1 DROP SCHEMA IF EXISTS vandelay CASCADE;
5 CREATE SCHEMA vandelay;
7 CREATE TABLE vandelay.match_set (
10 owner INT NOT NULL REFERENCES actor.org_unit (id) ON DELETE CASCADE,
11 mtype TEXT NOT NULL DEFAULT 'biblio', -- 'biblio','authority','mfhd'?, others?
12 CONSTRAINT name_once_per_owner_mtype UNIQUE (name, owner, mtype)
15 -- Table to define match points, either FF via SVF or tag+subfield
16 CREATE TABLE vandelay.match_set_point (
17 id SERIAL PRIMARY KEY,
18 match_set INT REFERENCES vandelay.match_set (id) ON DELETE CASCADE,
19 parent INT REFERENCES vandelay.match_set_point (id),
20 bool_op TEXT CHECK (bool_op IS NULL OR (bool_op IN ('AND','OR','NOT'))),
21 svf TEXT REFERENCES config.record_attr_definition (name),
24 negate BOOL DEFAULT FALSE,
25 quality INT NOT NULL DEFAULT 1, -- higher is better
26 CONSTRAINT vmsp_need_a_subfield_with_a_tag CHECK ((tag IS NOT NULL AND subfield IS NOT NULL) OR tag IS NULL),
27 CONSTRAINT vmsp_need_a_tag_or_a_ff_or_a_bo CHECK (
28 (tag IS NOT NULL AND svf IS NULL AND bool_op IS NULL) OR
29 (tag IS NULL AND svf IS NOT NULL AND bool_op IS NULL) OR
30 (tag IS NULL AND svf IS NULL AND bool_op IS NOT NULL)
34 CREATE TABLE vandelay.match_set_quality (
35 id SERIAL PRIMARY KEY,
36 match_set INT NOT NULL REFERENCES vandelay.match_set (id) ON DELETE CASCADE,
37 svf TEXT REFERENCES config.record_attr_definition,
41 quality INT NOT NULL DEFAULT 1, -- higher is better
42 CONSTRAINT vmsq_need_a_subfield_with_a_tag CHECK ((tag IS NOT NULL AND subfield IS NOT NULL) OR tag IS NULL),
43 CONSTRAINT vmsq_need_a_tag_or_a_ff CHECK ((tag IS NOT NULL AND svf IS NULL) OR (tag IS NULL AND svf IS NOT NULL))
45 CREATE UNIQUE INDEX vmsq_def_once_per_set ON vandelay.match_set_quality (match_set, COALESCE(tag,''), COALESCE(subfield,''), COALESCE(svf,''), value);
48 CREATE TABLE vandelay.queue (
49 id BIGSERIAL PRIMARY KEY,
50 owner INT NOT NULL REFERENCES actor.usr (id) DEFERRABLE INITIALLY DEFERRED,
52 complete BOOL NOT NULL DEFAULT FALSE,
53 match_set INT REFERENCES vandelay.match_set (id) ON UPDATE CASCADE ON DELETE SET NULL DEFERRABLE INITIALLY DEFERRED
56 CREATE TABLE vandelay.queued_record (
57 id BIGSERIAL PRIMARY KEY,
58 create_time TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT NOW(),
59 import_time TIMESTAMP WITH TIME ZONE,
60 purpose TEXT NOT NULL DEFAULT 'import' CHECK (purpose IN ('import','overlay')),
62 quality INT NOT NULL DEFAULT 0
67 /* Bib stuff at the top */
68 ----------------------------------------------------
70 CREATE TABLE vandelay.bib_attr_definition (
71 id SERIAL PRIMARY KEY,
72 code TEXT UNIQUE NOT NULL,
75 remove TEXT NOT NULL DEFAULT ''
78 -- Each TEXT field (other than 'name') should hold an XPath predicate for pulling the data needed
79 -- DROP TABLE vandelay.import_item_attr_definition CASCADE;
80 CREATE TABLE vandelay.import_item_attr_definition (
81 id BIGSERIAL PRIMARY KEY,
82 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
85 keep BOOL NOT NULL DEFAULT FALSE,
105 priv_note_title TEXT,
108 CONSTRAINT vand_import_item_attr_def_idx UNIQUE (owner,name)
111 CREATE TABLE vandelay.import_error (
112 code TEXT PRIMARY KEY,
113 description TEXT NOT NULL -- i18n
116 CREATE TYPE vandelay.bib_queue_queue_type AS ENUM ('bib', 'acq');
118 CREATE TABLE vandelay.bib_queue (
119 queue_type vandelay.bib_queue_queue_type NOT NULL DEFAULT 'bib',
120 item_attr_def BIGINT REFERENCES vandelay.import_item_attr_definition (id) ON DELETE SET NULL DEFERRABLE INITIALLY DEFERRED,
121 CONSTRAINT vand_bib_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
122 ) INHERITS (vandelay.queue);
123 ALTER TABLE vandelay.bib_queue ADD PRIMARY KEY (id);
125 CREATE TABLE vandelay.queued_bib_record (
126 queue INT NOT NULL REFERENCES vandelay.bib_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
127 bib_source INT REFERENCES config.bib_source (id) DEFERRABLE INITIALLY DEFERRED,
128 imported_as BIGINT REFERENCES biblio.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
