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),
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),
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,
164 deposit_amount NUMERIC(8,2),
177 CREATE TABLE vandelay.import_bib_trash_fields (
178 id BIGSERIAL PRIMARY KEY,
179 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
181 CONSTRAINT vand_import_bib_trash_fields_idx UNIQUE (owner,field)
184 CREATE TABLE vandelay.merge_profile (
185 id BIGSERIAL PRIMARY KEY,
186 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
193 CONSTRAINT vand_merge_prof_owner_name_idx UNIQUE (owner,name),
194 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))
197 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
204 retval config.marc21_rec_type_map%ROWTYPE;
206 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
208 IF ldr IS NULL OR ldr = '' THEN
209 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
213 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
214 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
217 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
218 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
220 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
222 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
225 IF retval.code IS NULL THEN
226 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
231 $func$ LANGUAGE PLPGSQL;
233 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
240 rtype := (vandelay.marc21_record_type( marc )).code;
241 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
242 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
243 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
246 val := REPEAT( ff_pos.default_val, ff_pos.length );
252 $func$ LANGUAGE PLPGSQL;
254 CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
255 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
260 output biblio.record_ff_map%ROWTYPE;
262 rtype := (vandelay.marc21_record_type( marc )).code;
264 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
265 output.ff_name := ff_pos.fixed_field;
266 output.ff_value := NULL;
268 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(tag) || '"]/text()', marc ) ) x(value) LOOP
269 output.ff_value := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
270 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
272 output.ff_value := NULL;
279 $func$ LANGUAGE PLPGSQL;
281 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
282 CREATE OR REPLACE FUNCTION vandelay.marc21_physical_characteristics( marc TEXT) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
286 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
287 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
288 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
289 retval biblio.marc21_physical_characteristics%ROWTYPE;
292 _007 := oils_xpath_string( '//*[@tag="007"]', marc );
294 IF _007 IS NOT NULL AND _007 <> '' THEN
295 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007, 1, 1 );
297 IF ptype.ptype_key IS NOT NULL THEN
298 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
299 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 );
301 IF pval.id IS NOT NULL THEN
304 retval.ptype := ptype.ptype_key;
305 retval.subfield := psf.id;
306 retval.value := pval.id;
316 $func$ LANGUAGE PLPGSQL;
318 CREATE TYPE vandelay.flat_marc AS ( tag CHAR(3), ind1 TEXT, ind2 TEXT, subfield TEXT, value TEXT );
319 CREATE OR REPLACE FUNCTION vandelay.flay_marc ( TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
322 use MARC::File::XML (BinaryEncoding => 'UTF-8');
326 MARC::Charset->assume_unicode(1);
329 my $r = MARC::Record->new_from_xml( $xml );
331 return_next( { tag => 'LDR', value => $r->leader } );
333 for my $f ( $r->fields ) {
334 if ($f->is_control_field) {
335 return_next({ tag => $f->tag, value => $f->data });
337 for my $s ($f->subfields) {
340 ind1 => $f->indicator(1),
341 ind2 => $f->indicator(2),
346 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
347 my $trim = $f->indicator(2) || 0;
350 ind1 => $f->indicator(1),
351 ind2 => $f->indicator(2),
353 value => substr( $s->[1], $trim )
362 $func$ LANGUAGE PLPERLU;
364 CREATE OR REPLACE FUNCTION vandelay.flatten_marc ( marc TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
366 output vandelay.flat_marc%ROWTYPE;
369 FOR field IN SELECT * FROM vandelay.flay_marc( marc ) LOOP
370 output.ind1 := field.ind1;
371 output.ind2 := field.ind2;
372 output.tag := field.tag;
373 output.subfield := field.subfield;
374 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
375 output.value := naco_normalize(field.value, field.subfield);
377 output.value := field.value;
380 CONTINUE WHEN output.value IS NULL;
385 $func$ LANGUAGE PLPGSQL;
387 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT, attr_defs TEXT[]) RETURNS hstore AS $_$
389 transformed_xml TEXT;
392 xfrm config.xml_transform%ROWTYPE;
394 new_attrs HSTORE := ''::HSTORE;
395 attr_def config.record_attr_definition%ROWTYPE;
398 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE name IN (SELECT * FROM UNNEST(attr_defs)) ORDER BY format LOOP
400 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
401 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(x.value), COALESCE(attr_def.joiner,' ')) INTO attr_value
402 FROM vandelay.flatten_marc(xml) AS x
403 WHERE x.tag LIKE attr_def.tag
405 WHEN attr_def.sf_list IS NOT NULL
406 THEN POSITION(x.subfield IN attr_def.sf_list) > 0
413 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
414 attr_value := vandelay.marc21_extract_fixed_field(xml, attr_def.fixed_field);
416 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
418 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
420 -- See if we can skip the XSLT ... it's expensive
421 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
422 -- Can't skip the transform
423 IF xfrm.xslt <> '---' THEN
424 transformed_xml := oils_xslt_process(xml,xfrm.xslt);
426 transformed_xml := xml;
429 prev_xfrm := xfrm.name;
432 IF xfrm.name IS NULL THEN
433 -- just grab the marcxml (empty) transform
434 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
435 prev_xfrm := xfrm.name;
438 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
440 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
441 SELECT value::TEXT INTO attr_value
442 FROM vandelay.marc21_physical_characteristics(xml)
443 WHERE subfield = attr_def.phys_char_sf
444 LIMIT 1; -- Just in case ...
448 -- apply index normalizers to attr_value
450 SELECT n.func AS func,
451 n.param_count AS param_count,
453 FROM config.index_normalizer n
454 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
455 WHERE attr = attr_def.name
457 EXECUTE 'SELECT ' || normalizer.func || '(' ||
458 quote_literal( attr_value ) ||
460 WHEN normalizer.param_count > 0
461 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
468 -- Add the new value to the hstore
469 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
475 $_$ LANGUAGE PLPGSQL;
477 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT ) RETURNS hstore AS $_$
478 SELECT vandelay.extract_rec_attrs( $1, (SELECT ARRAY_ACCUM(name) FROM config.record_attr_definition));
481 -- Everything between this comment and the beginning of the definition of
482 -- vandelay.match_bib_record() is strictly in service of that function.
