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 match_set INT REFERENCES vandelay.match_set (id) ON UPDATE CASCADE ON DELETE SET NULL DEFERRABLE INITIALLY DEFERRED
56 CREATE TABLE vandelay.queued_record (
57 id BIGSERIAL PRIMARY KEY,
58 create_time TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT NOW(),
59 import_time TIMESTAMP WITH TIME ZONE,
60 purpose TEXT NOT NULL DEFAULT 'import' CHECK (purpose IN ('import','overlay')),
62 quality INT NOT NULL DEFAULT 0
67 /* Bib stuff at the top */
68 ----------------------------------------------------
70 CREATE TABLE vandelay.bib_attr_definition (
71 id SERIAL PRIMARY KEY,
72 code TEXT UNIQUE NOT NULL,
75 remove TEXT NOT NULL DEFAULT ''
78 -- Each TEXT field (other than 'name') should hold an XPath predicate for pulling the data needed
79 -- DROP TABLE vandelay.import_item_attr_definition CASCADE;
80 CREATE TABLE vandelay.import_item_attr_definition (
81 id BIGSERIAL PRIMARY KEY,
82 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
85 keep BOOL NOT NULL DEFAULT FALSE,
105 priv_note_title TEXT,
108 CONSTRAINT vand_import_item_attr_def_idx UNIQUE (owner,name)
111 CREATE TABLE vandelay.import_error (
112 code TEXT PRIMARY KEY,
113 description TEXT NOT NULL -- i18n
116 CREATE TYPE vandelay.bib_queue_queue_type AS ENUM ('bib', 'acq');
118 CREATE TABLE vandelay.bib_queue (
119 queue_type vandelay.bib_queue_queue_type NOT NULL DEFAULT 'bib',
120 item_attr_def BIGINT REFERENCES vandelay.import_item_attr_definition (id) ON DELETE SET NULL DEFERRABLE INITIALLY DEFERRED,
121 CONSTRAINT vand_bib_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
122 ) INHERITS (vandelay.queue);
123 ALTER TABLE vandelay.bib_queue ADD PRIMARY KEY (id);
125 CREATE TABLE vandelay.queued_bib_record (
126 queue INT NOT NULL REFERENCES vandelay.bib_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
127 bib_source INT REFERENCES config.bib_source (id) DEFERRABLE INITIALLY DEFERRED,
128 imported_as BIGINT REFERENCES biblio.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
129 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
131 ) INHERITS (vandelay.queued_record);
132 ALTER TABLE vandelay.queued_bib_record ADD PRIMARY KEY (id);
133 CREATE INDEX queued_bib_record_queue_idx ON vandelay.queued_bib_record (queue);
135 CREATE TABLE vandelay.queued_bib_record_attr (
136 id BIGSERIAL PRIMARY KEY,
137 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) DEFERRABLE INITIALLY DEFERRED,
138 field INT NOT NULL REFERENCES vandelay.bib_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
139 attr_value TEXT NOT NULL
141 CREATE INDEX queued_bib_record_attr_record_idx ON vandelay.queued_bib_record_attr (record);
143 CREATE TABLE vandelay.bib_match (
144 id BIGSERIAL PRIMARY KEY,
145 queued_record BIGINT REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
146 eg_record BIGINT REFERENCES biblio.record_entry (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
147 quality INT NOT NULL DEFAULT 1,
148 match_score INT NOT NULL DEFAULT 0
151 CREATE TABLE vandelay.import_item (
152 id BIGSERIAL PRIMARY KEY,
153 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
154 definition BIGINT NOT NULL REFERENCES vandelay.import_item_attr_definition (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
155 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
158 import_time TIMESTAMP WITH TIME ZONE,
167 deposit_amount NUMERIC(8,2),
178 internal_id BIGINT -- queue_type == 'acq' ? acq.lineitem_detail.id : asset.copy.id
181 CREATE TABLE vandelay.import_bib_trash_fields (
182 id BIGSERIAL PRIMARY KEY,
183 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
185 CONSTRAINT vand_import_bib_trash_fields_idx UNIQUE (owner,field)
188 CREATE TABLE vandelay.merge_profile (
189 id BIGSERIAL PRIMARY KEY,
190 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
197 CONSTRAINT vand_merge_prof_owner_name_idx UNIQUE (owner,name),
198 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))
201 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
208 retval config.marc21_rec_type_map%ROWTYPE;
210 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
212 IF ldr IS NULL OR ldr = '' THEN
213 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
217 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
218 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
221 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
222 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
224 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
226 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
229 IF retval.code IS NULL THEN
230 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
235 $func$ LANGUAGE PLPGSQL;
237 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
244 rtype := (vandelay.marc21_record_type( marc )).code;
245 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
246 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
247 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
250 val := REPEAT( ff_pos.default_val, ff_pos.length );
256 $func$ LANGUAGE PLPGSQL;
258 CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
259 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
264 output biblio.record_ff_map%ROWTYPE;
266 rtype := (vandelay.marc21_record_type( marc )).code;
268 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
269 output.ff_name := ff_pos.fixed_field;
270 output.ff_value := NULL;
272 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(tag) || '"]/text()', marc ) ) x(value) LOOP
273 output.ff_value := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
274 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