129 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
131 ) INHERITS (vandelay.queued_record);
132 ALTER TABLE vandelay.queued_bib_record ADD PRIMARY KEY (id);
133 CREATE INDEX queued_bib_record_queue_idx ON vandelay.queued_bib_record (queue);
135 CREATE TABLE vandelay.queued_bib_record_attr (
136 id BIGSERIAL PRIMARY KEY,
137 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) DEFERRABLE INITIALLY DEFERRED,
138 field INT NOT NULL REFERENCES vandelay.bib_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
139 attr_value TEXT NOT NULL
141 CREATE INDEX queued_bib_record_attr_record_idx ON vandelay.queued_bib_record_attr (record);
143 CREATE TABLE vandelay.bib_match (
144 id BIGSERIAL PRIMARY KEY,
145 queued_record BIGINT REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
146 eg_record BIGINT REFERENCES biblio.record_entry (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
147 quality INT NOT NULL DEFAULT 1,
148 match_score INT NOT NULL DEFAULT 0
151 CREATE TABLE vandelay.import_item (
152 id BIGSERIAL PRIMARY KEY,
153 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
154 definition BIGINT NOT NULL REFERENCES vandelay.import_item_attr_definition (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
155 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
158 import_time TIMESTAMP WITH TIME ZONE,
167 deposit_amount NUMERIC(8,2),
178 internal_id BIGINT -- queue_type == 'acq' ? acq.lineitem_detail.id : asset.copy.id
181 CREATE TABLE vandelay.import_bib_trash_group(
182 id SERIAL PRIMARY KEY,
183 owner INTEGER NOT NULL REFERENCES actor.org_unit(id),
184 label TEXT NOT NULL, --i18n
185 always_apply BOOLEAN NOT NULL DEFAULT FALSE,
186 CONSTRAINT vand_import_bib_trash_grp_owner_label UNIQUE (owner, label)
189 CREATE TABLE vandelay.import_bib_trash_fields (
190 id BIGSERIAL PRIMARY KEY,
191 grp INTEGER NOT NULL REFERENCES vandelay.import_bib_trash_group,
193 CONSTRAINT vand_import_bib_trash_fields_once_per UNIQUE (grp, field)
196 CREATE TABLE vandelay.merge_profile (
197 id BIGSERIAL PRIMARY KEY,
198 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
205 CONSTRAINT vand_merge_prof_owner_name_idx UNIQUE (owner,name),
206 CONSTRAINT add_replace_strip_or_preserve CHECK ((preserve_spec IS NOT NULL OR replace_spec IS NOT NULL) OR (preserve_spec IS NULL AND replace_spec IS NULL))
209 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
216 retval config.marc21_rec_type_map%ROWTYPE;
218 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
220 IF ldr IS NULL OR ldr = '' THEN
221 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
225 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
226 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
229 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
230 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
232 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
234 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
237 IF retval.code IS NULL THEN
238 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
243 $func$ LANGUAGE PLPGSQL;
245 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
252 rtype := (vandelay.marc21_record_type( marc )).code;
253 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
254 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
255 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
258 val := REPEAT( ff_pos.default_val, ff_pos.length );
264 $func$ LANGUAGE PLPGSQL;
266 CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
267 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
272 output biblio.record_ff_map%ROWTYPE;
274 rtype := (vandelay.marc21_record_type( marc )).code;
276 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
277 output.ff_name := ff_pos.fixed_field;
278 output.ff_value := NULL;
280 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(tag) || '"]/text()', marc ) ) x(value) LOOP
281 output.ff_value := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
