1 DROP SCHEMA IF EXISTS vandelay CASCADE;
5 CREATE SCHEMA vandelay;
7 CREATE TABLE vandelay.match_set (
10 owner INT NOT NULL REFERENCES actor.org_unit (id) ON DELETE CASCADE,
11 mtype TEXT NOT NULL DEFAULT 'biblio', -- 'biblio','authority','mfhd'?, others?
12 CONSTRAINT name_once_per_owner_mtype UNIQUE (name, owner, mtype)
15 -- Table to define match points, either FF via SVF or tag+subfield
16 CREATE TABLE vandelay.match_set_point (
17 id SERIAL PRIMARY KEY,
18 match_set INT REFERENCES vandelay.match_set (id) ON DELETE CASCADE,
19 parent INT REFERENCES vandelay.match_set_point (id),
20 bool_op TEXT CHECK (bool_op IS NULL OR (bool_op IN ('AND','OR','NOT'))),
21 svf TEXT REFERENCES config.record_attr_definition (name),
24 negate BOOL DEFAULT FALSE,
25 quality INT NOT NULL DEFAULT 1, -- higher is better
26 CONSTRAINT vmsp_need_a_subfield_with_a_tag CHECK ((tag IS NOT NULL AND subfield IS NOT NULL) OR tag IS NULL),
27 CONSTRAINT vmsp_need_a_tag_or_a_ff_or_a_bo CHECK (
28 (tag IS NOT NULL AND svf IS NULL AND bool_op IS NULL) OR
29 (tag IS NULL AND svf IS NOT NULL AND bool_op IS NULL) OR
30 (tag IS NULL AND svf IS NULL AND bool_op IS NOT NULL)
34 CREATE TABLE vandelay.match_set_quality (
35 id SERIAL PRIMARY KEY,
36 match_set INT NOT NULL REFERENCES vandelay.match_set (id) ON DELETE CASCADE,
37 svf TEXT REFERENCES config.record_attr_definition,
41 quality INT NOT NULL DEFAULT 1, -- higher is better
42 CONSTRAINT vmsq_need_a_subfield_with_a_tag CHECK ((tag IS NOT NULL AND subfield IS NOT NULL) OR tag IS NULL),
43 CONSTRAINT vmsq_need_a_tag_or_a_ff CHECK ((tag IS NOT NULL AND svf IS NULL) OR (tag IS NULL AND svf IS NOT NULL))
45 CREATE UNIQUE INDEX vmsq_def_once_per_set ON vandelay.match_set_quality (match_set, COALESCE(tag,''), COALESCE(subfield,''), COALESCE(svf,''), value);
48 CREATE TABLE vandelay.queue (
49 id BIGSERIAL PRIMARY KEY,
50 owner INT NOT NULL REFERENCES actor.usr (id) DEFERRABLE INITIALLY DEFERRED,
52 complete BOOL NOT NULL DEFAULT FALSE,
53 queue_type TEXT NOT NULL DEFAULT 'bib' CHECK (queue_type IN ('bib','authority')),
54 match_set INT REFERENCES vandelay.match_set (id) ON UPDATE CASCADE ON DELETE SET NULL DEFERRABLE INITIALLY DEFERRED,
55 CONSTRAINT vand_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
58 CREATE TABLE vandelay.queued_record (
59 id BIGSERIAL PRIMARY KEY,
60 create_time TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT NOW(),
61 import_time TIMESTAMP WITH TIME ZONE,
62 purpose TEXT NOT NULL DEFAULT 'import' CHECK (purpose IN ('import','overlay')),
64 quality INT NOT NULL DEFAULT 0
69 /* Bib stuff at the top */
70 ----------------------------------------------------
72 CREATE TABLE vandelay.bib_attr_definition (
73 id SERIAL PRIMARY KEY,
74 code TEXT UNIQUE NOT NULL,
77 remove TEXT NOT NULL DEFAULT ''
80 -- Each TEXT field (other than 'name') should hold an XPath predicate for pulling the data needed
81 -- DROP TABLE vandelay.import_item_attr_definition CASCADE;
82 CREATE TABLE vandelay.import_item_attr_definition (
83 id BIGSERIAL PRIMARY KEY,
84 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
87 keep BOOL NOT NULL DEFAULT FALSE,
107 priv_note_title TEXT,
109 CONSTRAINT vand_import_item_attr_def_idx UNIQUE (owner,name)
112 CREATE TABLE vandelay.import_error (
113 code TEXT PRIMARY KEY,
114 description TEXT NOT NULL -- i18n
117 CREATE TABLE vandelay.bib_queue (
118 queue_type TEXT NOT NULL DEFAULT 'bib' CHECK (queue_type = 'bib'),
119 item_attr_def BIGINT REFERENCES vandelay.import_item_attr_definition (id) ON DELETE SET NULL DEFERRABLE INITIALLY DEFERRED,
120 CONSTRAINT vand_bib_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
121 ) INHERITS (vandelay.queue);
122 ALTER TABLE vandelay.bib_queue ADD PRIMARY KEY (id);
124 CREATE TABLE vandelay.queued_bib_record (
125 queue INT NOT NULL REFERENCES vandelay.bib_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
126 bib_source INT REFERENCES config.bib_source (id) DEFERRABLE INITIALLY DEFERRED,
127 imported_as BIGINT REFERENCES biblio.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
128 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
130 ) INHERITS (vandelay.queued_record);
131 ALTER TABLE vandelay.queued_bib_record ADD PRIMARY KEY (id);
132 CREATE INDEX queued_bib_record_queue_idx ON vandelay.queued_bib_record (queue);
134 CREATE TABLE vandelay.queued_bib_record_attr (
135 id BIGSERIAL PRIMARY KEY,
136 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) DEFERRABLE INITIALLY DEFERRED,
137 field INT NOT NULL REFERENCES vandelay.bib_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
138 attr_value TEXT NOT NULL
140 CREATE INDEX queued_bib_record_attr_record_idx ON vandelay.queued_bib_record_attr (record);
142 CREATE TABLE vandelay.bib_match (
143 id BIGSERIAL PRIMARY KEY,
144 queued_record BIGINT REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
145 eg_record BIGINT REFERENCES biblio.record_entry (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
146 quality INT NOT NULL DEFAULT 1,
147 match_score INT NOT NULL DEFAULT 0
150 CREATE TABLE vandelay.import_item (
151 id BIGSERIAL PRIMARY KEY,
152 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
153 definition BIGINT NOT NULL REFERENCES vandelay.import_item_attr_definition (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
154 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
157 import_time TIMESTAMP WITH TIME ZONE,
166 deposit_amount NUMERIC(8,2),
179 CREATE TABLE vandelay.import_bib_trash_fields (
180 id BIGSERIAL PRIMARY KEY,
181 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
183 CONSTRAINT vand_import_bib_trash_fields_idx UNIQUE (owner,field)
186 CREATE TABLE vandelay.merge_profile (
187 id BIGSERIAL PRIMARY KEY,
188 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
