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,
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 m.value::TEXT INTO attr_value
444 FROM vandelay.marc21_physical_characteristics(xml) v
445 JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
446 WHERE v.subfield = attr_def.phys_char_sf
447 LIMIT 1; -- Just in case ...
451 -- apply index normalizers to attr_value
453 SELECT n.func AS func,
454 n.param_count AS param_count,
456 FROM config.index_normalizer n
457 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
458 WHERE attr = attr_def.name
460 EXECUTE 'SELECT ' || normalizer.func || '(' ||
461 quote_literal( attr_value ) ||
463 WHEN normalizer.param_count > 0
464 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
471 -- Add the new value to the hstore
472 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
478 $_$ LANGUAGE PLPGSQL;
480 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT ) RETURNS hstore AS $_$
481 SELECT vandelay.extract_rec_attrs( $1, (SELECT ARRAY_ACCUM(name) FROM config.record_attr_definition));
484 -- Everything between this comment and the beginning of the definition of
485 -- vandelay.match_bib_record() is strictly in service of that function.
486 CREATE TYPE vandelay.match_set_test_result AS (record BIGINT, quality INTEGER);
488 CREATE OR REPLACE FUNCTION vandelay.match_set_test_marcxml(
489 match_set_id INTEGER, record_xml TEXT
490 ) RETURNS SETOF vandelay.match_set_test_result AS $$
501 tags_rstore := vandelay.flatten_marc_hstore(record_xml);
502 svf_rstore := vandelay.extract_rec_attrs(record_xml);
504 CREATE TEMPORARY TABLE _vandelay_tmp_qrows (q INTEGER);
505 CREATE TEMPORARY TABLE _vandelay_tmp_jrows (j TEXT);
507 -- generate the where clause and return that directly (into wq), and as
508 -- a side-effect, populate the _vandelay_tmp_[qj]rows tables.
509 wq := vandelay.get_expr_from_match_set(match_set_id);
511 query_ := 'SELECT bre.id AS record, ';
513 -- qrows table is for the quality bits we add to the SELECT clause
514 SELECT ARRAY_TO_STRING(
515 ARRAY_ACCUM('COALESCE(n' || q::TEXT || '.quality, 0)'), ' + '
516 ) INTO coal FROM _vandelay_tmp_qrows;
518 -- our query string so far is the SELECT clause and the inital FROM.
519 -- no JOINs yet nor the WHERE clause
520 query_ := query_ || coal || ' AS quality ' || E'\n' ||
521 'FROM biblio.record_entry bre ';
523 -- jrows table is for the joins we must make (and the real text conditions)
524 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(j), E'\n') INTO joins
525 FROM _vandelay_tmp_jrows;
527 -- add those joins and the where clause to our query.
528 query_ := query_ || joins || E'\n' || 'WHERE ' || wq || ' AND not bre.deleted';
530 -- this will return rows of record,quality
531 FOR rec IN EXECUTE query_ USING tags_rstore, svf_rstore LOOP
535 DROP TABLE _vandelay_tmp_qrows;
536 DROP TABLE _vandelay_tmp_jrows;
542 CREATE OR REPLACE FUNCTION vandelay.flatten_marc_hstore(
544 ) RETURNS HSTORE AS $$
548 ARRAY_ACCUM(tag || (COALESCE(subfield, ''))),
552 SELECT tag, subfield, ARRAY_ACCUM(value)::TEXT AS value
553 FROM vandelay.flatten_marc(record_xml)
554 GROUP BY tag, subfield ORDER BY tag, subfield
560 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set(
564 root vandelay.match_set_point;
566 SELECT * INTO root FROM vandelay.match_set_point
567 WHERE parent IS NULL AND match_set = match_set_id;
569 RETURN vandelay.get_expr_from_match_set_point(root);
573 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set_point(
574 node vandelay.match_set_point
581 child vandelay.match_set_point;
583 SELECT ARRAY_ACCUM(id) INTO children FROM vandelay.match_set_point
584 WHERE parent = node.id;
586 IF ARRAY_LENGTH(children, 1) > 0 THEN
587 this_op := vandelay._get_expr_render_one(node);
590 WHILE children[i] IS NOT NULL LOOP
591 SELECT * INTO child FROM vandelay.match_set_point
592 WHERE id = children[i];
594 q := q || ' ' || this_op || ' ';
597 q := q || vandelay.get_expr_from_match_set_point(child);
601 ELSIF node.bool_op IS NULL THEN
602 PERFORM vandelay._get_expr_push_qrow(node);
603 PERFORM vandelay._get_expr_push_jrow(node);
604 RETURN vandelay._get_expr_render_one(node);
611 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_qrow(
612 node vandelay.match_set_point
616 INSERT INTO _vandelay_tmp_qrows (q) VALUES (node.id);
620 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_jrow(
621 node vandelay.match_set_point
635 IF node.tag IS NOT NULL THEN
637 IF node.subfield IS NOT NULL THEN
638 tagkey := tagkey || node.subfield;
