $$ LANGUAGE PLPGSQL;
1561 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1562 SELECT vandelay.auto_overlay_bib_record_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1565 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1571 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1574 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1578 SELECT COUNT(*) INTO match_count FROM vandelay.bib_match WHERE queued_record = import_id;
1580 IF match_count <> 1 THEN
1581 -- RAISE NOTICE 'not an exact match';
1585 -- Check that the one match is on the first 901c
1586 SELECT m.eg_record INTO eg_id
1587 FROM vandelay.queued_bib_record q
1588 JOIN vandelay.bib_match m ON (m.queued_record = q.id)
1589 WHERE q.id = import_id
1590 AND m.eg_record = oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]',marc)::BIGINT;
1593 -- RAISE NOTICE 'not a 901c match';
1597 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1599 $$ LANGUAGE PLPGSQL;
1601 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1603 queued_record vandelay.queued_bib_record%ROWTYPE;
1606 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1608 IF vandelay.auto_overlay_bib_record( queued_record.id, merge_profile_id ) THEN
1609 RETURN NEXT queued_record.id;
1617 $$ LANGUAGE PLPGSQL;
1619 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue_with_best ( queue_id BIGINT, merge_profile_id INT, lwm_ratio_value NUMERIC ) RETURNS SETOF BIGINT AS $$
1621 queued_record vandelay.queued_bib_record%ROWTYPE;
1624 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1626 IF vandelay.auto_overlay_bib_record_with_best( queued_record.id, merge_profile_id, lwm_ratio_value ) THEN
1627 RETURN NEXT queued_record.id;
1635 $$ LANGUAGE PLPGSQL;
1637 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1638 SELECT vandelay.auto_overlay_bib_queue_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1641 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1642 SELECT * FROM vandelay.auto_overlay_bib_queue( $1, NULL );
1645 CREATE OR REPLACE FUNCTION vandelay.ingest_bib_marc ( ) RETURNS TRIGGER AS $$
1651 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1655 FOR adef IN SELECT * FROM vandelay.bib_attr_definition LOOP
1657 SELECT extract_marc_field('vandelay.queued_bib_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_bib_record WHERE id = NEW.id;
1658 IF (value IS NOT NULL AND value <> '') THEN
1659 INSERT INTO vandelay.queued_bib_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1666 $$ LANGUAGE PLPGSQL;
1668 CREATE OR REPLACE FUNCTION vandelay.cleanup_bib_marc ( ) RETURNS TRIGGER AS $$
1670 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1674 DELETE FROM vandelay.queued_bib_record_attr WHERE record = OLD.id;
1675 DELETE FROM vandelay.import_item WHERE record = OLD.id;
1677 IF TG_OP = 'UPDATE' THEN
1682 $$ LANGUAGE PLPGSQL;
1684 CREATE TRIGGER cleanup_bib_trigger
1685 BEFORE UPDATE OR DELETE ON vandelay.queued_bib_record
1686 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_bib_marc();
1688 CREATE TRIGGER ingest_bib_trigger
1689 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1690 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_bib_marc();
1692 CREATE TRIGGER zz_match_bibs_trigger
1693 BEFORE INSERT OR UPDATE ON vandelay.queued_bib_record
1694 FOR EACH ROW EXECUTE PROCEDURE vandelay.match_bib_record();
1697 /* Authority stuff down here */
1698 ---------------------------------------
1699 CREATE TABLE vandelay.authority_attr_definition (
1700 id SERIAL PRIMARY KEY,