195 CONSTRAINT vand_merge_prof_owner_name_idx UNIQUE (owner,name),
196 CONSTRAINT add_replace_strip_or_preserve CHECK ((preserve_spec IS NOT NULL OR replace_spec IS NOT NULL) OR (preserve_spec IS NULL AND replace_spec IS NULL))
199 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
206 retval config.marc21_rec_type_map%ROWTYPE;
208 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
210 IF ldr IS NULL OR ldr = '' THEN
211 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
215 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
216 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
219 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
220 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
222 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
224 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
227 IF retval.code IS NULL THEN
228 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
233 $func$ LANGUAGE PLPGSQL;
235 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
242 rtype := (vandelay.marc21_record_type( marc )).code;
243 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
244 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
245 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
248 val := REPEAT( ff_pos.default_val, ff_pos.length );
254 $func$ LANGUAGE PLPGSQL;
256 CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
257 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
262 output biblio.record_ff_map%ROWTYPE;
264 rtype := (vandelay.marc21_record_type( marc )).code;
266 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
267 output.ff_name := ff_pos.fixed_field;
268 output.ff_value := NULL;
270 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(tag) || '"]/text()', marc ) ) x(value) LOOP
271 output.ff_value := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
272 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
274 output.ff_value := NULL;
281 $func$ LANGUAGE PLPGSQL;
283 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
284 CREATE OR REPLACE FUNCTION vandelay.marc21_physical_characteristics( marc TEXT) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
288 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
289 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
290 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
291 retval biblio.marc21_physical_characteristics%ROWTYPE;
294 _007 := oils_xpath_string( '//*[@tag="007"]', marc );
296 IF _007 IS NOT NULL AND _007 <> '' THEN
297 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007, 1, 1 );
299 IF ptype.ptype_key IS NOT NULL THEN
300 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
301 SELECT * INTO pval FROM config.marc21_physical_characteristic_value_map WHERE ptype_subfield = psf.id AND value = SUBSTRING( _007, psf.start_pos + 1, psf.length );
303 IF pval.id IS NOT NULL THEN
306 retval.ptype := ptype.ptype_key;
307 retval.subfield := psf.id;
308 retval.value := pval.id;
318 $func$ LANGUAGE PLPGSQL;
320 CREATE TYPE vandelay.flat_marc AS ( tag CHAR(3), ind1 TEXT, ind2 TEXT, subfield TEXT, value TEXT );
321 CREATE OR REPLACE FUNCTION vandelay.flay_marc ( TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
324 use MARC::File::XML (BinaryEncoding => 'UTF-8');
328 MARC::Charset->assume_unicode(1);
331 my $r = MARC::Record->new_from_xml( $xml );
333 return_next( { tag => 'LDR', value => $r->leader } );
335 for my $f ( $r->fields ) {
336 if ($f->is_control_field) {
337 return_next({ tag => $f->tag, value => $f->data });
339 for my $s ($f->subfields) {
342 ind1 => $f->indicator(1),
343 ind2 => $f->indicator(2),
348 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
349 my $trim = $f->indicator(2) || 0;
352 ind1 => $f->indicator(1),
353 ind2 => $f->indicator(2),
355 value => substr( $s->[1], $trim )
364 $func$ LANGUAGE PLPERLU;
366 CREATE OR REPLACE FUNCTION vandelay.flatten_marc ( marc TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
368 output vandelay.flat_marc%ROWTYPE;
371 FOR field IN SELECT * FROM vandelay.flay_marc( marc ) LOOP
372 output.ind1 := field.ind1;
373 output.ind2 := field.ind2;
374 output.tag := field.tag;
375 output.subfield := field.subfield;
376 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
377 output.value := naco_normalize(field.value, field.subfield);
379 output.value := field.value;
382 CONTINUE WHEN output.value IS NULL;
387 $func$ LANGUAGE PLPGSQL;
389 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT, attr_defs TEXT[]) RETURNS hstore AS $_$
391 transformed_xml TEXT;
394 xfrm config.xml_transform%ROWTYPE;
396 new_attrs HSTORE := ''::HSTORE;
397 attr_def config.record_attr_definition%ROWTYPE;
400 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE name IN (SELECT * FROM UNNEST(attr_defs)) ORDER BY format LOOP
402 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
403 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(x.value), COALESCE(attr_def.joiner,' ')) INTO attr_value
404 FROM vandelay.flatten_marc(xml) AS x
405 WHERE x.tag LIKE attr_def.tag
407 WHEN attr_def.sf_list IS NOT NULL
408 THEN POSITION(x.subfield IN attr_def.sf_list) > 0
415 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
416 attr_value := vandelay.marc21_extract_fixed_field(xml, attr_def.fixed_field);
418 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
420 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
422 -- See if we can skip the XSLT ... it's expensive
423 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
424 -- Can't skip the transform
425 IF xfrm.xslt <> '---' THEN
426 transformed_xml := oils_xslt_process(xml,xfrm.xslt);
428 transformed_xml := xml;
431 prev_xfrm := xfrm.name;
434 IF xfrm.name IS NULL THEN
435 -- just grab the marcxml (empty) transform
436 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
437 prev_xfrm := xfrm.name;