483 CREATE TYPE vandelay.match_set_test_result AS (record BIGINT, quality INTEGER);
485 CREATE OR REPLACE FUNCTION vandelay.match_set_test_marcxml(
486 match_set_id INTEGER, record_xml TEXT
487 ) RETURNS SETOF vandelay.match_set_test_result AS $$
498 tags_rstore := vandelay.flatten_marc_hstore(record_xml);
499 svf_rstore := vandelay.extract_rec_attrs(record_xml);
501 CREATE TEMPORARY TABLE _vandelay_tmp_qrows (q INTEGER);
502 CREATE TEMPORARY TABLE _vandelay_tmp_jrows (j TEXT);
504 -- generate the where clause and return that directly (into wq), and as
505 -- a side-effect, populate the _vandelay_tmp_[qj]rows tables.
506 wq := vandelay.get_expr_from_match_set(match_set_id);
508 query_ := 'SELECT bre.id AS record, ';
510 -- qrows table is for the quality bits we add to the SELECT clause
511 SELECT ARRAY_TO_STRING(
512 ARRAY_ACCUM('COALESCE(n' || q::TEXT || '.quality, 0)'), ' + '
513 ) INTO coal FROM _vandelay_tmp_qrows;
515 -- our query string so far is the SELECT clause and the inital FROM.
516 -- no JOINs yet nor the WHERE clause
517 query_ := query_ || coal || ' AS quality ' || E'\n' ||
518 'FROM biblio.record_entry bre ';
520 -- jrows table is for the joins we must make (and the real text conditions)
521 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(j), E'\n') INTO joins
522 FROM _vandelay_tmp_jrows;
524 -- add those joins and the where clause to our query.
525 query_ := query_ || joins || E'\n' || 'WHERE ' || wq || ' AND not bre.deleted';
527 -- this will return rows of record,quality
528 FOR rec IN EXECUTE query_ USING tags_rstore, svf_rstore LOOP
532 DROP TABLE _vandelay_tmp_qrows;
533 DROP TABLE _vandelay_tmp_jrows;
539 CREATE OR REPLACE FUNCTION vandelay.flatten_marc_hstore(
541 ) RETURNS HSTORE AS $$
545 ARRAY_ACCUM(tag || (COALESCE(subfield, ''))),
549 SELECT tag, subfield, ARRAY_ACCUM(value)::TEXT AS value
550 FROM vandelay.flatten_marc(record_xml)
551 GROUP BY tag, subfield ORDER BY tag, subfield
557 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set(
561 root vandelay.match_set_point;
563 SELECT * INTO root FROM vandelay.match_set_point
564 WHERE parent IS NULL AND match_set = match_set_id;
566 RETURN vandelay.get_expr_from_match_set_point(root);
570 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set_point(
571 node vandelay.match_set_point
578 child vandelay.match_set_point;
580 SELECT ARRAY_ACCUM(id) INTO children FROM vandelay.match_set_point
581 WHERE parent = node.id;
583 IF ARRAY_LENGTH(children, 1) > 0 THEN
584 this_op := vandelay._get_expr_render_one(node);
587 WHILE children[i] IS NOT NULL LOOP
588 SELECT * INTO child FROM vandelay.match_set_point
589 WHERE id = children[i];
591 q := q || ' ' || this_op || ' ';
594 q := q || vandelay.get_expr_from_match_set_point(child);
598 ELSIF node.bool_op IS NULL THEN
599 PERFORM vandelay._get_expr_push_qrow(node);
600 PERFORM vandelay._get_expr_push_jrow(node);
601 RETURN vandelay._get_expr_render_one(node);
608 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_qrow(
609 node vandelay.match_set_point
613 INSERT INTO _vandelay_tmp_qrows (q) VALUES (node.id);
617 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_jrow(
618 node vandelay.match_set_point
632 IF node.tag IS NOT NULL THEN
634 IF node.subfield IS NOT NULL THEN
635 tagkey := tagkey || node.subfield;
639 my_alias := 'n' || node.id::TEXT;
641 jrow := 'LEFT JOIN (SELECT *, ' || node.quality ||
642 ' AS quality FROM metabib.';
643 IF node.tag IS NOT NULL THEN
644 jrow := jrow || 'full_rec) ' || my_alias || ' ON (' ||
645 my_alias || '.record = bre.id AND ' || my_alias || '.tag = ''' ||
647 IF node.subfield IS NOT NULL THEN
648 jrow := jrow || ' AND ' || my_alias || '.subfield = ''' ||
649 node.subfield || '''';
651 jrow := jrow || ' AND (' || my_alias || '.value ' || op ||
652 ' ANY(($1->''' || tagkey || ''')::TEXT[])))';
654 jrow := jrow || 'record_attr) ' || my_alias || ' ON (' ||
655 my_alias || '.id = bre.id AND (' ||
656 my_alias || '.attrs->''' || node.svf ||
657 ''' ' || op || ' $2->''' || node.svf || '''))';
659 INSERT INTO _vandelay_tmp_jrows (j) VALUES (jrow);
663 CREATE OR REPLACE FUNCTION vandelay._get_expr_render_one(
664 node vandelay.match_set_point
669 IF node.bool_op IS NOT NULL THEN
672 RETURN '(n' || node.id::TEXT || '.id IS NOT NULL)';
677 CREATE OR REPLACE FUNCTION vandelay.match_bib_record() RETURNS TRIGGER AS $func$
679 incoming_existing_id TEXT;
680 my_bib_queue vandelay.bib_queue%ROWTYPE;
681 test_result vandelay.match_set_test_result%ROWTYPE;
684 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
688 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
690 SELECT * INTO my_bib_queue FROM vandelay.bib_queue WHERE id = NEW.queue;
692 NEW.quality := vandelay.measure_record_quality( b.marc, my_bib_queue.match_set );
694 -- Perfect matches on 901$c exit early with a match with high quality.