282 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
284 output.ff_value := NULL;
291 $func$ LANGUAGE PLPGSQL;
293 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
294 CREATE OR REPLACE FUNCTION vandelay.marc21_physical_characteristics( marc TEXT) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
298 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
299 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
300 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
301 retval biblio.marc21_physical_characteristics%ROWTYPE;
304 _007 := oils_xpath_string( '//*[@tag="007"]', marc );
306 IF _007 IS NOT NULL AND _007 <> '' THEN
307 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007, 1, 1 );
309 IF ptype.ptype_key IS NOT NULL THEN
310 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
311 SELECT * INTO pval FROM config.marc21_physical_characteristic_value_map WHERE ptype_subfield = psf.id AND value = SUBSTRING( _007, psf.start_pos + 1, psf.length );
313 IF pval.id IS NOT NULL THEN
316 retval.ptype := ptype.ptype_key;
317 retval.subfield := psf.id;
318 retval.value := pval.id;
328 $func$ LANGUAGE PLPGSQL;
330 CREATE TYPE vandelay.flat_marc AS ( tag CHAR(3), ind1 TEXT, ind2 TEXT, subfield TEXT, value TEXT );
331 CREATE OR REPLACE FUNCTION vandelay.flay_marc ( TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
334 use MARC::File::XML (BinaryEncoding => 'UTF-8');
338 MARC::Charset->assume_unicode(1);
341 my $r = MARC::Record->new_from_xml( $xml );
343 return_next( { tag => 'LDR', value => $r->leader } );
345 for my $f ( $r->fields ) {
346 if ($f->is_control_field) {
347 return_next({ tag => $f->tag, value => $f->data });
349 for my $s ($f->subfields) {
352 ind1 => $f->indicator(1),
353 ind2 => $f->indicator(2),
358 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
359 my $trim = $f->indicator(2) || 0;
362 ind1 => $f->indicator(1),
363 ind2 => $f->indicator(2),
365 value => substr( $s->[1], $trim )
374 $func$ LANGUAGE PLPERLU;
376 CREATE OR REPLACE FUNCTION vandelay.flatten_marc ( marc TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
378 output vandelay.flat_marc%ROWTYPE;
381 FOR field IN SELECT * FROM vandelay.flay_marc( marc ) LOOP
382 output.ind1 := field.ind1;
383 output.ind2 := field.ind2;
384 output.tag := field.tag;
385 output.subfield := field.subfield;
386 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
387 output.value := naco_normalize(field.value, field.subfield);
389 output.value := field.value;
392 CONTINUE WHEN output.value IS NULL;
397 $func$ LANGUAGE PLPGSQL;
399 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT, attr_defs TEXT[]) RETURNS hstore AS $_$
401 transformed_xml TEXT;
404 xfrm config.xml_transform%ROWTYPE;
406 new_attrs HSTORE := ''::HSTORE;
407 attr_def config.record_attr_definition%ROWTYPE;
410 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE name IN (SELECT * FROM UNNEST(attr_defs)) ORDER BY format LOOP
412 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
413 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(x.value), COALESCE(attr_def.joiner,' ')) INTO attr_value
414 FROM vandelay.flatten_marc(xml) AS x
415 WHERE x.tag LIKE attr_def.tag
417 WHEN attr_def.sf_list IS NOT NULL
418 THEN POSITION(x.subfield IN attr_def.sf_list) > 0
425 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
426 attr_value := vandelay.marc21_extract_fixed_field(xml, attr_def.fixed_field);
428 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
430 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
432 -- See if we can skip the XSLT ... it's expensive
433 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
434 -- Can't skip the transform
435 IF xfrm.xslt <> '---' THEN
436 transformed_xml := oils_xslt_process(xml,xfrm.xslt);
438 transformed_xml := xml;
441 prev_xfrm := xfrm.name;
444 IF xfrm.name IS NULL THEN
445 -- just grab the marcxml (empty) transform
446 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
447 prev_xfrm := xfrm.name;