440 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
442 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
443 SELECT m.value::TEXT INTO attr_value
444 FROM vandelay.marc21_physical_characteristics(xml) v
445 JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
446 WHERE v.subfield = attr_def.phys_char_sf
447 LIMIT 1; -- Just in case ...
451 -- apply index normalizers to attr_value
453 SELECT n.func AS func,
454 n.param_count AS param_count,
456 FROM config.index_normalizer n
457 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
458 WHERE attr = attr_def.name
460 EXECUTE 'SELECT ' || normalizer.func || '(' ||
461 quote_literal( attr_value ) ||
463 WHEN normalizer.param_count > 0
464 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
471 -- Add the new value to the hstore
472 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
478 $_$ LANGUAGE PLPGSQL;
480 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT ) RETURNS hstore AS $_$
481 SELECT vandelay.extract_rec_attrs( $1, (SELECT ARRAY_ACCUM(name) FROM config.record_attr_definition));
484 -- Everything between this comment and the beginning of the definition of
485 -- vandelay.match_bib_record() is strictly in service of that function.
486 CREATE TYPE vandelay.match_set_test_result AS (record BIGINT, quality INTEGER);
488 CREATE OR REPLACE FUNCTION vandelay.match_set_test_marcxml(
489 match_set_id INTEGER, record_xml TEXT
490 ) RETURNS SETOF vandelay.match_set_test_result AS $$
501 tags_rstore := vandelay.flatten_marc_hstore(record_xml);
502 svf_rstore := vandelay.extract_rec_attrs(record_xml);
504 CREATE TEMPORARY TABLE _vandelay_tmp_qrows (q INTEGER);
505 CREATE TEMPORARY TABLE _vandelay_tmp_jrows (j TEXT);
507 -- generate the where clause and return that directly (into wq), and as
508 -- a side-effect, populate the _vandelay_tmp_[qj]rows tables.
509 wq := vandelay.get_expr_from_match_set(match_set_id);
511 query_ := 'SELECT DISTINCT(bre.id) AS record, ';
513 -- qrows table is for the quality bits we add to the SELECT clause
514 SELECT ARRAY_TO_STRING(
515 ARRAY_ACCUM('COALESCE(n' || q::TEXT || '.quality, 0)'), ' + '
516 ) INTO coal FROM _vandelay_tmp_qrows;
518 -- our query string so far is the SELECT clause and the inital FROM.
519 -- no JOINs yet nor the WHERE clause
520 query_ := query_ || coal || ' AS quality ' || E'\n' ||
521 'FROM biblio.record_entry bre ';
523 -- jrows table is for the joins we must make (and the real text conditions)
524 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(j), E'\n') INTO joins
525 FROM _vandelay_tmp_jrows;
527 -- add those joins and the where clause to our query.
528 query_ := query_ || joins || E'\n' || 'WHERE ' || wq || ' AND not bre.deleted';
530 -- this will return rows of record,quality
531 FOR rec IN EXECUTE query_ USING tags_rstore, svf_rstore LOOP
535 DROP TABLE _vandelay_tmp_qrows;
536 DROP TABLE _vandelay_tmp_jrows;
542 CREATE OR REPLACE FUNCTION vandelay.flatten_marc_hstore(
544 ) RETURNS HSTORE AS $$
548 ARRAY_ACCUM(tag || (COALESCE(subfield, ''))),
554 CASE WHEN tag IN ('020', '022', '024') THEN -- caseless
555 ARRAY_ACCUM(LOWER(value))::TEXT
557 ARRAY_ACCUM(value)::TEXT
559 FROM vandelay.flatten_marc(record_xml)
560 GROUP BY tag, subfield ORDER BY tag, subfield
566 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set(
570 root vandelay.match_set_point;
572 SELECT * INTO root FROM vandelay.match_set_point
573 WHERE parent IS NULL AND match_set = match_set_id;
575 RETURN vandelay.get_expr_from_match_set_point(root);
579 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set_point(
580 node vandelay.match_set_point
587 child vandelay.match_set_point;
589 SELECT ARRAY_ACCUM(id) INTO children FROM vandelay.match_set_point
590 WHERE parent = node.id;
592 IF ARRAY_LENGTH(children, 1) > 0 THEN
593 this_op := vandelay._get_expr_render_one(node);
596 WHILE children[i] IS NOT NULL LOOP
597 SELECT * INTO child FROM vandelay.match_set_point
598 WHERE id = children[i];
600 q := q || ' ' || this_op || ' ';
603 q := q || vandelay.get_expr_from_match_set_point(child);
607 ELSIF node.bool_op IS NULL THEN
608 PERFORM vandelay._get_expr_push_qrow(node);
609 PERFORM vandelay._get_expr_push_jrow(node);
610 RETURN vandelay._get_expr_render_one(node);
617 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_qrow(
618 node vandelay.match_set_point
622 INSERT INTO _vandelay_tmp_qrows (q) VALUES (node.id);
626 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_jrow(
627 node vandelay.match_set_point
636 -- remember $1 is tags_rstore, and $2 is svf_rstore
646 IF node.tag IS NOT NULL THEN
647 caseless := (node.tag IN ('020', '022', '024'));
649 IF node.subfield IS NOT NULL THEN
650 tagkey := tagkey || node.subfield;
654 my_alias := 'n' || node.id::TEXT;
656 jrow := 'LEFT JOIN (SELECT *, ' || node.quality ||
657 ' AS quality FROM metabib.';
658 IF node.tag IS NOT NULL THEN
659 jrow := jrow || 'full_rec) ' || my_alias || ' ON (' ||
660 my_alias || '.record = bre.id AND ' || my_alias || '.tag = ''' ||
662 IF node.subfield IS NOT NULL THEN
663 jrow := jrow || ' AND ' || my_alias || '.subfield = ''' ||
664 node.subfield || '''';
666 jrow := jrow || ' AND (';
669 jrow := jrow || 'LOWER(' || my_alias || '.value) ' || op;
671 jrow := jrow || my_alias || '.value ' || op;
674 jrow := jrow || ' ANY(($1->''' || tagkey || ''')::TEXT[])))';
676 jrow := jrow || 'record_attr) ' || my_alias || ' ON (' ||
677 my_alias || '.id = bre.id AND (' ||
678 my_alias || '.attrs->''' || node.svf ||
679 ''' ' || op || ' $2->''' || node.svf || '''))';