1701 code TEXT UNIQUE NOT NULL,
1703 xpath TEXT NOT NULL,
1704 remove TEXT NOT NULL DEFAULT ''
1707 CREATE TYPE vandelay.authority_queue_queue_type AS ENUM ('authority');
1708 CREATE TABLE vandelay.authority_queue (
1709 queue_type vandelay.authority_queue_queue_type NOT NULL DEFAULT 'authority',
1710 CONSTRAINT vand_authority_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
1711 ) INHERITS (vandelay.queue);
1712 ALTER TABLE vandelay.authority_queue ADD PRIMARY KEY (id);
1714 CREATE TABLE vandelay.queued_authority_record (
1715 queue INT NOT NULL REFERENCES vandelay.authority_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1716 imported_as INT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1717 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
1719 ) INHERITS (vandelay.queued_record);
1720 ALTER TABLE vandelay.queued_authority_record ADD PRIMARY KEY (id);
1721 CREATE INDEX queued_authority_record_queue_idx ON vandelay.queued_authority_record (queue);
1723 CREATE TABLE vandelay.queued_authority_record_attr (
1724 id BIGSERIAL PRIMARY KEY,
1725 record BIGINT NOT NULL REFERENCES vandelay.queued_authority_record (id) DEFERRABLE INITIALLY DEFERRED,
1726 field INT NOT NULL REFERENCES vandelay.authority_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
1727 attr_value TEXT NOT NULL
1729 CREATE INDEX queued_authority_record_attr_record_idx ON vandelay.queued_authority_record_attr (record);
1731 CREATE TABLE vandelay.authority_match (
1732 id BIGSERIAL PRIMARY KEY,
1733 queued_record BIGINT REFERENCES vandelay.queued_authority_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1734 eg_record BIGINT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1735 quality INT NOT NULL DEFAULT 0,
1736 match_score INT NOT NULL DEFAULT 0
1739 CREATE OR REPLACE FUNCTION vandelay.ingest_authority_marc ( ) RETURNS TRIGGER AS $$
1745 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1749 FOR adef IN SELECT * FROM vandelay.authority_attr_definition LOOP
1751 SELECT extract_marc_field('vandelay.queued_authority_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_authority_record WHERE id = NEW.id;
1752 IF (value IS NOT NULL AND value <> '') THEN
1753 INSERT INTO vandelay.queued_authority_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1760 $$ LANGUAGE PLPGSQL;
1762 CREATE OR REPLACE FUNCTION vandelay.cleanup_authority_marc ( ) RETURNS TRIGGER AS $$
1764 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1768 DELETE FROM vandelay.queued_authority_record_attr WHERE record = OLD.id;
1769 IF TG_OP = 'UPDATE' THEN
1774 $$ LANGUAGE PLPGSQL;
1776 CREATE TRIGGER cleanup_authority_trigger
1777 BEFORE UPDATE OR DELETE ON vandelay.queued_authority_record
1778 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_authority_marc();
1780 CREATE TRIGGER ingest_authority_trigger
1781 AFTER INSERT OR UPDATE ON vandelay.queued_authority_record
1782 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_authority_marc();
1784 CREATE OR REPLACE FUNCTION vandelay.overlay_authority_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1786 merge_profile vandelay.merge_profile%ROWTYPE;
1787 dyn_profile vandelay.compile_profile%ROWTYPE;
1796 SELECT b.marc INTO eg_marc
1797 FROM authority.record_entry b
1798 JOIN vandelay.authority_match m ON (m.eg_record = b.id AND m.queued_record = import_id)
1801 SELECT q.marc INTO v_marc
1802 FROM vandelay.queued_record q
1803 JOIN vandelay.authority_match m ON (m.queued_record = q.id AND q.id = import_id)
1806 IF eg_marc IS NULL OR v_marc IS NULL THEN
