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_group(
182 id SERIAL PRIMARY KEY,
183 owner INTEGER NOT NULL REFERENCES actor.org_unit(id),
184 label TEXT NOT NULL, --i18n
185 always_apply BOOLEAN NOT NULL DEFAULT FALSE,
186 CONSTRAINT vand_import_bib_trash_grp_owner_label UNIQUE (owner, label)
189 CREATE TABLE vandelay.import_bib_trash_fields (
190 id BIGSERIAL PRIMARY KEY,
191 grp INTEGER NOT NULL REFERENCES vandelay.import_bib_trash_group,
193 CONSTRAINT vand_import_bib_trash_fields_once_per UNIQUE (grp, field)
196 CREATE TABLE vandelay.merge_profile (
197 id BIGSERIAL PRIMARY KEY,
198 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
205 CONSTRAINT vand_merge_prof_owner_name_idx UNIQUE (owner,name),
206 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))
209 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
216 retval config.marc21_rec_type_map%ROWTYPE;
218 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
220 IF ldr IS NULL OR ldr = '' THEN
221 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
225 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
226 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
229 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
230 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
232 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
234 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
237 IF retval.code IS NULL THEN
238 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
243 $func$ LANGUAGE PLPGSQL;
245 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
252 rtype := (vandelay.marc21_record_type( marc )).code;
253 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
254 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
255 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
258 val := REPEAT( ff_pos.default_val, ff_pos.length );
264 $func$ LANGUAGE PLPGSQL;
266 CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
267 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
272 output biblio.record_ff_map%ROWTYPE;
274 rtype := (vandelay.marc21_record_type( marc )).code;
276 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
277 output.ff_name := ff_pos.fixed_field;
278 output.ff_value := NULL;
280 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(tag) || '"]/text()', marc ) ) x(value) LOOP
281 output.ff_value := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
282 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
284 output.ff_value := NULL;
291 $func$ LANGUAGE PLPGSQL;
293 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
294 CREATE OR REPLACE FUNCTION vandelay.marc21_physical_characteristics( marc TEXT) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
298 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
299 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
300 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
301 retval biblio.marc21_physical_characteristics%ROWTYPE;
304 _007 := oils_xpath_string( '//*[@tag="007"]', marc );
306 IF _007 IS NOT NULL AND _007 <> '' THEN
307 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007, 1, 1 );
309 IF ptype.ptype_key IS NOT NULL THEN
310 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
311 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 );
313 IF pval.id IS NOT NULL THEN
316 retval.ptype := ptype.ptype_key;
317 retval.subfield := psf.id;
318 retval.value := pval.id;
328 $func$ LANGUAGE PLPGSQL;
330 CREATE TYPE vandelay.flat_marc AS ( tag CHAR(3), ind1 TEXT, ind2 TEXT, subfield TEXT, value TEXT );
331 CREATE OR REPLACE FUNCTION vandelay.flay_marc ( TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
334 use MARC::File::XML (BinaryEncoding => 'UTF-8');
338 MARC::Charset->assume_unicode(1);
341 my $r = MARC::Record->new_from_xml( $xml );
343 return_next( { tag => 'LDR', value => $r->leader } );
345 for my $f ( $r->fields ) {
346 if ($f->is_control_field) {
347 return_next({ tag => $f->tag, value => $f->data });
349 for my $s ($f->subfields) {
352 ind1 => $f->indicator(1),
353 ind2 => $f->indicator(2),
358 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
359 my $trim = $f->indicator(2) || 0;
362 ind1 => $f->indicator(1),
363 ind2 => $f->indicator(2),
365 value => substr( $s->[1], $trim )
374 $func$ LANGUAGE PLPERLU;
376 CREATE OR REPLACE FUNCTION vandelay.flatten_marc ( marc TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
378 output vandelay.flat_marc%ROWTYPE;
381 FOR field IN SELECT * FROM vandelay.flay_marc( marc ) LOOP
382 output.ind1 := field.ind1;
383 output.ind2 := field.ind2;
384 output.tag := field.tag;
385 output.subfield := field.subfield;
386 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
387 output.value := naco_normalize(field.value, field.subfield);
389 output.value := field.value;
392 CONTINUE WHEN output.value IS NULL;
397 $func$ LANGUAGE PLPGSQL;
399 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT, attr_defs TEXT[]) RETURNS hstore AS $_$
401 transformed_xml TEXT;
404 xfrm config.xml_transform%ROWTYPE;
406 new_attrs HSTORE := ''::HSTORE;
407 attr_def config.record_attr_definition%ROWTYPE;
410 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE name IN (SELECT * FROM UNNEST(attr_defs)) ORDER BY format LOOP
412 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
413 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(x.value), COALESCE(attr_def.joiner,' ')) INTO attr_value
414 FROM vandelay.flatten_marc(xml) AS x
415 WHERE x.tag LIKE attr_def.tag
417 WHEN attr_def.sf_list IS NOT NULL
418 THEN POSITION(x.subfield IN attr_def.sf_list) > 0
425 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
426 attr_value := vandelay.marc21_extract_fixed_field(xml, attr_def.fixed_field);
428 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
430 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
432 -- See if we can skip the XSLT ... it's expensive
433 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
434 -- Can't skip the transform
435 IF xfrm.xslt <> '---' THEN
436 transformed_xml := oils_xslt_process(xml,xfrm.xslt);
438 transformed_xml := xml;
441 prev_xfrm := xfrm.name;
444 IF xfrm.name IS NULL THEN
445 -- just grab the marcxml (empty) transform
446 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
447 prev_xfrm := xfrm.name;