276 output.ff_value := NULL;
283 $func$ LANGUAGE PLPGSQL;
285 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
286 CREATE OR REPLACE FUNCTION vandelay.marc21_physical_characteristics( marc TEXT) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
290 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
291 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
292 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
293 retval biblio.marc21_physical_characteristics%ROWTYPE;
296 _007 := oils_xpath_string( '//*[@tag="007"]', marc );
298 IF _007 IS NOT NULL AND _007 <> '' THEN
299 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007, 1, 1 );
301 IF ptype.ptype_key IS NOT NULL THEN
302 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
303 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 );
305 IF pval.id IS NOT NULL THEN
308 retval.ptype := ptype.ptype_key;
309 retval.subfield := psf.id;
310 retval.value := pval.id;
320 $func$ LANGUAGE PLPGSQL;
322 CREATE TYPE vandelay.flat_marc AS ( tag CHAR(3), ind1 TEXT, ind2 TEXT, subfield TEXT, value TEXT );
323 CREATE OR REPLACE FUNCTION vandelay.flay_marc ( TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
326 use MARC::File::XML (BinaryEncoding => 'UTF-8');
330 MARC::Charset->assume_unicode(1);
333 my $r = MARC::Record->new_from_xml( $xml );
335 return_next( { tag => 'LDR', value => $r->leader } );
337 for my $f ( $r->fields ) {
338 if ($f->is_control_field) {
339 return_next({ tag => $f->tag, value => $f->data });
341 for my $s ($f->subfields) {
344 ind1 => $f->indicator(1),
345 ind2 => $f->indicator(2),
350 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
351 my $trim = $f->indicator(2) || 0;
354 ind1 => $f->indicator(1),
355 ind2 => $f->indicator(2),
357 value => substr( $s->[1], $trim )
366 $func$ LANGUAGE PLPERLU;
368 CREATE OR REPLACE FUNCTION vandelay.flatten_marc ( marc TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
370 output vandelay.flat_marc%ROWTYPE;
373 FOR field IN SELECT * FROM vandelay.flay_marc( marc ) LOOP
374 output.ind1 := field.ind1;
375 output.ind2 := field.ind2;
376 output.tag := field.tag;
377 output.subfield := field.subfield;
378 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
379 output.value := naco_normalize(field.value, field.subfield);
381 output.value := field.value;
384 CONTINUE WHEN output.value IS NULL;
389 $func$ LANGUAGE PLPGSQL;
391 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT, attr_defs TEXT[]) RETURNS hstore AS $_$
393 transformed_xml TEXT;
396 xfrm config.xml_transform%ROWTYPE;
398 new_attrs HSTORE := ''::HSTORE;
399 attr_def config.record_attr_definition%ROWTYPE;
402 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE name IN (SELECT * FROM UNNEST(attr_defs)) ORDER BY format LOOP
404 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
405 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(x.value), COALESCE(attr_def.joiner,' ')) INTO attr_value
406 FROM vandelay.flatten_marc(xml) AS x
407 WHERE x.tag LIKE attr_def.tag
409 WHEN attr_def.sf_list IS NOT NULL
410 THEN POSITION(x.subfield IN attr_def.sf_list) > 0
417 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
418 attr_value := vandelay.marc21_extract_fixed_field(xml, attr_def.fixed_field);
420 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
422 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
424 -- See if we can skip the XSLT ... it's expensive
425 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
426 -- Can't skip the transform
427 IF xfrm.xslt <> '---' THEN
428 transformed_xml := oils_xslt_process(xml,xfrm.xslt);
430 transformed_xml := xml;
433 prev_xfrm := xfrm.name;
436 IF xfrm.name IS NULL THEN
437 -- just grab the marcxml (empty) transform
438 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
439 prev_xfrm := xfrm.name;
442 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
444 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
445 SELECT m.value::TEXT INTO attr_value
446 FROM vandelay.marc21_physical_characteristics(xml) v
447 JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
448 WHERE v.subfield = attr_def.phys_char_sf
449 LIMIT 1; -- Just in case ...
453 -- apply index normalizers to attr_value
455 SELECT n.func AS func,
456 n.param_count AS param_count,
458 FROM config.index_normalizer n
459 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
460 WHERE attr = attr_def.name
462 EXECUTE 'SELECT ' || normalizer.func || '(' ||
463 quote_literal( attr_value ) ||
465 WHEN normalizer.param_count > 0
466 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
473 -- Add the new value to the hstore
474 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
480 $_$ LANGUAGE PLPGSQL;
482 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT ) RETURNS hstore AS $_$
483 SELECT vandelay.extract_rec_attrs( $1, (SELECT ARRAY_ACCUM(name) FROM config.record_attr_definition));
486 -- Everything between this comment and the beginning of the definition of
487 -- vandelay.match_bib_record() is strictly in service of that function.
488 CREATE TYPE vandelay.match_set_test_result AS (record BIGINT, quality INTEGER);
490 CREATE OR REPLACE FUNCTION vandelay.match_set_test_marcxml(
491 match_set_id INTEGER, record_xml TEXT
492 ) RETURNS SETOF vandelay.match_set_test_result AS $$
503 tags_rstore := vandelay.flatten_marc_hstore(record_xml);
504 svf_rstore := vandelay.extract_rec_attrs(record_xml);
506 CREATE TEMPORARY TABLE _vandelay_tmp_qrows (q INTEGER);
507 CREATE TEMPORARY TABLE _vandelay_tmp_jrows (j TEXT);
509 -- generate the where clause and return that directly (into wq), and as
510 -- a side-effect, populate the _vandelay_tmp_[qj]rows tables.