1807 -- RAISE NOTICE 'no marc for vandelay or authority record';
1811 dyn_profile := vandelay.compile_profile( v_marc );
1813 IF merge_profile_id IS NOT NULL THEN
1814 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1816 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1817 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1818 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1819 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1823 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1824 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1828 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1829 --Since we have nothing to do, just return a NOOP "we did it"
1831 ELSIF dyn_profile.replace_rule <> '' THEN
1832 source_marc = v_marc;
1833 target_marc = eg_marc;
1834 replace_rule = dyn_profile.replace_rule;
1836 source_marc = eg_marc;
1837 target_marc = v_marc;
1838 replace_rule = dyn_profile.preserve_rule;
1841 UPDATE authority.record_entry
1842 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1846 UPDATE vandelay.queued_authority_record
1847 SET imported_as = eg_id,
1849 WHERE id = import_id;
1853 -- RAISE NOTICE 'update of authority.record_entry failed';
1858 $$ LANGUAGE PLPGSQL;
1860 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1865 SELECT COUNT(*) INTO match_count FROM vandelay.authority_match WHERE queued_record = import_id;
1867 IF match_count <> 1 THEN
1868 -- RAISE NOTICE 'not an exact match';
1872 SELECT m.eg_record INTO eg_id
1873 FROM vandelay.authority_match m
1874 WHERE m.queued_record = import_id
1877 IF eg_id IS NULL THEN
1881 RETURN vandelay.overlay_authority_record( import_id, eg_id, merge_profile_id );
1883 $$ LANGUAGE PLPGSQL;
1885 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1887 queued_record vandelay.queued_authority_record%ROWTYPE;
1890 FOR queued_record IN SELECT * FROM vandelay.queued_authority_record WHERE queue = queue_id AND import_time IS NULL LOOP
1892 IF vandelay.auto_overlay_authority_record( queued_record.id, merge_profile_id ) THEN
1893 RETURN NEXT queued_record.id;
1901 $$ LANGUAGE PLPGSQL;
1903 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1904 SELECT * FROM vandelay.auto_overlay_authority_queue( $1, NULL );
1907 CREATE OR REPLACE FUNCTION vandelay.match_set_test_authxml(
1908 match_set_id INTEGER, record_xml TEXT
1909 ) RETURNS SETOF vandelay.match_set_test_result AS $$
1920 tags_rstore := vandelay.flatten_marc_hstore(record_xml);
1922 SELECT normalize_heading INTO heading
1923 FROM authority.normalize_heading(record_xml);
1925 CREATE TEMPORARY TABLE _vandelay_tmp_qrows (q INTEGER);
1926 CREATE TEMPORARY TABLE _vandelay_tmp_jrows (j TEXT);
1928 -- generate the where clause and return that directly (into wq), and as
1929 -- a side-effect, populate the _vandelay_tmp_[qj]rows tables.
1930 wq := vandelay.get_expr_from_match_set(
1931 match_set_id, tags_rstore, heading);
1933 query_ := 'SELECT DISTINCT(record), ';
1935 -- qrows table is for the quality bits we add to the SELECT clause
1937 'COALESCE(n' || q::TEXT || '.quality, 0)', ' + '
1938 ) INTO coal FROM _vandelay_tmp_qrows;
1940 -- our query string so far is the SELECT clause and the inital FROM.
1941 -- no JOINs yet nor the WHERE clause
1942 query_ := query_ || coal || ' AS quality ' || E'\n';
1944 -- jrows table is for the joins we must make (and the real text conditions)
1945 SELECT STRING_AGG(j, E'\n') INTO joins
1946 FROM _vandelay_tmp_jrows;
1948 -- add those joins and the where clause to our query.