695 incoming_existing_id :=
696 oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]', NEW.marc);
698 IF incoming_existing_id IS NOT NULL AND incoming_existing_id != '' THEN
699 SELECT id INTO tmp_rec FROM biblio.record_entry WHERE id = incoming_existing_id::bigint;
700 IF tmp_rec IS NOT NULL THEN
701 INSERT INTO vandelay.bib_match (queued_record, eg_record, quality) VALUES ( NEW.id, incoming_existing_id::bigint, 9999);
707 FOR test_result IN SELECT * FROM
708 vandelay.match_set_test_marcxml(my_bib_queue.match_set, NEW.marc) LOOP
710 INSERT INTO vandelay.bib_match ( queued_record, eg_record, match_score, quality )
714 vandelay.measure_record_quality( b.marc, my_bib_queue.match_set )
715 FROM biblio.record_entry b
716 WHERE id = test_result.record;
722 $func$ LANGUAGE PLPGSQL;
724 CREATE OR REPLACE FUNCTION vandelay.measure_record_quality ( xml TEXT, match_set_id INT ) RETURNS INT AS $_$
728 test vandelay.match_set_quality%ROWTYPE;
731 FOR test IN SELECT * FROM vandelay.match_set_quality WHERE match_set = match_set_id LOOP
732 IF test.tag IS NOT NULL THEN
733 FOR rvalue IN SELECT value FROM vandelay.flatten_marc( xml ) WHERE tag = test.tag AND subfield = test.subfield LOOP
734 IF test.value = rvalue THEN
735 out_q := out_q + test.quality;
739 IF test.value = vandelay.extract_rec_attrs(xml, ARRAY[test.svf]) THEN
740 out_q := out_q + test.quality;
747 $_$ LANGUAGE PLPGSQL;
749 CREATE TYPE vandelay.tcn_data AS (tcn TEXT, tcn_source TEXT, used BOOL);
750 CREATE OR REPLACE FUNCTION vandelay.find_bib_tcn_data ( xml TEXT ) RETURNS SETOF vandelay.tcn_data AS $_$
754 output vandelay.tcn_data%ROWTYPE;
758 eg_tcn := BTRIM((oils_xpath('//*[@tag="001"]/text()',xml))[1]);
759 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
761 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="003"]/text()',xml))[1]);
762 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
763 eg_tcn_source := 'System Local';
766 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
769 output.used := FALSE;
774 output.tcn := eg_tcn;
775 output.tcn_source := eg_tcn_source;
781 eg_tcn := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="a"]/text()',xml))[1]);
782 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
784 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="b"]/text()',xml))[1]);
785 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
786 eg_tcn_source := 'System Local';
789 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
792 output.used := FALSE;
797 output.tcn := eg_tcn;
798 output.tcn_source := eg_tcn_source;
804 eg_tcn := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="a"]/text()',xml))[1]);
805 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
807 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="b"]/text()',xml))[1]);
808 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
809 eg_tcn_source := 'System Local';
812 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
815 output.used := FALSE;
820 output.tcn := eg_tcn;
821 output.tcn_source := eg_tcn_source;
827 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="020"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
828 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
830 eg_tcn_source := 'ISBN';
832 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
835 output.used := FALSE;
840 output.tcn := eg_tcn;
841 output.tcn_source := eg_tcn_source;
847 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="022"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
848 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
850 eg_tcn_source := 'ISSN';
852 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
855 output.used := FALSE;
860 output.tcn := eg_tcn;
861 output.tcn_source := eg_tcn_source;
867 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="010"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
868 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
870 eg_tcn_source := 'LCCN';
872 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
875 output.used := FALSE;
880 output.tcn := eg_tcn;
881 output.tcn_source := eg_tcn_source;
887 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="035"]/*[@code="a"]/text()',xml))[1], $re$^.*?(\w+)$$re$, $re$\1$re$);
888 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
890 eg_tcn_source := 'System Legacy';
892 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
895 output.used := FALSE;
900 output.tcn := eg_tcn;
901 output.tcn_source := eg_tcn_source;
908 $_$ LANGUAGE PLPGSQL;
910 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT, force_add INT ) RETURNS TEXT AS $_$
913 use MARC::File::XML (BinaryEncoding => 'UTF-8');
917 MARC::Charset->assume_unicode(1);
919 my $target_xml = shift;
920 my $source_xml = shift;
921 my $field_spec = shift;
922 my $force_add = shift || 0;
924 my $target_r = MARC::Record->new_from_xml( $target_xml );
925 my $source_r = MARC::Record->new_from_xml( $source_xml );
927 return $target_xml unless ($target_r && $source_r);
929 my @field_list = split(',', $field_spec);
932 for my $f (@field_list) {
933 $f =~ s/^\s*//; $f =~ s/\s*$//;
934 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
940 $match =~ s/^\s*//; $match =~ s/\s*$//;
941 $fields{$field} = { sf => [ split('', $sf) ] };
943 my ($msf,$mre) = split('~', $match);
944 if (length($msf) > 0 and length($mre) > 0) {
945 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
946 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
947 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
953 for my $f ( keys %fields) {
954 if ( @{$fields{$f}{sf}} ) {
955 for my $from_field ($source_r->field( $f )) {
956 my @tos = $target_r->field( $f );
958 next if (exists($fields{$f}{match}) and !$force_add);
959 my @new_fields = map { $_->clone } $source_r->field( $f );
960 $target_r->insert_fields_ordered( @new_fields );
962 for my $to_field (@tos) {
963 if (exists($fields{$f}{match})) {
964 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
966 my @new_sf = map { ($_ => $from_field->subfield($_)) } @{$fields{$f}{sf}};
967 $to_field->add_subfields( @new_sf );
972 my @new_fields = map { $_->clone } $source_r->field( $f );
973 $target_r->insert_fields_ordered( @new_fields );
977 $target_xml = $target_r->as_xml_record;
978 $target_xml =~ s/^<\?.+?\?>$//mo;
979 $target_xml =~ s/\n//sgo;
980 $target_xml =~ s/>\s+</></sgo;
984 $_$ LANGUAGE PLPERLU;
