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 CONSTRAINT vmsp_need_a_subfield_with_a_tag CHECK ((tag IS NOT NULL AND subfield IS NOT NULL) OR tag IS NULL),
27 CONSTRAINT vmsp_need_a_tag_or_a_ff_or_a_bo CHECK (
28 (tag IS NOT NULL AND svf IS NULL AND bool_op IS NULL) OR
29 (tag IS NULL AND svf IS NOT NULL AND bool_op IS NULL) OR
30 (tag IS NULL AND svf IS NULL AND bool_op IS NOT NULL)
34 CREATE TABLE vandelay.match_set_quality (
35 id SERIAL PRIMARY KEY,
36 match_set INT NOT NULL REFERENCES vandelay.match_set (id) ON DELETE CASCADE,
37 svf TEXT REFERENCES config.record_attr_definition,
41 quality INT NOT NULL DEFAULT 1, -- higher is better
42 CONSTRAINT vmsq_need_a_subfield_with_a_tag CHECK ((tag IS NOT NULL AND subfield IS NOT NULL) OR tag IS NULL),
43 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))
45 CREATE UNIQUE INDEX vmsq_def_once_per_set ON vandelay.match_set_quality (match_set, COALESCE(tag,''), COALESCE(subfield,''), COALESCE(svf,''), value);
48 CREATE TABLE vandelay.queue (
49 id BIGSERIAL PRIMARY KEY,
50 owner INT NOT NULL REFERENCES actor.usr (id) DEFERRABLE INITIALLY DEFERRED,
52 complete BOOL NOT NULL DEFAULT FALSE,
53 queue_type TEXT NOT NULL DEFAULT 'bib' CHECK (queue_type IN ('bib','authority')),
54 match_set INT REFERENCES vandelay.match_set (id) ON UPDATE CASCADE ON DELETE SET NULL DEFERRABLE INITIALLY DEFERRED,
55 CONSTRAINT vand_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
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,
109 CONSTRAINT vand_import_item_attr_def_idx UNIQUE (owner,name)
112 CREATE TABLE vandelay.import_error (
113 code TEXT PRIMARY KEY,
114 description TEXT NOT NULL -- i18n
117 CREATE TABLE vandelay.bib_queue (
118 queue_type TEXT NOT NULL DEFAULT 'bib' CHECK (queue_type = 'bib'),
119 item_attr_def BIGINT REFERENCES vandelay.import_item_attr_definition (id) ON DELETE SET NULL DEFERRABLE INITIALLY DEFERRED,
120 CONSTRAINT vand_bib_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
121 ) INHERITS (vandelay.queue);
122 ALTER TABLE vandelay.bib_queue ADD PRIMARY KEY (id);
124 CREATE TABLE vandelay.queued_bib_record (
125 queue INT NOT NULL REFERENCES vandelay.bib_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
126 bib_source INT REFERENCES config.bib_source (id) DEFERRABLE INITIALLY DEFERRED,
127 imported_as BIGINT REFERENCES biblio.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
128 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
130 ) INHERITS (vandelay.queued_record);
131 ALTER TABLE vandelay.queued_bib_record ADD PRIMARY KEY (id);
132 CREATE INDEX queued_bib_record_queue_idx ON vandelay.queued_bib_record (queue);
134 CREATE TABLE vandelay.queued_bib_record_attr (
135 id BIGSERIAL PRIMARY KEY,
136 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) DEFERRABLE INITIALLY DEFERRED,
137 field INT NOT NULL REFERENCES vandelay.bib_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
138 attr_value TEXT NOT NULL
140 CREATE INDEX queued_bib_record_attr_record_idx ON vandelay.queued_bib_record_attr (record);
142 CREATE TABLE vandelay.bib_match (
143 id BIGSERIAL PRIMARY KEY,
144 queued_record BIGINT REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
145 eg_record BIGINT REFERENCES biblio.record_entry (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
146 quality INT NOT NULL DEFAULT 1,
147 match_score INT NOT NULL DEFAULT 0
150 CREATE TABLE vandelay.import_item (
151 id BIGSERIAL PRIMARY KEY,
152 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
153 definition BIGINT NOT NULL REFERENCES vandelay.import_item_attr_definition (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
154 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
156 imported_as BIGINT REFERENCES asset.copy (id) DEFERRABLE INITIALLY DEFERRED,
157 import_time TIMESTAMP WITH TIME ZONE,
166 deposit_amount NUMERIC(8,2),
179 CREATE TABLE vandelay.import_bib_trash_fields (
180 id BIGSERIAL PRIMARY KEY,
181 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
183 CONSTRAINT vand_import_bib_trash_fields_idx UNIQUE (owner,field)
186 CREATE TABLE vandelay.merge_profile (
187 id BIGSERIAL PRIMARY KEY,
188 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
195 CONSTRAINT vand_merge_prof_owner_name_idx UNIQUE (owner,name),
196 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))
199 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
206 retval config.marc21_rec_type_map%ROWTYPE;
208 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
210 IF ldr IS NULL OR ldr = '' THEN
211 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
215 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
216 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
219 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
220 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
222 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
224 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
227 IF retval.code IS NULL THEN
228 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
233 $func$ LANGUAGE PLPGSQL;
235 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
242 rtype := (vandelay.marc21_record_type( marc )).code;
243 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
244 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
245 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
248 val := REPEAT( ff_pos.default_val, ff_pos.length );
254 $func$ LANGUAGE PLPGSQL;
256 CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
257 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
262 output biblio.record_ff_map%ROWTYPE;
264 rtype := (vandelay.marc21_record_type( marc )).code;
266 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
267 output.ff_name := ff_pos.fixed_field;
268 output.ff_value := NULL;
270 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(tag) || '"]/text()', marc ) ) x(value) LOOP
271 output.ff_value := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
272 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
274 output.ff_value := NULL;
281 $func$ LANGUAGE PLPGSQL;
283 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
284 CREATE OR REPLACE FUNCTION vandelay.marc21_physical_characteristics( marc TEXT) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