450 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
452 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
453 SELECT m.value::TEXT INTO attr_value
454 FROM vandelay.marc21_physical_characteristics(xml) v
455 JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
456 WHERE v.subfield = attr_def.phys_char_sf
457 LIMIT 1; -- Just in case ...
461 -- apply index normalizers to attr_value
463 SELECT n.func AS func,
464 n.param_count AS param_count,
466 FROM config.index_normalizer n
467 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
468 WHERE attr = attr_def.name
470 EXECUTE 'SELECT ' || normalizer.func || '(' ||
471 quote_nullable( attr_value ) ||
473 WHEN normalizer.param_count > 0
474 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
481 -- Add the new value to the hstore
482 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
488 $_$ LANGUAGE PLPGSQL;
490 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT ) RETURNS hstore AS $_$
491 SELECT vandelay.extract_rec_attrs( $1, (SELECT ARRAY_ACCUM(name) FROM config.record_attr_definition));
494 -- Everything between this comment and the beginning of the definition of
495 -- vandelay.match_bib_record() is strictly in service of that function.
496 CREATE TYPE vandelay.match_set_test_result AS (record BIGINT, quality INTEGER);
498 CREATE OR REPLACE FUNCTION vandelay.match_set_test_marcxml(
499 match_set_id INTEGER, record_xml TEXT, bucket_id INTEGER
500 ) RETURNS SETOF vandelay.match_set_test_result AS $$
511 tags_rstore := vandelay.flatten_marc_hstore(record_xml);
512 svf_rstore := vandelay.extract_rec_attrs(record_xml);
514 CREATE TEMPORARY TABLE _vandelay_tmp_qrows (q INTEGER);
515 CREATE TEMPORARY TABLE _vandelay_tmp_jrows (j TEXT);
517 -- generate the where clause and return that directly (into wq), and as
518 -- a side-effect, populate the _vandelay_tmp_[qj]rows tables.
519 wq := vandelay.get_expr_from_match_set(match_set_id, tags_rstore);
521 query_ := 'SELECT DISTINCT(record), ';
523 -- qrows table is for the quality bits we add to the SELECT clause
524 SELECT ARRAY_TO_STRING(
525 ARRAY_ACCUM('COALESCE(n' || q::TEXT || '.quality, 0)'), ' + '
526 ) INTO coal FROM _vandelay_tmp_qrows;
528 -- our query string so far is the SELECT clause and the inital FROM.
529 -- no JOINs yet nor the WHERE clause
530 query_ := query_ || coal || ' AS quality ' || E'\n';
532 -- jrows table is for the joins we must make (and the real text conditions)
533 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(j), E'\n') INTO joins
534 FROM _vandelay_tmp_jrows;
536 -- add those joins and the where clause to our query.
537 query_ := query_ || joins || E'\n';
539 -- join the record bucket
540 IF bucket_id IS NOT NULL THEN
541 query_ := query_ || 'JOIN container.biblio_record_entry_bucket_item ' ||
542 'brebi ON (brebi.target_biblio_record_entry = record ' ||
543 'AND brebi.bucket = ' || bucket_id || E')\n';
546 query_ := query_ || 'JOIN biblio.record_entry bre ON (bre.id = record) ' || 'WHERE ' || wq || ' AND not bre.deleted';
548 -- this will return rows of record,quality
549 FOR rec IN EXECUTE query_ USING tags_rstore, svf_rstore LOOP
553 DROP TABLE _vandelay_tmp_qrows;
554 DROP TABLE _vandelay_tmp_jrows;
560 CREATE OR REPLACE FUNCTION vandelay.flatten_marc_hstore(
562 ) RETURNS HSTORE AS $func$
566 ARRAY_ACCUM(tag || (COALESCE(subfield, ''))),
570 SELECT tag, subfield, ARRAY_ACCUM(value)::TEXT AS value
573 CASE WHEN tag = '020' THEN -- caseless -- isbn
574 LOWER((REGEXP_MATCHES(value,$$^(\S{10,17})$$))[1] || '%')
575 WHEN tag = '022' THEN -- caseless -- issn
576 LOWER((REGEXP_MATCHES(value,$$^(\S{4}[- ]?\S{4})$$))[1] || '%')
577 WHEN tag = '024' THEN -- caseless -- upc (other)