986 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
987 SELECT vandelay.add_field( $1, $2, $3, 0 );
990 CREATE OR REPLACE FUNCTION vandelay.strip_field ( xml TEXT, field TEXT ) RETURNS TEXT AS $_$
993 use MARC::File::XML (BinaryEncoding => 'UTF-8');
997 MARC::Charset->assume_unicode(1);
1000 my $r = MARC::Record->new_from_xml( $xml );
1002 return $xml unless ($r);
1004 my $field_spec = shift;
1005 my @field_list = split(',', $field_spec);
1008 for my $f (@field_list) {
1009 $f =~ s/^\s*//; $f =~ s/\s*$//;
1010 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1016 $match =~ s/^\s*//; $match =~ s/\s*$//;
1017 $fields{$field} = { sf => [ split('', $sf) ] };
1019 my ($msf,$mre) = split('~', $match);
1020 if (length($msf) > 0 and length($mre) > 0) {
1021 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1022 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1023 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1029 for my $f ( keys %fields) {
1030 for my $to_field ($r->field( $f )) {
1031 if (exists($fields{$f}{match})) {
1032 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1035 if ( @{$fields{$f}{sf}} ) {
1036 $to_field->delete_subfield(code => $fields{$f}{sf});
1038 $r->delete_field( $to_field );
1043 $xml = $r->as_xml_record;
1044 $xml =~ s/^<\?.+?\?>$//mo;
1046 $xml =~ s/>\s+</></sgo;
1050 $_$ LANGUAGE PLPERLU;
1052 CREATE OR REPLACE FUNCTION vandelay.replace_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1059 parsed_target := vandelay.strip_field( target_xml, ''); -- this dance normalizes the format of the xml for the IF below
1061 FOR curr_field IN SELECT UNNEST( STRING_TO_ARRAY(field, ',') ) LOOP -- naive split, but it's the same we use in the perl
1063 xml_output := vandelay.strip_field( parsed_target, curr_field);
1065 IF xml_output <> parsed_target AND curr_field ~ E'~' THEN
1066 -- we removed something, and there was a regexp restriction in the curr_field definition, so proceed
1067 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 1 );
1068 ELSIF curr_field !~ E'~' THEN
1069 -- No regexp restriction, add the curr_field
1070 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 0 );
1073 parsed_target := xml_output; -- in prep for any following loop iterations
1079 $_$ LANGUAGE PLPGSQL;
1081 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 $_$
1082 SELECT vandelay.replace_field( vandelay.add_field( vandelay.strip_field( $1, $5) , $2, $3 ), $2, $4);
1085 CREATE TYPE vandelay.compile_profile AS (add_rule TEXT, replace_rule TEXT, preserve_rule TEXT, strip_rule TEXT);
1086 CREATE OR REPLACE FUNCTION vandelay.compile_profile ( incoming_xml TEXT ) RETURNS vandelay.compile_profile AS $_$
1088 output vandelay.compile_profile%ROWTYPE;
1089 profile vandelay.merge_profile%ROWTYPE;
1091 profile_tmpl_owner TEXT;
1092 add_rule TEXT := '';
1093 strip_rule TEXT := '';
1094 replace_rule TEXT := '';
1095 preserve_rule TEXT := '';
1099 profile_tmpl := (oils_xpath('//*[@tag="905"]/*[@code="t"]/text()',incoming_xml))[1];
1100 profile_tmpl_owner := (oils_xpath('//*[@tag="905"]/*[@code="o"]/text()',incoming_xml))[1];
1102 IF profile_tmpl IS NOT NULL AND profile_tmpl <> '' AND profile_tmpl_owner IS NOT NULL AND profile_tmpl_owner <> '' THEN
1103 SELECT p.* INTO profile
1104 FROM vandelay.merge_profile p
1105 JOIN actor.org_unit u ON (u.id = p.owner)
1106 WHERE p.name = profile_tmpl
1107 AND u.shortname = profile_tmpl_owner;
1109 IF profile.id IS NOT NULL THEN
1110 add_rule := COALESCE(profile.add_spec,'');
1111 strip_rule := COALESCE(profile.strip_spec,'');
1112 replace_rule := COALESCE(profile.replace_spec,'');
1113 preserve_rule := COALESCE(profile.preserve_spec,'');
1117 add_rule := add_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="a"]/text()',incoming_xml),','),'');
1118 strip_rule := strip_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="d"]/text()',incoming_xml),','),'');
1119 replace_rule := replace_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="r"]/text()',incoming_xml),','),'');
1120 preserve_rule := preserve_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="p"]/text()',incoming_xml),','),'');
1122 output.add_rule := BTRIM(add_rule,',');
1123 output.replace_rule := BTRIM(replace_rule,',');
1124 output.strip_rule := BTRIM(strip_rule,',');
1125 output.preserve_rule := BTRIM(preserve_rule,',');
1129 $_$ LANGUAGE PLPGSQL;
1131 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1133 merge_profile vandelay.merge_profile%ROWTYPE;
1134 dyn_profile vandelay.compile_profile%ROWTYPE;
1144 SELECT b.marc INTO eg_marc
1145 FROM biblio.record_entry b
1149 IF eg_marc IS NULL OR v_marc IS NULL THEN
1150 -- RAISE NOTICE 'no marc for template or bib record';
1154 dyn_profile := vandelay.compile_profile( v_marc );
1156 IF merge_profile_id IS NOT NULL THEN
1157 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1159 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1160 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1161 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1162 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1166 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1167 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1171 IF dyn_profile.replace_rule <> '' THEN
1172 source_marc = v_marc;
1173 target_marc = eg_marc;
1174 replace_rule = dyn_profile.replace_rule;
1176 source_marc = eg_marc;
1177 target_marc = v_marc;
1178 replace_rule = dyn_profile.preserve_rule;
1181 UPDATE biblio.record_entry
1182 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1186 -- RAISE NOTICE 'update of biblio.record_entry failed';
1193 $$ LANGUAGE PLPGSQL;
1195 CREATE OR REPLACE FUNCTION vandelay.merge_record_xml ( target_marc TEXT, template_marc TEXT ) RETURNS TEXT AS $$
1197 dyn_profile vandelay.compile_profile%ROWTYPE;
1205 IF target_marc IS NULL OR template_marc IS NULL THEN
1206 -- RAISE NOTICE 'no marc for target or template record';
1210 dyn_profile := vandelay.compile_profile( template_marc );
1212 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1213 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1217 IF dyn_profile.replace_rule <> '' THEN
1218 trgt_marc = target_marc;
1219 tmpl_marc = template_marc;
1220 replace_rule = dyn_profile.replace_rule;
1222 tmp_marc = target_marc;
1223 trgt_marc = template_marc;
1224 tmpl_marc = tmp_marc;
1225 replace_rule = dyn_profile.preserve_rule;