681 INSERT INTO _vandelay_tmp_jrows (j) VALUES (jrow);
685 CREATE OR REPLACE FUNCTION vandelay._get_expr_render_one(
686 node vandelay.match_set_point
691 IF node.bool_op IS NOT NULL THEN
694 RETURN '(n' || node.id::TEXT || '.id IS NOT NULL)';
699 CREATE OR REPLACE FUNCTION vandelay.match_bib_record() RETURNS TRIGGER AS $func$
701 incoming_existing_id TEXT;
702 test_result vandelay.match_set_test_result%ROWTYPE;
706 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
710 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
712 SELECT q.match_set INTO match_set FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
714 IF match_set IS NOT NULL THEN
715 NEW.quality := vandelay.measure_record_quality( NEW.marc, match_set );
718 -- Perfect matches on 901$c exit early with a match with high quality.
719 incoming_existing_id :=
720 oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]', NEW.marc);
722 IF incoming_existing_id IS NOT NULL AND incoming_existing_id != '' THEN
723 SELECT id INTO tmp_rec FROM biblio.record_entry WHERE id = incoming_existing_id::bigint;
724 IF tmp_rec IS NOT NULL THEN
725 INSERT INTO vandelay.bib_match (queued_record, eg_record, match_score, quality)
730 -- note: no match_set means quality==0
731 vandelay.measure_record_quality( b.marc, match_set )
732 FROM biblio.record_entry b
733 WHERE id = incoming_existing_id::bigint;
737 IF match_set IS NULL THEN
741 FOR test_result IN SELECT * FROM
742 vandelay.match_set_test_marcxml(match_set, NEW.marc) LOOP
744 INSERT INTO vandelay.bib_match ( queued_record, eg_record, match_score, quality )
749 vandelay.measure_record_quality( b.marc, match_set )
750 FROM biblio.record_entry b
751 WHERE id = test_result.record;
757 $func$ LANGUAGE PLPGSQL;
759 CREATE OR REPLACE FUNCTION vandelay.measure_record_quality ( xml TEXT, match_set_id INT ) RETURNS INT AS $_$
763 test vandelay.match_set_quality%ROWTYPE;
766 FOR test IN SELECT * FROM vandelay.match_set_quality WHERE match_set = match_set_id LOOP
767 IF test.tag IS NOT NULL THEN
768 FOR rvalue IN SELECT value FROM vandelay.flatten_marc( xml ) WHERE tag = test.tag AND subfield = test.subfield LOOP
769 IF test.value = rvalue THEN
770 out_q := out_q + test.quality;
774 IF test.value = vandelay.extract_rec_attrs(xml, ARRAY[test.svf]) -> test.svf THEN
775 out_q := out_q + test.quality;
782 $_$ LANGUAGE PLPGSQL;
784 CREATE TYPE vandelay.tcn_data AS (tcn TEXT, tcn_source TEXT, used BOOL);
785 CREATE OR REPLACE FUNCTION vandelay.find_bib_tcn_data ( xml TEXT ) RETURNS SETOF vandelay.tcn_data AS $_$
789 output vandelay.tcn_data%ROWTYPE;
793 eg_tcn := BTRIM((oils_xpath('//*[@tag="001"]/text()',xml))[1]);
794 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
796 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="003"]/text()',xml))[1]);
797 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
798 eg_tcn_source := 'System Local';
801 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
804 output.used := FALSE;
809 output.tcn := eg_tcn;
810 output.tcn_source := eg_tcn_source;
816 eg_tcn := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="a"]/text()',xml))[1]);
817 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
819 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="b"]/text()',xml))[1]);
820 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
821 eg_tcn_source := 'System Local';
824 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
827 output.used := FALSE;
832 output.tcn := eg_tcn;
833 output.tcn_source := eg_tcn_source;
839 eg_tcn := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="a"]/text()',xml))[1]);
840 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
842 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="b"]/text()',xml))[1]);
843 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
844 eg_tcn_source := 'System Local';
847 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
850 output.used := FALSE;
855 output.tcn := eg_tcn;
856 output.tcn_source := eg_tcn_source;
862 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="020"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
863 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
865 eg_tcn_source := 'ISBN';
867 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
870 output.used := FALSE;
875 output.tcn := eg_tcn;
876 output.tcn_source := eg_tcn_source;
882 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="022"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
883 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
885 eg_tcn_source := 'ISSN';
887 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
890 output.used := FALSE;
895 output.tcn := eg_tcn;
896 output.tcn_source := eg_tcn_source;
902 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="010"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
903 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
905 eg_tcn_source := 'LCCN';
907 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
910 output.used := FALSE;
915 output.tcn := eg_tcn;
916 output.tcn_source := eg_tcn_source;
922 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="035"]/*[@code="a"]/text()',xml))[1], $re$^.*?(\w+)$$re$, $re$\1$re$);
923 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
925 eg_tcn_source := 'System Legacy';
927 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
930 output.used := FALSE;
935 output.tcn := eg_tcn;
936 output.tcn_source := eg_tcn_source;
943 $_$ LANGUAGE PLPGSQL;
945 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT, force_add INT ) RETURNS TEXT AS $_$
