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 heading BOOLEAN NOT NULL DEFAULT FALSE, -- match on authority heading
27 CONSTRAINT vmsp_need_a_subfield_with_a_tag CHECK ((tag IS NOT NULL AND subfield IS NOT NULL) OR tag IS NULL),
28 CONSTRAINT vmsp_need_a_tag_or_a_ff_or_a_bo CHECK (
29 (tag IS NOT NULL AND svf IS NULL AND heading IS FALSE AND bool_op IS NULL) OR
30 (tag IS NULL AND svf IS NOT NULL AND heading IS FALSE AND bool_op IS NULL) OR
31 (tag IS NULL AND svf IS NULL AND heading IS TRUE AND bool_op IS NULL) OR
32 (tag IS NULL AND svf IS NULL AND heading IS FALSE AND bool_op IS NOT NULL)
36 CREATE TABLE vandelay.match_set_quality (
37 id SERIAL PRIMARY KEY,
38 match_set INT NOT NULL REFERENCES vandelay.match_set (id) ON DELETE CASCADE,
39 svf TEXT REFERENCES config.record_attr_definition,
43 quality INT NOT NULL DEFAULT 1, -- higher is better
44 CONSTRAINT vmsq_need_a_subfield_with_a_tag CHECK ((tag IS NOT NULL AND subfield IS NOT NULL) OR tag IS NULL),
45 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))
47 CREATE UNIQUE INDEX vmsq_def_once_per_set ON vandelay.match_set_quality (match_set, COALESCE(tag,''), COALESCE(subfield,''), COALESCE(svf,''), value);
50 CREATE TABLE vandelay.queue (
51 id BIGSERIAL PRIMARY KEY,
52 owner INT NOT NULL REFERENCES actor.usr (id) DEFERRABLE INITIALLY DEFERRED,
54 complete BOOL NOT NULL DEFAULT FALSE,
55 match_set INT REFERENCES vandelay.match_set (id) ON UPDATE CASCADE ON DELETE SET NULL DEFERRABLE INITIALLY DEFERRED
58 CREATE TABLE vandelay.queued_record (
59 id BIGSERIAL PRIMARY KEY,
60 create_time TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT NOW(),
61 import_time TIMESTAMP WITH TIME ZONE,
62 purpose TEXT NOT NULL DEFAULT 'import' CHECK (purpose IN ('import','overlay')),
64 quality INT NOT NULL DEFAULT 0
69 /* Bib stuff at the top */
70 ----------------------------------------------------
72 CREATE TABLE vandelay.bib_attr_definition (
73 id SERIAL PRIMARY KEY,
74 code TEXT UNIQUE NOT NULL,
77 remove TEXT NOT NULL DEFAULT ''
80 -- Each TEXT field (other than 'name') should hold an XPath predicate for pulling the data needed
81 -- DROP TABLE vandelay.import_item_attr_definition CASCADE;
82 CREATE TABLE vandelay.import_item_attr_definition (
83 id BIGSERIAL PRIMARY KEY,
84 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
87 keep BOOL NOT NULL DEFAULT FALSE,
107 priv_note_title TEXT,
112 CONSTRAINT vand_import_item_attr_def_idx UNIQUE (owner,name)
115 CREATE TABLE vandelay.import_error (
116 code TEXT PRIMARY KEY,
117 description TEXT NOT NULL -- i18n
120 CREATE TYPE vandelay.bib_queue_queue_type AS ENUM ('bib', 'acq');
122 CREATE TABLE vandelay.bib_queue (
123 queue_type vandelay.bib_queue_queue_type NOT NULL DEFAULT 'bib',
124 item_attr_def BIGINT REFERENCES vandelay.import_item_attr_definition (id) ON DELETE SET NULL DEFERRABLE INITIALLY DEFERRED,
125 match_bucket INTEGER, -- REFERENCES container.biblio_record_entry_bucket(id);
126 CONSTRAINT vand_bib_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
127 ) INHERITS (vandelay.queue);
128 ALTER TABLE vandelay.bib_queue ADD PRIMARY KEY (id);
130 CREATE TABLE vandelay.queued_bib_record (
131 queue INT NOT NULL REFERENCES vandelay.bib_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
132 bib_source INT REFERENCES config.bib_source (id) DEFERRABLE INITIALLY DEFERRED,
133 imported_as BIGINT REFERENCES biblio.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
134 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
136 ) INHERITS (vandelay.queued_record);
137 ALTER TABLE vandelay.queued_bib_record ADD PRIMARY KEY (id);
138 CREATE INDEX queued_bib_record_queue_idx ON vandelay.queued_bib_record (queue);
140 CREATE TABLE vandelay.queued_bib_record_attr (
141 id BIGSERIAL PRIMARY KEY,
142 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) DEFERRABLE INITIALLY DEFERRED,
143 field INT NOT NULL REFERENCES vandelay.bib_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
144 attr_value TEXT NOT NULL
146 CREATE INDEX queued_bib_record_attr_record_idx ON vandelay.queued_bib_record_attr (record);
148 CREATE TABLE vandelay.bib_match (
149 id BIGSERIAL PRIMARY KEY,
150 queued_record BIGINT REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
151 eg_record BIGINT REFERENCES biblio.record_entry (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
152 quality INT NOT NULL DEFAULT 1,
153 match_score INT NOT NULL DEFAULT 0
155 CREATE INDEX bib_match_queued_record_idx ON vandelay.bib_match (queued_record);
157 CREATE TABLE vandelay.import_item (
158 id BIGSERIAL PRIMARY KEY,
159 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
160 definition BIGINT NOT NULL REFERENCES vandelay.import_item_attr_definition (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
161 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
164 import_time TIMESTAMP WITH TIME ZONE,
173 deposit_amount NUMERIC(8,2),
186 internal_id BIGINT -- queue_type == 'acq' ? acq.lineitem_detail.id : asset.copy.id
188 CREATE INDEX import_item_record_idx ON vandelay.import_item (record);
190 CREATE TABLE vandelay.import_bib_trash_group(
191 id SERIAL PRIMARY KEY,
192 owner INTEGER NOT NULL REFERENCES actor.org_unit(id),
193 label TEXT NOT NULL, --i18n
194 always_apply BOOLEAN NOT NULL DEFAULT FALSE,
195 CONSTRAINT vand_import_bib_trash_grp_owner_label UNIQUE (owner, label)
198 CREATE TABLE vandelay.import_bib_trash_fields (
199 id BIGSERIAL PRIMARY KEY,
200 grp INTEGER NOT NULL REFERENCES vandelay.import_bib_trash_group,
202 CONSTRAINT vand_import_bib_trash_fields_once_per UNIQUE (grp, field)
205 CREATE TABLE vandelay.merge_profile (
206 id BIGSERIAL PRIMARY KEY,
207 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
213 update_bib_source BOOLEAN NOT NULL DEFAULT FALSE,
214 update_bib_editor BOOLEAN NOT NULL DEFAULT FALSE,
216 CONSTRAINT vand_merge_prof_owner_name_idx UNIQUE (owner,name),
217 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))
220 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
227 retval config.marc21_rec_type_map%ROWTYPE;
229 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
231 IF ldr IS NULL OR ldr = '' THEN
232 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
236 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
237 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
240 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
241 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
243 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
245 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
248 IF retval.code IS NULL THEN
249 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
254 $func$ LANGUAGE PLPGSQL;
256 CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
257 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT, use_default BOOL DEFAULT FALSE ) RETURNS SETOF biblio.record_ff_map AS $func$
262 output biblio.record_ff_map%ROWTYPE;
264 rtype := (vandelay.marc21_record_type( marc )).code;
266 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
267 output.ff_name := ff_pos.fixed_field;
268 output.ff_value := NULL;
270 IF ff_pos.tag = 'ldr' THEN
271 output.ff_value := oils_xpath_string('//*[local-name()="leader"]', marc);
272 IF output.ff_value IS NOT NULL THEN
273 output.ff_value := SUBSTRING( output.ff_value, ff_pos.start_pos + 1, ff_pos.length );
275 output.ff_value := NULL;
278 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
279 output.ff_value := SUBSTRING( tag_data, ff_pos.start_pos + 1, ff_pos.length );
280 CONTINUE WHEN output.ff_value IS NULL AND NOT use_default;
281 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
283 output.ff_value := NULL;
291 $func$ LANGUAGE PLPGSQL;
293 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field_list( marc TEXT, ff TEXT, use_default BOOL DEFAULT FALSE ) RETURNS TEXT[] AS $func$