511 wq := vandelay.get_expr_from_match_set(match_set_id, tags_rstore);
513 query_ := 'SELECT DISTINCT(bre.id) AS record, ';
515 -- qrows table is for the quality bits we add to the SELECT clause
516 SELECT ARRAY_TO_STRING(
517 ARRAY_ACCUM('COALESCE(n' || q::TEXT || '.quality, 0)'), ' + '
518 ) INTO coal FROM _vandelay_tmp_qrows;
520 -- our query string so far is the SELECT clause and the inital FROM.
521 -- no JOINs yet nor the WHERE clause
522 query_ := query_ || coal || ' AS quality ' || E'\n' ||
523 'FROM biblio.record_entry bre ';
525 -- jrows table is for the joins we must make (and the real text conditions)
526 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(j), E'\n') INTO joins
527 FROM _vandelay_tmp_jrows;
529 -- add those joins and the where clause to our query.
530 query_ := query_ || joins || E'\n' || 'WHERE ' || wq || ' AND not bre.deleted';
532 -- this will return rows of record,quality
533 FOR rec IN EXECUTE query_ USING tags_rstore, svf_rstore LOOP
537 DROP TABLE _vandelay_tmp_qrows;
538 DROP TABLE _vandelay_tmp_jrows;
544 CREATE OR REPLACE FUNCTION vandelay.flatten_marc_hstore(
546 ) RETURNS HSTORE AS $func$
550 ARRAY_ACCUM(tag || (COALESCE(subfield, ''))),
554 SELECT tag, subfield, ARRAY_ACCUM(value)::TEXT AS value
557 CASE WHEN tag = '020' THEN -- caseless -- isbn
558 LOWER((REGEXP_MATCHES(value,$$^(\S{10,17})$$))[1] || '%')
559 WHEN tag = '022' THEN -- caseless -- issn
560 LOWER((REGEXP_MATCHES(value,$$^(\S{4}[- ]?\S{4})$$))[1] || '%')
561 WHEN tag = '024' THEN -- caseless -- upc (other)
566 FROM vandelay.flatten_marc(record_xml)) x
567 GROUP BY tag, subfield ORDER BY tag, subfield
571 $func$ LANGUAGE PLPGSQL;
573 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set(
574 match_set_id INTEGER,
578 root vandelay.match_set_point;
580 SELECT * INTO root FROM vandelay.match_set_point
581 WHERE parent IS NULL AND match_set = match_set_id;
583 RETURN vandelay.get_expr_from_match_set_point(root, tags_rstore);
587 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set_point(
588 node vandelay.match_set_point,
596 child vandelay.match_set_point;
598 SELECT ARRAY_ACCUM(id) INTO children FROM vandelay.match_set_point
599 WHERE parent = node.id;
601 IF ARRAY_LENGTH(children, 1) > 0 THEN
602 this_op := vandelay._get_expr_render_one(node);
605 WHILE children[i] IS NOT NULL LOOP
606 SELECT * INTO child FROM vandelay.match_set_point
607 WHERE id = children[i];
609 q := q || ' ' || this_op || ' ';
612 q := q || vandelay.get_expr_from_match_set_point(child, tags_rstore);
616 ELSIF node.bool_op IS NULL THEN
617 PERFORM vandelay._get_expr_push_qrow(node);
618 PERFORM vandelay._get_expr_push_jrow(node, tags_rstore);
619 RETURN vandelay._get_expr_render_one(node);
626 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_qrow(
627 node vandelay.match_set_point
631 INSERT INTO _vandelay_tmp_qrows (q) VALUES (node.id);
635 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_jrow(
636 node vandelay.match_set_point,
646 -- remember $1 is tags_rstore, and $2 is svf_rstore
650 IF node.tag IS NOT NULL THEN
651 caseless := (node.tag IN ('020', '022', '024'));
653 IF node.subfield IS NOT NULL THEN
654 tagkey := tagkey || node.subfield;
672 my_alias := 'n' || node.id::TEXT;
674 jrow := 'LEFT JOIN (SELECT *, ' || node.quality ||
675 ' AS quality FROM metabib.';
676 IF node.tag IS NOT NULL THEN
677 jrow := jrow || 'full_rec) ' || my_alias || ' ON (' ||
678 my_alias || '.record = bre.id AND ' || my_alias || '.tag = ''' ||
680 IF node.subfield IS NOT NULL THEN
681 jrow := jrow || ' AND ' || my_alias || '.subfield = ''' ||
682 node.subfield || '''';
684 jrow := jrow || ' AND (';
686 jrow := jrow || vandelay._node_tag_comparisons(caseless, my_alias, op, tags_rstore, tagkey);
687 jrow := jrow || '))';
689 jrow := jrow || 'record_attr) ' || my_alias || ' ON (' ||
690 my_alias || '.id = bre.id AND (' ||
691 my_alias || '.attrs->''' || node.svf ||
692 ''' ' || op || ' $2->''' || node.svf || '''))';
694 INSERT INTO _vandelay_tmp_jrows (j) VALUES (jrow);
698 CREATE OR REPLACE FUNCTION vandelay._node_tag_comparisons(
711 vals := tags_rstore->tagkey;
716 IF vals[i] IS NULL THEN
719 result := result || ' OR ';
724 result := result || 'LOWER(' || my_alias || '.value) ' || op;
726 result := result || my_alias || '.value ' || op;
729 result := result || ' ' || COALESCE('''' || vals[i] || '''', 'NULL');
731 IF vals[i] IS NULL THEN
742 CREATE OR REPLACE FUNCTION vandelay._get_expr_render_one(
743 node vandelay.match_set_point
748 IF node.bool_op IS NOT NULL THEN
751 RETURN '(n' || node.id::TEXT || '.id IS NOT NULL)';
756 CREATE OR REPLACE FUNCTION vandelay.match_bib_record() RETURNS TRIGGER AS $func$
758 incoming_existing_id TEXT;
759 test_result vandelay.match_set_test_result%ROWTYPE;
763 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
767 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
769 SELECT q.match_set INTO match_set FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
771 IF match_set IS NOT NULL THEN
772 NEW.quality := vandelay.measure_record_quality( NEW.marc, match_set );
775 -- Perfect matches on 901$c exit early with a match with high quality.