288 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
289 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
290 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
291 retval biblio.marc21_physical_characteristics%ROWTYPE;
294 _007 := oils_xpath_string( '//*[@tag="007"]', marc );
296 IF _007 IS NOT NULL AND _007 <> '' THEN
297 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007, 1, 1 );
299 IF ptype.ptype_key IS NOT NULL THEN
300 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
301 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 );
303 IF pval.id IS NOT NULL THEN
306 retval.ptype := ptype.ptype_key;
307 retval.subfield := psf.id;
308 retval.value := pval.id;
318 $func$ LANGUAGE PLPGSQL;
320 CREATE TYPE vandelay.flat_marc AS ( tag CHAR(3), ind1 TEXT, ind2 TEXT, subfield TEXT, value TEXT );
321 CREATE OR REPLACE FUNCTION vandelay.flay_marc ( TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
324 use MARC::File::XML (BinaryEncoding => 'UTF-8');
328 MARC::Charset->assume_unicode(1);
331 my $r = MARC::Record->new_from_xml( $xml );
333 return_next( { tag => 'LDR', value => $r->leader } );
335 for my $f ( $r->fields ) {
336 if ($f->is_control_field) {
337 return_next({ tag => $f->tag, value => $f->data });
339 for my $s ($f->subfields) {
342 ind1 => $f->indicator(1),
343 ind2 => $f->indicator(2),
348 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
349 my $trim = $f->indicator(2) || 0;
352 ind1 => $f->indicator(1),
353 ind2 => $f->indicator(2),
355 value => substr( $s->[1], $trim )
364 $func$ LANGUAGE PLPERLU;
366 CREATE OR REPLACE FUNCTION vandelay.flatten_marc ( marc TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
368 output vandelay.flat_marc%ROWTYPE;
371 FOR field IN SELECT * FROM vandelay.flay_marc( marc ) LOOP
372 output.ind1 := field.ind1;
373 output.ind2 := field.ind2;
374 output.tag := field.tag;
375 output.subfield := field.subfield;
376 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
377 output.value := naco_normalize(field.value, field.subfield);
379 output.value := field.value;
382 CONTINUE WHEN output.value IS NULL;
387 $func$ LANGUAGE PLPGSQL;
389 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT, attr_defs TEXT[]) RETURNS hstore AS $_$
391 transformed_xml TEXT;
394 xfrm config.xml_transform%ROWTYPE;
396 new_attrs HSTORE := ''::HSTORE;
397 attr_def config.record_attr_definition%ROWTYPE;
400 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE name IN (SELECT * FROM UNNEST(attr_defs)) ORDER BY format LOOP
402 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
403 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(x.value), COALESCE(attr_def.joiner,' ')) INTO attr_value
404 FROM vandelay.flatten_marc(xml) AS x
405 WHERE x.tag LIKE attr_def.tag
407 WHEN attr_def.sf_list IS NOT NULL
408 THEN POSITION(x.subfield IN attr_def.sf_list) > 0
415 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
416 attr_value := vandelay.marc21_extract_fixed_field(xml, attr_def.fixed_field);
418 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
420 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
422 -- See if we can skip the XSLT ... it's expensive
423 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
424 -- Can't skip the transform
425 IF xfrm.xslt <> '---' THEN
426 transformed_xml := oils_xslt_process(xml,xfrm.xslt);
428 transformed_xml := xml;
431 prev_xfrm := xfrm.name;
434 IF xfrm.name IS NULL THEN
435 -- just grab the marcxml (empty) transform
436 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
437 prev_xfrm := xfrm.name;
440 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
442 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
443 SELECT value::TEXT INTO attr_value
444 FROM vandelay.marc21_physical_characteristics(xml)
445 WHERE subfield = attr_def.phys_char_sf
446 LIMIT 1; -- Just in case ...
450 -- apply index normalizers to attr_value
452 SELECT n.func AS func,
453 n.param_count AS param_count,
455 FROM config.index_normalizer n
456 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
457 WHERE attr = attr_def.name
459 EXECUTE 'SELECT ' || normalizer.func || '(' ||
460 quote_literal( attr_value ) ||
462 WHEN normalizer.param_count > 0
463 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
470 -- Add the new value to the hstore
471 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
477 $_$ LANGUAGE PLPGSQL;
479 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT ) RETURNS hstore AS $_$
480 SELECT vandelay.extract_rec_attrs( $1, (SELECT ARRAY_ACCUM(name) FROM config.record_attr_definition));
483 -- Everything between this comment and the beginning of the definition of
484 -- vandelay.match_bib_record() is strictly in service of that function.
485 CREATE TYPE vandelay.match_set_test_result AS (record BIGINT, quality INTEGER);
487 CREATE OR REPLACE FUNCTION vandelay.match_set_test_marcxml(
488 match_set_id INTEGER, record_xml TEXT
489 ) RETURNS SETOF vandelay.match_set_test_result AS $$
500 tags_rstore := vandelay.flatten_marc_hstore(record_xml);
501 svf_rstore := vandelay.extract_rec_attrs(record_xml);
503 CREATE TEMPORARY TABLE _vandelay_tmp_qrows (q INTEGER);
504 CREATE TEMPORARY TABLE _vandelay_tmp_jrows (j TEXT);
506 -- generate the where clause and return that directly (into wq), and as
507 -- a side-effect, populate the _vandelay_tmp_[qj]rows tables.
508 wq := vandelay.get_expr_from_match_set(match_set_id);
510 query_ := 'SELECT bre.id AS record, ';
512 -- qrows table is for the quality bits we add to the SELECT clause
513 SELECT ARRAY_TO_STRING(
514 ARRAY_ACCUM('COALESCE(n' || q::TEXT || '.quality, 0)'), ' + '
515 ) INTO coal FROM _vandelay_tmp_qrows;
517 -- our query string so far is the SELECT clause and the inital FROM.
518 -- no JOINs yet nor the WHERE clause
519 query_ := query_ || coal || ' AS quality ' || E'\n' ||
520 'FROM biblio.record_entry bre ';
522 -- jrows table is for the joins we must make (and the real text conditions)
523 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(j), E'\n') INTO joins
524 FROM _vandelay_tmp_jrows;
526 -- add those joins and the where clause to our query.
