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 $$
503 tags_rstore := vandelay.flatten_marc_hstore(record_xml);
504 svf_rstore := vandelay.extract_rec_attrs(record_xml);
506 CREATE TEMPORARY TABLE _vandelay_tmp_qrows (q INTEGER);
507 CREATE TEMPORARY TABLE _vandelay_tmp_jrows (j TEXT);
509 -- generate the where clause and return that directly (into wq), and as
510 -- a side-effect, populate the _vandelay_tmp_[qj]rows tables.
511 wq := vandelay.get_expr_from_match_set(match_set_id, tags_rstore);
513 query_ := 'SELECT DISTINCT(record), ';
515 -- qrows table is for the quality bits we add to the SELECT clause
516 SELECT ARRAY_TO_STRING(
517 ARRAY_ACCUM('COALESCE(n' || q::TEXT || '.quality, 0)'), ' + '
518 ) INTO coal FROM _vandelay_tmp_qrows;
520 -- our query string so far is the SELECT clause and the inital FROM.
521 -- no JOINs yet nor the WHERE clause
522 query_ := query_ || coal || ' AS quality ' || E'\n';
524 -- jrows table is for the joins we must make (and the real text conditions)
525 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(j), E'\n') INTO joins
526 FROM _vandelay_tmp_jrows;
528 -- add those joins and the where clause to our query.
529 query_ := query_ || joins || E'\n' || 'JOIN biblio.record_entry bre ON (bre.id = record) ' || 'WHERE ' || wq || ' AND not bre.deleted';
531 -- this will return rows of record,quality
532 FOR rec IN EXECUTE query_ USING tags_rstore, svf_rstore LOOP
536 DROP TABLE _vandelay_tmp_qrows;
537 DROP TABLE _vandelay_tmp_jrows;
543 CREATE OR REPLACE FUNCTION vandelay.flatten_marc_hstore(
545 ) RETURNS HSTORE AS $func$
549 ARRAY_ACCUM(tag || (COALESCE(subfield, ''))),
553 SELECT tag, subfield, ARRAY_ACCUM(value)::TEXT AS value
556 CASE WHEN tag = '020' THEN -- caseless -- isbn
557 LOWER((REGEXP_MATCHES(value,$$^(\S{10,17})$$))[1] || '%')
558 WHEN tag = '022' THEN -- caseless -- issn
559 LOWER((REGEXP_MATCHES(value,$$^(\S{4}[- ]?\S{4})$$))[1] || '%')
560 WHEN tag = '024' THEN -- caseless -- upc (other)
565 FROM vandelay.flatten_marc(record_xml)) x
566 GROUP BY tag, subfield ORDER BY tag, subfield
570 $func$ LANGUAGE PLPGSQL;
572 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set(
573 match_set_id INTEGER,
577 root vandelay.match_set_point;
579 SELECT * INTO root FROM vandelay.match_set_point
580 WHERE parent IS NULL AND match_set = match_set_id;
582 RETURN vandelay.get_expr_from_match_set_point(root, tags_rstore);
586 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set_point(
587 node vandelay.match_set_point,
595 child vandelay.match_set_point;
597 SELECT ARRAY_ACCUM(id) INTO children FROM vandelay.match_set_point
598 WHERE parent = node.id;
600 IF ARRAY_LENGTH(children, 1) > 0 THEN
601 this_op := vandelay._get_expr_render_one(node);
604 WHILE children[i] IS NOT NULL LOOP
605 SELECT * INTO child FROM vandelay.match_set_point
606 WHERE id = children[i];
608 q := q || ' ' || this_op || ' ';
611 q := q || vandelay.get_expr_from_match_set_point(child, tags_rstore);
615 ELSIF node.bool_op IS NULL THEN
616 PERFORM vandelay._get_expr_push_qrow(node);
617 PERFORM vandelay._get_expr_push_jrow(node, tags_rstore);
618 RETURN vandelay._get_expr_render_one(node);
625 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_qrow(
626 node vandelay.match_set_point
630 INSERT INTO _vandelay_tmp_qrows (q) VALUES (node.id);
634 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_jrow(
635 node vandelay.match_set_point,
648 -- remember $1 is tags_rstore, and $2 is svf_rstore
651 SELECT COUNT(*) INTO jrow_count FROM _vandelay_tmp_jrows;
652 IF jrow_count > 0 THEN
653 my_using := ' USING (record)';
654 my_join := 'FULL OUTER JOIN';
660 IF node.tag IS NOT NULL THEN
661 caseless := (node.tag IN ('020', '022', '024'));
663 IF node.subfield IS NOT NULL THEN
664 tagkey := tagkey || node.subfield;
682 my_alias := 'n' || node.id::TEXT;
684 jrow := my_join || ' (SELECT *, ';
685 IF node.tag IS NOT NULL THEN
686 jrow := jrow || node.quality ||
687 ' AS quality FROM metabib.full_rec mfr WHERE mfr.tag = ''' ||
689 IF node.subfield IS NOT NULL THEN
690 jrow := jrow || ' AND mfr.subfield = ''' ||
691 node.subfield || '''';