450 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
452 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
453 SELECT m.value::TEXT INTO attr_value
454 FROM vandelay.marc21_physical_characteristics(xml) v
455 JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
456 WHERE v.subfield = attr_def.phys_char_sf
457 LIMIT 1; -- Just in case ...
461 -- apply index normalizers to attr_value
463 SELECT n.func AS func,
464 n.param_count AS param_count,
466 FROM config.index_normalizer n
467 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
468 WHERE attr = attr_def.name
470 EXECUTE 'SELECT ' || normalizer.func || '(' ||
471 quote_nullable( attr_value ) ||
473 WHEN normalizer.param_count > 0
474 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
481 -- Add the new value to the hstore
482 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
488 $_$ LANGUAGE PLPGSQL;
490 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT ) RETURNS hstore AS $_$
491 SELECT vandelay.extract_rec_attrs( $1, (SELECT ARRAY_ACCUM(name) FROM config.record_attr_definition));
494 -- Everything between this comment and the beginning of the definition of
495 -- vandelay.match_bib_record() is strictly in service of that function.
496 CREATE TYPE vandelay.match_set_test_result AS (record BIGINT, quality INTEGER);
498 CREATE OR REPLACE FUNCTION vandelay.match_set_test_marcxml(
499 match_set_id INTEGER, record_xml TEXT
500 ) RETURNS SETOF vandelay.match_set_test_result AS $$
511 tags_rstore := vandelay.flatten_marc_hstore(record_xml);
512 svf_rstore := vandelay.extract_rec_attrs(record_xml);
514 CREATE TEMPORARY TABLE _vandelay_tmp_qrows (q INTEGER);
515 CREATE TEMPORARY TABLE _vandelay_tmp_jrows (j TEXT);
517 -- generate the where clause and return that directly (into wq), and as
518 -- a side-effect, populate the _vandelay_tmp_[qj]rows tables.
519 wq := vandelay.get_expr_from_match_set(match_set_id, tags_rstore);
521 query_ := 'SELECT DISTINCT(record), ';
523 -- qrows table is for the quality bits we add to the SELECT clause
524 SELECT ARRAY_TO_STRING(
525 ARRAY_ACCUM('COALESCE(n' || q::TEXT || '.quality, 0)'), ' + '
526 ) INTO coal FROM _vandelay_tmp_qrows;
528 -- our query string so far is the SELECT clause and the inital FROM.
529 -- no JOINs yet nor the WHERE clause
530 query_ := query_ || coal || ' AS quality ' || E'\n';
532 -- jrows table is for the joins we must make (and the real text conditions)
533 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(j), E'\n') INTO joins
534 FROM _vandelay_tmp_jrows;
536 -- add those joins and the where clause to our query.
537 query_ := query_ || joins || E'\n' || 'JOIN biblio.record_entry bre ON (bre.id = record) ' || 'WHERE ' || wq || ' AND not bre.deleted';
539 -- this will return rows of record,quality
540 FOR rec IN EXECUTE query_ USING tags_rstore, svf_rstore LOOP
544 DROP TABLE _vandelay_tmp_qrows;
545 DROP TABLE _vandelay_tmp_jrows;
551 CREATE OR REPLACE FUNCTION vandelay.flatten_marc_hstore(
553 ) RETURNS HSTORE AS $func$
557 ARRAY_ACCUM(tag || (COALESCE(subfield, ''))),
561 SELECT tag, subfield, ARRAY_ACCUM(value)::TEXT AS value
564 CASE WHEN tag = '020' THEN -- caseless -- isbn
565 LOWER((REGEXP_MATCHES(value,$$^(\S{10,17})$$))[1] || '%')
566 WHEN tag = '022' THEN -- caseless -- issn
567 LOWER((REGEXP_MATCHES(value,$$^(\S{4}[- ]?\S{4})$$))[1] || '%')
568 WHEN tag = '024' THEN -- caseless -- upc (other)
573 FROM vandelay.flatten_marc(record_xml)) x
574 GROUP BY tag, subfield ORDER BY tag, subfield
578 $func$ LANGUAGE PLPGSQL;
580 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set(
581 match_set_id INTEGER,
585 root vandelay.match_set_point;
587 SELECT * INTO root FROM vandelay.match_set_point
588 WHERE parent IS NULL AND match_set = match_set_id;
590 RETURN vandelay.get_expr_from_match_set_point(root, tags_rstore);
594 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set_point(
595 node vandelay.match_set_point,
603 child vandelay.match_set_point;
605 SELECT ARRAY_ACCUM(id) INTO children FROM vandelay.match_set_point
606 WHERE parent = node.id;
608 IF ARRAY_LENGTH(children, 1) > 0 THEN
609 this_op := vandelay._get_expr_render_one(node);
612 WHILE children[i] IS NOT NULL LOOP
613 SELECT * INTO child FROM vandelay.match_set_point
614 WHERE id = children[i];
616 q := q || ' ' || this_op || ' ';
619 q := q || vandelay.get_expr_from_match_set_point(child, tags_rstore);
623 ELSIF node.bool_op IS NULL THEN
624 PERFORM vandelay._get_expr_push_qrow(node);
625 PERFORM vandelay._get_expr_push_jrow(node, tags_rstore);
626 RETURN vandelay._get_expr_render_one(node);
633 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_qrow(
634 node vandelay.match_set_point
638 INSERT INTO _vandelay_tmp_qrows (q) VALUES (node.id);
642 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_jrow(
643 node vandelay.match_set_point,
656 -- remember $1 is tags_rstore, and $2 is svf_rstore
659 SELECT COUNT(*) INTO jrow_count FROM _vandelay_tmp_jrows;
660 IF jrow_count > 0 THEN
661 my_using := ' USING (record)';
662 my_join := 'FULL OUTER JOIN';