299 collection TEXT[] := '{}'::TEXT[];
301 rtype := (vandelay.marc21_record_type( marc )).code;
302 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
303 IF ff_pos.tag = 'ldr' THEN
304 val := oils_xpath_string('//*[local-name()="leader"]', marc);
305 IF val IS NOT NULL THEN
306 val := SUBSTRING( val, ff_pos.start_pos + 1, ff_pos.length );
307 collection := collection || val;
310 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
311 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
312 collection := collection || val;
315 CONTINUE WHEN NOT use_default;
316 CONTINUE WHEN ARRAY_UPPER(collection, 1) > 0;
317 val := REPEAT( ff_pos.default_val, ff_pos.length );
318 collection := collection || val;
323 $func$ LANGUAGE PLPGSQL;
325 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT, use_default BOOL DEFAULT FALSE ) RETURNS TEXT AS $func$
332 rtype := (vandelay.marc21_record_type( marc )).code;
333 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
334 IF ff_pos.tag = 'ldr' THEN
335 val := oils_xpath_string('//*[local-name()="leader"]', marc);
336 IF val IS NOT NULL THEN
337 val := SUBSTRING( val, ff_pos.start_pos + 1, ff_pos.length );
341 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
342 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
346 CONTINUE WHEN NOT use_default;
347 val := REPEAT( ff_pos.default_val, ff_pos.length );
353 $func$ LANGUAGE PLPGSQL;
355 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
356 CREATE OR REPLACE FUNCTION vandelay.marc21_physical_characteristics( marc TEXT) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
360 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
361 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
362 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
363 retval biblio.marc21_physical_characteristics%ROWTYPE;
366 FOR _007 IN SELECT oils_xpath_string('//*', value) FROM UNNEST(oils_xpath('//*[@tag="007"]', marc)) x(value) LOOP
367 IF _007 IS NOT NULL AND _007 <> '' THEN
368 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007, 1, 1 );
370 IF ptype.ptype_key IS NOT NULL THEN
371 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
372 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 );
374 IF pval.id IS NOT NULL THEN
377 retval.ptype := ptype.ptype_key;
378 retval.subfield := psf.id;
379 retval.value := pval.id;
390 $func$ LANGUAGE PLPGSQL;
392 CREATE TYPE vandelay.flat_marc AS ( tag CHAR(3), ind1 TEXT, ind2 TEXT, subfield TEXT, value TEXT );
393 CREATE OR REPLACE FUNCTION vandelay.flay_marc ( TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
396 use MARC::File::XML (BinaryEncoding => 'UTF-8');
400 MARC::Charset->assume_unicode(1);
403 my $r = MARC::Record->new_from_xml( $xml );
405 return_next( { tag => 'LDR', value => $r->leader } );
407 for my $f ( $r->fields ) {
408 if ($f->is_control_field) {
409 return_next({ tag => $f->tag, value => $f->data });
411 for my $s ($f->subfields) {
414 ind1 => $f->indicator(1),
415 ind2 => $f->indicator(2),
420 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
421 my $trim = $f->indicator(2) || 0;
424 ind1 => $f->indicator(1),
425 ind2 => $f->indicator(2),
427 value => substr( $s->[1], $trim )
436 $func$ LANGUAGE PLPERLU;
438 CREATE OR REPLACE FUNCTION vandelay.flatten_marc ( marc TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
440 output vandelay.flat_marc%ROWTYPE;
443 FOR field IN SELECT * FROM vandelay.flay_marc( marc ) LOOP
444 output.ind1 := field.ind1;
445 output.ind2 := field.ind2;
446 output.tag := field.tag;
447 output.subfield := field.subfield;
448 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
449 output.value := naco_normalize(field.value, field.subfield);
451 output.value := field.value;
454 CONTINUE WHEN output.value IS NULL;
459 $func$ LANGUAGE PLPGSQL;
461 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT, attr_defs TEXT[]) RETURNS hstore AS $_$
463 transformed_xml TEXT;
466 xfrm config.xml_transform%ROWTYPE;
468 new_attrs HSTORE := ''::HSTORE;
469 attr_def config.record_attr_definition%ROWTYPE;
472 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE name IN (SELECT * FROM UNNEST(attr_defs)) ORDER BY format LOOP
474 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
475 SELECT STRING_AGG(x.value, COALESCE(attr_def.joiner,' ')) INTO attr_value
476 FROM vandelay.flatten_marc(xml) AS x
477 WHERE x.tag LIKE attr_def.tag
479 WHEN attr_def.sf_list IS NOT NULL
480 THEN POSITION(x.subfield IN attr_def.sf_list) > 0
487 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
488 attr_value := vandelay.marc21_extract_fixed_field(xml, attr_def.fixed_field);
490 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
492 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
494 -- See if we can skip the XSLT ... it's expensive
495 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
496 -- Can't skip the transform
497 IF xfrm.xslt <> '---' THEN
498 transformed_xml := oils_xslt_process(xml,xfrm.xslt);
500 transformed_xml := xml;
503 prev_xfrm := xfrm.name;
506 IF xfrm.name IS NULL THEN
507 -- just grab the marcxml (empty) transform
508 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
509 prev_xfrm := xfrm.name;
512 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
514 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
515 SELECT m.value::TEXT INTO attr_value
516 FROM vandelay.marc21_physical_characteristics(xml) v
517 JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
518 WHERE v.subfield = attr_def.phys_char_sf
519 LIMIT 1; -- Just in case ...
523 -- apply index normalizers to attr_value
525 SELECT n.func AS func,
526 n.param_count AS param_count,
528 FROM config.index_normalizer n
529 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
530 WHERE attr = attr_def.name
532 EXECUTE 'SELECT ' || normalizer.func || '(' ||
533 quote_nullable( attr_value ) ||
535 WHEN normalizer.param_count > 0
536 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
543 -- Add the new value to the hstore
544 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
550 $_$ LANGUAGE PLPGSQL;
552 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT ) RETURNS hstore AS $_$
553 SELECT vandelay.extract_rec_attrs( $1, (SELECT ARRAY_AGG(name) FROM config.record_attr_definition));
556 -- Everything between this comment and the beginning of the definition of
557 -- vandelay.match_bib_record() is strictly in service of that function.
558 CREATE TYPE vandelay.match_set_test_result AS (record BIGINT, quality INTEGER);
560 CREATE OR REPLACE FUNCTION vandelay.match_set_test_marcxml(
561 match_set_id INTEGER, record_xml TEXT, bucket_id INTEGER
562 ) RETURNS SETOF vandelay.match_set_test_result AS $$
573 tags_rstore := vandelay.flatten_marc_hstore(record_xml);
574 svf_rstore := vandelay.extract_rec_attrs(record_xml);
576 CREATE TEMPORARY TABLE _vandelay_tmp_qrows (q INTEGER);
577 CREATE TEMPORARY TABLE _vandelay_tmp_jrows (j TEXT);
579 -- generate the where clause and return that directly (into wq), and as
580 -- a side-effect, populate the _vandelay_tmp_[qj]rows tables.
581 wq := vandelay.get_expr_from_match_set(match_set_id, tags_rstore);
583 query_ := 'SELECT DISTINCT(record), ';
585 -- qrows table is for the quality bits we add to the SELECT clause
587 'COALESCE(n' || q::TEXT || '.quality, 0)', ' + '
588 ) INTO coal FROM _vandelay_tmp_qrows;
590 -- our query string so far is the SELECT clause and the inital FROM.
591 -- no JOINs yet nor the WHERE clause
592 query_ := query_ || coal || ' AS quality ' || E'\n';
594 -- jrows table is for the joins we must make (and the real text conditions)
595 SELECT STRING_AGG(j, E'\n') INTO joins
596 FROM _vandelay_tmp_jrows;
598 -- add those joins and the where clause to our query.