693 jrow := jrow || ' AND (';
694 jrow := jrow || vandelay._node_tag_comparisons(caseless, op, tags_rstore, tagkey);
695 jrow := jrow || ')) ' || my_alias || my_using || E'\n';
697 jrow := jrow || 'id AS record, ' || node.quality ||
698 ' AS quality FROM metabib.record_attr mra WHERE mra.attrs->''' ||
699 node.svf || ''' ' || op || ' $2->''' || node.svf || ''') ' ||
700 my_alias || my_using || E'\n';
702 INSERT INTO _vandelay_tmp_jrows (j) VALUES (jrow);
706 CREATE OR REPLACE FUNCTION vandelay._node_tag_comparisons(
718 vals := tags_rstore->tagkey;
723 IF vals[i] IS NULL THEN
726 result := result || ' OR ';
731 result := result || 'LOWER(mfr.value) ' || op;
733 result := result || 'mfr.value ' || op;
736 result := result || ' ' || COALESCE('''' || vals[i] || '''', 'NULL');
738 IF vals[i] IS NULL THEN
749 CREATE OR REPLACE FUNCTION vandelay._get_expr_render_one(
750 node vandelay.match_set_point
755 IF node.bool_op IS NOT NULL THEN
758 RETURN '(n' || node.id::TEXT || '.id IS NOT NULL)';
763 CREATE OR REPLACE FUNCTION vandelay.match_bib_record() RETURNS TRIGGER AS $func$
765 incoming_existing_id TEXT;
766 test_result vandelay.match_set_test_result%ROWTYPE;
770 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
774 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
776 SELECT q.match_set INTO match_set FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
778 IF match_set IS NOT NULL THEN
779 NEW.quality := vandelay.measure_record_quality( NEW.marc, match_set );
782 -- Perfect matches on 901$c exit early with a match with high quality.
783 incoming_existing_id :=
784 oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]', NEW.marc);
786 IF incoming_existing_id IS NOT NULL AND incoming_existing_id != '' THEN
787 SELECT id INTO tmp_rec FROM biblio.record_entry WHERE id = incoming_existing_id::bigint;
788 IF tmp_rec IS NOT NULL THEN
789 INSERT INTO vandelay.bib_match (queued_record, eg_record, match_score, quality)
794 -- note: no match_set means quality==0
795 vandelay.measure_record_quality( b.marc, match_set )
796 FROM biblio.record_entry b
797 WHERE id = incoming_existing_id::bigint;
801 IF match_set IS NULL THEN
805 FOR test_result IN SELECT * FROM
806 vandelay.match_set_test_marcxml(match_set, NEW.marc) LOOP
808 INSERT INTO vandelay.bib_match ( queued_record, eg_record, match_score, quality )
813 vandelay.measure_record_quality( b.marc, match_set )
814 FROM biblio.record_entry b
815 WHERE id = test_result.record;
821 $func$ LANGUAGE PLPGSQL;
823 CREATE OR REPLACE FUNCTION vandelay.measure_record_quality ( xml TEXT, match_set_id INT ) RETURNS INT AS $_$
827 test vandelay.match_set_quality%ROWTYPE;
830 FOR test IN SELECT * FROM vandelay.match_set_quality WHERE match_set = match_set_id LOOP
831 IF test.tag IS NOT NULL THEN
832 FOR rvalue IN SELECT value FROM vandelay.flatten_marc( xml ) WHERE tag = test.tag AND subfield = test.subfield LOOP
833 IF test.value = rvalue THEN
834 out_q := out_q + test.quality;
838 IF test.value = vandelay.extract_rec_attrs(xml, ARRAY[test.svf]) -> test.svf THEN
839 out_q := out_q + test.quality;
846 $_$ LANGUAGE PLPGSQL;
848 CREATE TYPE vandelay.tcn_data AS (tcn TEXT, tcn_source TEXT, used BOOL);
849 CREATE OR REPLACE FUNCTION vandelay.find_bib_tcn_data ( xml TEXT ) RETURNS SETOF vandelay.tcn_data AS $_$
853 output vandelay.tcn_data%ROWTYPE;
857 eg_tcn := BTRIM((oils_xpath('//*[@tag="001"]/text()',xml))[1]);
858 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
860 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="003"]/text()',xml))[1]);
861 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
862 eg_tcn_source := 'System Local';
865 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
868 output.used := FALSE;
873 output.tcn := eg_tcn;
874 output.tcn_source := eg_tcn_source;
880 eg_tcn := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="a"]/text()',xml))[1]);
881 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
883 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="b"]/text()',xml))[1]);
884 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
885 eg_tcn_source := 'System Local';