668 IF node.tag IS NOT NULL THEN
669 caseless := (node.tag IN ('020', '022', '024'));
671 IF node.subfield IS NOT NULL THEN
672 tagkey := tagkey || node.subfield;
690 my_alias := 'n' || node.id::TEXT;
692 jrow := my_join || ' (SELECT *, ';
693 IF node.tag IS NOT NULL THEN
694 jrow := jrow || node.quality ||
695 ' AS quality FROM metabib.full_rec mfr WHERE mfr.tag = ''' ||
697 IF node.subfield IS NOT NULL THEN
698 jrow := jrow || ' AND mfr.subfield = ''' ||
699 node.subfield || '''';
701 jrow := jrow || ' AND (';
702 jrow := jrow || vandelay._node_tag_comparisons(caseless, op, tags_rstore, tagkey);
703 jrow := jrow || ')) ' || my_alias || my_using || E'\n';
705 jrow := jrow || 'id AS record, ' || node.quality ||
706 ' AS quality FROM metabib.record_attr mra WHERE mra.attrs->''' ||
707 node.svf || ''' ' || op || ' $2->''' || node.svf || ''') ' ||
708 my_alias || my_using || E'\n';
710 INSERT INTO _vandelay_tmp_jrows (j) VALUES (jrow);
714 CREATE OR REPLACE FUNCTION vandelay._node_tag_comparisons(
726 vals := tags_rstore->tagkey;
731 IF vals[i] IS NULL THEN
734 result := result || ' OR ';
739 result := result || 'LOWER(mfr.value) ' || op;
741 result := result || 'mfr.value ' || op;
744 result := result || ' ' || COALESCE('''' || vals[i] || '''', 'NULL');
746 IF vals[i] IS NULL THEN
757 CREATE OR REPLACE FUNCTION vandelay._get_expr_render_one(
758 node vandelay.match_set_point
763 IF node.bool_op IS NOT NULL THEN
766 RETURN '(n' || node.id::TEXT || '.id IS NOT NULL)';
771 CREATE OR REPLACE FUNCTION vandelay.match_bib_record() RETURNS TRIGGER AS $func$
773 incoming_existing_id TEXT;
774 test_result vandelay.match_set_test_result%ROWTYPE;
778 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
782 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
784 SELECT q.match_set INTO match_set FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
786 IF match_set IS NOT NULL THEN
787 NEW.quality := vandelay.measure_record_quality( NEW.marc, match_set );
790 -- Perfect matches on 901$c exit early with a match with high quality.
791 incoming_existing_id :=
792 oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]', NEW.marc);
794 IF incoming_existing_id IS NOT NULL AND incoming_existing_id != '' THEN
795 SELECT id INTO tmp_rec FROM biblio.record_entry WHERE id = incoming_existing_id::bigint;
796 IF tmp_rec IS NOT NULL THEN
797 INSERT INTO vandelay.bib_match (queued_record, eg_record, match_score, quality)
802 -- note: no match_set means quality==0
803 vandelay.measure_record_quality( b.marc, match_set )
804 FROM biblio.record_entry b
805 WHERE id = incoming_existing_id::bigint;
809 IF match_set IS NULL THEN
813 FOR test_result IN SELECT * FROM
814 vandelay.match_set_test_marcxml(match_set, NEW.marc) LOOP
816 INSERT INTO vandelay.bib_match ( queued_record, eg_record, match_score, quality )
821 vandelay.measure_record_quality( b.marc, match_set )
822 FROM biblio.record_entry b
823 WHERE id = test_result.record;
829 $func$ LANGUAGE PLPGSQL;
831 CREATE OR REPLACE FUNCTION vandelay.measure_record_quality ( xml TEXT, match_set_id INT ) RETURNS INT AS $_$
835 test vandelay.match_set_quality%ROWTYPE;
838 FOR test IN SELECT * FROM vandelay.match_set_quality WHERE match_set = match_set_id LOOP
839 IF test.tag IS NOT NULL THEN
840 FOR rvalue IN SELECT value FROM vandelay.flatten_marc( xml ) WHERE tag = test.tag AND subfield = test.subfield LOOP
841 IF test.value = rvalue THEN
842 out_q := out_q + test.quality;
846 IF test.value = vandelay.extract_rec_attrs(xml, ARRAY[test.svf]) -> test.svf THEN
847 out_q := out_q + test.quality;
854 $_$ LANGUAGE PLPGSQL;
856 CREATE TYPE vandelay.tcn_data AS (tcn TEXT, tcn_source TEXT, used BOOL);
857 CREATE OR REPLACE FUNCTION vandelay.find_bib_tcn_data ( xml TEXT ) RETURNS SETOF vandelay.tcn_data AS $_$
861 output vandelay.tcn_data%ROWTYPE;
865 eg_tcn := BTRIM((oils_xpath('//*[@tag="001"]/text()',xml))[1]);
866 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
868 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="003"]/text()',xml))[1]);
869 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
870 eg_tcn_source := 'System Local';
873 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
876 output.used := FALSE;
881 output.tcn := eg_tcn;
882 output.tcn_source := eg_tcn_source;
888 eg_tcn := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="a"]/text()',xml))[1]);
889 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
891 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="b"]/text()',xml))[1]);
892 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
893 eg_tcn_source := 'System Local';
896 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
899 output.used := FALSE;
904 output.tcn := eg_tcn;
905 output.tcn_source := eg_tcn_source;
911 eg_tcn := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="a"]/text()',xml))[1]);
912 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
914 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="b"]/text()',xml))[1]);
915 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