599 query_ := query_ || joins || E'\n';
601 -- join the record bucket
602 IF bucket_id IS NOT NULL THEN
603 query_ := query_ || 'JOIN container.biblio_record_entry_bucket_item ' ||
604 'brebi ON (brebi.target_biblio_record_entry = record ' ||
605 'AND brebi.bucket = ' || bucket_id || E')\n';
608 query_ := query_ || 'JOIN biblio.record_entry bre ON (bre.id = record) ' || 'WHERE ' || wq || ' AND not bre.deleted';
610 -- this will return rows of record,quality
611 FOR rec IN EXECUTE query_ USING tags_rstore, svf_rstore LOOP
615 DROP TABLE _vandelay_tmp_qrows;
616 DROP TABLE _vandelay_tmp_jrows;
622 CREATE OR REPLACE FUNCTION vandelay.flatten_marc_hstore(
624 ) RETURNS HSTORE AS $func$
628 ARRAY_AGG(tag || (COALESCE(subfield, ''))),
632 SELECT tag, subfield, ARRAY_AGG(value)::TEXT AS value
635 CASE WHEN tag = '020' THEN -- caseless -- isbn
636 LOWER((SELECT REGEXP_MATCHES(value,$$^(\S{10,17})$$))[1] || '%')
637 WHEN tag = '022' THEN -- caseless -- issn
638 LOWER((SELECT REGEXP_MATCHES(value,$$^(\S{4}[- ]?\S{4})$$))[1] || '%')
639 WHEN tag = '024' THEN -- caseless -- upc (other)
644 FROM vandelay.flatten_marc(record_xml)) x
645 GROUP BY tag, subfield ORDER BY tag, subfield
649 $func$ LANGUAGE PLPGSQL;
651 -- backwards compat version so we don't have
652 -- to modify vandelay.match_set_test_marcxml()
653 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set(
654 match_set_id INTEGER,
658 RETURN vandelay.get_expr_from_match_set(
659 match_set_id, tags_rstore, NULL);
663 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set(
664 match_set_id INTEGER,
669 root vandelay.match_set_point;
671 SELECT * INTO root FROM vandelay.match_set_point
672 WHERE parent IS NULL AND match_set = match_set_id;
674 RETURN vandelay.get_expr_from_match_set_point(
675 root, tags_rstore, auth_heading);
679 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set_point(
680 node vandelay.match_set_point,
689 child vandelay.match_set_point;
691 SELECT ARRAY_AGG(id) INTO children FROM vandelay.match_set_point
692 WHERE parent = node.id;
694 IF ARRAY_LENGTH(children, 1) > 0 THEN
695 this_op := vandelay._get_expr_render_one(node);
698 WHILE children[i] IS NOT NULL LOOP
699 SELECT * INTO child FROM vandelay.match_set_point
700 WHERE id = children[i];
702 q := q || ' ' || this_op || ' ';
705 q := q || vandelay.get_expr_from_match_set_point(
706 child, tags_rstore, auth_heading);
710 ELSIF node.bool_op IS NULL THEN
711 PERFORM vandelay._get_expr_push_qrow(node);
712 PERFORM vandelay._get_expr_push_jrow(node, tags_rstore, auth_heading);
713 RETURN vandelay._get_expr_render_one(node);
720 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_qrow(
721 node vandelay.match_set_point
725 INSERT INTO _vandelay_tmp_qrows (q) VALUES (node.id);
729 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_jrow(
730 node vandelay.match_set_point,
745 -- remember $1 is tags_rstore, and $2 is svf_rstore
746 -- a non-NULL auth_heading means we're matching authority records
748 IF auth_heading IS NOT NULL THEN
749 rec_table := 'authority.full_rec';
751 rec_table := 'metabib.full_rec';
755 SELECT COUNT(*) INTO jrow_count FROM _vandelay_tmp_jrows;
756 IF jrow_count > 0 THEN
757 my_using := ' USING (record)';
758 my_join := 'FULL OUTER JOIN';
764 IF node.tag IS NOT NULL THEN
765 caseless := (node.tag IN ('020', '022', '024'));
767 IF node.subfield IS NOT NULL THEN
768 tagkey := tagkey || node.subfield;
786 my_alias := 'n' || node.id::TEXT;
788 jrow := my_join || ' (SELECT *, ';
789 IF node.tag IS NOT NULL THEN
790 jrow := jrow || node.quality ||
791 ' AS quality FROM ' || rec_table || ' mfr WHERE mfr.tag = ''' ||
793 IF node.subfield IS NOT NULL THEN
794 jrow := jrow || ' AND mfr.subfield = ''' ||
795 node.subfield || '''';
797 jrow := jrow || ' AND (';
798 jrow := jrow || vandelay._node_tag_comparisons(caseless, op, tags_rstore, tagkey);
799 jrow := jrow || ')) ' || my_alias || my_using || E'\n';
801 IF auth_heading IS NOT NULL THEN -- authority record
802 IF node.heading AND auth_heading <> '' THEN
803 jrow := jrow || 'id AS record, ' || node.quality ||
804 ' AS quality FROM authority.record_entry are ' ||
805 ' WHERE are.heading = ''' || auth_heading || '''';
806 jrow := jrow || ') ' || my_alias || my_using || E'\n';
809 jrow := jrow || 'id AS record, ' || node.quality ||
810 ' AS quality FROM metabib.record_attr_flat mraf WHERE mraf.attr = ''' ||
811 node.svf || ''' AND mraf.value ' || op || ' $2->''' || node.svf || ''') ' ||
812 my_alias || my_using || E'\n';
815 INSERT INTO _vandelay_tmp_jrows (j) VALUES (jrow);
819 CREATE OR REPLACE FUNCTION vandelay._node_tag_comparisons(
831 vals := tags_rstore->tagkey;
836 IF vals[i] IS NULL THEN
839 result := result || ' OR ';
844 result := result || 'LOWER(mfr.value) ' || op;
846 result := result || 'mfr.value ' || op;
849 result := result || ' ' || COALESCE('''' || vals[i] || '''', 'NULL');
851 IF vals[i] IS NULL THEN
862 CREATE OR REPLACE FUNCTION vandelay._get_expr_render_one(
863 node vandelay.match_set_point
868 IF node.bool_op IS NOT NULL THEN
871 RETURN '(n' || node.id::TEXT || '.id IS NOT NULL)';
876 CREATE OR REPLACE FUNCTION vandelay.match_bib_record() RETURNS TRIGGER AS $func$
878 incoming_existing_id TEXT;
879 test_result vandelay.match_set_test_result%ROWTYPE;
884 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
888 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
890 SELECT q.match_set INTO match_set FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
892 IF match_set IS NOT NULL THEN
893 NEW.quality := vandelay.measure_record_quality( NEW.marc, match_set );
896 -- Perfect matches on 901$c exit early with a match with high quality.
897 incoming_existing_id :=
898 oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]', NEW.marc);
900 IF incoming_existing_id IS NOT NULL AND incoming_existing_id != '' THEN
901 SELECT id INTO tmp_rec FROM biblio.record_entry WHERE id = incoming_existing_id::bigint;
902 IF tmp_rec IS NOT NULL THEN
903 INSERT INTO vandelay.bib_match (queued_record, eg_record, match_score, quality)
908 -- note: no match_set means quality==0
909 vandelay.measure_record_quality( b.marc, match_set )
910 FROM biblio.record_entry b
911 WHERE id = incoming_existing_id::bigint;
915 IF match_set IS NULL THEN
919 SELECT q.match_bucket INTO match_bucket FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
921 FOR test_result IN SELECT * FROM
922 vandelay.match_set_test_marcxml(match_set, NEW.marc, match_bucket) LOOP
924 INSERT INTO vandelay.bib_match ( queued_record, eg_record, match_score, quality )
929 vandelay.measure_record_quality( b.marc, match_set )
930 FROM biblio.record_entry b
931 WHERE id = test_result.record;
937 $func$ LANGUAGE PLPGSQL;
939 CREATE OR REPLACE FUNCTION vandelay.measure_record_quality ( xml TEXT, match_set_id INT ) RETURNS INT AS $_$
943 test vandelay.match_set_quality%ROWTYPE;
946 FOR test IN SELECT * FROM vandelay.match_set_quality WHERE match_set = match_set_id LOOP
947 IF test.tag IS NOT NULL THEN
948 FOR rvalue IN SELECT value FROM vandelay.flatten_marc( xml ) WHERE tag = test.tag AND subfield = test.subfield LOOP
949 IF test.value = rvalue THEN
950 out_q := out_q + test.quality;
954 IF test.value = vandelay.extract_rec_attrs(xml, ARRAY[test.svf]) -> test.svf THEN
955 out_q := out_q + test.quality;
962 $_$ LANGUAGE PLPGSQL;
964 CREATE TYPE vandelay.tcn_data AS (tcn TEXT, tcn_source TEXT, used BOOL);
965 CREATE OR REPLACE FUNCTION vandelay.find_bib_tcn_data ( xml TEXT ) RETURNS SETOF vandelay.tcn_data AS $_$
969 output vandelay.tcn_data%ROWTYPE;
973 eg_tcn := BTRIM((oils_xpath('//*[@tag="001"]/text()',xml))[1]);
974 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
976 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="003"]/text()',xml))[1]);
977 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
978 eg_tcn_source := 'System Local';
981 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
984 output.used := FALSE;
989 output.tcn := eg_tcn;
990 output.tcn_source := eg_tcn_source;
996 eg_tcn := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="a"]/text()',xml))[1]);
997 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
999 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="b"]/text()',xml))[1]);
1000 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
1001 eg_tcn_source := 'System Local';
1004 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1007 output.used := FALSE;
1009 output.used := TRUE;
1012 output.tcn := eg_tcn;
1013 output.tcn_source := eg_tcn_source;
1019 eg_tcn := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="a"]/text()',xml))[1]);
1020 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
1022 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="b"]/text()',xml))[1]);
1023 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
1024 eg_tcn_source := 'System Local';
1027 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1030 output.used := FALSE;