948 use MARC::File::XML (BinaryEncoding => 'UTF-8');
952 MARC::Charset->assume_unicode(1);
954 my $target_xml = shift;
955 my $source_xml = shift;
956 my $field_spec = shift;
957 my $force_add = shift || 0;
959 my $target_r = MARC::Record->new_from_xml( $target_xml );
960 my $source_r = MARC::Record->new_from_xml( $source_xml );
962 return $target_xml unless ($target_r && $source_r);
964 my @field_list = split(',', $field_spec);
967 for my $f (@field_list) {
968 $f =~ s/^\s*//; $f =~ s/\s*$//;
969 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
975 $match =~ s/^\s*//; $match =~ s/\s*$//;
976 $fields{$field} = { sf => [ split('', $sf) ] };
978 my ($msf,$mre) = split('~', $match);
979 if (length($msf) > 0 and length($mre) > 0) {
980 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
981 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
982 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
988 for my $f ( keys %fields) {
989 if ( @{$fields{$f}{sf}} ) {
990 for my $from_field ($source_r->field( $f )) {
991 my @tos = $target_r->field( $f );
993 next if (exists($fields{$f}{match}) and !$force_add);
994 my @new_fields = map { $_->clone } $source_r->field( $f );
995 $target_r->insert_fields_ordered( @new_fields );
997 for my $to_field (@tos) {
998 if (exists($fields{$f}{match})) {
999 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1001 my @new_sf = map { ($_ => $from_field->subfield($_)) } grep { defined($from_field->subfield($_)) } @{$fields{$f}{sf}};
1002 $to_field->add_subfields( @new_sf );
1007 my @new_fields = map { $_->clone } $source_r->field( $f );
1008 $target_r->insert_fields_ordered( @new_fields );
1012 $target_xml = $target_r->as_xml_record;
1013 $target_xml =~ s/^<\?.+?\?>$//mo;
1014 $target_xml =~ s/\n//sgo;
1015 $target_xml =~ s/>\s+</></sgo;
1019 $_$ LANGUAGE PLPERLU;
1021 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1022 SELECT vandelay.add_field( $1, $2, $3, 0 );
1025 CREATE OR REPLACE FUNCTION vandelay.strip_field ( xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1028 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1032 MARC::Charset->assume_unicode(1);
1035 my $r = MARC::Record->new_from_xml( $xml );
1037 return $xml unless ($r);
1039 my $field_spec = shift;
1040 my @field_list = split(',', $field_spec);
1043 for my $f (@field_list) {
1044 $f =~ s/^\s*//; $f =~ s/\s*$//;
1045 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1051 $match =~ s/^\s*//; $match =~ s/\s*$//;
1052 $fields{$field} = { sf => [ split('', $sf) ] };
1054 my ($msf,$mre) = split('~', $match);
1055 if (length($msf) > 0 and length($mre) > 0) {
1056 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1057 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1058 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1064 for my $f ( keys %fields) {
1065 for my $to_field ($r->field( $f )) {
1066 if (exists($fields{$f}{match})) {
1067 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1070 if ( @{$fields{$f}{sf}} ) {
1071 $to_field->delete_subfield(code => $fields{$f}{sf});
1073 $r->delete_field( $to_field );
1078 $xml = $r->as_xml_record;
1079 $xml =~ s/^<\?.+?\?>$//mo;
1081 $xml =~ s/>\s+</></sgo;
1085 $_$ LANGUAGE PLPERLU;
1087 CREATE OR REPLACE FUNCTION vandelay.replace_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1094 parsed_target := vandelay.strip_field( target_xml, ''); -- this dance normalizes the format of the xml for the IF below
1095 xml_output := parsed_target; -- if there are no replace rules, just return the input
1097 FOR curr_field IN SELECT UNNEST( STRING_TO_ARRAY(field, ',') ) LOOP -- naive split, but it's the same we use in the perl
1099 xml_output := vandelay.strip_field( parsed_target, curr_field);
1101 IF xml_output <> parsed_target AND curr_field ~ E'~' THEN
1102 -- we removed something, and there was a regexp restriction in the curr_field definition, so proceed
1103 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 1 );
1104 ELSIF curr_field !~ E'~' THEN
1105 -- No regexp restriction, add the curr_field
1106 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 0 );
1109 parsed_target := xml_output; -- in prep for any following loop iterations
1115 $_$ LANGUAGE PLPGSQL;
1117 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 $_$
1118 SELECT vandelay.replace_field( vandelay.add_field( vandelay.strip_field( $1, $5) , $2, $3 ), $2, $4);
1121 CREATE TYPE vandelay.compile_profile AS (add_rule TEXT, replace_rule TEXT, preserve_rule TEXT, strip_rule TEXT);
1122 CREATE OR REPLACE FUNCTION vandelay.compile_profile ( incoming_xml TEXT ) RETURNS vandelay.compile_profile AS $_$
1124 output vandelay.compile_profile%ROWTYPE;
1125 profile vandelay.merge_profile%ROWTYPE;
1127 profile_tmpl_owner TEXT;
1128 add_rule TEXT := '';
1129 strip_rule TEXT := '';
1130 replace_rule TEXT := '';
1131 preserve_rule TEXT := '';
1135 profile_tmpl := (oils_xpath('//*[@tag="905"]/*[@code="t"]/text()',incoming_xml))[1];
1136 profile_tmpl_owner := (oils_xpath('//*[@tag="905"]/*[@code="o"]/text()',incoming_xml))[1];
1138 IF profile_tmpl IS NOT NULL AND profile_tmpl <> '' AND profile_tmpl_owner IS NOT NULL AND profile_tmpl_owner <> '' THEN
1139 SELECT p.* INTO profile
1140 FROM vandelay.merge_profile p
1141 JOIN actor.org_unit u ON (u.id = p.owner)
1142 WHERE p.name = profile_tmpl
1143 AND u.shortname = profile_tmpl_owner;
1145 IF profile.id IS NOT NULL THEN
1146 add_rule := COALESCE(profile.add_spec,'');
1147 strip_rule := COALESCE(profile.strip_spec,'');
1148 replace_rule := COALESCE(profile.replace_spec,'');