582 FROM vandelay.flatten_marc(record_xml)) x
583 GROUP BY tag, subfield ORDER BY tag, subfield
587 $func$ LANGUAGE PLPGSQL;
589 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set(
590 match_set_id INTEGER,
594 root vandelay.match_set_point;
596 SELECT * INTO root FROM vandelay.match_set_point
597 WHERE parent IS NULL AND match_set = match_set_id;
599 RETURN vandelay.get_expr_from_match_set_point(root, tags_rstore);
603 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set_point(
604 node vandelay.match_set_point,
612 child vandelay.match_set_point;
614 SELECT ARRAY_ACCUM(id) INTO children FROM vandelay.match_set_point
615 WHERE parent = node.id;
617 IF ARRAY_LENGTH(children, 1) > 0 THEN
618 this_op := vandelay._get_expr_render_one(node);
621 WHILE children[i] IS NOT NULL LOOP
622 SELECT * INTO child FROM vandelay.match_set_point
623 WHERE id = children[i];
625 q := q || ' ' || this_op || ' ';
628 q := q || vandelay.get_expr_from_match_set_point(child, tags_rstore);
632 ELSIF node.bool_op IS NULL THEN
633 PERFORM vandelay._get_expr_push_qrow(node);
634 PERFORM vandelay._get_expr_push_jrow(node, tags_rstore);
635 RETURN vandelay._get_expr_render_one(node);
642 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_qrow(
643 node vandelay.match_set_point
647 INSERT INTO _vandelay_tmp_qrows (q) VALUES (node.id);
651 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_jrow(
652 node vandelay.match_set_point,
665 -- remember $1 is tags_rstore, and $2 is svf_rstore
668 SELECT COUNT(*) INTO jrow_count FROM _vandelay_tmp_jrows;
669 IF jrow_count > 0 THEN
670 my_using := ' USING (record)';
671 my_join := 'FULL OUTER JOIN';
677 IF node.tag IS NOT NULL THEN
678 caseless := (node.tag IN ('020', '022', '024'));
680 IF node.subfield IS NOT NULL THEN
681 tagkey := tagkey || node.subfield;
699 my_alias := 'n' || node.id::TEXT;
701 jrow := my_join || ' (SELECT *, ';
702 IF node.tag IS NOT NULL THEN
703 jrow := jrow || node.quality ||
704 ' AS quality FROM metabib.full_rec mfr WHERE mfr.tag = ''' ||
706 IF node.subfield IS NOT NULL THEN
707 jrow := jrow || ' AND mfr.subfield = ''' ||
708 node.subfield || '''';
710 jrow := jrow || ' AND (';
711 jrow := jrow || vandelay._node_tag_comparisons(caseless, op, tags_rstore, tagkey);
712 jrow := jrow || ')) ' || my_alias || my_using || E'\n';
714 jrow := jrow || 'id AS record, ' || node.quality ||
715 ' AS quality FROM metabib.record_attr mra WHERE mra.attrs->''' ||
716 node.svf || ''' ' || op || ' $2->''' || node.svf || ''') ' ||
717 my_alias || my_using || E'\n';
719 INSERT INTO _vandelay_tmp_jrows (j) VALUES (jrow);
723 CREATE OR REPLACE FUNCTION vandelay._node_tag_comparisons(
735 vals := tags_rstore->tagkey;
740 IF vals[i] IS NULL THEN
743 result := result || ' OR ';
748 result := result || 'LOWER(mfr.value) ' || op;
750 result := result || 'mfr.value ' || op;
753 result := result || ' ' || COALESCE('''' || vals[i] || '''', 'NULL');
755 IF vals[i] IS NULL THEN
766 CREATE OR REPLACE FUNCTION vandelay._get_expr_render_one(
767 node vandelay.match_set_point
772 IF node.bool_op IS NOT NULL THEN
775 RETURN '(n' || node.id::TEXT || '.id IS NOT NULL)';
780 CREATE OR REPLACE FUNCTION vandelay.match_bib_record() RETURNS TRIGGER AS $func$
782 incoming_existing_id TEXT;
783 test_result vandelay.match_set_test_result%ROWTYPE;
788 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
792 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
794 SELECT q.match_set INTO match_set FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
796 IF match_set IS NOT NULL THEN
797 NEW.quality := vandelay.measure_record_quality( NEW.marc, match_set );
800 -- Perfect matches on 901$c exit early with a match with high quality.