1032 output.used := TRUE;
1035 output.tcn := eg_tcn;
1036 output.tcn_source := eg_tcn_source;
1042 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="020"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
1043 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
1045 eg_tcn_source := 'ISBN';
1047 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1050 output.used := FALSE;
1052 output.used := TRUE;
1055 output.tcn := eg_tcn;
1056 output.tcn_source := eg_tcn_source;
1062 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="022"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
1063 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
1065 eg_tcn_source := 'ISSN';
1067 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1070 output.used := FALSE;
1072 output.used := TRUE;
1075 output.tcn := eg_tcn;
1076 output.tcn_source := eg_tcn_source;
1082 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="010"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
1083 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
1085 eg_tcn_source := 'LCCN';
1087 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1090 output.used := FALSE;
1092 output.used := TRUE;
1095 output.tcn := eg_tcn;
1096 output.tcn_source := eg_tcn_source;
1102 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="035"]/*[@code="a"]/text()',xml))[1], $re$^.*?(\w+)$$re$, $re$\1$re$);
1103 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
1105 eg_tcn_source := 'System Legacy';
1107 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1110 output.used := FALSE;
1112 output.used := TRUE;
1115 output.tcn := eg_tcn;
1116 output.tcn_source := eg_tcn_source;
1123 $_$ LANGUAGE PLPGSQL;
1125 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT, force_add INT ) RETURNS TEXT AS $_$
1128 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1132 MARC::Charset->assume_unicode(1);
1134 my $target_xml = shift;
1135 my $source_xml = shift;
1136 my $field_spec = shift;
1137 my $force_add = shift || 0;
1139 my $target_r = MARC::Record->new_from_xml( $target_xml );
1140 my $source_r = MARC::Record->new_from_xml( $source_xml );
1142 return $target_xml unless ($target_r && $source_r);
1144 my @field_list = split(',', $field_spec);
1147 for my $f (@field_list) {
1148 $f =~ s/^\s*//; $f =~ s/\s*$//;
1149 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1155 $match =~ s/^\s*//; $match =~ s/\s*$//;
1156 $fields{$field} = { sf => [ split('', $sf) ] };
1158 my ($msf,$mre) = split('~', $match);
1159 if (length($msf) > 0 and length($mre) > 0) {
1160 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1161 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1162 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1168 for my $f ( keys %fields) {
1169 if ( @{$fields{$f}{sf}} ) {
1170 for my $from_field ($source_r->field( $f )) {
1171 my @tos = $target_r->field( $f );
1173 next if (exists($fields{$f}{match}) and !$force_add);
1174 my @new_fields = map { $_->clone } $source_r->field( $f );
1175 $target_r->insert_fields_ordered( @new_fields );
1177 for my $to_field (@tos) {
1178 if (exists($fields{$f}{match})) {
1179 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1181 for my $old_sf ($from_field->subfields) {
1182 $to_field->add_subfields( @$old_sf ) if grep(/$$old_sf[0]/,@{$fields{$f}{sf}});
1188 my @new_fields = map { $_->clone } $source_r->field( $f );
1189 $target_r->insert_fields_ordered( @new_fields );
1193 $target_xml = $target_r->as_xml_record;
1194 $target_xml =~ s/^<\?.+?\?>$//mo;
1195 $target_xml =~ s/\n//sgo;
1196 $target_xml =~ s/>\s+</></sgo;
1200 $_$ LANGUAGE PLPERLU;
1202 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1203 SELECT vandelay.add_field( $1, $2, $3, 0 );
1206 CREATE OR REPLACE FUNCTION vandelay.strip_field ( xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1209 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1213 MARC::Charset->assume_unicode(1);
1216 my $r = MARC::Record->new_from_xml( $xml );
1218 return $xml unless ($r);
1220 my $field_spec = shift;
1221 my @field_list = split(',', $field_spec);
1224 for my $f (@field_list) {
1225 $f =~ s/^\s*//; $f =~ s/\s*$//;
1226 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1232 $match =~ s/^\s*//; $match =~ s/\s*$//;
1233 $fields{$field} = { sf => [ split('', $sf) ] };
1235 my ($msf,$mre) = split('~', $match);
1236 if (length($msf) > 0 and length($mre) > 0) {
1237 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1238 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1239 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1245 for my $f ( keys %fields) {
1246 for my $to_field ($r->field( $f )) {
1247 if (exists($fields{$f}{match})) {
1248 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1251 if ( @{$fields{$f}{sf}} ) {
1252 $to_field->delete_subfield(code => $fields{$f}{sf});
1254 $r->delete_field( $to_field );
1259 $xml = $r->as_xml_record;
1260 $xml =~ s/^<\?.+?\?>$//mo;
1262 $xml =~ s/>\s+</></sgo;
1266 $_$ LANGUAGE PLPERLU;
1268 CREATE OR REPLACE FUNCTION vandelay.replace_field
1269 (target_xml TEXT, source_xml TEXT, field TEXT) RETURNS TEXT AS $_$
1274 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1277 MARC::Charset->assume_unicode(1);
1279 my $target_xml = shift;
1280 my $source_xml = shift;
1281 my $field_spec = shift;
1283 my $target_r = MARC::Record->new_from_xml($target_xml);
1284 my $source_r = MARC::Record->new_from_xml($source_xml);
1286 return $target_xml unless $target_r && $source_r;
1288 # Extract the field_spec components into MARC tags, subfields,
1289 # and regex matches. Copied wholesale from vandelay.strip_field()
1291 my @field_list = split(',', $field_spec);
1293 for my $f (@field_list) {
1294 $f =~ s/^\s*//; $f =~ s/\s*$//;
1295 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1301 $match =~ s/^\s*//; $match =~ s/\s*$//;
1302 $fields{$field} = { sf => [ split('', $sf) ] };
1304 my ($msf,$mre) = split('~', $match);
1305 if (length($msf) > 0 and length($mre) > 0) {
1306 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1307 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1308 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1314 # Returns a flat list of subfield (code, value, code, value, ...)
1315 # suitable for adding to a MARC::Field.
1316 sub generate_replacement_subfields {
1317 my ($source_field, $target_field, @controlled_subfields) = @_;
1319 # Performing a wholesale field replacment.
1320 # Use the entire source field as-is.
1321 return map {$_->[0], $_->[1]} $source_field->subfields
1322 unless @controlled_subfields;
1326 # Iterate over all target field subfields:
1327 # 1. Keep uncontrolled subfields as is.
1328 # 2. Replace values for controlled subfields when a
1329 # replacement value exists on the source record.
1330 # 3. Delete values for controlled subfields when no
1331 # replacement value exists on the source record.
1333 for my $target_sf ($target_field->subfields) {
1334 my $subfield = $target_sf->[0];
1335 my $target_val = $target_sf->[1];
1337 if (grep {$_ eq $subfield} @controlled_subfields) {
1338 if (my $source_val = $source_field->subfield($subfield)) {
1339 # We have a replacement value
1340 push(@new_subfields, $subfield, $source_val);
1342 # no replacement value for controlled subfield, drop it.
1345 # Field is not controlled. Copy it over as-is.
1346 push(@new_subfields, $subfield, $target_val);
1350 # Iterate over all subfields in the source field and back-fill
1351 # any values that exist only in the source field. Insert these
1352 # subfields in the same relative position they exist in the
1356 for my $source_sf ($source_field->subfields) {
1357 my $subfield = $source_sf->[0];
1358 my $source_val = $source_sf->[1];
1359 push(@seen_subfields, $subfield);
1361 # target field already contains this subfield,
1362 # so it would have been addressed above.
1363 next if $target_field->subfield($subfield);
1365 # Ignore uncontrolled subfields.
1366 next unless grep {$_ eq $subfield} @controlled_subfields;
1368 # Adding a new subfield. Find its relative position and add
1369 # it to the list under construction. Work backwards from
1370 # the list of already seen subfields to find the best slot.
1373 for my $seen_sf (reverse(@seen_subfields)) {
1374 my $idx = @new_subfields;
1375 for my $new_sf (reverse(@new_subfields)) {
1377 next if $idx % 2 == 1; # sf codes are in the even slots
1379 if ($new_subfields[$idx] eq $seen_sf) {
1380 splice(@new_subfields, $idx + 2, 0, $subfield, $source_val);
1388 # if no slot was found, add to the end of the list.
1389 push(@new_subfields, $subfield, $source_val) unless $done;
1392 return @new_subfields;
1396 for my $f (keys %fields) {
1398 for my $target_field ($target_r->field($f)) {
1400 # field spec contains a regex for this field. Confirm field on
1401 # target record matches the specified regex before replacing.