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 := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="a"]/text()',xml))[1]);
904 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
906 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="b"]/text()',xml))[1]);
907 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
908 eg_tcn_source := 'System Local';
911 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
914 output.used := FALSE;
919 output.tcn := eg_tcn;
920 output.tcn_source := eg_tcn_source;
926 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="020"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
927 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
929 eg_tcn_source := 'ISBN';
931 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
934 output.used := FALSE;
939 output.tcn := eg_tcn;
940 output.tcn_source := eg_tcn_source;
946 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="022"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
947 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
949 eg_tcn_source := 'ISSN';
951 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
954 output.used := FALSE;
959 output.tcn := eg_tcn;
960 output.tcn_source := eg_tcn_source;
966 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="010"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
967 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
969 eg_tcn_source := 'LCCN';
971 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
974 output.used := FALSE;
979 output.tcn := eg_tcn;
980 output.tcn_source := eg_tcn_source;
986 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="035"]/*[@code="a"]/text()',xml))[1], $re$^.*?(\w+)$$re$, $re$\1$re$);
987 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
989 eg_tcn_source := 'System Legacy';
991 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
994 output.used := FALSE;
999 output.tcn := eg_tcn;
1000 output.tcn_source := eg_tcn_source;
1007 $_$ LANGUAGE PLPGSQL;
1009 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT, force_add INT ) RETURNS TEXT AS $_$
1012 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1016 MARC::Charset->assume_unicode(1);
1018 my $target_xml = shift;
1019 my $source_xml = shift;
1020 my $field_spec = shift;
1021 my $force_add = shift || 0;
1023 my $target_r = MARC::Record->new_from_xml( $target_xml );
1024 my $source_r = MARC::Record->new_from_xml( $source_xml );
1026 return $target_xml unless ($target_r && $source_r);
1028 my @field_list = split(',', $field_spec);
1031 for my $f (@field_list) {
1032 $f =~ s/^\s*//; $f =~ s/\s*$//;
1033 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1039 $match =~ s/^\s*//; $match =~ s/\s*$//;
1040 $fields{$field} = { sf => [ split('', $sf) ] };
1042 my ($msf,$mre) = split('~', $match);
1043 if (length($msf) > 0 and length($mre) > 0) {
1044 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1045 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1046 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1052 for my $f ( keys %fields) {
1053 if ( @{$fields{$f}{sf}} ) {
1054 for my $from_field ($source_r->field( $f )) {
1055 my @tos = $target_r->field( $f );
1057 next if (exists($fields{$f}{match}) and !$force_add);
1058 my @new_fields = map { $_->clone } $source_r->field( $f );
1059 $target_r->insert_fields_ordered( @new_fields );
1061 for my $to_field (@tos) {
1062 if (exists($fields{$f}{match})) {
1063 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1065 my @new_sf = map { ($_ => $from_field->subfield($_)) } grep { defined($from_field->subfield($_)) } @{$fields{$f}{sf}};
1066 $to_field->add_subfields( @new_sf );
1071 my @new_fields = map { $_->clone } $source_r->field( $f );
1072 $target_r->insert_fields_ordered( @new_fields );
1076 $target_xml = $target_r->as_xml_record;
1077 $target_xml =~ s/^<\?.+?\?>$//mo;
1078 $target_xml =~ s/\n//sgo;
1079 $target_xml =~ s/>\s+</></sgo;
1083 $_$ LANGUAGE PLPERLU;
1085 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1086 SELECT vandelay.add_field( $1, $2, $3, 0 );