776 incoming_existing_id :=
777 oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]', NEW.marc);
779 IF incoming_existing_id IS NOT NULL AND incoming_existing_id != '' THEN
780 SELECT id INTO tmp_rec FROM biblio.record_entry WHERE id = incoming_existing_id::bigint;
781 IF tmp_rec IS NOT NULL THEN
782 INSERT INTO vandelay.bib_match (queued_record, eg_record, match_score, quality)
787 -- note: no match_set means quality==0
788 vandelay.measure_record_quality( b.marc, match_set )
789 FROM biblio.record_entry b
790 WHERE id = incoming_existing_id::bigint;
794 IF match_set IS NULL THEN
798 FOR test_result IN SELECT * FROM
799 vandelay.match_set_test_marcxml(match_set, NEW.marc) LOOP
801 INSERT INTO vandelay.bib_match ( queued_record, eg_record, match_score, quality )
806 vandelay.measure_record_quality( b.marc, match_set )
807 FROM biblio.record_entry b
808 WHERE id = test_result.record;
814 $func$ LANGUAGE PLPGSQL;
816 CREATE OR REPLACE FUNCTION vandelay.measure_record_quality ( xml TEXT, match_set_id INT ) RETURNS INT AS $_$
820 test vandelay.match_set_quality%ROWTYPE;
823 FOR test IN SELECT * FROM vandelay.match_set_quality WHERE match_set = match_set_id LOOP
824 IF test.tag IS NOT NULL THEN
825 FOR rvalue IN SELECT value FROM vandelay.flatten_marc( xml ) WHERE tag = test.tag AND subfield = test.subfield LOOP
826 IF test.value = rvalue THEN
827 out_q := out_q + test.quality;
831 IF test.value = vandelay.extract_rec_attrs(xml, ARRAY[test.svf]) -> test.svf THEN
832 out_q := out_q + test.quality;
839 $_$ LANGUAGE PLPGSQL;
841 CREATE TYPE vandelay.tcn_data AS (tcn TEXT, tcn_source TEXT, used BOOL);
842 CREATE OR REPLACE FUNCTION vandelay.find_bib_tcn_data ( xml TEXT ) RETURNS SETOF vandelay.tcn_data AS $_$
846 output vandelay.tcn_data%ROWTYPE;
850 eg_tcn := BTRIM((oils_xpath('//*[@tag="001"]/text()',xml))[1]);
851 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
853 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="003"]/text()',xml))[1]);
854 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
855 eg_tcn_source := 'System Local';
858 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
861 output.used := FALSE;
866 output.tcn := eg_tcn;
867 output.tcn_source := eg_tcn_source;
873 eg_tcn := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="a"]/text()',xml))[1]);
874 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
876 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="b"]/text()',xml))[1]);
877 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
878 eg_tcn_source := 'System Local';
881 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
884 output.used := FALSE;
889 output.tcn := eg_tcn;
890 output.tcn_source := eg_tcn_source;
896 eg_tcn := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="a"]/text()',xml))[1]);
897 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
899 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="b"]/text()',xml))[1]);
900 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
901 eg_tcn_source := 'System Local';
904 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
907 output.used := FALSE;
912 output.tcn := eg_tcn;
913 output.tcn_source := eg_tcn_source;
919 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="020"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
920 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
922 eg_tcn_source := 'ISBN';
924 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
927 output.used := FALSE;
932 output.tcn := eg_tcn;
933 output.tcn_source := eg_tcn_source;
939 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="022"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
940 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
942 eg_tcn_source := 'ISSN';
944 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
947 output.used := FALSE;
952 output.tcn := eg_tcn;
953 output.tcn_source := eg_tcn_source;
959 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="010"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
960 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
962 eg_tcn_source := 'LCCN';
964 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
967 output.used := FALSE;
972 output.tcn := eg_tcn;
973 output.tcn_source := eg_tcn_source;
979 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="035"]/*[@code="a"]/text()',xml))[1], $re$^.*?(\w+)$$re$, $re$\1$re$);
980 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
982 eg_tcn_source := 'System Legacy';
984 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
987 output.used := FALSE;
992 output.tcn := eg_tcn;
993 output.tcn_source := eg_tcn_source;
1000 $_$ LANGUAGE PLPGSQL;
1002 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT, force_add INT ) RETURNS TEXT AS $_$
1005 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1009 MARC::Charset->assume_unicode(1);
1011 my $target_xml = shift;
1012 my $source_xml = shift;
1013 my $field_spec = shift;
1014 my $force_add = shift || 0;
1016 my $target_r = MARC::Record->new_from_xml( $target_xml );
1017 my $source_r = MARC::Record->new_from_xml( $source_xml );
1019 return $target_xml unless ($target_r && $source_r);
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 $f ( keys %fields) {
1046 if ( @{$fields{$f}{sf}} ) {
1047 for my $from_field ($source_r->field( $f )) {