642 my_alias := 'n' || node.id::TEXT;
644 jrow := 'LEFT JOIN (SELECT *, ' || node.quality ||
645 ' AS quality FROM metabib.';
646 IF node.tag IS NOT NULL THEN
647 jrow := jrow || 'full_rec) ' || my_alias || ' ON (' ||
648 my_alias || '.record = bre.id AND ' || my_alias || '.tag = ''' ||
650 IF node.subfield IS NOT NULL THEN
651 jrow := jrow || ' AND ' || my_alias || '.subfield = ''' ||
652 node.subfield || '''';
654 jrow := jrow || ' AND (' || my_alias || '.value ' || op ||
655 ' ANY(($1->''' || tagkey || ''')::TEXT[])))';
657 jrow := jrow || 'record_attr) ' || my_alias || ' ON (' ||
658 my_alias || '.id = bre.id AND (' ||
659 my_alias || '.attrs->''' || node.svf ||
660 ''' ' || op || ' $2->''' || node.svf || '''))';
662 INSERT INTO _vandelay_tmp_jrows (j) VALUES (jrow);
666 CREATE OR REPLACE FUNCTION vandelay._get_expr_render_one(
667 node vandelay.match_set_point
672 IF node.bool_op IS NOT NULL THEN
675 RETURN '(n' || node.id::TEXT || '.id IS NOT NULL)';
680 CREATE OR REPLACE FUNCTION vandelay.match_bib_record() RETURNS TRIGGER AS $func$
682 incoming_existing_id TEXT;
683 test_result vandelay.match_set_test_result%ROWTYPE;
687 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
691 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
693 SELECT q.match_set INTO match_set FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
695 IF match_set IS NOT NULL THEN
696 NEW.quality := vandelay.measure_record_quality( NEW.marc, match_set );
699 -- Perfect matches on 901$c exit early with a match with high quality.
700 incoming_existing_id :=
701 oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]', NEW.marc);
703 IF incoming_existing_id IS NOT NULL AND incoming_existing_id != '' THEN
704 SELECT id INTO tmp_rec FROM biblio.record_entry WHERE id = incoming_existing_id::bigint;
705 IF tmp_rec IS NOT NULL THEN
706 INSERT INTO vandelay.bib_match (queued_record, eg_record, match_score, quality)
711 -- note: no match_set means quality==0
712 vandelay.measure_record_quality( b.marc, match_set )
713 FROM biblio.record_entry b
714 WHERE id = incoming_existing_id::bigint;
718 IF match_set IS NULL THEN
722 FOR test_result IN SELECT * FROM
723 vandelay.match_set_test_marcxml(match_set, NEW.marc) LOOP
725 INSERT INTO vandelay.bib_match ( queued_record, eg_record, match_score, quality )
730 vandelay.measure_record_quality( b.marc, match_set )
731 FROM biblio.record_entry b
732 WHERE id = test_result.record;
738 $func$ LANGUAGE PLPGSQL;
740 CREATE OR REPLACE FUNCTION vandelay.measure_record_quality ( xml TEXT, match_set_id INT ) RETURNS INT AS $_$
744 test vandelay.match_set_quality%ROWTYPE;
747 FOR test IN SELECT * FROM vandelay.match_set_quality WHERE match_set = match_set_id LOOP
748 IF test.tag IS NOT NULL THEN
749 FOR rvalue IN SELECT value FROM vandelay.flatten_marc( xml ) WHERE tag = test.tag AND subfield = test.subfield LOOP
750 IF test.value = rvalue THEN
751 out_q := out_q + test.quality;
755 IF test.value = vandelay.extract_rec_attrs(xml, ARRAY[test.svf]) -> test.svf THEN
756 out_q := out_q + test.quality;
763 $_$ LANGUAGE PLPGSQL;
765 CREATE TYPE vandelay.tcn_data AS (tcn TEXT, tcn_source TEXT, used BOOL);
766 CREATE OR REPLACE FUNCTION vandelay.find_bib_tcn_data ( xml TEXT ) RETURNS SETOF vandelay.tcn_data AS $_$
770 output vandelay.tcn_data%ROWTYPE;
774 eg_tcn := BTRIM((oils_xpath('//*[@tag="001"]/text()',xml))[1]);
775 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
777 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="003"]/text()',xml))[1]);
778 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
779 eg_tcn_source := 'System Local';
782 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
785 output.used := FALSE;
790 output.tcn := eg_tcn;
791 output.tcn_source := eg_tcn_source;
797 eg_tcn := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="a"]/text()',xml))[1]);
798 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
800 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="b"]/text()',xml))[1]);
801 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
802 eg_tcn_source := 'System Local';
805 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
808 output.used := FALSE;
813 output.tcn := eg_tcn;
814 output.tcn_source := eg_tcn_source;
820 eg_tcn := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="a"]/text()',xml))[1]);
821 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
823 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="b"]/text()',xml))[1]);