1228 RETURN vandelay.merge_record_xml( trgt_marc, tmpl_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule );
1231 $$ LANGUAGE PLPGSQL;
1233 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT) RETURNS BOOL AS $$
1234 SELECT vandelay.template_overlay_bib_record( $1, $2, NULL);
1237 CREATE OR REPLACE FUNCTION vandelay.overlay_bib_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1239 merge_profile vandelay.merge_profile%ROWTYPE;
1240 dyn_profile vandelay.compile_profile%ROWTYPE;
1250 SELECT q.marc INTO v_marc
1251 FROM vandelay.queued_record q
1252 JOIN vandelay.bib_match m ON (m.queued_record = q.id AND q.id = import_id)
1255 IF v_marc IS NULL THEN
1256 -- RAISE NOTICE 'no marc for vandelay or bib record';
1260 IF vandelay.template_overlay_bib_record( v_marc, eg_id, merge_profile_id) THEN
1261 UPDATE vandelay.queued_bib_record
1262 SET imported_as = eg_id,
1264 WHERE id = import_id;
1266 editor_string := (oils_xpath('//*[@tag="905"]/*[@code="u"]/text()',v_marc))[1];
1268 IF editor_string IS NOT NULL AND editor_string <> '' THEN
1269 SELECT usr INTO editor_id FROM actor.card WHERE barcode = editor_string;
1271 IF editor_id IS NULL THEN
1272 SELECT id INTO editor_id FROM actor.usr WHERE usrname = editor_string;
1275 IF editor_id IS NOT NULL THEN
1276 UPDATE biblio.record_entry SET editor = editor_id WHERE id = eg_id;
1283 -- RAISE NOTICE 'update of biblio.record_entry failed';
1288 $$ LANGUAGE PLPGSQL;
1290 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
1293 lwm_ratio_value NUMERIC;
1296 lwm_ratio_value := COALESCE(lwm_ratio_value_p, 0.0);
1298 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1301 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1305 SELECT m.eg_record INTO eg_id
1306 FROM vandelay.bib_match m
1307 JOIN vandelay.queued_bib_record qr ON (m.queued_record = qr.id)
1308 JOIN vandelay.bib_queue q ON (qr.queue = q.id)
1309 JOIN biblio.record_entry r ON (r.id = m.eg_record)
1310 WHERE m.queued_record = import_id
1311 AND qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC >= lwm_ratio_value
1312 ORDER BY m.match_score DESC, -- required match score
1313 qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC DESC, -- quality tie breaker
1314 m.id -- when in doubt, use the first match
1317 IF eg_id IS NULL THEN
1318 -- RAISE NOTICE 'incoming record is not of high enough quality';
1322 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1324 $$ LANGUAGE PLPGSQL;
1326 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1327 SELECT vandelay.auto_overlay_bib_record_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1330 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1336 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1339 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1343 SELECT COUNT(*) INTO match_count FROM vandelay.bib_match WHERE queued_record = import_id;
1345 IF match_count <> 1 THEN
1346 -- RAISE NOTICE 'not an exact match';
1350 -- Check that the one match is on the first 901c
1351 SELECT m.eg_record INTO eg_id
1352 FROM vandelay.queued_bib_record q
1353 JOIN vandelay.bib_match m ON (m.queued_record = q.id)
1354 WHERE q.id = import_id
1355 AND m.eg_record = oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]',marc)::BIGINT;
1358 -- RAISE NOTICE 'not a 901c match';
1362 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1364 $$ LANGUAGE PLPGSQL;
1366 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1368 queued_record vandelay.queued_bib_record%ROWTYPE;
1371 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1373 IF vandelay.auto_overlay_bib_record( queued_record.id, merge_profile_id ) THEN
1374 RETURN NEXT queued_record.id;
1382 $$ LANGUAGE PLPGSQL;
1384 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 $$
1386 queued_record vandelay.queued_bib_record%ROWTYPE;
1389 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1391 IF vandelay.auto_overlay_bib_record_with_best( queued_record.id, merge_profile_id, lwm_ratio_value ) THEN
1392 RETURN NEXT queued_record.id;
1400 $$ LANGUAGE PLPGSQL;
1402 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1403 SELECT vandelay.auto_overlay_bib_queue_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1406 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1407 SELECT * FROM vandelay.auto_overlay_bib_queue( $1, NULL );
1410 CREATE OR REPLACE FUNCTION vandelay.ingest_items ( import_id BIGINT, attr_def_id BIGINT ) RETURNS SETOF vandelay.import_item AS $$
1421 deposit_amount TEXT;
1434 tmp_attr_set RECORD;
1435 attr_set vandelay.import_item%ROWTYPE;
1441 SELECT * INTO attr_def FROM vandelay.import_item_attr_definition WHERE id = attr_def_id;
1445 attr_set.definition := attr_def.id;
1447 -- Build the combined XPath
1451 WHEN attr_def.owning_lib IS NULL THEN 'null()'
1452 WHEN LENGTH( attr_def.owning_lib ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.owning_lib || '"]'
1453 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.owning_lib
1458 WHEN attr_def.circ_lib IS NULL THEN 'null()'
1459 WHEN LENGTH( attr_def.circ_lib ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.circ_lib || '"]'
1460 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.circ_lib
1465 WHEN attr_def.call_number IS NULL THEN 'null()'
1466 WHEN LENGTH( attr_def.call_number ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.call_number || '"]'
1467 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.call_number
1472 WHEN attr_def.copy_number IS NULL THEN 'null()'
1473 WHEN LENGTH( attr_def.copy_number ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.copy_number || '"]'
1474 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.copy_number
1479 WHEN attr_def.status IS NULL THEN 'null()'
1480 WHEN LENGTH( attr_def.status ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.status || '"]'
1481 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.status
1486 WHEN attr_def.location IS NULL THEN 'null()'
1487 WHEN LENGTH( attr_def.location ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.location || '"]'
1488 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.location
1493 WHEN attr_def.circulate IS NULL THEN 'null()'
1494 WHEN LENGTH( attr_def.circulate ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.circulate || '"]'
1495 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.circulate
1500 WHEN attr_def.deposit IS NULL THEN 'null()'
1501 WHEN LENGTH( attr_def.deposit ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.deposit || '"]'