1048 my @tos = $target_r->field( $f );
1050 next if (exists($fields{$f}{match}) and !$force_add);
1051 my @new_fields = map { $_->clone } $source_r->field( $f );
1052 $target_r->insert_fields_ordered( @new_fields );
1054 for my $to_field (@tos) {
1055 if (exists($fields{$f}{match})) {
1056 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1058 my @new_sf = map { ($_ => $from_field->subfield($_)) } grep { defined($from_field->subfield($_)) } @{$fields{$f}{sf}};
1059 $to_field->add_subfields( @new_sf );
1064 my @new_fields = map { $_->clone } $source_r->field( $f );
1065 $target_r->insert_fields_ordered( @new_fields );
1069 $target_xml = $target_r->as_xml_record;
1070 $target_xml =~ s/^<\?.+?\?>$//mo;
1071 $target_xml =~ s/\n//sgo;
1072 $target_xml =~ s/>\s+</></sgo;
1076 $_$ LANGUAGE PLPERLU;
1078 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1079 SELECT vandelay.add_field( $1, $2, $3, 0 );
1082 CREATE OR REPLACE FUNCTION vandelay.strip_field ( xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1085 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1089 MARC::Charset->assume_unicode(1);
1092 my $r = MARC::Record->new_from_xml( $xml );
1094 return $xml unless ($r);
1096 my $field_spec = shift;
1097 my @field_list = split(',', $field_spec);
1100 for my $f (@field_list) {
1101 $f =~ s/^\s*//; $f =~ s/\s*$//;
1102 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1108 $match =~ s/^\s*//; $match =~ s/\s*$//;
1109 $fields{$field} = { sf => [ split('', $sf) ] };
1111 my ($msf,$mre) = split('~', $match);
1112 if (length($msf) > 0 and length($mre) > 0) {
1113 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1114 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1115 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1121 for my $f ( keys %fields) {
1122 for my $to_field ($r->field( $f )) {
1123 if (exists($fields{$f}{match})) {
1124 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1127 if ( @{$fields{$f}{sf}} ) {
1128 $to_field->delete_subfield(code => $fields{$f}{sf});
1130 $r->delete_field( $to_field );
1135 $xml = $r->as_xml_record;
1136 $xml =~ s/^<\?.+?\?>$//mo;
1138 $xml =~ s/>\s+</></sgo;
1142 $_$ LANGUAGE PLPERLU;
1144 CREATE OR REPLACE FUNCTION vandelay.replace_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1151 parsed_target := vandelay.strip_field( target_xml, ''); -- this dance normalizes the format of the xml for the IF below
1152 xml_output := parsed_target; -- if there are no replace rules, just return the input
1154 FOR curr_field IN SELECT UNNEST( STRING_TO_ARRAY(field, ',') ) LOOP -- naive split, but it's the same we use in the perl
1156 xml_output := vandelay.strip_field( parsed_target, curr_field);
1158 IF xml_output <> parsed_target AND curr_field ~ E'~' THEN
1159 -- we removed something, and there was a regexp restriction in the curr_field definition, so proceed
1160 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 1 );
1161 ELSIF curr_field !~ E'~' THEN
1162 -- No regexp restriction, add the curr_field
1163 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 0 );
1166 parsed_target := xml_output; -- in prep for any following loop iterations
1172 $_$ LANGUAGE PLPGSQL;
1174 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 $_$
1175 SELECT vandelay.replace_field( vandelay.add_field( vandelay.strip_field( $1, $5) , $2, $3 ), $2, $4);
1178 CREATE TYPE vandelay.compile_profile AS (add_rule TEXT, replace_rule TEXT, preserve_rule TEXT, strip_rule TEXT);
1179 CREATE OR REPLACE FUNCTION vandelay.compile_profile ( incoming_xml TEXT ) RETURNS vandelay.compile_profile AS $_$
1181 output vandelay.compile_profile%ROWTYPE;
1182 profile vandelay.merge_profile%ROWTYPE;
1184 profile_tmpl_owner TEXT;
1185 add_rule TEXT := '';
1186 strip_rule TEXT := '';
1187 replace_rule TEXT := '';
1188 preserve_rule TEXT := '';
1192 profile_tmpl := (oils_xpath('//*[@tag="905"]/*[@code="t"]/text()',incoming_xml))[1];
1193 profile_tmpl_owner := (oils_xpath('//*[@tag="905"]/*[@code="o"]/text()',incoming_xml))[1];
1195 IF profile_tmpl IS NOT NULL AND profile_tmpl <> '' AND profile_tmpl_owner IS NOT NULL AND profile_tmpl_owner <> '' THEN
1196 SELECT p.* INTO profile
1197 FROM vandelay.merge_profile p
1198 JOIN actor.org_unit u ON (u.id = p.owner)
1199 WHERE p.name = profile_tmpl
1200 AND u.shortname = profile_tmpl_owner;
1202 IF profile.id IS NOT NULL THEN
1203 add_rule := COALESCE(profile.add_spec,'');
1204 strip_rule := COALESCE(profile.strip_spec,'');
1205 replace_rule := COALESCE(profile.replace_spec,'');
1206 preserve_rule := COALESCE(profile.preserve_spec,'');
1210 add_rule := add_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="a"]/text()',incoming_xml),','),'');
1211 strip_rule := strip_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="d"]/text()',incoming_xml),','),'');
1212 replace_rule := replace_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="r"]/text()',incoming_xml),','),'');
1213 preserve_rule := preserve_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="p"]/text()',incoming_xml),','),'');
1215 output.add_rule := BTRIM(add_rule,',');
1216 output.replace_rule := BTRIM(replace_rule,',');
1217 output.strip_rule := BTRIM(strip_rule,',');
1218 output.preserve_rule := BTRIM(preserve_rule,',');
1222 $_$ LANGUAGE PLPGSQL;
1224 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1226 merge_profile vandelay.merge_profile%ROWTYPE;