824 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
825 eg_tcn_source := 'System Local';
828 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
831 output.used := FALSE;
836 output.tcn := eg_tcn;
837 output.tcn_source := eg_tcn_source;
843 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="020"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
844 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
846 eg_tcn_source := 'ISBN';
848 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
851 output.used := FALSE;
856 output.tcn := eg_tcn;
857 output.tcn_source := eg_tcn_source;
863 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="022"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
864 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
866 eg_tcn_source := 'ISSN';
868 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
871 output.used := FALSE;
876 output.tcn := eg_tcn;
877 output.tcn_source := eg_tcn_source;
883 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="010"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
884 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
886 eg_tcn_source := 'LCCN';
888 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
891 output.used := FALSE;
896 output.tcn := eg_tcn;
897 output.tcn_source := eg_tcn_source;
903 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="035"]/*[@code="a"]/text()',xml))[1], $re$^.*?(\w+)$$re$, $re$\1$re$);
904 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
906 eg_tcn_source := 'System Legacy';
908 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
911 output.used := FALSE;
916 output.tcn := eg_tcn;
917 output.tcn_source := eg_tcn_source;
924 $_$ LANGUAGE PLPGSQL;
926 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT, force_add INT ) RETURNS TEXT AS $_$
929 use MARC::File::XML (BinaryEncoding => 'UTF-8');
933 MARC::Charset->assume_unicode(1);
935 my $target_xml = shift;
936 my $source_xml = shift;
937 my $field_spec = shift;
938 my $force_add = shift || 0;
940 my $target_r = MARC::Record->new_from_xml( $target_xml );
941 my $source_r = MARC::Record->new_from_xml( $source_xml );
943 return $target_xml unless ($target_r && $source_r);
945 my @field_list = split(',', $field_spec);
948 for my $f (@field_list) {
949 $f =~ s/^\s*//; $f =~ s/\s*$//;
950 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
956 $match =~ s/^\s*//; $match =~ s/\s*$//;
957 $fields{$field} = { sf => [ split('', $sf) ] };
959 my ($msf,$mre) = split('~', $match);
960 if (length($msf) > 0 and length($mre) > 0) {
961 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
962 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
963 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
969 for my $f ( keys %fields) {
970 if ( @{$fields{$f}{sf}} ) {
971 for my $from_field ($source_r->field( $f )) {
972 my @tos = $target_r->field( $f );
974 next if (exists($fields{$f}{match}) and !$force_add);
975 my @new_fields = map { $_->clone } $source_r->field( $f );
976 $target_r->insert_fields_ordered( @new_fields );
978 for my $to_field (@tos) {
979 if (exists($fields{$f}{match})) {
980 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
982 my @new_sf = map { ($_ => $from_field->subfield($_)) } @{$fields{$f}{sf}};
983 $to_field->add_subfields( @new_sf );
988 my @new_fields = map { $_->clone } $source_r->field( $f );
989 $target_r->insert_fields_ordered( @new_fields );
993 $target_xml = $target_r->as_xml_record;
994 $target_xml =~ s/^<\?.+?\?>$//mo;
995 $target_xml =~ s/\n//sgo;
996 $target_xml =~ s/>\s+</></sgo;
1000 $_$ LANGUAGE PLPERLU;
1002 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1003 SELECT vandelay.add_field( $1, $2, $3, 0 );
1006 CREATE OR REPLACE FUNCTION vandelay.strip_field ( xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1009 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1013 MARC::Charset->assume_unicode(1);
1016 my $r = MARC::Record->new_from_xml( $xml );
1018 return $xml unless ($r);
1020 my $field_spec = shift;
1021 my @field_list = split(',', $field_spec);
1024 for my $f (@field_list) {
1025 $f =~ s/^\s*//; $f =~ s/\s*$//;
1026 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1032 $match =~ s/^\s*//; $match =~ s/\s*$//;
1033 $fields{$field} = { sf => [ split('', $sf) ] };
1035 my ($msf,$mre) = split('~', $match);
1036 if (length($msf) > 0 and length($mre) > 0) {
1037 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1038 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1039 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1045 for my $f ( keys %fields) {