1149 preserve_rule := COALESCE(profile.preserve_spec,'');
1153 add_rule := add_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="a"]/text()',incoming_xml),','),'');
1154 strip_rule := strip_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="d"]/text()',incoming_xml),','),'');
1155 replace_rule := replace_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="r"]/text()',incoming_xml),','),'');
1156 preserve_rule := preserve_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="p"]/text()',incoming_xml),','),'');
1158 output.add_rule := BTRIM(add_rule,',');
1159 output.replace_rule := BTRIM(replace_rule,',');
1160 output.strip_rule := BTRIM(strip_rule,',');
1161 output.preserve_rule := BTRIM(preserve_rule,',');
1165 $_$ LANGUAGE PLPGSQL;
1167 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1169 merge_profile vandelay.merge_profile%ROWTYPE;
1170 dyn_profile vandelay.compile_profile%ROWTYPE;
1180 SELECT b.marc INTO eg_marc
1181 FROM biblio.record_entry b
1185 IF eg_marc IS NULL OR v_marc IS NULL THEN
1186 -- RAISE NOTICE 'no marc for template or bib record';
1190 dyn_profile := vandelay.compile_profile( v_marc );
1192 IF merge_profile_id IS NOT NULL THEN
1193 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1195 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1196 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1197 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1198 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1202 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1203 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1207 IF dyn_profile.replace_rule <> '' THEN
1208 source_marc = v_marc;
1209 target_marc = eg_marc;
1210 replace_rule = dyn_profile.replace_rule;
1212 source_marc = eg_marc;
1213 target_marc = v_marc;
1214 replace_rule = dyn_profile.preserve_rule;
1217 UPDATE biblio.record_entry
1218 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1222 -- RAISE NOTICE 'update of biblio.record_entry failed';
1229 $$ LANGUAGE PLPGSQL;
1231 CREATE OR REPLACE FUNCTION vandelay.merge_record_xml ( target_marc TEXT, template_marc TEXT ) RETURNS TEXT AS $$
1233 dyn_profile vandelay.compile_profile%ROWTYPE;
1241 IF target_marc IS NULL OR template_marc IS NULL THEN
1242 -- RAISE NOTICE 'no marc for target or template record';
1246 dyn_profile := vandelay.compile_profile( template_marc );
1248 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1249 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1253 IF dyn_profile.replace_rule <> '' THEN
1254 trgt_marc = target_marc;
1255 tmpl_marc = template_marc;
1256 replace_rule = dyn_profile.replace_rule;
1258 tmp_marc = target_marc;
1259 trgt_marc = template_marc;
1260 tmpl_marc = tmp_marc;
1261 replace_rule = dyn_profile.preserve_rule;
1264 RETURN vandelay.merge_record_xml( trgt_marc, tmpl_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule );
1267 $$ LANGUAGE PLPGSQL;
1269 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT) RETURNS BOOL AS $$
1270 SELECT vandelay.template_overlay_bib_record( $1, $2, NULL);
1273 CREATE OR REPLACE FUNCTION vandelay.overlay_bib_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1275 merge_profile vandelay.merge_profile%ROWTYPE;
1276 dyn_profile vandelay.compile_profile%ROWTYPE;
1286 SELECT q.marc INTO v_marc
1287 FROM vandelay.queued_record q
1288 JOIN vandelay.bib_match m ON (m.queued_record = q.id AND q.id = import_id)
1291 IF v_marc IS NULL THEN
1292 -- RAISE NOTICE 'no marc for vandelay or bib record';
1296 IF vandelay.template_overlay_bib_record( v_marc, eg_id, merge_profile_id) THEN
1297 UPDATE vandelay.queued_bib_record
1298 SET imported_as = eg_id,
1300 WHERE id = import_id;
1302 editor_string := (oils_xpath('//*[@tag="905"]/*[@code="u"]/text()',v_marc))[1];
1304 IF editor_string IS NOT NULL AND editor_string <> '' THEN
1305 SELECT usr INTO editor_id FROM actor.card WHERE barcode = editor_string;
1307 IF editor_id IS NULL THEN
1308 SELECT id INTO editor_id FROM actor.usr WHERE usrname = editor_string;
1311 IF editor_id IS NOT NULL THEN
1312 UPDATE biblio.record_entry SET editor = editor_id WHERE id = eg_id;
1319 -- RAISE NOTICE 'update of biblio.record_entry failed';
1324 $$ LANGUAGE PLPGSQL;
1326 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
1329 lwm_ratio_value NUMERIC;
1332 lwm_ratio_value := COALESCE(lwm_ratio_value_p, 0.0);
1334 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1337 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1341 SELECT m.eg_record INTO eg_id
1342 FROM vandelay.bib_match m
1343 JOIN vandelay.queued_bib_record qr ON (m.queued_record = qr.id)
1344 JOIN vandelay.bib_queue q ON (qr.queue = q.id)
1345 JOIN biblio.record_entry r ON (r.id = m.eg_record)
1346 WHERE m.queued_record = import_id
1347 AND qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC >= lwm_ratio_value
1348 ORDER BY m.match_score DESC, -- required match score
1349 qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC DESC, -- quality tie breaker
1350 m.id -- when in doubt, use the first match
1353 IF eg_id IS NULL THEN
1354 -- RAISE NOTICE 'incoming record is not of high enough quality';
1358 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1360 $$ LANGUAGE PLPGSQL;
1362 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1363 SELECT vandelay.auto_overlay_bib_record_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1366 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1372 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1375 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1379 SELECT COUNT(*) INTO match_count FROM vandelay.bib_match WHERE queued_record = import_id;