1227 dyn_profile vandelay.compile_profile%ROWTYPE;
1237 SELECT b.marc INTO eg_marc
1238 FROM biblio.record_entry b
1242 IF eg_marc IS NULL OR v_marc IS NULL THEN
1243 -- RAISE NOTICE 'no marc for template or bib record';
1247 dyn_profile := vandelay.compile_profile( v_marc );
1249 IF merge_profile_id IS NOT NULL THEN
1250 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1252 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1253 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1254 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1255 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1259 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1260 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1264 IF dyn_profile.replace_rule <> '' THEN
1265 source_marc = v_marc;
1266 target_marc = eg_marc;
1267 replace_rule = dyn_profile.replace_rule;
1269 source_marc = eg_marc;
1270 target_marc = v_marc;
1271 replace_rule = dyn_profile.preserve_rule;
1274 UPDATE biblio.record_entry
1275 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1279 -- RAISE NOTICE 'update of biblio.record_entry failed';
1286 $$ LANGUAGE PLPGSQL;
1288 CREATE OR REPLACE FUNCTION vandelay.merge_record_xml ( target_marc TEXT, template_marc TEXT ) RETURNS TEXT AS $$
1290 dyn_profile vandelay.compile_profile%ROWTYPE;
1298 IF target_marc IS NULL OR template_marc IS NULL THEN
1299 -- RAISE NOTICE 'no marc for target or template record';
1303 dyn_profile := vandelay.compile_profile( template_marc );
1305 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1306 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1310 IF dyn_profile.replace_rule <> '' THEN
1311 trgt_marc = target_marc;
1312 tmpl_marc = template_marc;
1313 replace_rule = dyn_profile.replace_rule;
1315 tmp_marc = target_marc;
1316 trgt_marc = template_marc;
1317 tmpl_marc = tmp_marc;
1318 replace_rule = dyn_profile.preserve_rule;
1321 RETURN vandelay.merge_record_xml( trgt_marc, tmpl_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule );
1324 $$ LANGUAGE PLPGSQL;
1326 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT) RETURNS BOOL AS $$
1327 SELECT vandelay.template_overlay_bib_record( $1, $2, NULL);
1330 CREATE OR REPLACE FUNCTION vandelay.overlay_bib_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1332 merge_profile vandelay.merge_profile%ROWTYPE;
1333 dyn_profile vandelay.compile_profile%ROWTYPE;
1343 SELECT q.marc INTO v_marc
1344 FROM vandelay.queued_record q
1345 JOIN vandelay.bib_match m ON (m.queued_record = q.id AND q.id = import_id)
1348 IF v_marc IS NULL THEN
1349 -- RAISE NOTICE 'no marc for vandelay or bib record';
1353 IF vandelay.template_overlay_bib_record( v_marc, eg_id, merge_profile_id) THEN
1354 UPDATE vandelay.queued_bib_record
1355 SET imported_as = eg_id,
1357 WHERE id = import_id;
1359 editor_string := (oils_xpath('//*[@tag="905"]/*[@code="u"]/text()',v_marc))[1];
1361 IF editor_string IS NOT NULL AND editor_string <> '' THEN
1362 SELECT usr INTO editor_id FROM actor.card WHERE barcode = editor_string;
1364 IF editor_id IS NULL THEN
1365 SELECT id INTO editor_id FROM actor.usr WHERE usrname = editor_string;
1368 IF editor_id IS NOT NULL THEN
1369 UPDATE biblio.record_entry SET editor = editor_id WHERE id = eg_id;
1376 -- RAISE NOTICE 'update of biblio.record_entry failed';
1381 $$ LANGUAGE PLPGSQL;
1383 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
1386 lwm_ratio_value NUMERIC;
1389 lwm_ratio_value := COALESCE(lwm_ratio_value_p, 0.0);
1391 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1394 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1398 SELECT m.eg_record INTO eg_id
1399 FROM vandelay.bib_match m
1400 JOIN vandelay.queued_bib_record qr ON (m.queued_record = qr.id)
1401 JOIN vandelay.bib_queue q ON (qr.queue = q.id)
1402 JOIN biblio.record_entry r ON (r.id = m.eg_record)
1403 WHERE m.queued_record = import_id
1404 AND qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC >= lwm_ratio_value
1405 ORDER BY m.match_score DESC, -- required match score
1406 qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC DESC, -- quality tie breaker
1407 m.id -- when in doubt, use the first match
1410 IF eg_id IS NULL THEN
1411 -- RAISE NOTICE 'incoming record is not of high enough quality';
1415 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1417 $$ LANGUAGE PLPGSQL;
1419 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1420 SELECT vandelay.auto_overlay_bib_record_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1423 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1429 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1432 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1436 SELECT COUNT(*) INTO match_count FROM vandelay.bib_match WHERE queued_record = import_id;
1438 IF match_count <> 1 THEN
1439 -- RAISE NOTICE 'not an exact match';
1443 -- Check that the one match is on the first 901c
1444 SELECT m.eg_record INTO eg_id
1445 FROM vandelay.queued_bib_record q
1446 JOIN vandelay.bib_match m ON (m.queued_record = q.id)
1447 WHERE q.id = import_id
1448 AND m.eg_record = oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]',marc)::BIGINT;
1451 -- RAISE NOTICE 'not a 901c match';
1455 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1457 $$ LANGUAGE PLPGSQL;