1046 for my $to_field ($r->field( $f )) {
1047 if (exists($fields{$f}{match})) {
1048 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1051 if ( @{$fields{$f}{sf}} ) {
1052 $to_field->delete_subfield(code => $fields{$f}{sf});
1054 $r->delete_field( $to_field );
1059 $xml = $r->as_xml_record;
1060 $xml =~ s/^<\?.+?\?>$//mo;
1062 $xml =~ s/>\s+</></sgo;
1066 $_$ LANGUAGE PLPERLU;
1068 CREATE OR REPLACE FUNCTION vandelay.replace_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1075 parsed_target := vandelay.strip_field( target_xml, ''); -- this dance normalizes the format of the xml for the IF below
1077 FOR curr_field IN SELECT UNNEST( STRING_TO_ARRAY(field, ',') ) LOOP -- naive split, but it's the same we use in the perl
1079 xml_output := vandelay.strip_field( parsed_target, curr_field);
1081 IF xml_output <> parsed_target AND curr_field ~ E'~' THEN
1082 -- we removed something, and there was a regexp restriction in the curr_field definition, so proceed
1083 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 1 );
1084 ELSIF curr_field !~ E'~' THEN
1085 -- No regexp restriction, add the curr_field
1086 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 0 );
1089 parsed_target := xml_output; -- in prep for any following loop iterations
1095 $_$ LANGUAGE PLPGSQL;
1097 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 $_$
1098 SELECT vandelay.replace_field( vandelay.add_field( vandelay.strip_field( $1, $5) , $2, $3 ), $2, $4);
1101 CREATE TYPE vandelay.compile_profile AS (add_rule TEXT, replace_rule TEXT, preserve_rule TEXT, strip_rule TEXT);
1102 CREATE OR REPLACE FUNCTION vandelay.compile_profile ( incoming_xml TEXT ) RETURNS vandelay.compile_profile AS $_$
1104 output vandelay.compile_profile%ROWTYPE;
1105 profile vandelay.merge_profile%ROWTYPE;
1107 profile_tmpl_owner TEXT;
1108 add_rule TEXT := '';
1109 strip_rule TEXT := '';
1110 replace_rule TEXT := '';
1111 preserve_rule TEXT := '';
1115 profile_tmpl := (oils_xpath('//*[@tag="905"]/*[@code="t"]/text()',incoming_xml))[1];
1116 profile_tmpl_owner := (oils_xpath('//*[@tag="905"]/*[@code="o"]/text()',incoming_xml))[1];
1118 IF profile_tmpl IS NOT NULL AND profile_tmpl <> '' AND profile_tmpl_owner IS NOT NULL AND profile_tmpl_owner <> '' THEN
1119 SELECT p.* INTO profile
1120 FROM vandelay.merge_profile p
1121 JOIN actor.org_unit u ON (u.id = p.owner)
1122 WHERE p.name = profile_tmpl
1123 AND u.shortname = profile_tmpl_owner;
1125 IF profile.id IS NOT NULL THEN
1126 add_rule := COALESCE(profile.add_spec,'');
1127 strip_rule := COALESCE(profile.strip_spec,'');
1128 replace_rule := COALESCE(profile.replace_spec,'');
1129 preserve_rule := COALESCE(profile.preserve_spec,'');
1133 add_rule := add_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="a"]/text()',incoming_xml),','),'');
1134 strip_rule := strip_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="d"]/text()',incoming_xml),','),'');
1135 replace_rule := replace_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="r"]/text()',incoming_xml),','),'');
1136 preserve_rule := preserve_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="p"]/text()',incoming_xml),','),'');
1138 output.add_rule := BTRIM(add_rule,',');
1139 output.replace_rule := BTRIM(replace_rule,',');
1140 output.strip_rule := BTRIM(strip_rule,',');
1141 output.preserve_rule := BTRIM(preserve_rule,',');
1145 $_$ LANGUAGE PLPGSQL;
1147 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1149 merge_profile vandelay.merge_profile%ROWTYPE;
1150 dyn_profile vandelay.compile_profile%ROWTYPE;
1160 SELECT b.marc INTO eg_marc
1161 FROM biblio.record_entry b
1165 IF eg_marc IS NULL OR v_marc IS NULL THEN
1166 -- RAISE NOTICE 'no marc for template or bib record';
1170 dyn_profile := vandelay.compile_profile( v_marc );
1172 IF merge_profile_id IS NOT NULL THEN
1173 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1175 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1176 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1177 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1178 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1182 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1183 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1187 IF dyn_profile.replace_rule <> '' THEN
1188 source_marc = v_marc;
1189 target_marc = eg_marc;
1190 replace_rule = dyn_profile.replace_rule;
1192 source_marc = eg_marc;
1193 target_marc = v_marc;