916 eg_tcn_source := 'System Local';
919 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
922 output.used := FALSE;
927 output.tcn := eg_tcn;
928 output.tcn_source := eg_tcn_source;
934 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="020"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
935 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
937 eg_tcn_source := 'ISBN';
939 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
942 output.used := FALSE;
947 output.tcn := eg_tcn;
948 output.tcn_source := eg_tcn_source;
954 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="022"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
955 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
957 eg_tcn_source := 'ISSN';
959 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
962 output.used := FALSE;
967 output.tcn := eg_tcn;
968 output.tcn_source := eg_tcn_source;
974 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="010"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
975 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
977 eg_tcn_source := 'LCCN';
979 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
982 output.used := FALSE;
987 output.tcn := eg_tcn;
988 output.tcn_source := eg_tcn_source;
994 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="035"]/*[@code="a"]/text()',xml))[1], $re$^.*?(\w+)$$re$, $re$\1$re$);
995 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
997 eg_tcn_source := 'System Legacy';
999 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1002 output.used := FALSE;
1004 output.used := TRUE;
1007 output.tcn := eg_tcn;
1008 output.tcn_source := eg_tcn_source;
1015 $_$ LANGUAGE PLPGSQL;
1017 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT, force_add INT ) RETURNS TEXT AS $_$
1020 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1024 MARC::Charset->assume_unicode(1);
1026 my $target_xml = shift;
1027 my $source_xml = shift;
1028 my $field_spec = shift;
1029 my $force_add = shift || 0;
1031 my $target_r = MARC::Record->new_from_xml( $target_xml );
1032 my $source_r = MARC::Record->new_from_xml( $source_xml );
1034 return $target_xml unless ($target_r && $source_r);
1036 my @field_list = split(',', $field_spec);
1039 for my $f (@field_list) {
1040 $f =~ s/^\s*//; $f =~ s/\s*$//;
1041 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1047 $match =~ s/^\s*//; $match =~ s/\s*$//;
1048 $fields{$field} = { sf => [ split('', $sf) ] };
1050 my ($msf,$mre) = split('~', $match);
1051 if (length($msf) > 0 and length($mre) > 0) {
1052 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1053 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1054 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1060 for my $f ( keys %fields) {
1061 if ( @{$fields{$f}{sf}} ) {
1062 for my $from_field ($source_r->field( $f )) {
1063 my @tos = $target_r->field( $f );
1065 next if (exists($fields{$f}{match}) and !$force_add);
1066 my @new_fields = map { $_->clone } $source_r->field( $f );
1067 $target_r->insert_fields_ordered( @new_fields );
1069 for my $to_field (@tos) {
1070 if (exists($fields{$f}{match})) {
1071 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1073 my @new_sf = map { ($_ => $from_field->subfield($_)) } grep { defined($from_field->subfield($_)) } @{$fields{$f}{sf}};
1074 $to_field->add_subfields( @new_sf );
1079 my @new_fields = map { $_->clone } $source_r->field( $f );
1080 $target_r->insert_fields_ordered( @new_fields );
1084 $target_xml = $target_r->as_xml_record;
1085 $target_xml =~ s/^<\?.+?\?>$//mo;
1086 $target_xml =~ s/\n//sgo;
1087 $target_xml =~ s/>\s+</></sgo;
1091 $_$ LANGUAGE PLPERLU;
1093 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1094 SELECT vandelay.add_field( $1, $2, $3, 0 );
1097 CREATE OR REPLACE FUNCTION vandelay.strip_field ( xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1100 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1104 MARC::Charset->assume_unicode(1);
1107 my $r = MARC::Record->new_from_xml( $xml );
1109 return $xml unless ($r);
1111 my $field_spec = shift;
1112 my @field_list = split(',', $field_spec);
1115 for my $f (@field_list) {
1116 $f =~ s/^\s*//; $f =~ s/\s*$//;
1117 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1123 $match =~ s/^\s*//; $match =~ s/\s*$//;
1124 $fields{$field} = { sf => [ split('', $sf) ] };
1126 my ($msf,$mre) = split('~', $match);
1127 if (length($msf) > 0 and length($mre) > 0) {
1128 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1129 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1130 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1136 for my $f ( keys %fields) {
1137 for my $to_field ($r->field( $f )) {
1138 if (exists($fields{$f}{match})) {
1139 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1142 if ( @{$fields{$f}{sf}} ) {
1143 $to_field->delete_subfield(code => $fields{$f}{sf});
1145 $r->delete_field( $to_field );
1150 $xml = $r->as_xml_record;
1151 $xml =~ s/^<\?.+?\?>$//mo;
1153 $xml =~ s/>\s+</></sgo;
1157 $_$ LANGUAGE PLPERLU;