801 incoming_existing_id :=
802 oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]', NEW.marc);
804 IF incoming_existing_id IS NOT NULL AND incoming_existing_id != '' THEN
805 SELECT id INTO tmp_rec FROM biblio.record_entry WHERE id = incoming_existing_id::bigint;
806 IF tmp_rec IS NOT NULL THEN
807 INSERT INTO vandelay.bib_match (queued_record, eg_record, match_score, quality)
812 -- note: no match_set means quality==0
813 vandelay.measure_record_quality( b.marc, match_set )
814 FROM biblio.record_entry b
815 WHERE id = incoming_existing_id::bigint;
819 IF match_set IS NULL THEN
823 SELECT q.match_bucket INTO match_bucket FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
825 FOR test_result IN SELECT * FROM
826 vandelay.match_set_test_marcxml(match_set, NEW.marc, match_bucket) LOOP
828 INSERT INTO vandelay.bib_match ( queued_record, eg_record, match_score, quality )
833 vandelay.measure_record_quality( b.marc, match_set )
834 FROM biblio.record_entry b
835 WHERE id = test_result.record;
841 $func$ LANGUAGE PLPGSQL;
843 CREATE OR REPLACE FUNCTION vandelay.measure_record_quality ( xml TEXT, match_set_id INT ) RETURNS INT AS $_$
847 test vandelay.match_set_quality%ROWTYPE;
850 FOR test IN SELECT * FROM vandelay.match_set_quality WHERE match_set = match_set_id LOOP
851 IF test.tag IS NOT NULL THEN
852 FOR rvalue IN SELECT value FROM vandelay.flatten_marc( xml ) WHERE tag = test.tag AND subfield = test.subfield LOOP
853 IF test.value = rvalue THEN
854 out_q := out_q + test.quality;
858 IF test.value = vandelay.extract_rec_attrs(xml, ARRAY[test.svf]) -> test.svf THEN
859 out_q := out_q + test.quality;
866 $_$ LANGUAGE PLPGSQL;
868 CREATE TYPE vandelay.tcn_data AS (tcn TEXT, tcn_source TEXT, used BOOL);
869 CREATE OR REPLACE FUNCTION vandelay.find_bib_tcn_data ( xml TEXT ) RETURNS SETOF vandelay.tcn_data AS $_$
873 output vandelay.tcn_data%ROWTYPE;
877 eg_tcn := BTRIM((oils_xpath('//*[@tag="001"]/text()',xml))[1]);
878 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
880 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="003"]/text()',xml))[1]);
881 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
882 eg_tcn_source := 'System Local';
885 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
888 output.used := FALSE;
893 output.tcn := eg_tcn;
894 output.tcn_source := eg_tcn_source;
900 eg_tcn := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="a"]/text()',xml))[1]);
901 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
903 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="b"]/text()',xml))[1]);
904 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
905 eg_tcn_source := 'System Local';
908 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
911 output.used := FALSE;
916 output.tcn := eg_tcn;
917 output.tcn_source := eg_tcn_source;
923 eg_tcn := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="a"]/text()',xml))[1]);
924 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
926 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="b"]/text()',xml))[1]);
927 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
928 eg_tcn_source := 'System Local';
931 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
934 output.used := FALSE;
939 output.tcn := eg_tcn;
940 output.tcn_source := eg_tcn_source;
946 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="020"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
947 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
949 eg_tcn_source := 'ISBN';
951 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
954 output.used := FALSE;
959 output.tcn := eg_tcn;
960 output.tcn_source := eg_tcn_source;
966 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="022"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
967 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
969 eg_tcn_source := 'ISSN';
971 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
974 output.used := FALSE;
979 output.tcn := eg_tcn;
980 output.tcn_source := eg_tcn_source;
986 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="010"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
987 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
989 eg_tcn_source := 'LCCN';
991 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
994 output.used := FALSE;
999 output.tcn := eg_tcn;
1000 output.tcn_source := eg_tcn_source;
1006 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="035"]/*[@code="a"]/text()',xml))[1], $re$^.*?(\w+)$$re$, $re$\1$re$);