527 query_ := query_ || joins || E'\n' || 'WHERE ' || wq || ' AND not bre.deleted';
529 -- this will return rows of record,quality
530 FOR rec IN EXECUTE query_ USING tags_rstore, svf_rstore LOOP
534 DROP TABLE _vandelay_tmp_qrows;
535 DROP TABLE _vandelay_tmp_jrows;
541 CREATE OR REPLACE FUNCTION vandelay.flatten_marc_hstore(
543 ) RETURNS HSTORE AS $$
547 ARRAY_ACCUM(tag || (COALESCE(subfield, ''))),
551 SELECT tag, subfield, ARRAY_ACCUM(value)::TEXT AS value
552 FROM vandelay.flatten_marc(record_xml)
553 GROUP BY tag, subfield ORDER BY tag, subfield
559 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set(
563 root vandelay.match_set_point;
565 SELECT * INTO root FROM vandelay.match_set_point
566 WHERE parent IS NULL AND match_set = match_set_id;
568 RETURN vandelay.get_expr_from_match_set_point(root);
572 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set_point(
573 node vandelay.match_set_point
580 child vandelay.match_set_point;
582 SELECT ARRAY_ACCUM(id) INTO children FROM vandelay.match_set_point
583 WHERE parent = node.id;
585 IF ARRAY_LENGTH(children, 1) > 0 THEN
586 this_op := vandelay._get_expr_render_one(node);
589 WHILE children[i] IS NOT NULL LOOP
590 SELECT * INTO child FROM vandelay.match_set_point
591 WHERE id = children[i];
593 q := q || ' ' || this_op || ' ';
596 q := q || vandelay.get_expr_from_match_set_point(child);
600 ELSIF node.bool_op IS NULL THEN
601 PERFORM vandelay._get_expr_push_qrow(node);
602 PERFORM vandelay._get_expr_push_jrow(node);
603 RETURN vandelay._get_expr_render_one(node);
610 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_qrow(
611 node vandelay.match_set_point
615 INSERT INTO _vandelay_tmp_qrows (q) VALUES (node.id);
619 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_jrow(
620 node vandelay.match_set_point
634 IF node.tag IS NOT NULL THEN
636 IF node.subfield IS NOT NULL THEN
637 tagkey := tagkey || node.subfield;
641 my_alias := 'n' || node.id::TEXT;
643 jrow := 'LEFT JOIN (SELECT *, ' || node.quality ||
644 ' AS quality FROM metabib.';
645 IF node.tag IS NOT NULL THEN
646 jrow := jrow || 'full_rec) ' || my_alias || ' ON (' ||
647 my_alias || '.record = bre.id AND ' || my_alias || '.tag = ''' ||
649 IF node.subfield IS NOT NULL THEN
650 jrow := jrow || ' AND ' || my_alias || '.subfield = ''' ||
651 node.subfield || '''';
653 jrow := jrow || ' AND (' || my_alias || '.value ' || op ||
654 ' ANY(($1->''' || tagkey || ''')::TEXT[])))';
656 jrow := jrow || 'record_attr) ' || my_alias || ' ON (' ||
657 my_alias || '.id = bre.id AND (' ||
658 my_alias || '.attrs->''' || node.svf ||
659 ''' ' || op || ' $2->''' || node.svf || '''))';
661 INSERT INTO _vandelay_tmp_jrows (j) VALUES (jrow);
665 CREATE OR REPLACE FUNCTION vandelay._get_expr_render_one(
666 node vandelay.match_set_point
671 IF node.bool_op IS NOT NULL THEN
674 RETURN '(n' || node.id::TEXT || '.id IS NOT NULL)';
679 CREATE OR REPLACE FUNCTION vandelay.match_bib_record() RETURNS TRIGGER AS $func$
681 incoming_existing_id TEXT;
682 test_result vandelay.match_set_test_result%ROWTYPE;
686 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
690 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
692 SELECT q.match_set INTO match_set FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
694 IF match_set IS NOT NULL THEN
695 NEW.quality := vandelay.measure_record_quality( NEW.marc, match_set );
698 -- Perfect matches on 901$c exit early with a match with high quality.
699 incoming_existing_id :=
700 oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]', NEW.marc);
702 IF incoming_existing_id IS NOT NULL AND incoming_existing_id != '' THEN
703 SELECT id INTO tmp_rec FROM biblio.record_entry WHERE id = incoming_existing_id::bigint;
704 IF tmp_rec IS NOT NULL THEN
705 INSERT INTO vandelay.bib_match (queued_record, eg_record, match_score, quality)
710 -- note: no match_set means quality==0
711 vandelay.measure_record_quality( b.marc, match_set )
712 FROM biblio.record_entry b
713 WHERE id = incoming_existing_id::bigint;
717 IF match_set IS NULL THEN
721 FOR test_result IN SELECT * FROM
722 vandelay.match_set_test_marcxml(match_set, NEW.marc) LOOP
724 INSERT INTO vandelay.bib_match ( queued_record, eg_record, match_score, quality )
729 vandelay.measure_record_quality( b.marc, match_set )
730 FROM biblio.record_entry b
731 WHERE id = test_result.record;
737 $func$ LANGUAGE PLPGSQL;
739 CREATE OR REPLACE FUNCTION vandelay.measure_record_quality ( xml TEXT, match_set_id INT ) RETURNS INT AS $_$
743 test vandelay.match_set_quality%ROWTYPE;
746 FOR test IN SELECT * FROM vandelay.match_set_quality WHERE match_set = match_set_id LOOP
747 IF test.tag IS NOT NULL THEN
748 FOR rvalue IN SELECT value FROM vandelay.flatten_marc( xml ) WHERE tag = test.tag AND subfield = test.subfield LOOP
749 IF test.value = rvalue THEN
750 out_q := out_q + test.quality;
754 IF test.value = vandelay.extract_rec_attrs(xml, ARRAY[test.svf]) -> test.svf THEN
755 out_q := out_q + test.quality;
762 $_$ LANGUAGE PLPGSQL;
764 CREATE TYPE vandelay.tcn_data AS (tcn TEXT, tcn_source TEXT, used BOOL);
765 CREATE OR REPLACE FUNCTION vandelay.find_bib_tcn_data ( xml TEXT ) RETURNS SETOF vandelay.tcn_data AS $_$
769 output vandelay.tcn_data%ROWTYPE;
773 eg_tcn := BTRIM((oils_xpath('//*[@tag="001"]/text()',xml))[1]);
774 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
776 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="003"]/text()',xml))[1]);
777 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
778 eg_tcn_source := 'System Local';
781 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
784 output.used := FALSE;
789 output.tcn := eg_tcn;
790 output.tcn_source := eg_tcn_source;
796 eg_tcn := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="a"]/text()',xml))[1]);
797 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
799 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="b"]/text()',xml))[1]);
800 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
801 eg_tcn_source := 'System Local';
804 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
807 output.used := FALSE;
812 output.tcn := eg_tcn;
813 output.tcn_source := eg_tcn_source;
819 eg_tcn := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="a"]/text()',xml))[1]);
820 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
822 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="b"]/text()',xml))[1]);
823 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
824 eg_tcn_source := 'System Local';
827 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
830 output.used := FALSE;