1159 CREATE OR REPLACE FUNCTION vandelay.replace_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1166 parsed_target := vandelay.strip_field( target_xml, ''); -- this dance normalizes the format of the xml for the IF below
1167 xml_output := parsed_target; -- if there are no replace rules, just return the input
1169 FOR curr_field IN SELECT UNNEST( STRING_TO_ARRAY(field, ',') ) LOOP -- naive split, but it's the same we use in the perl
1171 xml_output := vandelay.strip_field( parsed_target, curr_field);
1173 IF xml_output <> parsed_target AND curr_field ~ E'~' THEN
1174 -- we removed something, and there was a regexp restriction in the curr_field definition, so proceed
1175 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 1 );
1176 ELSIF curr_field !~ E'~' THEN
1177 -- No regexp restriction, add the curr_field
1178 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 0 );
1181 parsed_target := xml_output; -- in prep for any following loop iterations
1187 $_$ LANGUAGE PLPGSQL;
1189 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 $_$
1190 SELECT vandelay.replace_field( vandelay.add_field( vandelay.strip_field( $1, $5) , $2, $3 ), $2, $4);
1193 CREATE TYPE vandelay.compile_profile AS (add_rule TEXT, replace_rule TEXT, preserve_rule TEXT, strip_rule TEXT);
1194 CREATE OR REPLACE FUNCTION vandelay.compile_profile ( incoming_xml TEXT ) RETURNS vandelay.compile_profile AS $_$
1196 output vandelay.compile_profile%ROWTYPE;
1197 profile vandelay.merge_profile%ROWTYPE;
1199 profile_tmpl_owner TEXT;
1200 add_rule TEXT := '';
1201 strip_rule TEXT := '';
1202 replace_rule TEXT := '';
1203 preserve_rule TEXT := '';
1207 profile_tmpl := (oils_xpath('//*[@tag="905"]/*[@code="t"]/text()',incoming_xml))[1];
1208 profile_tmpl_owner := (oils_xpath('//*[@tag="905"]/*[@code="o"]/text()',incoming_xml))[1];
1210 IF profile_tmpl IS NOT NULL AND profile_tmpl <> '' AND profile_tmpl_owner IS NOT NULL AND profile_tmpl_owner <> '' THEN
1211 SELECT p.* INTO profile
1212 FROM vandelay.merge_profile p
1213 JOIN actor.org_unit u ON (u.id = p.owner)
1214 WHERE p.name = profile_tmpl
1215 AND u.shortname = profile_tmpl_owner;
1217 IF profile.id IS NOT NULL THEN
1218 add_rule := COALESCE(profile.add_spec,'');
1219 strip_rule := COALESCE(profile.strip_spec,'');
1220 replace_rule := COALESCE(profile.replace_spec,'');
1221 preserve_rule := COALESCE(profile.preserve_spec,'');
1225 add_rule := add_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="a"]/text()',incoming_xml),','),'');
1226 strip_rule := strip_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="d"]/text()',incoming_xml),','),'');
1227 replace_rule := replace_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="r"]/text()',incoming_xml),','),'');
1228 preserve_rule := preserve_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="p"]/text()',incoming_xml),','),'');
1230 output.add_rule := BTRIM(add_rule,',');
1231 output.replace_rule := BTRIM(replace_rule,',');
1232 output.strip_rule := BTRIM(strip_rule,',');
1233 output.preserve_rule := BTRIM(preserve_rule,',');
1237 $_$ LANGUAGE PLPGSQL;
1239 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1241 merge_profile vandelay.merge_profile%ROWTYPE;
1242 dyn_profile vandelay.compile_profile%ROWTYPE;
1252 SELECT b.marc INTO eg_marc
1253 FROM biblio.record_entry b
1257 IF eg_marc IS NULL OR v_marc IS NULL THEN
1258 -- RAISE NOTICE 'no marc for template or bib record';
1262 dyn_profile := vandelay.compile_profile( v_marc );
1264 IF merge_profile_id IS NOT NULL THEN
1265 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1267 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1268 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1269 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1270 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1274 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1275 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1279 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1280 --Since we have nothing to do, just return a NOOP "we did it"
1282 ELSIF dyn_profile.replace_rule <> '' THEN
1283 source_marc = v_marc;
1284 target_marc = eg_marc;
1285 replace_rule = dyn_profile.replace_rule;
1287 source_marc = eg_marc;
1288 target_marc = v_marc;
1289 replace_rule = dyn_profile.preserve_rule;
1292 UPDATE biblio.record_entry
1293 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1297 -- RAISE NOTICE 'update of biblio.record_entry failed';
1304 $$ LANGUAGE PLPGSQL;
1306 CREATE OR REPLACE FUNCTION vandelay.merge_record_xml ( target_marc TEXT, template_marc TEXT ) RETURNS TEXT AS $$
1308 dyn_profile vandelay.compile_profile%ROWTYPE;
1316 IF target_marc IS NULL OR template_marc IS NULL THEN
1317 -- RAISE NOTICE 'no marc for target or template record';
1321 dyn_profile := vandelay.compile_profile( template_marc );
1323 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1324 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1328 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1329 --Since we have nothing to do, just return what we were given.