1089 CREATE OR REPLACE FUNCTION vandelay.strip_field ( xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1092 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1096 MARC::Charset->assume_unicode(1);
1099 my $r = MARC::Record->new_from_xml( $xml );
1101 return $xml unless ($r);
1103 my $field_spec = shift;
1104 my @field_list = split(',', $field_spec);
1107 for my $f (@field_list) {
1108 $f =~ s/^\s*//; $f =~ s/\s*$//;
1109 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1115 $match =~ s/^\s*//; $match =~ s/\s*$//;
1116 $fields{$field} = { sf => [ split('', $sf) ] };
1118 my ($msf,$mre) = split('~', $match);
1119 if (length($msf) > 0 and length($mre) > 0) {
1120 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1121 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1122 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1128 for my $f ( keys %fields) {
1129 for my $to_field ($r->field( $f )) {
1130 if (exists($fields{$f}{match})) {
1131 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1134 if ( @{$fields{$f}{sf}} ) {
1135 $to_field->delete_subfield(code => $fields{$f}{sf});
1137 $r->delete_field( $to_field );
1142 $xml = $r->as_xml_record;
1143 $xml =~ s/^<\?.+?\?>$//mo;
1145 $xml =~ s/>\s+</></sgo;
1149 $_$ LANGUAGE PLPERLU;
1151 CREATE OR REPLACE FUNCTION vandelay.replace_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1158 parsed_target := vandelay.strip_field( target_xml, ''); -- this dance normalizes the format of the xml for the IF below
1159 xml_output := parsed_target; -- if there are no replace rules, just return the input
1161 FOR curr_field IN SELECT UNNEST( STRING_TO_ARRAY(field, ',') ) LOOP -- naive split, but it's the same we use in the perl
1163 xml_output := vandelay.strip_field( parsed_target, curr_field);
1165 IF xml_output <> parsed_target AND curr_field ~ E'~' THEN
1166 -- we removed something, and there was a regexp restriction in the curr_field definition, so proceed
1167 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 1 );
1168 ELSIF curr_field !~ E'~' THEN
1169 -- No regexp restriction, add the curr_field
1170 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 0 );
1173 parsed_target := xml_output; -- in prep for any following loop iterations
1179 $_$ LANGUAGE PLPGSQL;
1181 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 $_$
1182 SELECT vandelay.replace_field( vandelay.add_field( vandelay.strip_field( $1, $5) , $2, $3 ), $2, $4);
1185 CREATE TYPE vandelay.compile_profile AS (add_rule TEXT, replace_rule TEXT, preserve_rule TEXT, strip_rule TEXT);
1186 CREATE OR REPLACE FUNCTION vandelay.compile_profile ( incoming_xml TEXT ) RETURNS vandelay.compile_profile AS $_$
1188 output vandelay.compile_profile%ROWTYPE;
1189 profile vandelay.merge_profile%ROWTYPE;
1191 profile_tmpl_owner TEXT;
1192 add_rule TEXT := '';
1193 strip_rule TEXT := '';
1194 replace_rule TEXT := '';
1195 preserve_rule TEXT := '';
1199 profile_tmpl := (oils_xpath('//*[@tag="905"]/*[@code="t"]/text()',incoming_xml))[1];
1200 profile_tmpl_owner := (oils_xpath('//*[@tag="905"]/*[@code="o"]/text()',incoming_xml))[1];
1202 IF profile_tmpl IS NOT NULL AND profile_tmpl <> '' AND profile_tmpl_owner IS NOT NULL AND profile_tmpl_owner <> '' THEN
1203 SELECT p.* INTO profile
1204 FROM vandelay.merge_profile p
1205 JOIN actor.org_unit u ON (u.id = p.owner)
1206 WHERE p.name = profile_tmpl
1207 AND u.shortname = profile_tmpl_owner;
1209 IF profile.id IS NOT NULL THEN
1210 add_rule := COALESCE(profile.add_spec,'');
1211 strip_rule := COALESCE(profile.strip_spec,'');
1212 replace_rule := COALESCE(profile.replace_spec,'');
1213 preserve_rule := COALESCE(profile.preserve_spec,'');
1217 add_rule := add_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="a"]/text()',incoming_xml),','),'');
1218 strip_rule := strip_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="d"]/text()',incoming_xml),','),'');
1219 replace_rule := replace_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="r"]/text()',incoming_xml),','),'');
1220 preserve_rule := preserve_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="p"]/text()',incoming_xml),','),'');
1222 output.add_rule := BTRIM(add_rule,',');
1223 output.replace_rule := BTRIM(replace_rule,',');