835 output.tcn := eg_tcn;
836 output.tcn_source := eg_tcn_source;
842 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="020"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
843 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
845 eg_tcn_source := 'ISBN';
847 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
850 output.used := FALSE;
855 output.tcn := eg_tcn;
856 output.tcn_source := eg_tcn_source;
862 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="022"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
863 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
865 eg_tcn_source := 'ISSN';
867 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
870 output.used := FALSE;
875 output.tcn := eg_tcn;
876 output.tcn_source := eg_tcn_source;
882 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="010"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
883 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
885 eg_tcn_source := 'LCCN';
887 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
890 output.used := FALSE;
895 output.tcn := eg_tcn;
896 output.tcn_source := eg_tcn_source;
902 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="035"]/*[@code="a"]/text()',xml))[1], $re$^.*?(\w+)$$re$, $re$\1$re$);
903 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
905 eg_tcn_source := 'System Legacy';
907 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
910 output.used := FALSE;
915 output.tcn := eg_tcn;
916 output.tcn_source := eg_tcn_source;
923 $_$ LANGUAGE PLPGSQL;
925 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT, force_add INT ) RETURNS TEXT AS $_$
928 use MARC::File::XML (BinaryEncoding => 'UTF-8');
932 MARC::Charset->assume_unicode(1);
934 my $target_xml = shift;
935 my $source_xml = shift;
936 my $field_spec = shift;
937 my $force_add = shift || 0;
939 my $target_r = MARC::Record->new_from_xml( $target_xml );
940 my $source_r = MARC::Record->new_from_xml( $source_xml );
942 return $target_xml unless ($target_r && $source_r);
944 my @field_list = split(',', $field_spec);
947 for my $f (@field_list) {
948 $f =~ s/^\s*//; $f =~ s/\s*$//;
949 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
955 $match =~ s/^\s*//; $match =~ s/\s*$//;
956 $fields{$field} = { sf => [ split('', $sf) ] };
958 my ($msf,$mre) = split('~', $match);
959 if (length($msf) > 0 and length($mre) > 0) {
960 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
961 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
962 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
968 for my $f ( keys %fields) {
969 if ( @{$fields{$f}{sf}} ) {
970 for my $from_field ($source_r->field( $f )) {
971 my @tos = $target_r->field( $f );
973 next if (exists($fields{$f}{match}) and !$force_add);
974 my @new_fields = map { $_->clone } $source_r->field( $f );
975 $target_r->insert_fields_ordered( @new_fields );
977 for my $to_field (@tos) {
978 if (exists($fields{$f}{match})) {
979 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
981 my @new_sf = map { ($_ => $from_field->subfield($_)) } @{$fields{$f}{sf}};
982 $to_field->add_subfields( @new_sf );
987 my @new_fields = map { $_->clone } $source_r->field( $f );
988 $target_r->insert_fields_ordered( @new_fields );
992 $target_xml = $target_r->as_xml_record;
993 $target_xml =~ s/^<\?.+?\?>$//mo;
994 $target_xml =~ s/\n//sgo;
995 $target_xml =~ s/>\s+</></sgo;
999 $_$ LANGUAGE PLPERLU;
1001 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1002 SELECT vandelay.add_field( $1, $2, $3, 0 );
1005 CREATE OR REPLACE FUNCTION vandelay.strip_field ( xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1008 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1012 MARC::Charset->assume_unicode(1);
1015 my $r = MARC::Record->new_from_xml( $xml );
1017 return $xml unless ($r);
1019 my $field_spec = shift;
1020 my @field_list = split(',', $field_spec);
1023 for my $f (@field_list) {
1024 $f =~ s/^\s*//; $f =~ s/\s*$//;
1025 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1031 $match =~ s/^\s*//; $match =~ s/\s*$//;
1032 $fields{$field} = { sf => [ split('', $sf) ] };
1034 my ($msf,$mre) = split('~', $match);
1035 if (length($msf) > 0 and length($mre) > 0) {
1036 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1037 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1038 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1044 for my $f ( keys %fields) {
1045 for my $to_field ($r->field( $f )) {
1046 if (exists($fields{$f}{match})) {
1047 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1050 if ( @{$fields{$f}{sf}} ) {
1051 $to_field->delete_subfield(code => $fields{$f}{sf});
1053 $r->delete_field( $to_field );
1058 $xml = $r->as_xml_record;
1059 $xml =~ s/^<\?.+?\?>$//mo;
1061 $xml =~ s/>\s+</></sgo;
1065 $_$ LANGUAGE PLPERLU;
1067 CREATE OR REPLACE FUNCTION vandelay.replace_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1074 parsed_target := vandelay.strip_field( target_xml, ''); -- this dance normalizes the format of the xml for the IF below
1076 FOR curr_field IN SELECT UNNEST( STRING_TO_ARRAY(field, ',') ) LOOP -- naive split, but it's the same we use in the perl
1078 xml_output := vandelay.strip_field( parsed_target, curr_field);
1080 IF xml_output <> parsed_target AND curr_field ~ E'~' THEN
1081 -- we removed something, and there was a regexp restriction in the curr_field definition, so proceed
1082 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 1 );
1083 ELSIF curr_field !~ E'~' THEN
1084 -- No regexp restriction, add the curr_field
1085 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 0 );
1088 parsed_target := xml_output; -- in prep for any following loop iterations
1094 $_$ LANGUAGE PLPGSQL;
1096 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 $_$
1097 SELECT vandelay.replace_field( vandelay.add_field( vandelay.strip_field( $1, $5) , $2, $3 ), $2, $4);
1100 CREATE TYPE vandelay.compile_profile AS (add_rule TEXT, replace_rule TEXT, preserve_rule TEXT, strip_rule TEXT);
1101 CREATE OR REPLACE FUNCTION vandelay.compile_profile ( incoming_xml TEXT ) RETURNS vandelay.compile_profile AS $_$
1103 output vandelay.compile_profile%ROWTYPE;
1104 profile vandelay.merge_profile%ROWTYPE;
1106 profile_tmpl_owner TEXT;
1107 add_rule TEXT := '';
1108 strip_rule TEXT := '';
1109 replace_rule TEXT := '';
1110 preserve_rule TEXT := '';
1114 profile_tmpl := (oils_xpath('//*[@tag="905"]/*[@code="t"]/text()',incoming_xml))[1];
1115 profile_tmpl_owner := (oils_xpath('//*[@tag="905"]/*[@code="o"]/text()',incoming_xml))[1];