1331 ELSIF dyn_profile.replace_rule <> '' THEN
1332 trgt_marc = target_marc;
1333 tmpl_marc = template_marc;
1334 replace_rule = dyn_profile.replace_rule;
1336 tmp_marc = target_marc;
1337 trgt_marc = template_marc;
1338 tmpl_marc = tmp_marc;
1339 replace_rule = dyn_profile.preserve_rule;
1342 RETURN vandelay.merge_record_xml( trgt_marc, tmpl_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule );
1345 $$ LANGUAGE PLPGSQL;
1347 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT) RETURNS BOOL AS $$
1348 SELECT vandelay.template_overlay_bib_record( $1, $2, NULL);
1351 CREATE OR REPLACE FUNCTION vandelay.overlay_bib_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1353 merge_profile vandelay.merge_profile%ROWTYPE;
1354 dyn_profile vandelay.compile_profile%ROWTYPE;
1364 SELECT q.marc INTO v_marc
1365 FROM vandelay.queued_record q
1366 JOIN vandelay.bib_match m ON (m.queued_record = q.id AND q.id = import_id)
1369 IF v_marc IS NULL THEN
1370 -- RAISE NOTICE 'no marc for vandelay or bib record';
1374 IF vandelay.template_overlay_bib_record( v_marc, eg_id, merge_profile_id) THEN
1375 UPDATE vandelay.queued_bib_record
1376 SET imported_as = eg_id,
1378 WHERE id = import_id;
1380 editor_string := (oils_xpath('//*[@tag="905"]/*[@code="u"]/text()',v_marc))[1];
1382 IF editor_string IS NOT NULL AND editor_string <> '' THEN
1383 SELECT usr INTO editor_id FROM actor.card WHERE barcode = editor_string;
1385 IF editor_id IS NULL THEN
1386 SELECT id INTO editor_id FROM actor.usr WHERE usrname = editor_string;
1389 IF editor_id IS NOT NULL THEN
1390 UPDATE biblio.record_entry SET editor = editor_id WHERE id = eg_id;
1397 -- RAISE NOTICE 'update of biblio.record_entry failed';
1402 $$ LANGUAGE PLPGSQL;
1404 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
1407 lwm_ratio_value NUMERIC;
1410 lwm_ratio_value := COALESCE(lwm_ratio_value_p, 0.0);
1412 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1415 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1419 SELECT m.eg_record INTO eg_id
1420 FROM vandelay.bib_match m
1421 JOIN vandelay.queued_bib_record qr ON (m.queued_record = qr.id)
1422 JOIN vandelay.bib_queue q ON (qr.queue = q.id)
1423 JOIN biblio.record_entry r ON (r.id = m.eg_record)
1424 WHERE m.queued_record = import_id
1425 AND qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC >= lwm_ratio_value
1426 ORDER BY m.match_score DESC, -- required match score
1427 qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC DESC, -- quality tie breaker
1428 m.id -- when in doubt, use the first match
1431 IF eg_id IS NULL THEN
1432 -- RAISE NOTICE 'incoming record is not of high enough quality';
1436 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1438 $$ LANGUAGE PLPGSQL;
1440 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1441 SELECT vandelay.auto_overlay_bib_record_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1444 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1450 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1453 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1457 SELECT COUNT(*) INTO match_count FROM vandelay.bib_match WHERE queued_record = import_id;
1459 IF match_count <> 1 THEN
1460 -- RAISE NOTICE 'not an exact match';
1464 -- Check that the one match is on the first 901c
1465 SELECT m.eg_record INTO eg_id
1466 FROM vandelay.queued_bib_record q
1467 JOIN vandelay.bib_match m ON (m.queued_record = q.id)
1468 WHERE q.id = import_id
1469 AND m.eg_record = oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]',marc)::BIGINT;
1472 -- RAISE NOTICE 'not a 901c match';
1476 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1478 $$ LANGUAGE PLPGSQL;
1480 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1482 queued_record vandelay.queued_bib_record%ROWTYPE;
1485 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1487 IF vandelay.auto_overlay_bib_record( queued_record.id, merge_profile_id ) THEN
1488 RETURN NEXT queued_record.id;
1496 $$ LANGUAGE PLPGSQL;
1498 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 $$
1500 queued_record vandelay.queued_bib_record%ROWTYPE;
1503 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1505 IF vandelay.auto_overlay_bib_record_with_best( queued_record.id, merge_profile_id, lwm_ratio_value ) THEN
1506 RETURN NEXT queued_record.id;
1514 $$ LANGUAGE PLPGSQL;
1516 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1517 SELECT vandelay.auto_overlay_bib_queue_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1520 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1521 SELECT * FROM vandelay.auto_overlay_bib_queue( $1, NULL );
1524 CREATE OR REPLACE FUNCTION vandelay.ingest_bib_marc ( ) RETURNS TRIGGER AS $$
1530 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1534 FOR adef IN SELECT * FROM vandelay.bib_attr_definition LOOP