1459 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1461 queued_record vandelay.queued_bib_record%ROWTYPE;
1464 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1466 IF vandelay.auto_overlay_bib_record( queued_record.id, merge_profile_id ) THEN
1467 RETURN NEXT queued_record.id;
1475 $$ LANGUAGE PLPGSQL;
1477 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 $$
1479 queued_record vandelay.queued_bib_record%ROWTYPE;
1482 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1484 IF vandelay.auto_overlay_bib_record_with_best( queued_record.id, merge_profile_id, lwm_ratio_value ) THEN
1485 RETURN NEXT queued_record.id;
1493 $$ LANGUAGE PLPGSQL;
1495 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1496 SELECT vandelay.auto_overlay_bib_queue_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1499 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1500 SELECT * FROM vandelay.auto_overlay_bib_queue( $1, NULL );
1503 CREATE OR REPLACE FUNCTION vandelay.ingest_bib_marc ( ) RETURNS TRIGGER AS $$
1509 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1513 FOR adef IN SELECT * FROM vandelay.bib_attr_definition LOOP
1515 SELECT extract_marc_field('vandelay.queued_bib_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_bib_record WHERE id = NEW.id;
1516 IF (value IS NOT NULL AND value <> '') THEN
1517 INSERT INTO vandelay.queued_bib_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1524 $$ LANGUAGE PLPGSQL;
1526 CREATE OR REPLACE FUNCTION vandelay.cleanup_bib_marc ( ) RETURNS TRIGGER AS $$
1528 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1532 DELETE FROM vandelay.queued_bib_record_attr WHERE record = OLD.id;
1533 DELETE FROM vandelay.import_item WHERE record = OLD.id;
1535 IF TG_OP = 'UPDATE' THEN
1540 $$ LANGUAGE PLPGSQL;
1542 CREATE TRIGGER cleanup_bib_trigger
1543 BEFORE UPDATE OR DELETE ON vandelay.queued_bib_record
1544 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_bib_marc();
1546 CREATE TRIGGER ingest_bib_trigger
1547 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1548 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_bib_marc();
1550 CREATE TRIGGER zz_match_bibs_trigger
1551 BEFORE INSERT OR UPDATE ON vandelay.queued_bib_record
1552 FOR EACH ROW EXECUTE PROCEDURE vandelay.match_bib_record();
1555 /* Authority stuff down here */
1556 ---------------------------------------
1557 CREATE TABLE vandelay.authority_attr_definition (
1558 id SERIAL PRIMARY KEY,
1559 code TEXT UNIQUE NOT NULL,
1561 xpath TEXT NOT NULL,
1562 remove TEXT NOT NULL DEFAULT ''
1565 CREATE TYPE vandelay.authority_queue_queue_type AS ENUM ('authority');
1566 CREATE TABLE vandelay.authority_queue (
1567 queue_type vandelay.authority_queue_queue_type NOT NULL DEFAULT 'authority',
1568 CONSTRAINT vand_authority_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
1569 ) INHERITS (vandelay.queue);
1570 ALTER TABLE vandelay.authority_queue ADD PRIMARY KEY (id);
1572 CREATE TABLE vandelay.queued_authority_record (
1573 queue INT NOT NULL REFERENCES vandelay.authority_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1574 imported_as INT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1575 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
1577 ) INHERITS (vandelay.queued_record);
1578 ALTER TABLE vandelay.queued_authority_record ADD PRIMARY KEY (id);
1579 CREATE INDEX queued_authority_record_queue_idx ON vandelay.queued_authority_record (queue);
1581 CREATE TABLE vandelay.queued_authority_record_attr (
1582 id BIGSERIAL PRIMARY KEY,
1583 record BIGINT NOT NULL REFERENCES vandelay.queued_authority_record (id) DEFERRABLE INITIALLY DEFERRED,
1584 field INT NOT NULL REFERENCES vandelay.authority_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
1585 attr_value TEXT NOT NULL
1587 CREATE INDEX queued_authority_record_attr_record_idx ON vandelay.queued_authority_record_attr (record);
1589 CREATE TABLE vandelay.authority_match (
1590 id BIGSERIAL PRIMARY KEY,
1591 queued_record BIGINT REFERENCES vandelay.queued_authority_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1592 eg_record BIGINT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1593 quality INT NOT NULL DEFAULT 0
1596 CREATE OR REPLACE FUNCTION vandelay.ingest_authority_marc ( ) RETURNS TRIGGER AS $$
1602 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1606 FOR adef IN SELECT * FROM vandelay.authority_attr_definition LOOP
1608 SELECT extract_marc_field('vandelay.queued_authority_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_authority_record WHERE id = NEW.id;
1609 IF (value IS NOT NULL AND value <> '') THEN
1610 INSERT INTO vandelay.queued_authority_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1617 $$ LANGUAGE PLPGSQL;
1619 CREATE OR REPLACE FUNCTION vandelay.cleanup_authority_marc ( ) RETURNS TRIGGER AS $$
1621 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1625 DELETE FROM vandelay.queued_authority_record_attr WHERE record = OLD.id;
1626 IF TG_OP = 'UPDATE' THEN
1631 $$ LANGUAGE PLPGSQL;
1633 CREATE TRIGGER cleanup_authority_trigger
1634 BEFORE UPDATE OR DELETE ON vandelay.queued_authority_record
1635 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_authority_marc();
1637 CREATE TRIGGER ingest_authority_trigger
1638 AFTER INSERT OR UPDATE ON vandelay.queued_authority_record