1117 IF profile_tmpl IS NOT NULL AND profile_tmpl <> '' AND profile_tmpl_owner IS NOT NULL AND profile_tmpl_owner <> '' THEN
1118 SELECT p.* INTO profile
1119 FROM vandelay.merge_profile p
1120 JOIN actor.org_unit u ON (u.id = p.owner)
1121 WHERE p.name = profile_tmpl
1122 AND u.shortname = profile_tmpl_owner;
1124 IF profile.id IS NOT NULL THEN
1125 add_rule := COALESCE(profile.add_spec,'');
1126 strip_rule := COALESCE(profile.strip_spec,'');
1127 replace_rule := COALESCE(profile.replace_spec,'');
1128 preserve_rule := COALESCE(profile.preserve_spec,'');
1132 add_rule := add_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="a"]/text()',incoming_xml),','),'');
1133 strip_rule := strip_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="d"]/text()',incoming_xml),','),'');
1134 replace_rule := replace_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="r"]/text()',incoming_xml),','),'');
1135 preserve_rule := preserve_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="p"]/text()',incoming_xml),','),'');
1137 output.add_rule := BTRIM(add_rule,',');
1138 output.replace_rule := BTRIM(replace_rule,',');
1139 output.strip_rule := BTRIM(strip_rule,',');
1140 output.preserve_rule := BTRIM(preserve_rule,',');
1144 $_$ LANGUAGE PLPGSQL;
1146 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1148 merge_profile vandelay.merge_profile%ROWTYPE;
1149 dyn_profile vandelay.compile_profile%ROWTYPE;
1159 SELECT b.marc INTO eg_marc
1160 FROM biblio.record_entry b
1164 IF eg_marc IS NULL OR v_marc IS NULL THEN
1165 -- RAISE NOTICE 'no marc for template or bib record';
1169 dyn_profile := vandelay.compile_profile( v_marc );
1171 IF merge_profile_id IS NOT NULL THEN
1172 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1174 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1175 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1176 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1177 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1181 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1182 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1186 IF dyn_profile.replace_rule <> '' THEN
1187 source_marc = v_marc;
1188 target_marc = eg_marc;
1189 replace_rule = dyn_profile.replace_rule;
1191 source_marc = eg_marc;
1192 target_marc = v_marc;
1193 replace_rule = dyn_profile.preserve_rule;
1196 UPDATE biblio.record_entry
1197 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1201 -- RAISE NOTICE 'update of biblio.record_entry failed';
1208 $$ LANGUAGE PLPGSQL;
1210 CREATE OR REPLACE FUNCTION vandelay.merge_record_xml ( target_marc TEXT, template_marc TEXT ) RETURNS TEXT AS $$
1212 dyn_profile vandelay.compile_profile%ROWTYPE;
1220 IF target_marc IS NULL OR template_marc IS NULL THEN
1221 -- RAISE NOTICE 'no marc for target or template record';
1225 dyn_profile := vandelay.compile_profile( template_marc );
1227 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1228 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1232 IF dyn_profile.replace_rule <> '' THEN
1233 trgt_marc = target_marc;
1234 tmpl_marc = template_marc;
1235 replace_rule = dyn_profile.replace_rule;
1237 tmp_marc = target_marc;
1238 trgt_marc = template_marc;
1239 tmpl_marc = tmp_marc;
1240 replace_rule = dyn_profile.preserve_rule;
1243 RETURN vandelay.merge_record_xml( trgt_marc, tmpl_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule );
1246 $$ LANGUAGE PLPGSQL;
1248 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT) RETURNS BOOL AS $$
1249 SELECT vandelay.template_overlay_bib_record( $1, $2, NULL);
1252 CREATE OR REPLACE FUNCTION vandelay.overlay_bib_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1254 merge_profile vandelay.merge_profile%ROWTYPE;
1255 dyn_profile vandelay.compile_profile%ROWTYPE;
1265 SELECT q.marc INTO v_marc
1266 FROM vandelay.queued_record q
1267 JOIN vandelay.bib_match m ON (m.queued_record = q.id AND q.id = import_id)
1270 IF v_marc IS NULL THEN
1271 -- RAISE NOTICE 'no marc for vandelay or bib record';
1275 IF vandelay.template_overlay_bib_record( v_marc, eg_id, merge_profile_id) THEN
1276 UPDATE vandelay.queued_bib_record
1277 SET imported_as = eg_id,
1279 WHERE id = import_id;
1281 editor_string := (oils_xpath('//*[@tag="905"]/*[@code="u"]/text()',v_marc))[1];
1283 IF editor_string IS NOT NULL AND editor_string <> '' THEN
1284 SELECT usr INTO editor_id FROM actor.card WHERE barcode = editor_string;
1286 IF editor_id IS NULL THEN
1287 SELECT id INTO editor_id FROM actor.usr WHERE usrname = editor_string;
1290 IF editor_id IS NOT NULL THEN
1291 UPDATE biblio.record_entry SET editor = editor_id WHERE id = eg_id;
1298 -- RAISE NOTICE 'update of biblio.record_entry failed';
1303 $$ LANGUAGE PLPGSQL;
1305 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
1308 lwm_ratio_value NUMERIC;
1311 lwm_ratio_value := COALESCE(lwm_ratio_value_p, 0.0);
1313 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1316 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1320 SELECT m.eg_record INTO eg_id
1321 FROM vandelay.bib_match m
1322 JOIN vandelay.queued_bib_record qr ON (m.queued_record = qr.id)
1323 JOIN vandelay.bib_queue q ON (qr.queue = q.id)
1324 JOIN biblio.record_entry r ON (r.id = m.eg_record)
1325 WHERE m.queued_record = import_id
1326 AND qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC >= lwm_ratio_value
1327 ORDER BY m.match_score DESC, -- required match score
1328 qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC DESC, -- quality tie breaker
1329 m.id -- when in doubt, use the first match
1332 IF eg_id IS NULL THEN
1333 -- RAISE NOTICE 'incoming record is not of high enough quality';
1337 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1339 $$ LANGUAGE PLPGSQL;
1341 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1342 SELECT vandelay.auto_overlay_bib_record_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1345 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1351 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1354 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1358 SELECT COUNT(*) INTO match_count FROM vandelay.bib_match WHERE queued_record = import_id;
1360 IF match_count <> 1 THEN
1361 -- RAISE NOTICE 'not an exact match';
1365 -- Check that the one match is on the first 901c
1366 SELECT m.eg_record INTO eg_id
1367 FROM vandelay.queued_bib_record q
1368 JOIN vandelay.bib_match m ON (m.queued_record = q.id)
1369 WHERE q.id = import_id
1370 AND m.eg_record = oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]',marc)::BIGINT;
1373 -- RAISE NOTICE 'not a 901c match';
1377 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1379 $$ LANGUAGE PLPGSQL;