1536 SELECT extract_marc_field('vandelay.queued_bib_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_bib_record WHERE id = NEW.id;
1537 IF (value IS NOT NULL AND value <> '') THEN
1538 INSERT INTO vandelay.queued_bib_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1545 $$ LANGUAGE PLPGSQL;
1547 CREATE OR REPLACE FUNCTION vandelay.cleanup_bib_marc ( ) RETURNS TRIGGER AS $$
1549 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1553 DELETE FROM vandelay.queued_bib_record_attr WHERE record = OLD.id;
1554 DELETE FROM vandelay.import_item WHERE record = OLD.id;
1556 IF TG_OP = 'UPDATE' THEN
1561 $$ LANGUAGE PLPGSQL;
1563 CREATE TRIGGER cleanup_bib_trigger
1564 BEFORE UPDATE OR DELETE ON vandelay.queued_bib_record
1565 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_bib_marc();
1567 CREATE TRIGGER ingest_bib_trigger
1568 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1569 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_bib_marc();
1571 CREATE TRIGGER zz_match_bibs_trigger
1572 BEFORE INSERT OR UPDATE ON vandelay.queued_bib_record
1573 FOR EACH ROW EXECUTE PROCEDURE vandelay.match_bib_record();
1576 /* Authority stuff down here */
1577 ---------------------------------------
1578 CREATE TABLE vandelay.authority_attr_definition (
1579 id SERIAL PRIMARY KEY,
1580 code TEXT UNIQUE NOT NULL,
1582 xpath TEXT NOT NULL,
1583 remove TEXT NOT NULL DEFAULT ''
1586 CREATE TYPE vandelay.authority_queue_queue_type AS ENUM ('authority');
1587 CREATE TABLE vandelay.authority_queue (
1588 queue_type vandelay.authority_queue_queue_type NOT NULL DEFAULT 'authority',
1589 CONSTRAINT vand_authority_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
1590 ) INHERITS (vandelay.queue);
1591 ALTER TABLE vandelay.authority_queue ADD PRIMARY KEY (id);
1593 CREATE TABLE vandelay.queued_authority_record (
1594 queue INT NOT NULL REFERENCES vandelay.authority_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1595 imported_as INT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1596 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
1598 ) INHERITS (vandelay.queued_record);
1599 ALTER TABLE vandelay.queued_authority_record ADD PRIMARY KEY (id);
1600 CREATE INDEX queued_authority_record_queue_idx ON vandelay.queued_authority_record (queue);
1602 CREATE TABLE vandelay.queued_authority_record_attr (
1603 id BIGSERIAL PRIMARY KEY,
1604 record BIGINT NOT NULL REFERENCES vandelay.queued_authority_record (id) DEFERRABLE INITIALLY DEFERRED,
1605 field INT NOT NULL REFERENCES vandelay.authority_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
1606 attr_value TEXT NOT NULL
1608 CREATE INDEX queued_authority_record_attr_record_idx ON vandelay.queued_authority_record_attr (record);
1610 CREATE TABLE vandelay.authority_match (
1611 id BIGSERIAL PRIMARY KEY,
1612 queued_record BIGINT REFERENCES vandelay.queued_authority_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1613 eg_record BIGINT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1614 quality INT NOT NULL DEFAULT 0
1617 CREATE OR REPLACE FUNCTION vandelay.ingest_authority_marc ( ) RETURNS TRIGGER AS $$
1623 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1627 FOR adef IN SELECT * FROM vandelay.authority_attr_definition LOOP
1629 SELECT extract_marc_field('vandelay.queued_authority_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_authority_record WHERE id = NEW.id;
1630 IF (value IS NOT NULL AND value <> '') THEN
1631 INSERT INTO vandelay.queued_authority_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1638 $$ LANGUAGE PLPGSQL;
1640 CREATE OR REPLACE FUNCTION vandelay.cleanup_authority_marc ( ) RETURNS TRIGGER AS $$
1642 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1646 DELETE FROM vandelay.queued_authority_record_attr WHERE record = OLD.id;
1647 IF TG_OP = 'UPDATE' THEN
1652 $$ LANGUAGE PLPGSQL;
1654 CREATE TRIGGER cleanup_authority_trigger
1655 BEFORE UPDATE OR DELETE ON vandelay.queued_authority_record
1656 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_authority_marc();
1658 CREATE TRIGGER ingest_authority_trigger
1659 AFTER INSERT OR UPDATE ON vandelay.queued_authority_record
1660 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_authority_marc();
1662 CREATE OR REPLACE FUNCTION vandelay.overlay_authority_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1664 merge_profile vandelay.merge_profile%ROWTYPE;
1665 dyn_profile vandelay.compile_profile%ROWTYPE;
1674 SELECT b.marc INTO eg_marc
1675 FROM authority.record_entry b
1676 JOIN vandelay.authority_match m ON (m.eg_record = b.id AND m.queued_record = import_id)
1679 SELECT q.marc INTO v_marc
1680 FROM vandelay.queued_record q
1681 JOIN vandelay.authority_match m ON (m.queued_record = q.id AND q.id = import_id)
1684 IF eg_marc IS NULL OR v_marc IS NULL THEN