1402 if (exists($fields{$f}{match})) {
1403 next unless (grep { $_ =~ $fields{$f}{match}{re} }
1404 $target_field->subfield($fields{$f}{match}{sf}));
1408 my @controlled_subfields = @{$fields{$f}{sf}};
1410 # If the target record has multiple matching bib fields,
1411 # replace them from matching fields on the source record
1412 # in a predictable order to avoid replacing with them with
1413 # same source field repeatedly.
1414 my @source_fields = $source_r->field($f);
1415 my $source_field = $source_fields[++$tag_idx];
1417 if (!$source_field && @controlled_subfields) {
1418 # When there are more target fields than source fields
1419 # and we are replacing values for subfields and not
1420 # performing wholesale field replacment, use the last
1421 # available source field as the input for all remaining
1423 $source_field = $source_fields[$#source_fields];
1426 if (!$source_field) {
1427 # No source field exists. Delete all affected target
1428 # data. This is a little bit counterintuitive, but is
1429 # backwards compatible with the previous version of this
1430 # function which first deleted all affected data, then
1431 # replaced values where possible.
1432 if (@controlled_subfields) {
1433 $target_field->delete_subfield($_) for @controlled_subfields;
1435 $target_r->delete_field($target_field);
1440 my @new_subfields = generate_replacement_subfields(
1441 $source_field, $target_field, @controlled_subfields);
1443 # Build the replacement field from scratch.
1444 my $replacement_field = MARC::Field->new(
1446 $target_field->indicator(1),
1447 $target_field->indicator(2),
1451 $target_field->replace_with($replacement_field);
1455 $target_xml = $target_r->as_xml_record;
1456 $target_xml =~ s/^<\?.+?\?>$//mo;
1457 $target_xml =~ s/\n//sgo;
1458 $target_xml =~ s/>\s+</></sgo;
1462 $_$ LANGUAGE PLPERLU;
1464 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 $_$
1465 SELECT vandelay.replace_field( vandelay.add_field( vandelay.strip_field( $1, $5) , $2, $3 ), $2, $4);
1468 CREATE TYPE vandelay.compile_profile AS (add_rule TEXT, replace_rule TEXT, preserve_rule TEXT, strip_rule TEXT);
1469 CREATE OR REPLACE FUNCTION vandelay.compile_profile ( incoming_xml TEXT ) RETURNS vandelay.compile_profile AS $_$
1471 output vandelay.compile_profile%ROWTYPE;
1472 profile vandelay.merge_profile%ROWTYPE;
1474 profile_tmpl_owner TEXT;
1475 add_rule TEXT := '';
1476 strip_rule TEXT := '';
1477 replace_rule TEXT := '';
1478 preserve_rule TEXT := '';
1482 profile_tmpl := (oils_xpath('//*[@tag="905"]/*[@code="t"]/text()',incoming_xml))[1];
1483 profile_tmpl_owner := (oils_xpath('//*[@tag="905"]/*[@code="o"]/text()',incoming_xml))[1];
1485 IF profile_tmpl IS NOT NULL AND profile_tmpl <> '' AND profile_tmpl_owner IS NOT NULL AND profile_tmpl_owner <> '' THEN
1486 SELECT p.* INTO profile
1487 FROM vandelay.merge_profile p
1488 JOIN actor.org_unit u ON (u.id = p.owner)
1489 WHERE p.name = profile_tmpl
1490 AND u.shortname = profile_tmpl_owner;
1492 IF profile.id IS NOT NULL THEN
1493 add_rule := COALESCE(profile.add_spec,'');
1494 strip_rule := COALESCE(profile.strip_spec,'');
1495 replace_rule := COALESCE(profile.replace_spec,'');
1496 preserve_rule := COALESCE(profile.preserve_spec,'');
1500 add_rule := add_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="a"]/text()',incoming_xml),','),'');
1501 strip_rule := strip_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="d"]/text()',incoming_xml),','),'');
1502 replace_rule := replace_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="r"]/text()',incoming_xml),','),'');
1503 preserve_rule := preserve_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="p"]/text()',incoming_xml),','),'');
1505 output.add_rule := BTRIM(add_rule,',');
1506 output.replace_rule := BTRIM(replace_rule,',');
1507 output.strip_rule := BTRIM(strip_rule,',');
1508 output.preserve_rule := BTRIM(preserve_rule,',');
1512 $_$ LANGUAGE PLPGSQL;
1514 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1516 merge_profile vandelay.merge_profile%ROWTYPE;
1517 dyn_profile vandelay.compile_profile%ROWTYPE;
1527 SELECT b.marc INTO eg_marc
1528 FROM biblio.record_entry b
1532 IF eg_marc IS NULL OR v_marc IS NULL THEN
1533 -- RAISE NOTICE 'no marc for template or bib record';
1537 dyn_profile := vandelay.compile_profile( v_marc );
1539 IF merge_profile_id IS NOT NULL THEN
1540 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1542 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1543 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1544 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1545 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1549 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1550 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1554 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1555 --Since we have nothing to do, just return a NOOP "we did it"
1557 ELSIF dyn_profile.replace_rule <> '' THEN
1558 source_marc = v_marc;
1559 target_marc = eg_marc;
1560 replace_rule = dyn_profile.replace_rule;
1562 source_marc = eg_marc;
1563 target_marc = v_marc;
1564 replace_rule = dyn_profile.preserve_rule;
1567 UPDATE biblio.record_entry
1568 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1572 -- RAISE NOTICE 'update of biblio.record_entry failed';
1579 $$ LANGUAGE PLPGSQL;
1581 CREATE OR REPLACE FUNCTION vandelay.merge_record_xml ( target_marc TEXT, template_marc TEXT ) RETURNS TEXT AS $$
1583 dyn_profile vandelay.compile_profile%ROWTYPE;
1591 IF target_marc IS NULL OR template_marc IS NULL THEN
1592 -- RAISE NOTICE 'no marc for target or template record';
1596 dyn_profile := vandelay.compile_profile( template_marc );
1598 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1599 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1603 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1604 --Since we have nothing to do, just return what we were given.
1606 ELSIF dyn_profile.replace_rule <> '' THEN
1607 trgt_marc = target_marc;
1608 tmpl_marc = template_marc;
1609 replace_rule = dyn_profile.replace_rule;
1611 tmp_marc = target_marc;
1612 trgt_marc = template_marc;
1613 tmpl_marc = tmp_marc;
1614 replace_rule = dyn_profile.preserve_rule;
1617 RETURN vandelay.merge_record_xml( trgt_marc, tmpl_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule );
1620 $$ LANGUAGE PLPGSQL;
1622 CREATE OR REPLACE FUNCTION vandelay.merge_record_xml_using_profile ( incoming_marc TEXT, existing_marc TEXT, merge_profile_id BIGINT ) RETURNS TEXT AS $$
1624 merge_profile vandelay.merge_profile%ROWTYPE;
1625 dyn_profile vandelay.compile_profile%ROWTYPE;
1632 IF existing_marc IS NULL OR incoming_marc IS NULL THEN
1633 -- RAISE NOTICE 'no marc for source or target records';
1637 IF merge_profile_id IS NOT NULL THEN
1638 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1640 dyn_profile.add_rule := COALESCE(merge_profile.add_spec,'');
1641 dyn_profile.strip_rule := COALESCE(merge_profile.strip_spec,'');
1642 dyn_profile.replace_rule := COALESCE(merge_profile.replace_spec,'');
1643 dyn_profile.preserve_rule := COALESCE(merge_profile.preserve_spec,'');
1645 -- RAISE NOTICE 'merge profile not found';
1649 -- RAISE NOTICE 'no merge profile specified';
1653 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1654 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1658 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1659 -- Since we have nothing to do, just return a target record as is
1660 RETURN existing_marc;
1661 ELSIF dyn_profile.preserve_rule <> '' THEN
1662 source_marc = existing_marc;
1663 target_marc = incoming_marc;
1664 replace_rule = dyn_profile.preserve_rule;
1666 source_marc = incoming_marc;
1667 target_marc = existing_marc;
1668 replace_rule = dyn_profile.replace_rule;
1671 RETURN vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule );
1674 $$ LANGUAGE PLPGSQL;
1676 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT) RETURNS BOOL AS $$
1677 SELECT vandelay.template_overlay_bib_record( $1, $2, NULL);
1680 CREATE OR REPLACE FUNCTION vandelay.overlay_bib_record
1681 ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1687 update_fields TEXT[];
1689 update_bib_source BOOL;
1690 update_bib_editor BOOL;
1693 SELECT q.marc, q.bib_source INTO v_marc, v_bib_source
1694 FROM vandelay.queued_bib_record q
1695 JOIN vandelay.bib_match m ON (m.queued_record = q.id AND q.id = import_id)
1698 IF v_marc IS NULL THEN
1699 -- RAISE NOTICE 'no marc for vandelay or bib record';
1703 IF NOT vandelay.template_overlay_bib_record( v_marc, eg_id, merge_profile_id) THEN
1704 -- no update happened, get outta here.