1194 replace_rule = dyn_profile.preserve_rule;
1197 UPDATE biblio.record_entry
1198 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1202 -- RAISE NOTICE 'update of biblio.record_entry failed';
1209 $$ LANGUAGE PLPGSQL;
1211 CREATE OR REPLACE FUNCTION vandelay.merge_record_xml ( target_marc TEXT, template_marc TEXT ) RETURNS TEXT AS $$
1213 dyn_profile vandelay.compile_profile%ROWTYPE;
1221 IF target_marc IS NULL OR template_marc IS NULL THEN
1222 -- RAISE NOTICE 'no marc for target or template record';
1226 dyn_profile := vandelay.compile_profile( template_marc );
1228 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1229 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1233 IF dyn_profile.replace_rule <> '' THEN
1234 trgt_marc = target_marc;
1235 tmpl_marc = template_marc;
1236 replace_rule = dyn_profile.replace_rule;
1238 tmp_marc = target_marc;
1239 trgt_marc = template_marc;
1240 tmpl_marc = tmp_marc;
1241 replace_rule = dyn_profile.preserve_rule;
1244 RETURN vandelay.merge_record_xml( trgt_marc, tmpl_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule );
1247 $$ LANGUAGE PLPGSQL;
1249 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT) RETURNS BOOL AS $$
1250 SELECT vandelay.template_overlay_bib_record( $1, $2, NULL);
1253 CREATE OR REPLACE FUNCTION vandelay.overlay_bib_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1255 merge_profile vandelay.merge_profile%ROWTYPE;
1256 dyn_profile vandelay.compile_profile%ROWTYPE;
1266 SELECT q.marc INTO v_marc
1267 FROM vandelay.queued_record q
1268 JOIN vandelay.bib_match m ON (m.queued_record = q.id AND q.id = import_id)
1271 IF v_marc IS NULL THEN
1272 -- RAISE NOTICE 'no marc for vandelay or bib record';
1276 IF vandelay.template_overlay_bib_record( v_marc, eg_id, merge_profile_id) THEN
1277 UPDATE vandelay.queued_bib_record
1278 SET imported_as = eg_id,
1280 WHERE id = import_id;
1282 editor_string := (oils_xpath('//*[@tag="905"]/*[@code="u"]/text()',v_marc))[1];
1284 IF editor_string IS NOT NULL AND editor_string <> '' THEN
1285 SELECT usr INTO editor_id FROM actor.card WHERE barcode = editor_string;
1287 IF editor_id IS NULL THEN
1288 SELECT id INTO editor_id FROM actor.usr WHERE usrname = editor_string;
1291 IF editor_id IS NOT NULL THEN
1292 UPDATE biblio.record_entry SET editor = editor_id WHERE id = eg_id;
1299 -- RAISE NOTICE 'update of biblio.record_entry failed';
1304 $$ LANGUAGE PLPGSQL;
1306 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
1309 lwm_ratio_value NUMERIC;
1312 lwm_ratio_value := COALESCE(lwm_ratio_value_p, 0.0);
1314 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1317 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1321 SELECT m.eg_record INTO eg_id
1322 FROM vandelay.bib_match m
1323 JOIN vandelay.queued_bib_record qr ON (m.queued_record = qr.id)
1324 JOIN vandelay.bib_queue q ON (qr.queue = q.id)
1325 JOIN biblio.record_entry r ON (r.id = m.eg_record)
1326 WHERE m.queued_record = import_id
1327 AND qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC >= lwm_ratio_value
1328 ORDER BY m.match_score DESC, -- required match score
1329 qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC DESC, -- quality tie breaker
1330 m.id -- when in doubt, use the first match
1333 IF eg_id IS NULL THEN
1334 -- RAISE NOTICE 'incoming record is not of high enough quality';
1338 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1340 $$ LANGUAGE PLPGSQL;
1342 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1343 SELECT vandelay.auto_overlay_bib_record_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1346 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1352 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1355 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1359 SELECT COUNT(*) INTO match_count FROM vandelay.bib_match WHERE queued_record = import_id;
1361 IF match_count <> 1 THEN
1362 -- RAISE NOTICE 'not an exact match';
1366 -- Check that the one match is on the first 901c
1367 SELECT m.eg_record INTO eg_id
1368 FROM vandelay.queued_bib_record q
1369 JOIN vandelay.bib_match m ON (m.queued_record = q.id)
1370 WHERE q.id = import_id
1371 AND m.eg_record = oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]',marc)::BIGINT;
1374 -- RAISE NOTICE 'not a 901c match';
1378 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1380 $$ LANGUAGE PLPGSQL;
1382 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1384 queued_record vandelay.queued_bib_record%ROWTYPE;