1685 -- RAISE NOTICE 'no marc for vandelay or authority record';
1689 dyn_profile := vandelay.compile_profile( v_marc );
1691 IF merge_profile_id IS NOT NULL THEN
1692 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1694 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1695 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1696 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1697 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1701 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1702 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1706 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1707 --Since we have nothing to do, just return a NOOP "we did it"
1709 ELSIF dyn_profile.replace_rule <> '' THEN
1710 source_marc = v_marc;
1711 target_marc = eg_marc;
1712 replace_rule = dyn_profile.replace_rule;
1714 source_marc = eg_marc;
1715 target_marc = v_marc;
1716 replace_rule = dyn_profile.preserve_rule;
1719 UPDATE authority.record_entry
1720 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1724 UPDATE vandelay.queued_authority_record
1725 SET imported_as = eg_id,
1727 WHERE id = import_id;
1731 -- RAISE NOTICE 'update of authority.record_entry failed';
1736 $$ LANGUAGE PLPGSQL;
1738 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1743 SELECT COUNT(*) INTO match_count FROM vandelay.authority_match WHERE queued_record = import_id;
1745 IF match_count <> 1 THEN
1746 -- RAISE NOTICE 'not an exact match';
1750 SELECT m.eg_record INTO eg_id
1751 FROM vandelay.authority_match m
1752 WHERE m.queued_record = import_id
1755 IF eg_id IS NULL THEN
1759 RETURN vandelay.overlay_authority_record( import_id, eg_id, merge_profile_id );
1761 $$ LANGUAGE PLPGSQL;
1763 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1765 queued_record vandelay.queued_authority_record%ROWTYPE;
1768 FOR queued_record IN SELECT * FROM vandelay.queued_authority_record WHERE queue = queue_id AND import_time IS NULL LOOP
1770 IF vandelay.auto_overlay_authority_record( queued_record.id, merge_profile_id ) THEN
1771 RETURN NEXT queued_record.id;
1779 $$ LANGUAGE PLPGSQL;
1781 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1782 SELECT * FROM vandelay.auto_overlay_authority_queue( $1, NULL );
1786 -- Vandelay (for importing and exporting records) 012.schema.vandelay.sql
1787 --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)]');
1788 --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)]');
1789 --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]');
1790 --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]');
1791 --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$);
1792 --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$);
1793 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (7, 'price',oils_i18n_gettext(7, 'vqbrad', 'Price', 'description'),'//*[@tag="020" or @tag="022"]/*[@code="c"][1]');
1794 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (8, 'rec_identifier',oils_i18n_gettext(8, 'vqbrad', 'Accession Number', 'description'),'//*[@tag="001"]', TRUE);
1795 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (9, 'eg_tcn',oils_i18n_gettext(9, 'vqbrad', 'TCN Value', 'description'),'//*[@tag="901"]/*[@code="a"]', TRUE);
1796 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (10, 'eg_tcn_source',oils_i18n_gettext(10, 'vqbrad', 'TCN Source', 'description'),'//*[@tag="901"]/*[@code="b"]', TRUE);
1797 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (11, 'eg_identifier',oils_i18n_gettext(11, 'vqbrad', 'Internal ID', 'description'),'//*[@tag="901"]/*[@code="c"]', TRUE);
1798 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (12, 'publisher',oils_i18n_gettext(12, 'vqbrad', 'Publisher', 'description'),'//*[@tag="260"]/*[@code="b"][1]');
1799 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, remove ) VALUES (13, 'pubdate',oils_i18n_gettext(13, 'vqbrad', 'Publication Date', 'description'),'//*[@tag="260"]/*[@code="c"][1]',$r$\D$r$);
1800 --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]');
1802 --INSERT INTO vandelay.import_item_attr_definition (
1803 -- owner, name, tag, owning_lib, circ_lib, location,
1804 -- call_number, circ_modifier, barcode, price, copy_number,
1805 -- circulate, ref, holdable, opac_visible, status
1808 -- 'Evergreen 852 export format',
1810 -- '[@code = "b"][1]',
1811 -- '[@code = "b"][2]',
1818 -- '[@code = "x" and text() = "circulating"]',
1819 -- '[@code = "x" and text() = "reference"]',
1820 -- '[@code = "x" and text() = "holdable"]',
1821 -- '[@code = "x" and text() = "visible"]',
1825 --INSERT INTO vandelay.import_item_attr_definition (
1838 -- 'Unicorn Import format -- 999',
1849 --INSERT INTO vandelay.authority_attr_definition ( code, description, xpath, ident ) VALUES ('rec_identifier','Identifier','//*[@tag="001"]', TRUE);