1949 query_ := query_ || joins || E'\n';
1951 query_ := query_ || 'JOIN authority.record_entry are ON (are.id = record) '
1952 || 'WHERE ' || wq || ' AND not are.deleted';
1954 -- this will return rows of record,quality
1955 FOR rec IN EXECUTE query_ USING tags_rstore LOOP
1959 DROP TABLE _vandelay_tmp_qrows;
1960 DROP TABLE _vandelay_tmp_jrows;
1963 $$ LANGUAGE PLPGSQL;
1965 CREATE OR REPLACE FUNCTION vandelay.measure_auth_record_quality
1966 ( xml TEXT, match_set_id INT ) RETURNS INT AS $_$
1970 test vandelay.match_set_quality%ROWTYPE;
1973 FOR test IN SELECT * FROM vandelay.match_set_quality
1974 WHERE match_set = match_set_id LOOP
1975 IF test.tag IS NOT NULL THEN
1976 FOR rvalue IN SELECT value FROM vandelay.flatten_marc( xml )
1977 WHERE tag = test.tag AND subfield = test.subfield LOOP
1978 IF test.value = rvalue THEN
1979 out_q := out_q + test.quality;
1987 $_$ LANGUAGE PLPGSQL;
1991 CREATE OR REPLACE FUNCTION vandelay.match_authority_record() RETURNS TRIGGER AS $func$
1993 incoming_existing_id TEXT;
1994 test_result vandelay.match_set_test_result%ROWTYPE;
1998 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
2002 DELETE FROM vandelay.authority_match WHERE queued_record = NEW.id;
2004 SELECT q.match_set INTO match_set FROM vandelay.authority_queue q WHERE q.id = NEW.queue;
2006 IF match_set IS NOT NULL THEN
2007 NEW.quality := vandelay.measure_auth_record_quality( NEW.marc, match_set );
2010 -- Perfect matches on 901$c exit early with a match with high quality.
2011 incoming_existing_id :=
2012 oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]', NEW.marc);
2014 IF incoming_existing_id IS NOT NULL AND incoming_existing_id != '' THEN
2015 SELECT id INTO tmp_rec FROM authority.record_entry WHERE id = incoming_existing_id::bigint;
2016 IF tmp_rec IS NOT NULL THEN
2017 INSERT INTO vandelay.authority_match (queued_record, eg_record, match_score, quality)
2022 -- note: no match_set means quality==0
2023 vandelay.measure_auth_record_quality( b.marc, match_set )
2024 FROM authority.record_entry b
2025 WHERE id = incoming_existing_id::bigint;
2029 IF match_set IS NULL THEN
2033 FOR test_result IN SELECT * FROM
2034 vandelay.match_set_test_authxml(match_set, NEW.marc) LOOP
2036 INSERT INTO vandelay.authority_match ( queued_record, eg_record, match_score, quality )
2040 test_result.quality,
2041 vandelay.measure_auth_record_quality( b.marc, match_set )
2042 FROM authority.record_entry b
2043 WHERE id = test_result.record;
2049 $func$ LANGUAGE PLPGSQL;
2051 CREATE TRIGGER zz_match_auths_trigger
2052 BEFORE INSERT OR UPDATE ON vandelay.queued_authority_record
2053 FOR EACH ROW EXECUTE PROCEDURE vandelay.match_authority_record();
2055 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_record_with_best ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
2058 lwm_ratio_value NUMERIC;
2061 lwm_ratio_value := COALESCE(lwm_ratio_value_p, 0.0);
2063 PERFORM * FROM vandelay.queued_authority_record WHERE import_time IS NOT NULL AND id = import_id;
2066 -- RAISE NOTICE 'already imported, cannot auto-overlay'
2070 SELECT m.eg_record INTO eg_id
2071 FROM vandelay.authority_match m
2072 JOIN vandelay.queued_authority_record qr ON (m.queued_record = qr.id)
2073 JOIN vandelay.authority_queue q ON (qr.queue = q.id)
2074 JOIN authority.record_entry r ON (r.id = m.eg_record)
2075 WHERE m.queued_record = import_id
2076 AND qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC >= lwm_ratio_value
2077 ORDER BY m.match_score DESC, -- required match score
2078 qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC DESC, -- quality tie breaker
2079 m.id -- when in doubt, use the first match