1708 UPDATE vandelay.queued_bib_record
1709 SET imported_as = eg_id,
1711 WHERE id = import_id;
1713 SELECT q.update_bib_source INTO update_bib_source
1714 FROM vandelay.merge_profile q where q.id = merge_profile_Id;
1716 IF update_bib_source AND v_bib_source IS NOT NULL THEN
1717 update_fields := ARRAY_APPEND(update_fields, 'source = ' || v_bib_source);
1720 SELECT q.update_bib_editor INTO update_bib_editor
1721 FROM vandelay.merge_profile q where q.id = merge_profile_Id;
1723 IF update_bib_editor THEN
1725 editor_string := (oils_xpath('//*[@tag="905"]/*[@code="u"]/text()',v_marc))[1];
1727 IF editor_string IS NOT NULL AND editor_string <> '' THEN
1728 SELECT usr INTO editor_id FROM actor.card WHERE barcode = editor_string;
1730 IF editor_id IS NULL THEN
1731 SELECT id INTO editor_id FROM actor.usr WHERE usrname = editor_string;
1734 IF editor_id IS NOT NULL THEN
1735 --only update the edit date if we have a valid editor
1736 update_fields := ARRAY_APPEND(
1737 update_fields, 'editor = ' || editor_id || ', edit_date = NOW()');
1742 IF ARRAY_LENGTH(update_fields, 1) > 0 THEN
1743 update_query := 'UPDATE biblio.record_entry SET ' ||
1744 ARRAY_TO_STRING(update_fields, ',') || ' WHERE id = ' || eg_id || ';';
1745 EXECUTE update_query;
1750 $$ LANGUAGE PLPGSQL;
1753 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
1756 lwm_ratio_value NUMERIC;
1759 lwm_ratio_value := COALESCE(lwm_ratio_value_p, 0.0);
1761 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1764 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1768 SELECT m.eg_record INTO eg_id
1769 FROM vandelay.bib_match m
1770 JOIN vandelay.queued_bib_record qr ON (m.queued_record = qr.id)
1771 JOIN vandelay.bib_queue q ON (qr.queue = q.id)
1772 JOIN biblio.record_entry r ON (r.id = m.eg_record)
1773 WHERE m.queued_record = import_id
1774 AND qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC >= lwm_ratio_value
1775 ORDER BY m.match_score DESC, -- required match score
1776 qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC DESC, -- quality tie breaker
1777 m.id -- when in doubt, use the first match
1780 IF eg_id IS NULL THEN
1781 -- RAISE NOTICE 'incoming record is not of high enough quality';
1785 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1787 $$ LANGUAGE PLPGSQL;
1789 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1790 SELECT vandelay.auto_overlay_bib_record_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1793 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1799 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1802 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1806 SELECT COUNT(*) INTO match_count FROM vandelay.bib_match WHERE queued_record = import_id;
1808 IF match_count <> 1 THEN
1809 -- RAISE NOTICE 'not an exact match';
1813 -- Check that the one match is on the first 901c
1814 SELECT m.eg_record INTO eg_id
1815 FROM vandelay.queued_bib_record q
1816 JOIN vandelay.bib_match m ON (m.queued_record = q.id)
1817 WHERE q.id = import_id
1818 AND m.eg_record = oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]',marc)::BIGINT;
1821 -- RAISE NOTICE 'not a 901c match';
1825 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1827 $$ LANGUAGE PLPGSQL;
1829 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1831 queued_record vandelay.queued_bib_record%ROWTYPE;
1834 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1836 IF vandelay.auto_overlay_bib_record( queued_record.id, merge_profile_id ) THEN
1837 RETURN NEXT queued_record.id;
1845 $$ LANGUAGE PLPGSQL;
1847 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 $$
1849 queued_record vandelay.queued_bib_record%ROWTYPE;
1852 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1854 IF vandelay.auto_overlay_bib_record_with_best( queued_record.id, merge_profile_id, lwm_ratio_value ) THEN
1855 RETURN NEXT queued_record.id;
1863 $$ LANGUAGE PLPGSQL;
1865 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1866 SELECT vandelay.auto_overlay_bib_queue_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1869 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1870 SELECT * FROM vandelay.auto_overlay_bib_queue( $1, NULL );
1873 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_org_unit_copies ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
1877 rec vandelay.bib_match%ROWTYPE;
1881 copy_count INT := 0;
1882 max_copy_count INT := 0;
1885 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1888 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1892 -- Gather all the owning libs for our import items.
1893 -- These are our initial scope_orgs.
1894 SELECT ARRAY_AGG(DISTINCT owning_lib) INTO scope_orgs
1895 FROM vandelay.import_item
1896 WHERE record = import_id;
1898 WHILE CARDINALITY(scope_orgs) > 0 LOOP
1899 FOR scope_org IN SELECT * FROM UNNEST(scope_orgs) LOOP
1900 -- For each match, get a count of all copies at descendants of our scope org.
1901 FOR rec IN SELECT * FROM vandelay.bib_match AS vbm
1902 WHERE queued_record = import_id
1903 ORDER BY vbm.eg_record DESC
1905 SELECT COUNT(acp.id) INTO copy_count
1906 FROM asset.copy AS acp
1907 INNER JOIN asset.call_number AS acn
1908 ON acp.call_number = acn.id
1909 WHERE acn.owning_lib IN (SELECT id FROM
1910 actor.org_unit_descendants(scope_org))
1911 AND acn.record = rec.eg_record
1912 AND acp.deleted = FALSE;
1913 IF copy_count > max_copy_count THEN
1914 max_copy_count := copy_count;
1915 eg_id := rec.eg_record;
1920 -- If no matching bibs had holdings, gather our next set of orgs to check, and iterate.