1381 IF match_count <> 1 THEN
1382 -- RAISE NOTICE 'not an exact match';
1386 -- Check that the one match is on the first 901c
1387 SELECT m.eg_record INTO eg_id
1388 FROM vandelay.queued_bib_record q
1389 JOIN vandelay.bib_match m ON (m.queued_record = q.id)
1390 WHERE q.id = import_id
1391 AND m.eg_record = oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]',marc)::BIGINT;
1394 -- RAISE NOTICE 'not a 901c match';
1398 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1400 $$ LANGUAGE PLPGSQL;
1402 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1404 queued_record vandelay.queued_bib_record%ROWTYPE;
1407 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1409 IF vandelay.auto_overlay_bib_record( queued_record.id, merge_profile_id ) THEN
1410 RETURN NEXT queued_record.id;
1418 $$ LANGUAGE PLPGSQL;
1420 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 $$
1422 queued_record vandelay.queued_bib_record%ROWTYPE;
1425 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1427 IF vandelay.auto_overlay_bib_record_with_best( queued_record.id, merge_profile_id, lwm_ratio_value ) THEN
1428 RETURN NEXT queued_record.id;
1436 $$ LANGUAGE PLPGSQL;
1438 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1439 SELECT vandelay.auto_overlay_bib_queue_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1442 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1443 SELECT * FROM vandelay.auto_overlay_bib_queue( $1, NULL );
1446 CREATE OR REPLACE FUNCTION vandelay.ingest_bib_marc ( ) RETURNS TRIGGER AS $$
1452 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1456 FOR adef IN SELECT * FROM vandelay.bib_attr_definition LOOP
1458 SELECT extract_marc_field('vandelay.queued_bib_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_bib_record WHERE id = NEW.id;
1459 IF (value IS NOT NULL AND value <> '') THEN
1460 INSERT INTO vandelay.queued_bib_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1467 $$ LANGUAGE PLPGSQL;
1469 CREATE OR REPLACE FUNCTION vandelay.cleanup_bib_marc ( ) RETURNS TRIGGER AS $$
1471 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1475 DELETE FROM vandelay.queued_bib_record_attr WHERE record = OLD.id;
1476 DELETE FROM vandelay.import_item WHERE record = OLD.id;
1478 IF TG_OP = 'UPDATE' THEN
1483 $$ LANGUAGE PLPGSQL;
1485 CREATE TRIGGER cleanup_bib_trigger
1486 BEFORE UPDATE OR DELETE ON vandelay.queued_bib_record
1487 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_bib_marc();
1489 CREATE TRIGGER ingest_bib_trigger
1490 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1491 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_bib_marc();
1493 CREATE TRIGGER zz_match_bibs_trigger
1494 BEFORE INSERT OR UPDATE ON vandelay.queued_bib_record
1495 FOR EACH ROW EXECUTE PROCEDURE vandelay.match_bib_record();
1498 /* Authority stuff down here */
1499 ---------------------------------------
1500 CREATE TABLE vandelay.authority_attr_definition (
1501 id SERIAL PRIMARY KEY,
1502 code TEXT UNIQUE NOT NULL,
1504 xpath TEXT NOT NULL,
1505 remove TEXT NOT NULL DEFAULT ''
1508 CREATE TABLE vandelay.authority_queue (
1509 queue_type TEXT NOT NULL DEFAULT 'authority' CHECK (queue_type = 'authority'),
1510 CONSTRAINT vand_authority_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
1511 ) INHERITS (vandelay.queue);
1512 ALTER TABLE vandelay.authority_queue ADD PRIMARY KEY (id);
1514 CREATE TABLE vandelay.queued_authority_record (
1515 queue INT NOT NULL REFERENCES vandelay.authority_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1516 imported_as INT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1517 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
1519 ) INHERITS (vandelay.queued_record);
1520 ALTER TABLE vandelay.queued_authority_record ADD PRIMARY KEY (id);
1521 CREATE INDEX queued_authority_record_queue_idx ON vandelay.queued_authority_record (queue);
1523 CREATE TABLE vandelay.queued_authority_record_attr (
1524 id BIGSERIAL PRIMARY KEY,
1525 record BIGINT NOT NULL REFERENCES vandelay.queued_authority_record (id) DEFERRABLE INITIALLY DEFERRED,
1526 field INT NOT NULL REFERENCES vandelay.authority_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
1527 attr_value TEXT NOT NULL
1529 CREATE INDEX queued_authority_record_attr_record_idx ON vandelay.queued_authority_record_attr (record);
1531 CREATE TABLE vandelay.authority_match (
1532 id BIGSERIAL PRIMARY KEY,
1533 queued_record BIGINT REFERENCES vandelay.queued_authority_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1534 eg_record BIGINT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1535 quality INT NOT NULL DEFAULT 0
1538 CREATE OR REPLACE FUNCTION vandelay.ingest_authority_marc ( ) RETURNS TRIGGER AS $$
1544 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1548 FOR adef IN SELECT * FROM vandelay.authority_attr_definition LOOP
1550 SELECT extract_marc_field('vandelay.queued_authority_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_authority_record WHERE id = NEW.id;
1551 IF (value IS NOT NULL AND value <> '') THEN
1552 INSERT INTO vandelay.queued_authority_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1559 $$ LANGUAGE PLPGSQL;
1561 CREATE OR REPLACE FUNCTION vandelay.cleanup_authority_marc ( ) RETURNS TRIGGER AS $$
1563 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1567 DELETE FROM vandelay.queued_authority_record_attr WHERE record = OLD.id;
1568 IF TG_OP = 'UPDATE' THEN
1573 $$ LANGUAGE PLPGSQL;
1575 CREATE TRIGGER cleanup_authority_trigger