1381 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1383 queued_record vandelay.queued_bib_record%ROWTYPE;
1386 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1388 IF vandelay.auto_overlay_bib_record( queued_record.id, merge_profile_id ) THEN
1389 RETURN NEXT queued_record.id;
1397 $$ LANGUAGE PLPGSQL;
1399 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 $$
1401 queued_record vandelay.queued_bib_record%ROWTYPE;
1404 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1406 IF vandelay.auto_overlay_bib_record_with_best( queued_record.id, merge_profile_id, lwm_ratio_value ) THEN
1407 RETURN NEXT queued_record.id;
1415 $$ LANGUAGE PLPGSQL;
1417 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1418 SELECT vandelay.auto_overlay_bib_queue_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1421 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1422 SELECT * FROM vandelay.auto_overlay_bib_queue( $1, NULL );
1425 CREATE OR REPLACE FUNCTION vandelay.ingest_items ( import_id BIGINT, attr_def_id BIGINT ) RETURNS SETOF vandelay.import_item AS $$
1436 deposit_amount TEXT;
1449 tmp_attr_set RECORD;
1450 attr_set vandelay.import_item%ROWTYPE;
1456 SELECT * INTO attr_def FROM vandelay.import_item_attr_definition WHERE id = attr_def_id;
1460 attr_set.definition := attr_def.id;
1462 -- Build the combined XPath
1466 WHEN attr_def.owning_lib IS NULL THEN 'null()'
1467 WHEN LENGTH( attr_def.owning_lib ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.owning_lib || '"]'
1468 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.owning_lib
1473 WHEN attr_def.circ_lib IS NULL THEN 'null()'
1474 WHEN LENGTH( attr_def.circ_lib ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.circ_lib || '"]'
1475 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.circ_lib
1480 WHEN attr_def.call_number IS NULL THEN 'null()'
1481 WHEN LENGTH( attr_def.call_number ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.call_number || '"]'
1482 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.call_number
1487 WHEN attr_def.copy_number IS NULL THEN 'null()'
1488 WHEN LENGTH( attr_def.copy_number ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.copy_number || '"]'
1489 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.copy_number
1494 WHEN attr_def.status IS NULL THEN 'null()'
1495 WHEN LENGTH( attr_def.status ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.status || '"]'
1496 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.status
1501 WHEN attr_def.location IS NULL THEN 'null()'
1502 WHEN LENGTH( attr_def.location ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.location || '"]'
1503 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.location
1508 WHEN attr_def.circulate IS NULL THEN 'null()'
1509 WHEN LENGTH( attr_def.circulate ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.circulate || '"]'
1510 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.circulate
1515 WHEN attr_def.deposit IS NULL THEN 'null()'
1516 WHEN LENGTH( attr_def.deposit ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.deposit || '"]'
1517 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.deposit
1522 WHEN attr_def.deposit_amount IS NULL THEN 'null()'
1523 WHEN LENGTH( attr_def.deposit_amount ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.deposit_amount || '"]'
1524 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.deposit_amount
1529 WHEN attr_def.ref IS NULL THEN 'null()'
1530 WHEN LENGTH( attr_def.ref ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.ref || '"]'
1531 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.ref
1536 WHEN attr_def.holdable IS NULL THEN 'null()'
1537 WHEN LENGTH( attr_def.holdable ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.holdable || '"]'
1538 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.holdable
1543 WHEN attr_def.price IS NULL THEN 'null()'
1544 WHEN LENGTH( attr_def.price ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.price || '"]'
1545 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.price
1550 WHEN attr_def.barcode IS NULL THEN 'null()'
1551 WHEN LENGTH( attr_def.barcode ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.barcode || '"]'
1552 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.barcode
1557 WHEN attr_def.circ_modifier IS NULL THEN 'null()'
1558 WHEN LENGTH( attr_def.circ_modifier ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.circ_modifier || '"]'
1559 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.circ_modifier
1564 WHEN attr_def.circ_as_type IS NULL THEN 'null()'
1565 WHEN LENGTH( attr_def.circ_as_type ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.circ_as_type || '"]'
1566 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.circ_as_type
1571 WHEN attr_def.alert_message IS NULL THEN 'null()'
1572 WHEN LENGTH( attr_def.alert_message ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.alert_message || '"]'
1573 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.alert_message
1578 WHEN attr_def.opac_visible IS NULL THEN 'null()'
1579 WHEN LENGTH( attr_def.opac_visible ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.opac_visible || '"]'
1580 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.opac_visible
1585 WHEN attr_def.pub_note IS NULL THEN 'null()'
1586 WHEN LENGTH( attr_def.pub_note ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.pub_note || '"]'
1587 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.pub_note
1591 WHEN attr_def.priv_note IS NULL THEN 'null()'
1592 WHEN LENGTH( attr_def.priv_note ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.priv_note || '"]'
1593 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.priv_note
1598 owning_lib || '|' ||
1600 call_number || '|' ||
1601 copy_number || '|' ||
1606 deposit_amount || '|' ||
1611 circ_modifier || '|' ||
1612 circ_as_type || '|' ||
1613 alert_message || '|' ||
1618 -- RAISE NOTICE 'XPath: %', xpath;
1622 FROM oils_xpath_table( 'id', 'marc', 'vandelay.queued_bib_record', xpath, 'id = ' || import_id )
1623 AS t( id INT, ol TEXT, clib TEXT, cn TEXT, cnum TEXT, cs TEXT, cl TEXT, circ TEXT,
1624 dep TEXT, dep_amount TEXT, r TEXT, hold TEXT, pr TEXT, bc TEXT, circ_mod TEXT,
1625 circ_as TEXT, amessage TEXT, note TEXT, pnote TEXT, opac_vis TEXT )
1628 tmp_attr_set.pr = REGEXP_REPLACE(tmp_attr_set.pr, E'[^0-9\\.]', '', 'g');
1629 tmp_attr_set.dep_amount = REGEXP_REPLACE(tmp_attr_set.dep_amount, E'[^0-9\\.]', '', 'g');
1631 tmp_attr_set.pr := NULLIF( tmp_attr_set.pr, '' );