1921 IF max_copy_count = 0 THEN
1922 SELECT ARRAY_AGG(DISTINCT parent_ou) INTO scope_orgs
1924 WHERE id IN (SELECT * FROM UNNEST(scope_orgs))
1925 AND parent_ou IS NOT NULL;
1929 IF eg_id IS NULL THEN
1930 -- Could not determine best match via copy count
1931 -- fall back to default best match
1932 IF (SELECT * FROM vandelay.auto_overlay_bib_record_with_best( import_id, merge_profile_id, lwm_ratio_value_p )) THEN
1939 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1941 $$ LANGUAGE PLPGSQL;
1943 CREATE OR REPLACE FUNCTION vandelay.ingest_bib_marc ( ) RETURNS TRIGGER AS $$
1949 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1953 FOR adef IN SELECT * FROM vandelay.bib_attr_definition LOOP
1955 SELECT extract_marc_field('vandelay.queued_bib_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_bib_record WHERE id = NEW.id;
1956 IF (value IS NOT NULL AND value <> '') THEN
1957 INSERT INTO vandelay.queued_bib_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1964 $$ LANGUAGE PLPGSQL;
1966 CREATE OR REPLACE FUNCTION vandelay.cleanup_bib_marc ( ) RETURNS TRIGGER AS $$
1968 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1972 DELETE FROM vandelay.queued_bib_record_attr WHERE record = OLD.id;
1973 DELETE FROM vandelay.import_item WHERE record = OLD.id;
1975 IF TG_OP = 'UPDATE' THEN
1980 $$ LANGUAGE PLPGSQL;
1982 CREATE TRIGGER cleanup_bib_trigger
1983 BEFORE UPDATE OR DELETE ON vandelay.queued_bib_record
1984 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_bib_marc();
1986 CREATE TRIGGER ingest_bib_trigger
1987 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1988 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_bib_marc();
1990 CREATE TRIGGER zz_match_bibs_trigger
1991 BEFORE INSERT OR UPDATE ON vandelay.queued_bib_record
1992 FOR EACH ROW EXECUTE PROCEDURE vandelay.match_bib_record();
1995 /* Authority stuff down here */
1996 ---------------------------------------
1997 CREATE TABLE vandelay.authority_attr_definition (
1998 id SERIAL PRIMARY KEY,
1999 code TEXT UNIQUE NOT NULL,
2001 xpath TEXT NOT NULL,
2002 remove TEXT NOT NULL DEFAULT ''
2005 CREATE TYPE vandelay.authority_queue_queue_type AS ENUM ('authority');
2006 CREATE TABLE vandelay.authority_queue (
2007 queue_type vandelay.authority_queue_queue_type NOT NULL DEFAULT 'authority',
2008 CONSTRAINT vand_authority_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
2009 ) INHERITS (vandelay.queue);
2010 ALTER TABLE vandelay.authority_queue ADD PRIMARY KEY (id);
2012 CREATE TABLE vandelay.queued_authority_record (
2013 queue INT NOT NULL REFERENCES vandelay.authority_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
2014 imported_as INT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
2015 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
2017 ) INHERITS (vandelay.queued_record);
2018 ALTER TABLE vandelay.queued_authority_record ADD PRIMARY KEY (id);
2019 CREATE INDEX queued_authority_record_queue_idx ON vandelay.queued_authority_record (queue);
2021 CREATE TABLE vandelay.queued_authority_record_attr (
2022 id BIGSERIAL PRIMARY KEY,
2023 record BIGINT NOT NULL REFERENCES vandelay.queued_authority_record (id) DEFERRABLE INITIALLY DEFERRED,
2024 field INT NOT NULL REFERENCES vandelay.authority_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
2025 attr_value TEXT NOT NULL
2027 CREATE INDEX queued_authority_record_attr_record_idx ON vandelay.queued_authority_record_attr (record);
2029 CREATE TABLE vandelay.authority_match (
2030 id BIGSERIAL PRIMARY KEY,
2031 queued_record BIGINT REFERENCES vandelay.queued_authority_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
2032 eg_record BIGINT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
2033 quality INT NOT NULL DEFAULT 0,
2034 match_score INT NOT NULL DEFAULT 0
2037 CREATE OR REPLACE FUNCTION vandelay.ingest_authority_marc ( ) RETURNS TRIGGER AS $$
2043 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
2047 FOR adef IN SELECT * FROM vandelay.authority_attr_definition LOOP
2049 SELECT extract_marc_field('vandelay.queued_authority_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_authority_record WHERE id = NEW.id;
2050 IF (value IS NOT NULL AND value <> '') THEN
2051 INSERT INTO vandelay.queued_authority_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
2058 $$ LANGUAGE PLPGSQL;
2060 CREATE OR REPLACE FUNCTION vandelay.cleanup_authority_marc ( ) RETURNS TRIGGER AS $$
2062 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
2066 DELETE FROM vandelay.queued_authority_record_attr WHERE record = OLD.id;
2067 IF TG_OP = 'UPDATE' THEN
2072 $$ LANGUAGE PLPGSQL;
2074 CREATE TRIGGER cleanup_authority_trigger
2075 BEFORE UPDATE OR DELETE ON vandelay.queued_authority_record
2076 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_authority_marc();
2078 CREATE TRIGGER ingest_authority_trigger
2079 AFTER INSERT OR UPDATE ON vandelay.queued_authority_record
2080 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_authority_marc();
2082 CREATE OR REPLACE FUNCTION vandelay.overlay_authority_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
2084 merge_profile vandelay.merge_profile%ROWTYPE;
2085 dyn_profile vandelay.compile_profile%ROWTYPE;
2088 new_edit_date TIMESTAMPTZ;
2091 eg_marc_row authority.record_entry%ROWTYPE;
2099 SELECT * INTO eg_marc_row
2100 FROM authority.record_entry b
2101 JOIN vandelay.authority_match m ON (m.eg_record = b.id AND m.queued_record = import_id)
2104 SELECT q.marc INTO v_marc
2105 FROM vandelay.queued_record q
2106 JOIN vandelay.authority_match m ON (m.queued_record = q.id AND q.id = import_id)
2109 eg_marc := eg_marc_row.marc;
2111 IF eg_marc IS NULL OR v_marc IS NULL THEN
2112 -- RAISE NOTICE 'no marc for vandelay or authority record';
2116 -- Extract the editor string before any modification to the vandelay
2119 (oils_xpath('//*[@tag="905"]/*[@code="u"]/text()',v_marc))[1];
2121 -- If an editor value can be found, update the authority record
2122 -- editor and edit_date values.
2123 IF editor_string IS NOT NULL AND editor_string <> '' THEN
2125 -- Vandelay.pm sets the value to 'usrname' when needed.
2126 SELECT id INTO new_editor
2127 FROM actor.usr WHERE usrname = editor_string;
2129 IF new_editor IS NULL THEN
2130 SELECT usr INTO new_editor
2131 FROM actor.card WHERE barcode = editor_string;
2134 IF new_editor IS NOT NULL THEN
2135 new_edit_date := NOW();
2136 ELSE -- No valid editor, use current values
2137 new_editor = eg_marc_row.editor;
2138 new_edit_date = eg_marc_row.edit_date;
2141 new_editor = eg_marc_row.editor;
2142 new_edit_date = eg_marc_row.edit_date;
2145 dyn_profile := vandelay.compile_profile( v_marc );
2147 IF merge_profile_id IS NOT NULL THEN
2148 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
2150 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
2151 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
2152 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
2153 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
2157 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
2158 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
2162 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
2163 --Since we have nothing to do, just return a NOOP "we did it"
2165 ELSIF dyn_profile.replace_rule <> '' THEN
2166 source_marc = v_marc;
2167 target_marc = eg_marc;
2168 replace_rule = dyn_profile.replace_rule;
2170 source_marc = eg_marc;
2171 target_marc = v_marc;
2172 replace_rule = dyn_profile.preserve_rule;
2175 UPDATE authority.record_entry
2176 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule ),
2177 editor = new_editor,
2178 edit_date = new_edit_date
2182 -- Import/merge failed. Nothing left to do.
2186 -- Authority record successfully merged / imported.
2188 -- Update the vandelay record to show the successful import.
2189 UPDATE vandelay.queued_authority_record
2190 SET imported_as = eg_id,
2192 WHERE id = import_id;
2197 $$ LANGUAGE PLPGSQL;
2199 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
2204 SELECT COUNT(*) INTO match_count FROM vandelay.authority_match WHERE queued_record = import_id;
2206 IF match_count <> 1 THEN
2207 -- RAISE NOTICE 'not an exact match';
2211 SELECT m.eg_record INTO eg_id
2212 FROM vandelay.authority_match m
2213 WHERE m.queued_record = import_id
2216 IF eg_id IS NULL THEN
2220 RETURN vandelay.overlay_authority_record( import_id, eg_id, merge_profile_id );
2222 $$ LANGUAGE PLPGSQL;
2224 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
2226 queued_record vandelay.queued_authority_record%ROWTYPE;
2229 FOR queued_record IN SELECT * FROM vandelay.queued_authority_record WHERE queue = queue_id AND import_time IS NULL LOOP
2231 IF vandelay.auto_overlay_authority_record( queued_record.id, merge_profile_id ) THEN
2232 RETURN NEXT queued_record.id;
2240 $$ LANGUAGE PLPGSQL;
2242 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
2243 SELECT * FROM vandelay.auto_overlay_authority_queue( $1, NULL );
2246 CREATE OR REPLACE FUNCTION vandelay.match_set_test_authxml(
2247 match_set_id INTEGER, record_xml TEXT
2248 ) RETURNS SETOF vandelay.match_set_test_result AS $$
2259 tags_rstore := vandelay.flatten_marc_hstore(record_xml);
2261 SELECT normalize_heading INTO heading
2262 FROM authority.normalize_heading(record_xml);
2264 CREATE TEMPORARY TABLE _vandelay_tmp_qrows (q INTEGER);
2265 CREATE TEMPORARY TABLE _vandelay_tmp_jrows (j TEXT);
2267 -- generate the where clause and return that directly (into wq), and as
2268 -- a side-effect, populate the _vandelay_tmp_[qj]rows tables.
2269 wq := vandelay.get_expr_from_match_set(
2270 match_set_id, tags_rstore, heading);
2272 query_ := 'SELECT DISTINCT(record), ';
2274 -- qrows table is for the quality bits we add to the SELECT clause
2276 'COALESCE(n' || q::TEXT || '.quality, 0)', ' + '
2277 ) INTO coal FROM _vandelay_tmp_qrows;
2279 -- our query string so far is the SELECT clause and the inital FROM.
2280 -- no JOINs yet nor the WHERE clause
2281 query_ := query_ || coal || ' AS quality ' || E'\n';
2283 -- jrows table is for the joins we must make (and the real text conditions)
2284 SELECT STRING_AGG(j, E'\n') INTO joins
2285 FROM _vandelay_tmp_jrows;
2287 -- add those joins and the where clause to our query.