2082 IF eg_id IS NULL THEN
2083 -- RAISE NOTICE 'incoming record is not of high enough quality';
2087 RETURN vandelay.overlay_authority_record( import_id, eg_id, merge_profile_id );
2089 $$ LANGUAGE PLPGSQL;
2094 -- Vandelay (for importing and exporting records) 012.schema.vandelay.sql
2095 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (1, 'title', oils_i18n_gettext(1, 'vqbrad', 'Title of work', 'description'),'//*[@tag="245"]/*[contains("abcmnopr",@code)]');
2096 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (2, 'author', oils_i18n_gettext(1, 'vqbrad', 'Author of work', 'description'),'//*[@tag="100" or @tag="110" or @tag="113"]/*[contains("ad",@code)]');
2097 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (3, 'language', oils_i18n_gettext(3, 'vqbrad', 'Language of work', 'description'),'//*[@tag="240"]/*[@code="l"][1]');
2098 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (4, 'pagination', oils_i18n_gettext(4, 'vqbrad', 'Pagination', 'description'),'//*[@tag="300"]/*[@code="a"][1]');
2099 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident, remove ) VALUES (5, 'isbn',oils_i18n_gettext(5, 'vqbrad', 'ISBN', 'description'),'//*[@tag="020"]/*[@code="a"]', TRUE, $r$(?:-|\s.+$)$r$);
2100 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident, remove ) VALUES (6, 'issn',oils_i18n_gettext(6, 'vqbrad', 'ISSN', 'description'),'//*[@tag="022"]/*[@code="a"]', TRUE, $r$(?:-|\s.+$)$r$);
2101 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (7, 'price',oils_i18n_gettext(7, 'vqbrad', 'Price', 'description'),'//*[@tag="020" or @tag="022"]/*[@code="c"][1]');
2102 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (8, 'rec_identifier',oils_i18n_gettext(8, 'vqbrad', 'Accession Number', 'description'),'//*[@tag="001"]', TRUE);
2103 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (9, 'eg_tcn',oils_i18n_gettext(9, 'vqbrad', 'TCN Value', 'description'),'//*[@tag="901"]/*[@code="a"]', TRUE);
2104 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (10, 'eg_tcn_source',oils_i18n_gettext(10, 'vqbrad', 'TCN Source', 'description'),'//*[@tag="901"]/*[@code="b"]', TRUE);
2105 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (11, 'eg_identifier',oils_i18n_gettext(11, 'vqbrad', 'Internal ID', 'description'),'//*[@tag="901"]/*[@code="c"]', TRUE);
2106 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (12, 'publisher',oils_i18n_gettext(12, 'vqbrad', 'Publisher', 'description'),'//*[@tag="260"]/*[@code="b"][1]');
2107 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, remove ) VALUES (13, 'pubdate',oils_i18n_gettext(13, 'vqbrad', 'Publication Date', 'description'),'//*[@tag="260"]/*[@code="c"][1]',$r$\D$r$);
2108 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (14, 'edition',oils_i18n_gettext(14, 'vqbrad', 'Edition', 'description'),'//*[@tag="250"]/*[@code="a"][1]');
2110 --INSERT INTO vandelay.import_item_attr_definition (
2111 -- owner, name, tag, owning_lib, circ_lib, location,
2112 -- call_number, circ_modifier, barcode, price, copy_number,
2113 -- circulate, ref, holdable, opac_visible, status
2116 -- 'Evergreen 852 export format',
2118 -- '[@code = "b"][1]',
2119 -- '[@code = "b"][2]',
2126 -- '[@code = "x" and text() = "circulating"]',
2127 -- '[@code = "x" and text() = "reference"]',
2128 -- '[@code = "x" and text() = "holdable"]',
2129 -- '[@code = "x" and text() = "visible"]',
2133 --INSERT INTO vandelay.import_item_attr_definition (
2146 -- 'Unicorn Import format -- 999',
2157 --INSERT INTO vandelay.authority_attr_definition ( code, description, xpath, ident ) VALUES ('rec_identifier','Identifier','//*[@tag="001"]', TRUE);