1632 tmp_attr_set.dep_amount := NULLIF( tmp_attr_set.dep_amount, '' );
1634 SELECT id INTO attr_set.owning_lib FROM actor.org_unit WHERE shortname = UPPER(tmp_attr_set.ol); -- INT
1635 SELECT id INTO attr_set.circ_lib FROM actor.org_unit WHERE shortname = UPPER(tmp_attr_set.clib); -- INT
1636 SELECT id INTO attr_set.status FROM config.copy_status WHERE LOWER(name) = LOWER(tmp_attr_set.cs); -- INT
1638 SELECT id INTO attr_set.location
1639 FROM asset.copy_location
1640 WHERE LOWER(name) = LOWER(tmp_attr_set.cl)
1641 AND asset.copy_location.owning_lib = COALESCE(attr_set.owning_lib, attr_set.circ_lib); -- INT
1643 attr_set.circulate :=
1644 LOWER( SUBSTRING( tmp_attr_set.circ, 1, 1)) IN ('t','y','1')
1645 OR LOWER(tmp_attr_set.circ) = 'circulating'; -- BOOL
1648 LOWER( SUBSTRING( tmp_attr_set.dep, 1, 1 ) ) IN ('t','y','1')
1649 OR LOWER(tmp_attr_set.dep) = 'deposit'; -- BOOL
1651 attr_set.holdable :=
1652 LOWER( SUBSTRING( tmp_attr_set.hold, 1, 1 ) ) IN ('t','y','1')
1653 OR LOWER(tmp_attr_set.hold) = 'holdable'; -- BOOL
1655 attr_set.opac_visible :=
1656 LOWER( SUBSTRING( tmp_attr_set.opac_vis, 1, 1 ) ) IN ('t','y','1')
1657 OR LOWER(tmp_attr_set.opac_vis) = 'visible'; -- BOOL
1660 LOWER( SUBSTRING( tmp_attr_set.r, 1, 1 ) ) IN ('t','y','1')
1661 OR LOWER(tmp_attr_set.r) = 'reference'; -- BOOL
1663 attr_set.copy_number := tmp_attr_set.cnum::INT; -- INT,
1664 attr_set.deposit_amount := tmp_attr_set.dep_amount::NUMERIC(6,2); -- NUMERIC(6,2),
1665 attr_set.price := tmp_attr_set.pr::NUMERIC(8,2); -- NUMERIC(8,2),
1667 attr_set.call_number := tmp_attr_set.cn; -- TEXT
1668 attr_set.barcode := tmp_attr_set.bc; -- TEXT,
1669 attr_set.circ_modifier := tmp_attr_set.circ_mod; -- TEXT,
1670 attr_set.circ_as_type := tmp_attr_set.circ_as; -- TEXT,
1671 attr_set.alert_message := tmp_attr_set.amessage; -- TEXT,
1672 attr_set.pub_note := tmp_attr_set.note; -- TEXT,
1673 attr_set.priv_note := tmp_attr_set.pnote; -- TEXT,
1674 attr_set.alert_message := tmp_attr_set.amessage; -- TEXT,
1676 RETURN NEXT attr_set;
1685 $$ LANGUAGE PLPGSQL;
1688 CREATE OR REPLACE FUNCTION vandelay.ingest_bib_marc ( ) RETURNS TRIGGER AS $$
1694 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1698 FOR adef IN SELECT * FROM vandelay.bib_attr_definition LOOP
1700 SELECT extract_marc_field('vandelay.queued_bib_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_bib_record WHERE id = NEW.id;
1701 IF (value IS NOT NULL AND value <> '') THEN
1702 INSERT INTO vandelay.queued_bib_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1709 $$ LANGUAGE PLPGSQL;
1711 CREATE OR REPLACE FUNCTION vandelay.ingest_bib_items ( ) RETURNS TRIGGER AS $func$
1714 item_data vandelay.import_item%ROWTYPE;
1717 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1721 SELECT item_attr_def INTO attr_def FROM vandelay.bib_queue WHERE id = NEW.queue;
1723 FOR item_data IN SELECT * FROM vandelay.ingest_items( NEW.id::BIGINT, attr_def ) LOOP
1724 INSERT INTO vandelay.import_item (
1748 item_data.definition,
1749 item_data.owning_lib,
1751 item_data.call_number,
1752 item_data.copy_number,
1755 item_data.circulate,
1757 item_data.deposit_amount,
1762 item_data.circ_modifier,
1763 item_data.circ_as_type,
1764 item_data.alert_message,
1766 item_data.priv_note,
1767 item_data.opac_visible
1773 $func$ LANGUAGE PLPGSQL;
1775 CREATE OR REPLACE FUNCTION vandelay.match_bib_record ( ) RETURNS TRIGGER AS $func$
1784 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
1786 SELECT * INTO attr_def FROM vandelay.bib_attr_definition WHERE xpath = '//*[@tag="901"]/*[@code="c"]' ORDER BY id LIMIT 1;
1788 IF attr_def IS NOT NULL AND attr_def.id IS NOT NULL THEN
1789 id_value := extract_marc_field('vandelay.queued_bib_record', NEW.id, attr_def.xpath, attr_def.remove);
1791 IF id_value IS NOT NULL AND id_value <> '' AND id_value ~ $r$^\d+$$r$ THEN
1792 SELECT id INTO exact_id FROM biblio.record_entry WHERE id = id_value::BIGINT AND NOT deleted;
1793 SELECT * INTO attr FROM vandelay.queued_bib_record_attr WHERE record = NEW.id and field = attr_def.id LIMIT 1;
1794 IF exact_id IS NOT NULL THEN
1795 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('id', attr.id, NEW.id, exact_id);
1800 IF exact_id IS NULL THEN
1801 FOR attr IN SELECT a.* FROM vandelay.queued_bib_record_attr a JOIN vandelay.bib_attr_definition d ON (d.id = a.field) WHERE record = NEW.id AND d.ident IS TRUE LOOP
1803 -- All numbers? check for an id match
1804 IF (attr.attr_value ~ $r$^\d+$$r$) THEN
1805 FOR eg_rec IN SELECT * FROM biblio.record_entry WHERE id = attr.attr_value::BIGINT AND deleted IS FALSE LOOP
1806 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('id', attr.id, NEW.id, eg_rec.id);
1810 -- Looks like an ISBN? check for an isbn match
1811 IF (attr.attr_value ~* $r$^[0-9x]+$$r$ AND character_length(attr.attr_value) IN (10,13)) THEN
1812 FOR eg_rec IN EXECUTE $$SELECT * FROM metabib.full_rec fr WHERE fr.value LIKE evergreen.lowercase('$$ || attr.attr_value || $$%') AND fr.tag = '020' AND fr.subfield = 'a'$$ LOOP
1813 PERFORM id FROM biblio.record_entry WHERE id = eg_rec.record AND deleted IS FALSE;
1815 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('isbn', attr.id, NEW.id, eg_rec.record);
1819 -- subcheck for isbn-as-tcn
1820 FOR eg_rec IN SELECT * FROM biblio.record_entry WHERE tcn_value = 'i' || attr.attr_value AND deleted IS FALSE LOOP
1821 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('tcn_value', attr.id, NEW.id, eg_rec.id);
1825 -- check for an OCLC tcn_value match
1826 IF (attr.attr_value ~ $r$^o\d+$$r$) THEN
1827 FOR eg_rec IN SELECT * FROM biblio.record_entry WHERE tcn_value = regexp_replace(attr.attr_value,'^o','ocm') AND deleted IS FALSE LOOP
1828 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('tcn_value', attr.id, NEW.id, eg_rec.id);
1832 -- check for a direct tcn_value match
1833 FOR eg_rec IN SELECT * FROM biblio.record_entry WHERE tcn_value = attr.attr_value AND deleted IS FALSE LOOP
1834 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('tcn_value', attr.id, NEW.id, eg_rec.id);
1837 -- check for a direct item barcode match
1840 FROM biblio.record_entry b
1841 JOIN asset.call_number cn ON (cn.record = b.id)
1842 JOIN asset.copy cp ON (cp.call_number = cn.id)
1843 WHERE cp.barcode = attr.attr_value AND cp.deleted IS FALSE
1845 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('id', attr.id, NEW.id, eg_rec.id);
1853 $func$ LANGUAGE PLPGSQL;
1855 CREATE OR REPLACE FUNCTION vandelay.cleanup_bib_marc ( ) RETURNS TRIGGER AS $$
1857 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1861 DELETE FROM vandelay.queued_bib_record_attr WHERE record = OLD.id;