1576 BEFORE UPDATE OR DELETE ON vandelay.queued_authority_record
1577 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_authority_marc();
1579 CREATE TRIGGER ingest_authority_trigger
1580 AFTER INSERT OR UPDATE ON vandelay.queued_authority_record
1581 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_authority_marc();
1583 CREATE OR REPLACE FUNCTION vandelay.overlay_authority_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1585 merge_profile vandelay.merge_profile%ROWTYPE;
1586 dyn_profile vandelay.compile_profile%ROWTYPE;
1595 SELECT b.marc INTO eg_marc
1596 FROM authority.record_entry b
1597 JOIN vandelay.authority_match m ON (m.eg_record = b.id AND m.queued_record = import_id)
1600 SELECT q.marc INTO v_marc
1601 FROM vandelay.queued_record q
1602 JOIN vandelay.authority_match m ON (m.queued_record = q.id AND q.id = import_id)
1605 IF eg_marc IS NULL OR v_marc IS NULL THEN
1606 -- RAISE NOTICE 'no marc for vandelay or authority record';
1610 dyn_profile := vandelay.compile_profile( v_marc );
1612 IF merge_profile_id IS NOT NULL THEN
1613 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1615 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1616 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1617 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1618 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1622 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1623 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1627 IF dyn_profile.replace_rule <> '' THEN
1628 source_marc = v_marc;
1629 target_marc = eg_marc;
1630 replace_rule = dyn_profile.replace_rule;
1632 source_marc = eg_marc;
1633 target_marc = v_marc;
1634 replace_rule = dyn_profile.preserve_rule;
1637 UPDATE authority.record_entry
1638 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1642 UPDATE vandelay.queued_authority_record
1643 SET imported_as = eg_id,
1645 WHERE id = import_id;
1649 -- RAISE NOTICE 'update of authority.record_entry failed';
1654 $$ LANGUAGE PLPGSQL;
1656 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1661 SELECT COUNT(*) INTO match_count FROM vandelay.authority_match WHERE queued_record = import_id;
1663 IF match_count <> 1 THEN
1664 -- RAISE NOTICE 'not an exact match';
1668 SELECT m.eg_record INTO eg_id
1669 FROM vandelay.authority_match m
1670 WHERE m.queued_record = import_id
1673 IF eg_id IS NULL THEN
1677 RETURN vandelay.overlay_authority_record( import_id, eg_id, merge_profile_id );
1679 $$ LANGUAGE PLPGSQL;
1681 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1683 queued_record vandelay.queued_authority_record%ROWTYPE;
1686 FOR queued_record IN SELECT * FROM vandelay.queued_authority_record WHERE queue = queue_id AND import_time IS NULL LOOP
1688 IF vandelay.auto_overlay_authority_record( queued_record.id, merge_profile_id ) THEN
1689 RETURN NEXT queued_record.id;
1697 $$ LANGUAGE PLPGSQL;
1699 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1700 SELECT * FROM vandelay.auto_overlay_authority_queue( $1, NULL );
1704 -- Vandelay (for importing and exporting records) 012.schema.vandelay.sql
1705 --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)]');
1706 --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)]');
1707 --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]');
1708 --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]');
1709 --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$);
1710 --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$);
1711 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (7, 'price',oils_i18n_gettext(7, 'vqbrad', 'Price', 'description'),'//*[@tag="020" or @tag="022"]/*[@code="c"][1]');
1712 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (8, 'rec_identifier',oils_i18n_gettext(8, 'vqbrad', 'Accession Number', 'description'),'//*[@tag="001"]', TRUE);
1713 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (9, 'eg_tcn',oils_i18n_gettext(9, 'vqbrad', 'TCN Value', 'description'),'//*[@tag="901"]/*[@code="a"]', TRUE);
1714 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (10, 'eg_tcn_source',oils_i18n_gettext(10, 'vqbrad', 'TCN Source', 'description'),'//*[@tag="901"]/*[@code="b"]', TRUE);
1715 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (11, 'eg_identifier',oils_i18n_gettext(11, 'vqbrad', 'Internal ID', 'description'),'//*[@tag="901"]/*[@code="c"]', TRUE);
1716 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (12, 'publisher',oils_i18n_gettext(12, 'vqbrad', 'Publisher', 'description'),'//*[@tag="260"]/*[@code="b"][1]');
1717 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, remove ) VALUES (13, 'pubdate',oils_i18n_gettext(13, 'vqbrad', 'Publication Date', 'description'),'//*[@tag="260"]/*[@code="c"][1]',$r$\D$r$);
1718 --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]');
1720 --INSERT INTO vandelay.import_item_attr_definition (
1721 -- owner, name, tag, owning_lib, circ_lib, location,
1722 -- call_number, circ_modifier, barcode, price, copy_number,
1723 -- circulate, ref, holdable, opac_visible, status
1726 -- 'Evergreen 852 export format',
1728 -- '[@code = "b"][1]',
1729 -- '[@code = "b"][2]',
1736 -- '[@code = "x" and text() = "circulating"]',
1737 -- '[@code = "x" and text() = "reference"]',
1738 -- '[@code = "x" and text() = "holdable"]',
1739 -- '[@code = "x" and text() = "visible"]',
1743 --INSERT INTO vandelay.import_item_attr_definition (
1756 -- 'Unicorn Import format -- 999',
1767 --INSERT INTO vandelay.authority_attr_definition ( code, description, xpath, ident ) VALUES ('rec_identifier','Identifier','//*[@tag="001"]', TRUE);