1502 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.deposit
1507 WHEN attr_def.deposit_amount IS NULL THEN 'null()'
1508 WHEN LENGTH( attr_def.deposit_amount ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.deposit_amount || '"]'
1509 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.deposit_amount
1514 WHEN attr_def.ref IS NULL THEN 'null()'
1515 WHEN LENGTH( attr_def.ref ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.ref || '"]'
1516 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.ref
1521 WHEN attr_def.holdable IS NULL THEN 'null()'
1522 WHEN LENGTH( attr_def.holdable ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.holdable || '"]'
1523 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.holdable
1528 WHEN attr_def.price IS NULL THEN 'null()'
1529 WHEN LENGTH( attr_def.price ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.price || '"]'
1530 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.price
1535 WHEN attr_def.barcode IS NULL THEN 'null()'
1536 WHEN LENGTH( attr_def.barcode ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.barcode || '"]'
1537 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.barcode
1542 WHEN attr_def.circ_modifier IS NULL THEN 'null()'
1543 WHEN LENGTH( attr_def.circ_modifier ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.circ_modifier || '"]'
1544 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.circ_modifier
1549 WHEN attr_def.circ_as_type IS NULL THEN 'null()'
1550 WHEN LENGTH( attr_def.circ_as_type ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.circ_as_type || '"]'
1551 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.circ_as_type
1556 WHEN attr_def.alert_message IS NULL THEN 'null()'
1557 WHEN LENGTH( attr_def.alert_message ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.alert_message || '"]'
1558 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.alert_message
1563 WHEN attr_def.opac_visible IS NULL THEN 'null()'
1564 WHEN LENGTH( attr_def.opac_visible ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.opac_visible || '"]'
1565 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.opac_visible
1570 WHEN attr_def.pub_note IS NULL THEN 'null()'
1571 WHEN LENGTH( attr_def.pub_note ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.pub_note || '"]'
1572 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.pub_note
1576 WHEN attr_def.priv_note IS NULL THEN 'null()'
1577 WHEN LENGTH( attr_def.priv_note ) = 1 THEN '//*[@tag="' || attr_def.tag || '"]/*[@code="' || attr_def.priv_note || '"]'
1578 ELSE '//*[@tag="' || attr_def.tag || '"]/*' || attr_def.priv_note
1583 owning_lib || '|' ||
1585 call_number || '|' ||
1586 copy_number || '|' ||
1591 deposit_amount || '|' ||
1596 circ_modifier || '|' ||
1597 circ_as_type || '|' ||
1598 alert_message || '|' ||
1603 -- RAISE NOTICE 'XPath: %', xpath;
1607 FROM oils_xpath_table( 'id', 'marc', 'vandelay.queued_bib_record', xpath, 'id = ' || import_id )
1608 AS t( id INT, ol TEXT, clib TEXT, cn TEXT, cnum TEXT, cs TEXT, cl TEXT, circ TEXT,
1609 dep TEXT, dep_amount TEXT, r TEXT, hold TEXT, pr TEXT, bc TEXT, circ_mod TEXT,
1610 circ_as TEXT, amessage TEXT, note TEXT, pnote TEXT, opac_vis TEXT )
1613 tmp_attr_set.pr = REGEXP_REPLACE(tmp_attr_set.pr, E'[^0-9\\.]', '', 'g');
1614 tmp_attr_set.dep_amount = REGEXP_REPLACE(tmp_attr_set.dep_amount, E'[^0-9\\.]', '', 'g');
1616 tmp_attr_set.pr := NULLIF( tmp_attr_set.pr, '' );
1617 tmp_attr_set.dep_amount := NULLIF( tmp_attr_set.dep_amount, '' );
1619 SELECT id INTO attr_set.owning_lib FROM actor.org_unit WHERE shortname = UPPER(tmp_attr_set.ol); -- INT
1620 SELECT id INTO attr_set.circ_lib FROM actor.org_unit WHERE shortname = UPPER(tmp_attr_set.clib); -- INT
1621 SELECT id INTO attr_set.status FROM config.copy_status WHERE LOWER(name) = LOWER(tmp_attr_set.cs); -- INT
1623 SELECT id INTO attr_set.location
1624 FROM asset.copy_location
1625 WHERE LOWER(name) = LOWER(tmp_attr_set.cl)
1626 AND asset.copy_location.owning_lib = COALESCE(attr_set.owning_lib, attr_set.circ_lib); -- INT
1628 attr_set.circulate :=
1629 LOWER( SUBSTRING( tmp_attr_set.circ, 1, 1)) IN ('t','y','1')
1630 OR LOWER(tmp_attr_set.circ) = 'circulating'; -- BOOL
1633 LOWER( SUBSTRING( tmp_attr_set.dep, 1, 1 ) ) IN ('t','y','1')
1634 OR LOWER(tmp_attr_set.dep) = 'deposit'; -- BOOL
1636 attr_set.holdable :=
1637 LOWER( SUBSTRING( tmp_attr_set.hold, 1, 1 ) ) IN ('t','y','1')
1638 OR LOWER(tmp_attr_set.hold) = 'holdable'; -- BOOL
1640 attr_set.opac_visible :=
1641 LOWER( SUBSTRING( tmp_attr_set.opac_vis, 1, 1 ) ) IN ('t','y','1')
1642 OR LOWER(tmp_attr_set.opac_vis) = 'visible'; -- BOOL
1645 LOWER( SUBSTRING( tmp_attr_set.r, 1, 1 ) ) IN ('t','y','1')
1646 OR LOWER(tmp_attr_set.r) = 'reference'; -- BOOL
1648 attr_set.copy_number := tmp_attr_set.cnum::INT; -- INT,
1649 attr_set.deposit_amount := tmp_attr_set.dep_amount::NUMERIC(6,2); -- NUMERIC(6,2),
1650 attr_set.price := tmp_attr_set.pr::NUMERIC(8,2); -- NUMERIC(8,2),
1652 attr_set.call_number := tmp_attr_set.cn; -- TEXT
1653 attr_set.barcode := tmp_attr_set.bc; -- TEXT,
1654 attr_set.circ_modifier := tmp_attr_set.circ_mod; -- TEXT,
1655 attr_set.circ_as_type := tmp_attr_set.circ_as; -- TEXT,
1656 attr_set.alert_message := tmp_attr_set.amessage; -- TEXT,
1657 attr_set.pub_note := tmp_attr_set.note; -- TEXT,
1658 attr_set.priv_note := tmp_attr_set.pnote; -- TEXT,
1659 attr_set.alert_message := tmp_attr_set.amessage; -- TEXT,
1661 RETURN NEXT attr_set;
1670 $$ LANGUAGE PLPGSQL;
1673 CREATE OR REPLACE FUNCTION vandelay.ingest_bib_marc ( ) RETURNS TRIGGER AS $$
1679 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1683 FOR adef IN SELECT * FROM vandelay.bib_attr_definition LOOP
1685 SELECT extract_marc_field('vandelay.queued_bib_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_bib_record WHERE id = NEW.id;
1686 IF (value IS NOT NULL AND value <> '') THEN
1687 INSERT INTO vandelay.queued_bib_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1694 $$ LANGUAGE PLPGSQL;
1696 CREATE OR REPLACE FUNCTION vandelay.ingest_bib_items ( ) RETURNS TRIGGER AS $func$
1699 item_data vandelay.import_item%ROWTYPE;
1702 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1706 SELECT item_attr_def INTO attr_def FROM vandelay.bib_queue WHERE id = NEW.queue;
1708 FOR item_data IN SELECT * FROM vandelay.ingest_items( NEW.id::BIGINT, attr_def ) LOOP