1224 output.strip_rule := BTRIM(strip_rule,',');
1225 output.preserve_rule := BTRIM(preserve_rule,',');
1229 $_$ LANGUAGE PLPGSQL;
1231 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1233 merge_profile vandelay.merge_profile%ROWTYPE;
1234 dyn_profile vandelay.compile_profile%ROWTYPE;
1244 SELECT b.marc INTO eg_marc
1245 FROM biblio.record_entry b
1249 IF eg_marc IS NULL OR v_marc IS NULL THEN
1250 -- RAISE NOTICE 'no marc for template or bib record';
1254 dyn_profile := vandelay.compile_profile( v_marc );
1256 IF merge_profile_id IS NOT NULL THEN
1257 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1259 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1260 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1261 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1262 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1266 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1267 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1271 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1272 --Since we have nothing to do, just return a NOOP "we did it"
1274 ELSIF dyn_profile.replace_rule <> '' THEN
1275 source_marc = v_marc;
1276 target_marc = eg_marc;
1277 replace_rule = dyn_profile.replace_rule;
1279 source_marc = eg_marc;
1280 target_marc = v_marc;
1281 replace_rule = dyn_profile.preserve_rule;
1284 UPDATE biblio.record_entry
1285 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1289 -- RAISE NOTICE 'update of biblio.record_entry failed';
1296 $$ LANGUAGE PLPGSQL;
1298 CREATE OR REPLACE FUNCTION vandelay.merge_record_xml ( target_marc TEXT, template_marc TEXT ) RETURNS TEXT AS $$
1300 dyn_profile vandelay.compile_profile%ROWTYPE;
1308 IF target_marc IS NULL OR template_marc IS NULL THEN
1309 -- RAISE NOTICE 'no marc for target or template record';
1313 dyn_profile := vandelay.compile_profile( template_marc );
1315 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1316 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1320 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1321 --Since we have nothing to do, just return what we were given.
1323 ELSIF dyn_profile.replace_rule <> '' THEN
1324 trgt_marc = target_marc;
1325 tmpl_marc = template_marc;
1326 replace_rule = dyn_profile.replace_rule;
1328 tmp_marc = target_marc;
1329 trgt_marc = template_marc;
1330 tmpl_marc = tmp_marc;
1331 replace_rule = dyn_profile.preserve_rule;
1334 RETURN vandelay.merge_record_xml( trgt_marc, tmpl_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule );
1337 $$ LANGUAGE PLPGSQL;
1339 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT) RETURNS BOOL AS $$
1340 SELECT vandelay.template_overlay_bib_record( $1, $2, NULL);
1343 CREATE OR REPLACE FUNCTION vandelay.overlay_bib_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1345 merge_profile vandelay.merge_profile%ROWTYPE;
1346 dyn_profile vandelay.compile_profile%ROWTYPE;
1356 SELECT q.marc INTO v_marc
1357 FROM vandelay.queued_record q
1358 JOIN vandelay.bib_match m ON (m.queued_record = q.id AND q.id = import_id)
1361 IF v_marc IS NULL THEN
1362 -- RAISE NOTICE 'no marc for vandelay or bib record';
1366 IF vandelay.template_overlay_bib_record( v_marc, eg_id, merge_profile_id) THEN
1367 UPDATE vandelay.queued_bib_record
1368 SET imported_as = eg_id,
1370 WHERE id = import_id;
1372 editor_string := (oils_xpath('//*[@tag="905"]/*[@code="u"]/text()',v_marc))[1];
1374 IF editor_string IS NOT NULL AND editor_string <> '' THEN
1375 SELECT usr INTO editor_id FROM actor.card WHERE barcode = editor_string;
1377 IF editor_id IS NULL THEN
1378 SELECT id INTO editor_id FROM actor.usr WHERE usrname = editor_string;
1381 IF editor_id IS NOT NULL THEN
1382 UPDATE biblio.record_entry SET editor = editor_id WHERE id = eg_id;
1389 -- RAISE NOTICE 'update of biblio.record_entry failed';
1394 $$ LANGUAGE PLPGSQL;
1396 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
1399 lwm_ratio_value NUMERIC;
1402 lwm_ratio_value := COALESCE(lwm_ratio_value_p, 0.0);
1404 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1407 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1411 SELECT m.eg_record INTO eg_id