1387 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1389 IF vandelay.auto_overlay_bib_record( queued_record.id, merge_profile_id ) THEN
1390 RETURN NEXT queued_record.id;
1398 $$ LANGUAGE PLPGSQL;
1400 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 $$
1402 queued_record vandelay.queued_bib_record%ROWTYPE;
1405 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1407 IF vandelay.auto_overlay_bib_record_with_best( queued_record.id, merge_profile_id, lwm_ratio_value ) THEN
1408 RETURN NEXT queued_record.id;
1416 $$ LANGUAGE PLPGSQL;
1418 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1419 SELECT vandelay.auto_overlay_bib_queue_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1422 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1423 SELECT * FROM vandelay.auto_overlay_bib_queue( $1, NULL );
1426 CREATE OR REPLACE FUNCTION vandelay.ingest_bib_marc ( ) RETURNS TRIGGER AS $$
1432 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1436 FOR adef IN SELECT * FROM vandelay.bib_attr_definition LOOP
1438 SELECT extract_marc_field('vandelay.queued_bib_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_bib_record WHERE id = NEW.id;
1439 IF (value IS NOT NULL AND value <> '') THEN
1440 INSERT INTO vandelay.queued_bib_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1447 $$ LANGUAGE PLPGSQL;
1449 CREATE OR REPLACE FUNCTION vandelay.match_bib_record ( ) RETURNS TRIGGER AS $func$
1458 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
1460 SELECT * INTO attr_def FROM vandelay.bib_attr_definition WHERE xpath = '//*[@tag="901"]/*[@code="c"]' ORDER BY id LIMIT 1;
1462 IF attr_def IS NOT NULL AND attr_def.id IS NOT NULL THEN
1463 id_value := extract_marc_field('vandelay.queued_bib_record', NEW.id, attr_def.xpath, attr_def.remove);
1465 IF id_value IS NOT NULL AND id_value <> '' AND id_value ~ $r$^\d+$$r$ THEN
1466 SELECT id INTO exact_id FROM biblio.record_entry WHERE id = id_value::BIGINT AND NOT deleted;
1467 SELECT * INTO attr FROM vandelay.queued_bib_record_attr WHERE record = NEW.id and field = attr_def.id LIMIT 1;
1468 IF exact_id IS NOT NULL THEN
1469 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('id', attr.id, NEW.id, exact_id);
1474 IF exact_id IS NULL THEN
1475 FOR attr IN SELECT a.* FROM vandelay.queued_bib_record_attr a JOIN vandelay.bib_attr_definition d ON (d.id = a.field) WHERE record = NEW.id AND d.ident IS TRUE LOOP
1477 -- All numbers? check for an id match
1478 IF (attr.attr_value ~ $r$^\d+$$r$) THEN
1479 FOR eg_rec IN SELECT * FROM biblio.record_entry WHERE id = attr.attr_value::BIGINT AND deleted IS FALSE LOOP
1480 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('id', attr.id, NEW.id, eg_rec.id);
1484 -- Looks like an ISBN? check for an isbn match
1485 IF (attr.attr_value ~* $r$^[0-9x]+$$r$ AND character_length(attr.attr_value) IN (10,13)) THEN
1486 FOR eg_rec IN EXECUTE $$SELECT * FROM metabib.full_rec fr WHERE fr.value LIKE evergreen.lowercase('$$ || attr.attr_value || $$%') AND fr.tag = '020' AND fr.subfield = 'a'$$ LOOP
1487 PERFORM id FROM biblio.record_entry WHERE id = eg_rec.record AND deleted IS FALSE;
1489 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('isbn', attr.id, NEW.id, eg_rec.record);
1493 -- subcheck for isbn-as-tcn
1494 FOR eg_rec IN SELECT * FROM biblio.record_entry WHERE tcn_value = 'i' || attr.attr_value AND deleted IS FALSE LOOP
1495 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('tcn_value', attr.id, NEW.id, eg_rec.id);
1499 -- check for an OCLC tcn_value match
1500 IF (attr.attr_value ~ $r$^o\d+$$r$) THEN
1501 FOR eg_rec IN SELECT * FROM biblio.record_entry WHERE tcn_value = regexp_replace(attr.attr_value,'^o','ocm') AND deleted IS FALSE LOOP
1502 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('tcn_value', attr.id, NEW.id, eg_rec.id);
1506 -- check for a direct tcn_value match
1507 FOR eg_rec IN SELECT * FROM biblio.record_entry WHERE tcn_value = attr.attr_value AND deleted IS FALSE LOOP
1508 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('tcn_value', attr.id, NEW.id, eg_rec.id);
1511 -- check for a direct item barcode match
1514 FROM biblio.record_entry b
1515 JOIN asset.call_number cn ON (cn.record = b.id)
1516 JOIN asset.copy cp ON (cp.call_number = cn.id)
1517 WHERE cp.barcode = attr.attr_value AND cp.deleted IS FALSE
1519 INSERT INTO vandelay.bib_match (field_type, matched_attr, queued_record, eg_record) VALUES ('id', attr.id, NEW.id, eg_rec.id);
1527 $func$ LANGUAGE PLPGSQL;
1529 CREATE OR REPLACE FUNCTION vandelay.cleanup_bib_marc ( ) RETURNS TRIGGER AS $$