1709 INSERT INTO vandelay.import_item (
1733 item_data.definition,
1734 item_data.owning_lib,
1736 item_data.call_number,
1737 item_data.copy_number,
1740 item_data.circulate,
1742 item_data.deposit_amount,
1747 item_data.circ_modifier,
1748 item_data.circ_as_type,
1749 item_data.alert_message,
1751 item_data.priv_note,
1752 item_data.opac_visible
1758 $func$ LANGUAGE PLPGSQL;
1760 CREATE OR REPLACE FUNCTION vandelay.match_bib_record ( ) RETURNS TRIGGER AS $func$
1769 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
1771 SELECT * INTO attr_def FROM vandelay.bib_attr_definition WHERE xpath = '//*[@tag="901"]/*[@code="c"]' ORDER BY id LIMIT 1;
1773 IF attr_def IS NOT NULL AND attr_def.id IS NOT NULL THEN
1774 id_value := extract_marc_field('vandelay.queued_bib_record', NEW.id, attr_def.xpath, attr_def.remove);
1776 IF id_value IS NOT NULL AND id_value <> '' AND id_value ~ $r$^\d+$$r$ THEN
1777 SELECT id INTO exact_id FROM biblio.record_entry WHERE id = id_value::BIGINT AND NOT deleted;
1778 SELECT * INTO attr FROM vandelay.queued_bib_record_attr WHERE record = NEW.id and field = attr_def.id LIMIT 1;
1779 IF exact_id IS NOT NULL THEN
1780 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('id', attr.id, NEW.id, exact_id);
1785 IF exact_id IS NULL THEN
1786 FOR attr IN SELECT a.* FROM vandelay.queued_bib_record_attr a JOIN vandelay.bib_attr_definition d ON (d.id = a.field) WHERE record = NEW.id AND d.ident IS TRUE LOOP
1788 -- All numbers? check for an id match
1789 IF (attr.attr_value ~ $r$^\d+$$r$) THEN
1790 FOR eg_rec IN SELECT * FROM biblio.record_entry WHERE id = attr.attr_value::BIGINT AND deleted IS FALSE LOOP
1791 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('id', attr.id, NEW.id, eg_rec.id);
1795 -- Looks like an ISBN? check for an isbn match
1796 IF (attr.attr_value ~* $r$^[0-9x]+$$r$ AND character_length(attr.attr_value) IN (10,13)) THEN
1797 FOR eg_rec IN EXECUTE $$SELECT * FROM metabib.full_rec fr WHERE fr.value LIKE evergreen.lowercase('$$ || attr.attr_value || $$%') AND fr.tag = '020' AND fr.subfield = 'a'$$ LOOP
1798 PERFORM id FROM biblio.record_entry WHERE id = eg_rec.record AND deleted IS FALSE;
1800 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('isbn', attr.id, NEW.id, eg_rec.record);
1804 -- subcheck for isbn-as-tcn
1805 FOR eg_rec IN SELECT * FROM biblio.record_entry WHERE tcn_value = 'i' || attr.attr_value AND deleted IS FALSE LOOP
1806 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('tcn_value', attr.id, NEW.id, eg_rec.id);
1810 -- check for an OCLC tcn_value match
1811 IF (attr.attr_value ~ $r$^o\d+$$r$) THEN
1812 FOR eg_rec IN SELECT * FROM biblio.record_entry WHERE tcn_value = regexp_replace(attr.attr_value,'^o','ocm') AND deleted IS FALSE LOOP
1813 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('tcn_value', attr.id, NEW.id, eg_rec.id);
1817 -- check for a direct tcn_value match
1818 FOR eg_rec IN SELECT * FROM biblio.record_entry WHERE tcn_value = attr.attr_value AND deleted IS FALSE LOOP
1819 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('tcn_value', attr.id, NEW.id, eg_rec.id);
1822 -- check for a direct item barcode match
1825 FROM biblio.record_entry b
1826 JOIN asset.call_number cn ON (cn.record = b.id)
1827 JOIN asset.copy cp ON (cp.call_number = cn.id)
1828 WHERE cp.barcode = attr.attr_value AND cp.deleted IS FALSE
1830 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('id', attr.id, NEW.id, eg_rec.id);
1838 $func$ LANGUAGE PLPGSQL;
1840 CREATE OR REPLACE FUNCTION vandelay.cleanup_bib_marc ( ) RETURNS TRIGGER AS $$
1842 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1846 DELETE FROM vandelay.queued_bib_record_attr WHERE record = OLD.id;
1847 DELETE FROM vandelay.import_item WHERE record = OLD.id;
1849 IF TG_OP = 'UPDATE' THEN
1854 $$ LANGUAGE PLPGSQL;
1856 CREATE TRIGGER cleanup_bib_trigger
1857 BEFORE UPDATE OR DELETE ON vandelay.queued_bib_record
1858 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_bib_marc();
1860 CREATE TRIGGER ingest_bib_trigger
1861 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1862 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_bib_marc();
1864 CREATE TRIGGER ingest_item_trigger
1865 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1866 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_bib_items();
1868 CREATE TRIGGER zz_match_bibs_trigger
1869 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1870 FOR EACH ROW EXECUTE PROCEDURE vandelay.match_bib_record();
1873 /* Authority stuff down here */
1874 ---------------------------------------
1875 CREATE TABLE vandelay.authority_attr_definition (
1876 id SERIAL PRIMARY KEY,
1877 code TEXT UNIQUE NOT NULL,
1879 xpath TEXT NOT NULL,
1880 remove TEXT NOT NULL DEFAULT ''
1883 CREATE TABLE vandelay.authority_queue (
1884 queue_type TEXT NOT NULL DEFAULT 'authority' CHECK (queue_type = 'authority'),
1885 CONSTRAINT vand_authority_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
1886 ) INHERITS (vandelay.queue);
1887 ALTER TABLE vandelay.authority_queue ADD PRIMARY KEY (id);
1889 CREATE TABLE vandelay.queued_authority_record (
1890 queue INT NOT NULL REFERENCES vandelay.authority_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1891 imported_as INT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1892 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
1894 ) INHERITS (vandelay.queued_record);
1895 ALTER TABLE vandelay.queued_authority_record ADD PRIMARY KEY (id);
1896 CREATE INDEX queued_authority_record_queue_idx ON vandelay.queued_authority_record (queue);
1898 CREATE TABLE vandelay.queued_authority_record_attr (
1899 id BIGSERIAL PRIMARY KEY,
1900 record BIGINT NOT NULL REFERENCES vandelay.queued_authority_record (id) DEFERRABLE INITIALLY DEFERRED,
1901 field INT NOT NULL REFERENCES vandelay.authority_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
1902 attr_value TEXT NOT NULL
1904 CREATE INDEX queued_authority_record_attr_record_idx ON vandelay.queued_authority_record_attr (record);
1906 CREATE TABLE vandelay.authority_match (
1907 id BIGSERIAL PRIMARY KEY,
1908 queued_record BIGINT REFERENCES vandelay.queued_authority_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1909 eg_record BIGINT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1910 quality INT NOT NULL DEFAULT 0
1913 CREATE OR REPLACE FUNCTION vandelay.ingest_authority_marc ( ) RETURNS TRIGGER AS $$