1862 DELETE FROM vandelay.import_item WHERE record = OLD.id;
1864 IF TG_OP = 'UPDATE' THEN
1869 $$ LANGUAGE PLPGSQL;
1871 CREATE TRIGGER cleanup_bib_trigger
1872 BEFORE UPDATE OR DELETE ON vandelay.queued_bib_record
1873 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_bib_marc();
1875 CREATE TRIGGER ingest_bib_trigger
1876 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1877 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_bib_marc();
1879 CREATE TRIGGER ingest_item_trigger
1880 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1881 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_bib_items();
1883 CREATE TRIGGER zz_match_bibs_trigger
1884 BEFORE INSERT OR UPDATE ON vandelay.queued_bib_record
1885 FOR EACH ROW EXECUTE PROCEDURE vandelay.match_bib_record();
1888 /* Authority stuff down here */
1889 ---------------------------------------
1890 CREATE TABLE vandelay.authority_attr_definition (
1891 id SERIAL PRIMARY KEY,
1892 code TEXT UNIQUE NOT NULL,
1894 xpath TEXT NOT NULL,
1895 remove TEXT NOT NULL DEFAULT ''
1898 CREATE TABLE vandelay.authority_queue (
1899 queue_type TEXT NOT NULL DEFAULT 'authority' CHECK (queue_type = 'authority'),
1900 CONSTRAINT vand_authority_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
1901 ) INHERITS (vandelay.queue);
1902 ALTER TABLE vandelay.authority_queue ADD PRIMARY KEY (id);
1904 CREATE TABLE vandelay.queued_authority_record (
1905 queue INT NOT NULL REFERENCES vandelay.authority_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1906 imported_as INT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1907 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
1909 ) INHERITS (vandelay.queued_record);
1910 ALTER TABLE vandelay.queued_authority_record ADD PRIMARY KEY (id);
1911 CREATE INDEX queued_authority_record_queue_idx ON vandelay.queued_authority_record (queue);
1913 CREATE TABLE vandelay.queued_authority_record_attr (
1914 id BIGSERIAL PRIMARY KEY,
1915 record BIGINT NOT NULL REFERENCES vandelay.queued_authority_record (id) DEFERRABLE INITIALLY DEFERRED,
1916 field INT NOT NULL REFERENCES vandelay.authority_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
1917 attr_value TEXT NOT NULL
1919 CREATE INDEX queued_authority_record_attr_record_idx ON vandelay.queued_authority_record_attr (record);
1921 CREATE TABLE vandelay.authority_match (
1922 id BIGSERIAL PRIMARY KEY,
1923 queued_record BIGINT REFERENCES vandelay.queued_authority_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1924 eg_record BIGINT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1925 quality INT NOT NULL DEFAULT 0
1928 CREATE OR REPLACE FUNCTION vandelay.ingest_authority_marc ( ) RETURNS TRIGGER AS $$
1934 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1938 FOR adef IN SELECT * FROM vandelay.authority_attr_definition LOOP
1940 SELECT extract_marc_field('vandelay.queued_authority_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_authority_record WHERE id = NEW.id;
1941 IF (value IS NOT NULL AND value <> '') THEN
1942 INSERT INTO vandelay.queued_authority_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1949 $$ LANGUAGE PLPGSQL;
1951 CREATE OR REPLACE FUNCTION vandelay.cleanup_authority_marc ( ) RETURNS TRIGGER AS $$
1953 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1957 DELETE FROM vandelay.queued_authority_record_attr WHERE record = OLD.id;
1958 IF TG_OP = 'UPDATE' THEN
1963 $$ LANGUAGE PLPGSQL;
1965 CREATE TRIGGER cleanup_authority_trigger
1966 BEFORE UPDATE OR DELETE ON vandelay.queued_authority_record
1967 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_authority_marc();
1969 CREATE TRIGGER ingest_authority_trigger
1970 AFTER INSERT OR UPDATE ON vandelay.queued_authority_record
1971 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_authority_marc();
1973 CREATE OR REPLACE FUNCTION vandelay.overlay_authority_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1975 merge_profile vandelay.merge_profile%ROWTYPE;
1976 dyn_profile vandelay.compile_profile%ROWTYPE;
1985 SELECT b.marc INTO eg_marc
1986 FROM authority.record_entry b
1987 JOIN vandelay.authority_match m ON (m.eg_record = b.id AND m.queued_record = import_id)
1990 SELECT q.marc INTO v_marc
1991 FROM vandelay.queued_record q
1992 JOIN vandelay.authority_match m ON (m.queued_record = q.id AND q.id = import_id)
1995 IF eg_marc IS NULL OR v_marc IS NULL THEN
1996 -- RAISE NOTICE 'no marc for vandelay or authority record';
2000 dyn_profile := vandelay.compile_profile( v_marc );
2002 IF merge_profile_id IS NOT NULL THEN
2003 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
2005 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
2006 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
2007 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
2008 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
2012 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
2013 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
2017 IF dyn_profile.replace_rule <> '' THEN
2018 source_marc = v_marc;
2019 target_marc = eg_marc;
2020 replace_rule = dyn_profile.replace_rule;
2022 source_marc = eg_marc;
2023 target_marc = v_marc;
2024 replace_rule = dyn_profile.preserve_rule;
2027 UPDATE authority.record_entry
2028 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
2032 UPDATE vandelay.queued_authority_record
2033 SET imported_as = eg_id,
2035 WHERE id = import_id;
2039 -- RAISE NOTICE 'update of authority.record_entry failed';
2044 $$ LANGUAGE PLPGSQL;
2046 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
2051 SELECT COUNT(*) INTO match_count FROM vandelay.authority_match WHERE queued_record = import_id;
2053 IF match_count <> 1 THEN
2054 -- RAISE NOTICE 'not an exact match';
2058 SELECT m.eg_record INTO eg_id
2059 FROM vandelay.authority_match m
2060 WHERE m.queued_record = import_id
2063 IF eg_id IS NULL THEN
2067 RETURN vandelay.overlay_authority_record( import_id, eg_id, merge_profile_id );
2069 $$ LANGUAGE PLPGSQL;
2071 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
2073 queued_record vandelay.queued_authority_record%ROWTYPE;
2076 FOR queued_record IN SELECT * FROM vandelay.queued_authority_record WHERE queue = queue_id AND import_time IS NULL LOOP
2078 IF vandelay.auto_overlay_authority_record( queued_record.id, merge_profile_id ) THEN
2079 RETURN NEXT queued_record.id;
2087 $$ LANGUAGE PLPGSQL;
2089 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
2090 SELECT * FROM vandelay.auto_overlay_authority_queue( $1, NULL );
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);