1531 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1535 DELETE FROM vandelay.queued_bib_record_attr WHERE record = OLD.id;
1536 DELETE FROM vandelay.import_item WHERE record = OLD.id;
1538 IF TG_OP = 'UPDATE' THEN
1543 $$ LANGUAGE PLPGSQL;
1545 CREATE TRIGGER cleanup_bib_trigger
1546 BEFORE UPDATE OR DELETE ON vandelay.queued_bib_record
1547 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_bib_marc();
1549 CREATE TRIGGER ingest_bib_trigger
1550 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1551 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_bib_marc();
1553 CREATE TRIGGER zz_match_bibs_trigger
1554 BEFORE INSERT OR UPDATE ON vandelay.queued_bib_record
1555 FOR EACH ROW EXECUTE PROCEDURE vandelay.match_bib_record();
1558 /* Authority stuff down here */
1559 ---------------------------------------
1560 CREATE TABLE vandelay.authority_attr_definition (
1561 id SERIAL PRIMARY KEY,
1562 code TEXT UNIQUE NOT NULL,
1564 xpath TEXT NOT NULL,
1565 remove TEXT NOT NULL DEFAULT ''
1568 CREATE TABLE vandelay.authority_queue (
1569 queue_type TEXT NOT NULL DEFAULT 'authority' CHECK (queue_type = 'authority'),
1570 CONSTRAINT vand_authority_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
1571 ) INHERITS (vandelay.queue);
1572 ALTER TABLE vandelay.authority_queue ADD PRIMARY KEY (id);
1574 CREATE TABLE vandelay.queued_authority_record (
1575 queue INT NOT NULL REFERENCES vandelay.authority_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1576 imported_as INT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1577 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
1579 ) INHERITS (vandelay.queued_record);
1580 ALTER TABLE vandelay.queued_authority_record ADD PRIMARY KEY (id);
1581 CREATE INDEX queued_authority_record_queue_idx ON vandelay.queued_authority_record (queue);
1583 CREATE TABLE vandelay.queued_authority_record_attr (
1584 id BIGSERIAL PRIMARY KEY,
1585 record BIGINT NOT NULL REFERENCES vandelay.queued_authority_record (id) DEFERRABLE INITIALLY DEFERRED,
1586 field INT NOT NULL REFERENCES vandelay.authority_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
1587 attr_value TEXT NOT NULL
1589 CREATE INDEX queued_authority_record_attr_record_idx ON vandelay.queued_authority_record_attr (record);
1591 CREATE TABLE vandelay.authority_match (
1592 id BIGSERIAL PRIMARY KEY,
1593 queued_record BIGINT REFERENCES vandelay.queued_authority_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1594 eg_record BIGINT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1595 quality INT NOT NULL DEFAULT 0
1598 CREATE OR REPLACE FUNCTION vandelay.ingest_authority_marc ( ) RETURNS TRIGGER AS $$
1604 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1608 FOR adef IN SELECT * FROM vandelay.authority_attr_definition LOOP
1610 SELECT extract_marc_field('vandelay.queued_authority_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_authority_record WHERE id = NEW.id;
1611 IF (value IS NOT NULL AND value <> '') THEN
1612 INSERT INTO vandelay.queued_authority_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1619 $$ LANGUAGE PLPGSQL;
1621 CREATE OR REPLACE FUNCTION vandelay.cleanup_authority_marc ( ) RETURNS TRIGGER AS $$
1623 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1627 DELETE FROM vandelay.queued_authority_record_attr WHERE record = OLD.id;
1628 IF TG_OP = 'UPDATE' THEN
1633 $$ LANGUAGE PLPGSQL;
1635 CREATE TRIGGER cleanup_authority_trigger
1636 BEFORE UPDATE OR DELETE ON vandelay.queued_authority_record
1637 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_authority_marc();
1639 CREATE TRIGGER ingest_authority_trigger
1640 AFTER INSERT OR UPDATE ON vandelay.queued_authority_record
1641 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_authority_marc();
1643 CREATE OR REPLACE FUNCTION vandelay.overlay_authority_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1645 merge_profile vandelay.merge_profile%ROWTYPE;
1646 dyn_profile vandelay.compile_profile%ROWTYPE;
1655 SELECT b.marc INTO eg_marc
1656 FROM authority.record_entry b
1657 JOIN vandelay.authority_match m ON (m.eg_record = b.id AND m.queued_record = import_id)
1660 SELECT q.marc INTO v_marc
1661 FROM vandelay.queued_record q
1662 JOIN vandelay.authority_match m ON (m.queued_record = q.id AND q.id = import_id)
1665 IF eg_marc IS NULL OR v_marc IS NULL THEN
1666 -- RAISE NOTICE 'no marc for vandelay or authority record';
1670 dyn_profile := vandelay.compile_profile( v_marc );
1672 IF merge_profile_id IS NOT NULL THEN