1412 FROM vandelay.bib_match m
1413 JOIN vandelay.queued_bib_record qr ON (m.queued_record = qr.id)
1414 JOIN vandelay.bib_queue q ON (qr.queue = q.id)
1415 JOIN biblio.record_entry r ON (r.id = m.eg_record)
1416 WHERE m.queued_record = import_id
1417 AND qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC >= lwm_ratio_value
1418 ORDER BY m.match_score DESC, -- required match score
1419 qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC DESC, -- quality tie breaker
1420 m.id -- when in doubt, use the first match
1423 IF eg_id IS NULL THEN
1424 -- RAISE NOTICE 'incoming record is not of high enough quality';
1428 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1430 $$ LANGUAGE PLPGSQL;
1432 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1433 SELECT vandelay.auto_overlay_bib_record_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1436 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1442 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1445 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1449 SELECT COUNT(*) INTO match_count FROM vandelay.bib_match WHERE queued_record = import_id;
1451 IF match_count <> 1 THEN
1452 -- RAISE NOTICE 'not an exact match';
1456 -- Check that the one match is on the first 901c
1457 SELECT m.eg_record INTO eg_id
1458 FROM vandelay.queued_bib_record q
1459 JOIN vandelay.bib_match m ON (m.queued_record = q.id)
1460 WHERE q.id = import_id
1461 AND m.eg_record = oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]',marc)::BIGINT;
1464 -- RAISE NOTICE 'not a 901c match';
1468 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1470 $$ LANGUAGE PLPGSQL;
1472 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1474 queued_record vandelay.queued_bib_record%ROWTYPE;
1477 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1479 IF vandelay.auto_overlay_bib_record( queued_record.id, merge_profile_id ) THEN
1480 RETURN NEXT queued_record.id;
1488 $$ LANGUAGE PLPGSQL;
1490 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 $$
1492 queued_record vandelay.queued_bib_record%ROWTYPE;
1495 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1497 IF vandelay.auto_overlay_bib_record_with_best( queued_record.id, merge_profile_id, lwm_ratio_value ) THEN
1498 RETURN NEXT queued_record.id;
1506 $$ LANGUAGE PLPGSQL;
1508 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1509 SELECT vandelay.auto_overlay_bib_queue_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1512 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1513 SELECT * FROM vandelay.auto_overlay_bib_queue( $1, NULL );
1516 CREATE OR REPLACE FUNCTION vandelay.ingest_bib_marc ( ) RETURNS TRIGGER AS $$
1522 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1526 FOR adef IN SELECT * FROM vandelay.bib_attr_definition LOOP
1528 SELECT extract_marc_field('vandelay.queued_bib_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_bib_record WHERE id = NEW.id;
1529 IF (value IS NOT NULL AND value <> '') THEN
1530 INSERT INTO vandelay.queued_bib_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1537 $$ LANGUAGE PLPGSQL;
1539 CREATE OR REPLACE FUNCTION vandelay.cleanup_bib_marc ( ) RETURNS TRIGGER AS $$
1541 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1545 DELETE FROM vandelay.queued_bib_record_attr WHERE record = OLD.id;
1546 DELETE FROM vandelay.import_item WHERE record = OLD.id;
1548 IF TG_OP = 'UPDATE' THEN
1553 $$ LANGUAGE PLPGSQL;
1555 CREATE TRIGGER cleanup_bib_trigger
1556 BEFORE UPDATE OR DELETE ON vandelay.queued_bib_record
1557 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_bib_marc();
1559 CREATE TRIGGER ingest_bib_trigger
1560 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1561 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_bib_marc();
1563 CREATE TRIGGER zz_match_bibs_trigger
1564 BEFORE INSERT OR UPDATE ON vandelay.queued_bib_record
1565 FOR EACH ROW EXECUTE PROCEDURE vandelay.match_bib_record();
1568 /* Authority stuff down here */
1569 ---------------------------------------
1570 CREATE TABLE vandelay.authority_attr_definition (
1571 id SERIAL PRIMARY KEY,
1572 code TEXT UNIQUE NOT NULL,