1919 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1923 FOR adef IN SELECT * FROM vandelay.authority_attr_definition LOOP
1925 SELECT extract_marc_field('vandelay.queued_authority_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_authority_record WHERE id = NEW.id;
1926 IF (value IS NOT NULL AND value <> '') THEN
1927 INSERT INTO vandelay.queued_authority_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1934 $$ LANGUAGE PLPGSQL;
1936 CREATE OR REPLACE FUNCTION vandelay.cleanup_authority_marc ( ) RETURNS TRIGGER AS $$
1938 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1942 DELETE FROM vandelay.queued_authority_record_attr WHERE record = OLD.id;
1943 IF TG_OP = 'UPDATE' THEN
1948 $$ LANGUAGE PLPGSQL;
1950 CREATE TRIGGER cleanup_authority_trigger
1951 BEFORE UPDATE OR DELETE ON vandelay.queued_authority_record
1952 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_authority_marc();
1954 CREATE TRIGGER ingest_authority_trigger
1955 AFTER INSERT OR UPDATE ON vandelay.queued_authority_record
1956 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_authority_marc();
1958 CREATE OR REPLACE FUNCTION vandelay.overlay_authority_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1960 merge_profile vandelay.merge_profile%ROWTYPE;
1961 dyn_profile vandelay.compile_profile%ROWTYPE;
1970 SELECT b.marc INTO eg_marc
1971 FROM authority.record_entry b
1972 JOIN vandelay.authority_match m ON (m.eg_record = b.id AND m.queued_record = import_id)
1975 SELECT q.marc INTO v_marc
1976 FROM vandelay.queued_record q
1977 JOIN vandelay.authority_match m ON (m.queued_record = q.id AND q.id = import_id)
1980 IF eg_marc IS NULL OR v_marc IS NULL THEN
1981 -- RAISE NOTICE 'no marc for vandelay or authority record';
1985 dyn_profile := vandelay.compile_profile( v_marc );
1987 IF merge_profile_id IS NOT NULL THEN
1988 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1990 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1991 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1992 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1993 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1997 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1998 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
2002 IF dyn_profile.replace_rule <> '' THEN
2003 source_marc = v_marc;
2004 target_marc = eg_marc;
2005 replace_rule = dyn_profile.replace_rule;
2007 source_marc = eg_marc;
2008 target_marc = v_marc;
2009 replace_rule = dyn_profile.preserve_rule;
2012 UPDATE authority.record_entry
2013 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
2017 UPDATE vandelay.queued_authority_record
2018 SET imported_as = eg_id,
2020 WHERE id = import_id;
2024 -- RAISE NOTICE 'update of authority.record_entry failed';
2029 $$ LANGUAGE PLPGSQL;
2031 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
2036 SELECT COUNT(*) INTO match_count FROM vandelay.authority_match WHERE queued_record = import_id;
2038 IF match_count <> 1 THEN
2039 -- RAISE NOTICE 'not an exact match';
2043 SELECT m.eg_record INTO eg_id
2044 FROM vandelay.authority_match m
2045 WHERE m.queued_record = import_id
2048 IF eg_id IS NULL THEN
2052 RETURN vandelay.overlay_authority_record( import_id, eg_id, merge_profile_id );
2054 $$ LANGUAGE PLPGSQL;
2056 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
2058 queued_record vandelay.queued_authority_record%ROWTYPE;
2061 FOR queued_record IN SELECT * FROM vandelay.queued_authority_record WHERE queue = queue_id AND import_time IS NULL LOOP
2063 IF vandelay.auto_overlay_authority_record( queued_record.id, merge_profile_id ) THEN
2064 RETURN NEXT queued_record.id;
2072 $$ LANGUAGE PLPGSQL;
2074 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
2075 SELECT * FROM vandelay.auto_overlay_authority_queue( $1, NULL );
2079 -- Vandelay (for importing and exporting records) 012.schema.vandelay.sql
2080 --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)]');
2081 --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)]');
2082 --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]');
2083 --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]');
2084 --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$);
2085 --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$);
2086 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (7, 'price',oils_i18n_gettext(7, 'vqbrad', 'Price', 'description'),'//*[@tag="020" or @tag="022"]/*[@code="c"][1]');
2087 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (8, 'rec_identifier',oils_i18n_gettext(8, 'vqbrad', 'Accession Number', 'description'),'//*[@tag="001"]', TRUE);
2088 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (9, 'eg_tcn',oils_i18n_gettext(9, 'vqbrad', 'TCN Value', 'description'),'//*[@tag="901"]/*[@code="a"]', TRUE);
2089 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (10, 'eg_tcn_source',oils_i18n_gettext(10, 'vqbrad', 'TCN Source', 'description'),'//*[@tag="901"]/*[@code="b"]', TRUE);
2090 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (11, 'eg_identifier',oils_i18n_gettext(11, 'vqbrad', 'Internal ID', 'description'),'//*[@tag="901"]/*[@code="c"]', TRUE);
2091 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (12, 'publisher',oils_i18n_gettext(12, 'vqbrad', 'Publisher', 'description'),'//*[@tag="260"]/*[@code="b"][1]');
2092 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, remove ) VALUES (13, 'pubdate',oils_i18n_gettext(13, 'vqbrad', 'Publication Date', 'description'),'//*[@tag="260"]/*[@code="c"][1]',$r$\D$r$);
2093 --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]');
2095 --INSERT INTO vandelay.import_item_attr_definition (
2096 -- owner, name, tag, owning_lib, circ_lib, location,
2097 -- call_number, circ_modifier, barcode, price, copy_number,
2098 -- circulate, ref, holdable, opac_visible, status
2101 -- 'Evergreen 852 export format',
2103 -- '[@code = "b"][1]',
2104 -- '[@code = "b"][2]',
2111 -- '[@code = "x" and text() = "circulating"]',
2112 -- '[@code = "x" and text() = "reference"]',
2113 -- '[@code = "x" and text() = "holdable"]',
2114 -- '[@code = "x" and text() = "visible"]',
2118 --INSERT INTO vandelay.import_item_attr_definition (
2131 -- 'Unicorn Import format -- 999',
2142 --INSERT INTO vandelay.authority_attr_definition ( code, description, xpath, ident ) VALUES ('rec_identifier','Identifier','//*[@tag="001"]', TRUE);