1673 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1675 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1676 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1677 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1678 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1682 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1683 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1687 IF dyn_profile.replace_rule <> '' THEN
1688 source_marc = v_marc;
1689 target_marc = eg_marc;
1690 replace_rule = dyn_profile.replace_rule;
1692 source_marc = eg_marc;
1693 target_marc = v_marc;
1694 replace_rule = dyn_profile.preserve_rule;
1697 UPDATE authority.record_entry
1698 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1702 UPDATE vandelay.queued_authority_record
1703 SET imported_as = eg_id,
1705 WHERE id = import_id;
1709 -- RAISE NOTICE 'update of authority.record_entry failed';
1714 $$ LANGUAGE PLPGSQL;
1716 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1721 SELECT COUNT(*) INTO match_count FROM vandelay.authority_match WHERE queued_record = import_id;
1723 IF match_count <> 1 THEN
1724 -- RAISE NOTICE 'not an exact match';
1728 SELECT m.eg_record INTO eg_id
1729 FROM vandelay.authority_match m
1730 WHERE m.queued_record = import_id
1733 IF eg_id IS NULL THEN
1737 RETURN vandelay.overlay_authority_record( import_id, eg_id, merge_profile_id );
1739 $$ LANGUAGE PLPGSQL;
1741 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1743 queued_record vandelay.queued_authority_record%ROWTYPE;
1746 FOR queued_record IN SELECT * FROM vandelay.queued_authority_record WHERE queue = queue_id AND import_time IS NULL LOOP
1748 IF vandelay.auto_overlay_authority_record( queued_record.id, merge_profile_id ) THEN
1749 RETURN NEXT queued_record.id;
1757 $$ LANGUAGE PLPGSQL;
1759 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1760 SELECT * FROM vandelay.auto_overlay_authority_queue( $1, NULL );
1764 -- Vandelay (for importing and exporting records) 012.schema.vandelay.sql
1765 --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)]');
1766 --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)]');
1767 --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]');
1768 --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]');
1769 --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$);
1770 --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$);
1771 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (7, 'price',oils_i18n_gettext(7, 'vqbrad', 'Price', 'description'),'//*[@tag="020" or @tag="022"]/*[@code="c"][1]');
1772 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (8, 'rec_identifier',oils_i18n_gettext(8, 'vqbrad', 'Accession Number', 'description'),'//*[@tag="001"]', TRUE);
1773 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (9, 'eg_tcn',oils_i18n_gettext(9, 'vqbrad', 'TCN Value', 'description'),'//*[@tag="901"]/*[@code="a"]', TRUE);
1774 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (10, 'eg_tcn_source',oils_i18n_gettext(10, 'vqbrad', 'TCN Source', 'description'),'//*[@tag="901"]/*[@code="b"]', TRUE);
1775 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (11, 'eg_identifier',oils_i18n_gettext(11, 'vqbrad', 'Internal ID', 'description'),'//*[@tag="901"]/*[@code="c"]', TRUE);
1776 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (12, 'publisher',oils_i18n_gettext(12, 'vqbrad', 'Publisher', 'description'),'//*[@tag="260"]/*[@code="b"][1]');
1777 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, remove ) VALUES (13, 'pubdate',oils_i18n_gettext(13, 'vqbrad', 'Publication Date', 'description'),'//*[@tag="260"]/*[@code="c"][1]',$r$\D$r$);
1778 --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]');
1780 --INSERT INTO vandelay.import_item_attr_definition (
1781 -- owner, name, tag, owning_lib, circ_lib, location,
1782 -- call_number, circ_modifier, barcode, price, copy_number,
1783 -- circulate, ref, holdable, opac_visible, status
1786 -- 'Evergreen 852 export format',
1788 -- '[@code = "b"][1]',
1789 -- '[@code = "b"][2]',
1796 -- '[@code = "x" and text() = "circulating"]',
1797 -- '[@code = "x" and text() = "reference"]',
1798 -- '[@code = "x" and text() = "holdable"]',
1799 -- '[@code = "x" and text() = "visible"]',
1803 --INSERT INTO vandelay.import_item_attr_definition (
1816 -- 'Unicorn Import format -- 999',
1827 --INSERT INTO vandelay.authority_attr_definition ( code, description, xpath, ident ) VALUES ('rec_identifier','Identifier','//*[@tag="001"]', TRUE);