2288 query_ := query_ || joins || E'\n';
2290 query_ := query_ || 'JOIN authority.record_entry are ON (are.id = record) '
2291 || 'WHERE ' || wq || ' AND not are.deleted';
2293 -- this will return rows of record,quality
2294 FOR rec IN EXECUTE query_ USING tags_rstore LOOP
2298 DROP TABLE _vandelay_tmp_qrows;
2299 DROP TABLE _vandelay_tmp_jrows;
2302 $$ LANGUAGE PLPGSQL;
2304 CREATE OR REPLACE FUNCTION vandelay.measure_auth_record_quality
2305 ( xml TEXT, match_set_id INT ) RETURNS INT AS $_$
2309 test vandelay.match_set_quality%ROWTYPE;
2312 FOR test IN SELECT * FROM vandelay.match_set_quality
2313 WHERE match_set = match_set_id LOOP
2314 IF test.tag IS NOT NULL THEN
2315 FOR rvalue IN SELECT value FROM vandelay.flatten_marc( xml )
2316 WHERE tag = test.tag AND subfield = test.subfield LOOP
2317 IF test.value = rvalue THEN
2318 out_q := out_q + test.quality;
2326 $_$ LANGUAGE PLPGSQL;
2330 CREATE OR REPLACE FUNCTION vandelay.match_authority_record() RETURNS TRIGGER AS $func$
2332 incoming_existing_id TEXT;
2333 test_result vandelay.match_set_test_result%ROWTYPE;
2337 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
2341 DELETE FROM vandelay.authority_match WHERE queued_record = NEW.id;
2343 SELECT q.match_set INTO match_set FROM vandelay.authority_queue q WHERE q.id = NEW.queue;
2345 IF match_set IS NOT NULL THEN
2346 NEW.quality := vandelay.measure_auth_record_quality( NEW.marc, match_set );
2349 -- Perfect matches on 901$c exit early with a match with high quality.
2350 incoming_existing_id :=
2351 oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]', NEW.marc);
2353 IF incoming_existing_id IS NOT NULL AND incoming_existing_id != '' THEN
2354 SELECT id INTO tmp_rec FROM authority.record_entry WHERE id = incoming_existing_id::bigint;
2355 IF tmp_rec IS NOT NULL THEN
2356 INSERT INTO vandelay.authority_match (queued_record, eg_record, match_score, quality)
2361 -- note: no match_set means quality==0
2362 vandelay.measure_auth_record_quality( b.marc, match_set )
2363 FROM authority.record_entry b
2364 WHERE id = incoming_existing_id::bigint;
2368 IF match_set IS NULL THEN
2372 FOR test_result IN SELECT * FROM
2373 vandelay.match_set_test_authxml(match_set, NEW.marc) LOOP
2375 INSERT INTO vandelay.authority_match ( queued_record, eg_record, match_score, quality )
2379 test_result.quality,
2380 vandelay.measure_auth_record_quality( b.marc, match_set )
2381 FROM authority.record_entry b
2382 WHERE id = test_result.record;
2388 $func$ LANGUAGE PLPGSQL;
2390 CREATE TRIGGER zz_match_auths_trigger
2391 BEFORE INSERT OR UPDATE ON vandelay.queued_authority_record
2392 FOR EACH ROW EXECUTE PROCEDURE vandelay.match_authority_record();
2394 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_record_with_best ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
2397 lwm_ratio_value NUMERIC;
2400 lwm_ratio_value := COALESCE(lwm_ratio_value_p, 0.0);
2402 PERFORM * FROM vandelay.queued_authority_record WHERE import_time IS NOT NULL AND id = import_id;
2405 -- RAISE NOTICE 'already imported, cannot auto-overlay'
2409 SELECT m.eg_record INTO eg_id
2410 FROM vandelay.authority_match m
2411 JOIN vandelay.queued_authority_record qr ON (m.queued_record = qr.id)
2412 JOIN vandelay.authority_queue q ON (qr.queue = q.id)
2413 JOIN authority.record_entry r ON (r.id = m.eg_record)
2414 WHERE m.queued_record = import_id
2415 AND qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC >= lwm_ratio_value
2416 ORDER BY m.match_score DESC, -- required match score
2417 qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC DESC, -- quality tie breaker
2418 m.id -- when in doubt, use the first match
2421 IF eg_id IS NULL THEN
2422 -- RAISE NOTICE 'incoming record is not of high enough quality';
2426 RETURN vandelay.overlay_authority_record( import_id, eg_id, merge_profile_id );
2428 $$ LANGUAGE PLPGSQL;
2433 -- Vandelay (for importing and exporting records) 012.schema.vandelay.sql
2434 --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)]');
2435 --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)]');
2436 --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]');
2437 --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]');
2438 --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$);
2439 --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$);
2440 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (7, 'price',oils_i18n_gettext(7, 'vqbrad', 'Price', 'description'),'//*[@tag="020" or @tag="022"]/*[@code="c"][1]');
2441 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (8, 'rec_identifier',oils_i18n_gettext(8, 'vqbrad', 'Accession Number', 'description'),'//*[@tag="001"]', TRUE);
2442 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (9, 'eg_tcn',oils_i18n_gettext(9, 'vqbrad', 'TCN Value', 'description'),'//*[@tag="901"]/*[@code="a"]', TRUE);
2443 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (10, 'eg_tcn_source',oils_i18n_gettext(10, 'vqbrad', 'TCN Source', 'description'),'//*[@tag="901"]/*[@code="b"]', TRUE);
2444 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (11, 'eg_identifier',oils_i18n_gettext(11, 'vqbrad', 'Internal ID', 'description'),'//*[@tag="901"]/*[@code="c"]', TRUE);
2445 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (12, 'publisher',oils_i18n_gettext(12, 'vqbrad', 'Publisher', 'description'),'//*[@tag="260"]/*[@code="b"][1]');
2446 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, remove ) VALUES (13, 'pubdate',oils_i18n_gettext(13, 'vqbrad', 'Publication Date', 'description'),'//*[@tag="260"]/*[@code="c"][1]',$r$\D$r$);
2447 --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]');
2449 --INSERT INTO vandelay.import_item_attr_definition (
2450 -- owner, name, tag, owning_lib, circ_lib, location,
2451 -- call_number, circ_modifier, barcode, price, copy_number,
2452 -- circulate, ref, holdable, opac_visible, status
2455 -- 'Evergreen 852 export format',
2457 -- '[@code = "b"][1]',
2458 -- '[@code = "b"][2]',
2465 -- '[@code = "x" and text() = "circulating"]',
2466 -- '[@code = "x" and text() = "reference"]',
2467 -- '[@code = "x" and text() = "holdable"]',
2468 -- '[@code = "x" and text() = "visible"]',
2472 --INSERT INTO vandelay.import_item_attr_definition (
2485 -- 'Unicorn Import format -- 999',
2496 --INSERT INTO vandelay.authority_attr_definition ( code, description, xpath, ident ) VALUES ('rec_identifier','Identifier','//*[@tag="001"]', TRUE);
2499 CREATE TABLE vandelay.session_tracker (
2500 id BIGSERIAL PRIMARY KEY,
2502 -- string of characters (e.g. md5) used for linking trackers
2503 -- of different actions into a series. There can be multiple
2504 -- session_keys of each action type, creating the opportunity
2505 -- to link multiple action trackers into a single session.
2506 session_key TEXT NOT NULL,
2508 -- optional user-supplied name
2511 usr INTEGER NOT NULL REFERENCES actor.usr(id)
2512 DEFERRABLE INITIALLY DEFERRED,
2514 -- org unit can be derived from WS
2515 workstation INTEGER NOT NULL REFERENCES actor.workstation(id)
2516 ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
2519 record_type TEXT NOT NULL DEFAULT 'bib',
2521 -- Queue defines the source of the data, it does not necessarily
2522 -- mean that an action is being performed against an entire queue.
2523 -- E.g. some imports are misc. lists of record IDs, but they always
2524 -- come from one queue.
2525 -- No foreign key -- could be auth or bib queue.
2526 queue BIGINT NOT NULL,
2528 create_time TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT NOW(),
2529 update_time TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT NOW(),
2531 state TEXT NOT NULL DEFAULT 'active',
2533 action_type TEXT NOT NULL DEFAULT 'enqueue', -- import
2535 -- total number of tasks to perform / loosely defined
2536 -- could be # of recs to import or # of recs + # of copies
2537 -- depending on the import context
2538 total_actions INTEGER NOT NULL DEFAULT 0,
2540 -- total number of tasked performed so far
2541 actions_performed INTEGER NOT NULL DEFAULT 0,
2543 CONSTRAINT vand_tracker_valid_state
2544 CHECK (state IN ('active','error','complete')),
2546 CONSTRAINT vand_tracker_valid_action_type
2547 CHECK (action_type IN ('upload', 'enqueue', 'import')),
2549 CONSTRAINT vand_tracker_valid_record_type
2550 CHECK (record_type IN ('bib', 'authority'))