1639 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_authority_marc();
1641 CREATE OR REPLACE FUNCTION vandelay.overlay_authority_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1643 merge_profile vandelay.merge_profile%ROWTYPE;
1644 dyn_profile vandelay.compile_profile%ROWTYPE;
1653 SELECT b.marc INTO eg_marc
1654 FROM authority.record_entry b
1655 JOIN vandelay.authority_match m ON (m.eg_record = b.id AND m.queued_record = import_id)
1658 SELECT q.marc INTO v_marc
1659 FROM vandelay.queued_record q
1660 JOIN vandelay.authority_match m ON (m.queued_record = q.id AND q.id = import_id)
1663 IF eg_marc IS NULL OR v_marc IS NULL THEN
1664 -- RAISE NOTICE 'no marc for vandelay or authority record';
1668 dyn_profile := vandelay.compile_profile( v_marc );
1670 IF merge_profile_id IS NOT NULL THEN
1671 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1673 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1674 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1675 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1676 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1680 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1681 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1685 IF dyn_profile.replace_rule <> '' THEN
1686 source_marc = v_marc;
1687 target_marc = eg_marc;
1688 replace_rule = dyn_profile.replace_rule;
1690 source_marc = eg_marc;
1691 target_marc = v_marc;
1692 replace_rule = dyn_profile.preserve_rule;
1695 UPDATE authority.record_entry
1696 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1700 UPDATE vandelay.queued_authority_record
1701 SET imported_as = eg_id,
1703 WHERE id = import_id;
1707 -- RAISE NOTICE 'update of authority.record_entry failed';
1712 $$ LANGUAGE PLPGSQL;
1714 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1719 SELECT COUNT(*) INTO match_count FROM vandelay.authority_match WHERE queued_record = import_id;
1721 IF match_count <> 1 THEN
1722 -- RAISE NOTICE 'not an exact match';
1726 SELECT m.eg_record INTO eg_id
1727 FROM vandelay.authority_match m
1728 WHERE m.queued_record = import_id
1731 IF eg_id IS NULL THEN
1735 RETURN vandelay.overlay_authority_record( import_id, eg_id, merge_profile_id );
1737 $$ LANGUAGE PLPGSQL;
1739 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1741 queued_record vandelay.queued_authority_record%ROWTYPE;
1744 FOR queued_record IN SELECT * FROM vandelay.queued_authority_record WHERE queue = queue_id AND import_time IS NULL LOOP
1746 IF vandelay.auto_overlay_authority_record( queued_record.id, merge_profile_id ) THEN
1747 RETURN NEXT queued_record.id;
1755 $$ LANGUAGE PLPGSQL;
1757 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1758 SELECT * FROM vandelay.auto_overlay_authority_queue( $1, NULL );
1762 -- Vandelay (for importing and exporting records) 012.schema.vandelay.sql
1763 --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)]');
1764 --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)]');
1765 --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]');
1766 --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]');
1767 --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$);
1768 --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$);
1769 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (7, 'price',oils_i18n_gettext(7, 'vqbrad', 'Price', 'description'),'//*[@tag="020" or @tag="022"]/*[@code="c"][1]');
1770 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (8, 'rec_identifier',oils_i18n_gettext(8, 'vqbrad', 'Accession Number', 'description'),'//*[@tag="001"]', TRUE);
1771 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (9, 'eg_tcn',oils_i18n_gettext(9, 'vqbrad', 'TCN Value', 'description'),'//*[@tag="901"]/*[@code="a"]', TRUE);
1772 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (10, 'eg_tcn_source',oils_i18n_gettext(10, 'vqbrad', 'TCN Source', 'description'),'//*[@tag="901"]/*[@code="b"]', TRUE);
1773 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (11, 'eg_identifier',oils_i18n_gettext(11, 'vqbrad', 'Internal ID', 'description'),'//*[@tag="901"]/*[@code="c"]', TRUE);
1774 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (12, 'publisher',oils_i18n_gettext(12, 'vqbrad', 'Publisher', 'description'),'//*[@tag="260"]/*[@code="b"][1]');
1775 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, remove ) VALUES (13, 'pubdate',oils_i18n_gettext(13, 'vqbrad', 'Publication Date', 'description'),'//*[@tag="260"]/*[@code="c"][1]',$r$\D$r$);
1776 --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]');
1778 --INSERT INTO vandelay.import_item_attr_definition (
1779 -- owner, name, tag, owning_lib, circ_lib, location,
1780 -- call_number, circ_modifier, barcode, price, copy_number,
1781 -- circulate, ref, holdable, opac_visible, status
1784 -- 'Evergreen 852 export format',
1786 -- '[@code = "b"][1]',
1787 -- '[@code = "b"][2]',
1794 -- '[@code = "x" and text() = "circulating"]',
1795 -- '[@code = "x" and text() = "reference"]',
1796 -- '[@code = "x" and text() = "holdable"]',
1797 -- '[@code = "x" and text() = "visible"]',
1801 --INSERT INTO vandelay.import_item_attr_definition (
1814 -- 'Unicorn Import format -- 999',
1825 --INSERT INTO vandelay.authority_attr_definition ( code, description, xpath, ident ) VALUES ('rec_identifier','Identifier','//*[@tag="001"]', TRUE);