1574 xpath TEXT NOT NULL,
1575 remove TEXT NOT NULL DEFAULT ''
1578 CREATE TYPE vandelay.authority_queue_queue_type AS ENUM ('authority');
1579 CREATE TABLE vandelay.authority_queue (
1580 queue_type vandelay.authority_queue_queue_type NOT NULL DEFAULT 'authority',
1581 CONSTRAINT vand_authority_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
1582 ) INHERITS (vandelay.queue);
1583 ALTER TABLE vandelay.authority_queue ADD PRIMARY KEY (id);
1585 CREATE TABLE vandelay.queued_authority_record (
1586 queue INT NOT NULL REFERENCES vandelay.authority_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1587 imported_as INT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1588 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
1590 ) INHERITS (vandelay.queued_record);
1591 ALTER TABLE vandelay.queued_authority_record ADD PRIMARY KEY (id);
1592 CREATE INDEX queued_authority_record_queue_idx ON vandelay.queued_authority_record (queue);
1594 CREATE TABLE vandelay.queued_authority_record_attr (
1595 id BIGSERIAL PRIMARY KEY,
1596 record BIGINT NOT NULL REFERENCES vandelay.queued_authority_record (id) DEFERRABLE INITIALLY DEFERRED,
1597 field INT NOT NULL REFERENCES vandelay.authority_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
1598 attr_value TEXT NOT NULL
1600 CREATE INDEX queued_authority_record_attr_record_idx ON vandelay.queued_authority_record_attr (record);
1602 CREATE TABLE vandelay.authority_match (
1603 id BIGSERIAL PRIMARY KEY,
1604 queued_record BIGINT REFERENCES vandelay.queued_authority_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1605 eg_record BIGINT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1606 quality INT NOT NULL DEFAULT 0
1609 CREATE OR REPLACE FUNCTION vandelay.ingest_authority_marc ( ) RETURNS TRIGGER AS $$
1615 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1619 FOR adef IN SELECT * FROM vandelay.authority_attr_definition LOOP
1621 SELECT extract_marc_field('vandelay.queued_authority_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_authority_record WHERE id = NEW.id;
1622 IF (value IS NOT NULL AND value <> '') THEN
1623 INSERT INTO vandelay.queued_authority_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1630 $$ LANGUAGE PLPGSQL;
1632 CREATE OR REPLACE FUNCTION vandelay.cleanup_authority_marc ( ) RETURNS TRIGGER AS $$
1634 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1638 DELETE FROM vandelay.queued_authority_record_attr WHERE record = OLD.id;
1639 IF TG_OP = 'UPDATE' THEN
1644 $$ LANGUAGE PLPGSQL;
1646 CREATE TRIGGER cleanup_authority_trigger
1647 BEFORE UPDATE OR DELETE ON vandelay.queued_authority_record
1648 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_authority_marc();
1650 CREATE TRIGGER ingest_authority_trigger
1651 AFTER INSERT OR UPDATE ON vandelay.queued_authority_record
1652 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_authority_marc();
1654 CREATE OR REPLACE FUNCTION vandelay.overlay_authority_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1656 merge_profile vandelay.merge_profile%ROWTYPE;
1657 dyn_profile vandelay.compile_profile%ROWTYPE;
1666 SELECT b.marc INTO eg_marc
1667 FROM authority.record_entry b
1668 JOIN vandelay.authority_match m ON (m.eg_record = b.id AND m.queued_record = import_id)
1671 SELECT q.marc INTO v_marc
1672 FROM vandelay.queued_record q
1673 JOIN vandelay.authority_match m ON (m.queued_record = q.id AND q.id = import_id)
1676 IF eg_marc IS NULL OR v_marc IS NULL THEN
1677 -- RAISE NOTICE 'no marc for vandelay or authority record';
1681 dyn_profile := vandelay.compile_profile( v_marc );
1683 IF merge_profile_id IS NOT NULL THEN
1684 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1686 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1687 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1688 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1689 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1693 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1694 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1698 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1699 --Since we have nothing to do, just return a NOOP "we did it"