1007 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
1009 eg_tcn_source := 'System Legacy';
1011 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1014 output.used := FALSE;
1016 output.used := TRUE;
1019 output.tcn := eg_tcn;
1020 output.tcn_source := eg_tcn_source;
1027 $_$ LANGUAGE PLPGSQL;
1029 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT, force_add INT ) RETURNS TEXT AS $_$
1032 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1036 MARC::Charset->assume_unicode(1);
1038 my $target_xml = shift;
1039 my $source_xml = shift;
1040 my $field_spec = shift;
1041 my $force_add = shift || 0;
1043 my $target_r = MARC::Record->new_from_xml( $target_xml );
1044 my $source_r = MARC::Record->new_from_xml( $source_xml );
1046 return $target_xml unless ($target_r && $source_r);
1048 my @field_list = split(',', $field_spec);
1051 for my $f (@field_list) {
1052 $f =~ s/^\s*//; $f =~ s/\s*$//;
1053 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1059 $match =~ s/^\s*//; $match =~ s/\s*$//;
1060 $fields{$field} = { sf => [ split('', $sf) ] };
1062 my ($msf,$mre) = split('~', $match);
1063 if (length($msf) > 0 and length($mre) > 0) {
1064 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1065 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1066 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1072 for my $f ( keys %fields) {
1073 if ( @{$fields{$f}{sf}} ) {
1074 for my $from_field ($source_r->field( $f )) {
1075 my @tos = $target_r->field( $f );
1077 next if (exists($fields{$f}{match}) and !$force_add);
1078 my @new_fields = map { $_->clone } $source_r->field( $f );
1079 $target_r->insert_fields_ordered( @new_fields );
1081 for my $to_field (@tos) {
1082 if (exists($fields{$f}{match})) {
1083 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1085 my @new_sf = map { ($_ => $from_field->subfield($_)) } grep { defined($from_field->subfield($_)) } @{$fields{$f}{sf}};
1086 $to_field->add_subfields( @new_sf );
1091 my @new_fields = map { $_->clone } $source_r->field( $f );
1092 $target_r->insert_fields_ordered( @new_fields );
1096 $target_xml = $target_r->as_xml_record;
1097 $target_xml =~ s/^<\?.+?\?>$//mo;
1098 $target_xml =~ s/\n//sgo;
1099 $target_xml =~ s/>\s+</></sgo;
1103 $_$ LANGUAGE PLPERLU;
1105 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1106 SELECT vandelay.add_field( $1, $2, $3, 0 );
1109 CREATE OR REPLACE FUNCTION vandelay.strip_field ( xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1112 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1116 MARC::Charset->assume_unicode(1);
1119 my $r = MARC::Record->new_from_xml( $xml );
1121 return $xml unless ($r);
1123 my $field_spec = shift;
1124 my @field_list = split(',', $field_spec);
1127 for my $f (@field_list) {
1128 $f =~ s/^\s*//; $f =~ s/\s*$//;
1129 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1135 $match =~ s/^\s*//; $match =~ s/\s*$//;
1136 $fields{$field} = { sf => [ split('', $sf) ] };
1138 my ($msf,$mre) = split('~', $match);
1139 if (length($msf) > 0 and length($mre) > 0) {
1140 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1141 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1142 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1148 for my $f ( keys %fields) {
1149 for my $to_field ($r->field( $f )) {
1150 if (exists($fields{$f}{match})) {
1151 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1154 if ( @{$fields{$f}{sf}} ) {
1155 $to_field->delete_subfield(code => $fields{$f}{sf});
1157 $r->delete_field( $to_field );
1162 $xml = $r->as_xml_record;
1163 $xml =~ s/^<\?.+?\?>$//mo;
1165 $xml =~ s/>\s+</></sgo;
1169 $_$ LANGUAGE PLPERLU;
1171 CREATE OR REPLACE FUNCTION vandelay.replace_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1178 parsed_target := vandelay.strip_field( target_xml, ''); -- this dance normalizes the format of the xml for the IF below
1179 xml_output := parsed_target; -- if there are no replace rules, just return the input
1181 FOR curr_field IN SELECT UNNEST( STRING_TO_ARRAY(field, ',') ) LOOP -- naive split, but it's the same we use in the perl
1183 xml_output := vandelay.strip_field( parsed_target, curr_field);
1185 IF xml_output <> parsed_target AND curr_field ~ E'~' THEN
1186 -- we removed something, and there was a regexp restriction in the curr_field definition, so proceed
1187 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 1 );
1188 ELSIF curr_field !~ E'~' THEN
1189 -- No regexp restriction, add the curr_field
1190 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 0 );