1701 ELSIF dyn_profile.replace_rule <> '' THEN
1702 source_marc = v_marc;
1703 target_marc = eg_marc;
1704 replace_rule = dyn_profile.replace_rule;
1706 source_marc = eg_marc;
1707 target_marc = v_marc;
1708 replace_rule = dyn_profile.preserve_rule;
1711 UPDATE authority.record_entry
1712 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1716 UPDATE vandelay.queued_authority_record
1717 SET imported_as = eg_id,
1719 WHERE id = import_id;
1723 -- RAISE NOTICE 'update of authority.record_entry failed';
1728 $$ LANGUAGE PLPGSQL;
1730 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1735 SELECT COUNT(*) INTO match_count FROM vandelay.authority_match WHERE queued_record = import_id;
1737 IF match_count <> 1 THEN
1738 -- RAISE NOTICE 'not an exact match';
1742 SELECT m.eg_record INTO eg_id
1743 FROM vandelay.authority_match m
1744 WHERE m.queued_record = import_id
1747 IF eg_id IS NULL THEN
1751 RETURN vandelay.overlay_authority_record( import_id, eg_id, merge_profile_id );
1753 $$ LANGUAGE PLPGSQL;
1755 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1757 queued_record vandelay.queued_authority_record%ROWTYPE;
1760 FOR queued_record IN SELECT * FROM vandelay.queued_authority_record WHERE queue = queue_id AND import_time IS NULL LOOP
1762 IF vandelay.auto_overlay_authority_record( queued_record.id, merge_profile_id ) THEN
1763 RETURN NEXT queued_record.id;
1771 $$ LANGUAGE PLPGSQL;
1773 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1774 SELECT * FROM vandelay.auto_overlay_authority_queue( $1, NULL );
1778 -- Vandelay (for importing and exporting records) 012.schema.vandelay.sql
1779 --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)]');
1780 --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)]');
1781 --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]');
1782 --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]');
1783 --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$);
1784 --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$);
1785 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (7, 'price',oils_i18n_gettext(7, 'vqbrad', 'Price', 'description'),'//*[@tag="020" or @tag="022"]/*[@code="c"][1]');
1786 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (8, 'rec_identifier',oils_i18n_gettext(8, 'vqbrad', 'Accession Number', 'description'),'//*[@tag="001"]', TRUE);
1787 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (9, 'eg_tcn',oils_i18n_gettext(9, 'vqbrad', 'TCN Value', 'description'),'//*[@tag="901"]/*[@code="a"]', TRUE);
1788 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (10, 'eg_tcn_source',oils_i18n_gettext(10, 'vqbrad', 'TCN Source', 'description'),'//*[@tag="901"]/*[@code="b"]', TRUE);
1789 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (11, 'eg_identifier',oils_i18n_gettext(11, 'vqbrad', 'Internal ID', 'description'),'//*[@tag="901"]/*[@code="c"]', TRUE);
1790 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (12, 'publisher',oils_i18n_gettext(12, 'vqbrad', 'Publisher', 'description'),'//*[@tag="260"]/*[@code="b"][1]');
1791 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, remove ) VALUES (13, 'pubdate',oils_i18n_gettext(13, 'vqbrad', 'Publication Date', 'description'),'//*[@tag="260"]/*[@code="c"][1]',$r$\D$r$);
1792 --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]');
1794 --INSERT INTO vandelay.import_item_attr_definition (
1795 -- owner, name, tag, owning_lib, circ_lib, location,
1796 -- call_number, circ_modifier, barcode, price, copy_number,
1797 -- circulate, ref, holdable, opac_visible, status
1800 -- 'Evergreen 852 export format',
1802 -- '[@code = "b"][1]',
1803 -- '[@code = "b"][2]',
1810 -- '[@code = "x" and text() = "circulating"]',
1811 -- '[@code = "x" and text() = "reference"]',
1812 -- '[@code = "x" and text() = "holdable"]',
1813 -- '[@code = "x" and text() = "visible"]',
1817 --INSERT INTO vandelay.import_item_attr_definition (
1830 -- 'Unicorn Import format -- 999',
1841 --INSERT INTO vandelay.authority_attr_definition ( code, description, xpath, ident ) VALUES ('rec_identifier','Identifier','//*[@tag="001"]', TRUE);