1193 parsed_target := xml_output; -- in prep for any following loop iterations
1199 $_$ LANGUAGE PLPGSQL;
1201 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 $_$
1202 SELECT vandelay.replace_field( vandelay.add_field( vandelay.strip_field( $1, $5) , $2, $3 ), $2, $4);
1205 CREATE TYPE vandelay.compile_profile AS (add_rule TEXT, replace_rule TEXT, preserve_rule TEXT, strip_rule TEXT);
1206 CREATE OR REPLACE FUNCTION vandelay.compile_profile ( incoming_xml TEXT ) RETURNS vandelay.compile_profile AS $_$
1208 output vandelay.compile_profile%ROWTYPE;
1209 profile vandelay.merge_profile%ROWTYPE;
1211 profile_tmpl_owner TEXT;
1212 add_rule TEXT := '';
1213 strip_rule TEXT := '';
1214 replace_rule TEXT := '';
1215 preserve_rule TEXT := '';
1219 profile_tmpl := (oils_xpath('//*[@tag="905"]/*[@code="t"]/text()',incoming_xml))[1];
1220 profile_tmpl_owner := (oils_xpath('//*[@tag="905"]/*[@code="o"]/text()',incoming_xml))[1];
1222 IF profile_tmpl IS NOT NULL AND profile_tmpl <> '' AND profile_tmpl_owner IS NOT NULL AND profile_tmpl_owner <> '' THEN
1223 SELECT p.* INTO profile
1224 FROM vandelay.merge_profile p
1225 JOIN actor.org_unit u ON (u.id = p.owner)
1226 WHERE p.name = profile_tmpl
1227 AND u.shortname = profile_tmpl_owner;
1229 IF profile.id IS NOT NULL THEN
1230 add_rule := COALESCE(profile.add_spec,'');
1231 strip_rule := COALESCE(profile.strip_spec,'');
1232 replace_rule := COALESCE(profile.replace_spec,'');
1233 preserve_rule := COALESCE(profile.preserve_spec,'');
1237 add_rule := add_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="a"]/text()',incoming_xml),','),'');
1238 strip_rule := strip_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="d"]/text()',incoming_xml),','),'');
1239 replace_rule := replace_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="r"]/text()',incoming_xml),','),'');
1240 preserve_rule := preserve_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="p"]/text()',incoming_xml),','),'');
1242 output.add_rule := BTRIM(add_rule,',');
1243 output.replace_rule := BTRIM(replace_rule,',');
1244 output.strip_rule := BTRIM(strip_rule,',');
1245 output.preserve_rule := BTRIM(preserve_rule,',');
1249 $_$ LANGUAGE PLPGSQL;
1251 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1253 merge_profile vandelay.merge_profile%ROWTYPE;
1254 dyn_profile vandelay.compile_profile%ROWTYPE;
1264 SELECT b.marc INTO eg_marc
1265 FROM biblio.record_entry b
1269 IF eg_marc IS NULL OR v_marc IS NULL THEN
1270 -- RAISE NOTICE 'no marc for template or bib record';
1274 dyn_profile := vandelay.compile_profile( v_marc );
1276 IF merge_profile_id IS NOT NULL THEN
1277 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1279 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1280 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1281 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1282 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1286 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1287 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1291 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1292 --Since we have nothing to do, just return a NOOP "we did it"
1294 ELSIF dyn_profile.replace_rule <> '' THEN
1295 source_marc = v_marc;
1296 target_marc = eg_marc;
1297 replace_rule = dyn_profile.replace_rule;
1299 source_marc = eg_marc;
1300 target_marc = v_marc;
1301 replace_rule = dyn_profile.preserve_rule;
1304 UPDATE biblio.record_entry
1305 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1309 -- RAISE NOTICE 'update of biblio.record_entry failed';
1316 $$ LANGUAGE PLPGSQL;
1318 CREATE OR REPLACE FUNCTION vandelay.merge_record_xml ( target_marc TEXT, template_marc TEXT ) RETURNS TEXT AS $$
1320 dyn_profile vandelay.compile_profile%ROWTYPE;
1328 IF target_marc IS NULL OR template_marc IS NULL THEN
1329 -- RAISE NOTICE 'no marc for target or template record';
1333 dyn_profile := vandelay.compile_profile( template_marc );
1335 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1336 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1340 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1341 --Since we have nothing to do, just return what we were given.
1343 ELSIF dyn_profile.replace_rule <> '' THEN
1344 trgt_marc = target_marc;
1345 tmpl_marc = template_marc;
1346 replace_rule = dyn_profile.replace_rule;
1348 tmp_marc = target_marc;
1349 trgt_marc = template_marc;
1350 tmpl_marc = tmp_marc;
1351 replace_rule = dyn_profile.preserve_rule;
1354 RETURN vandelay.merge_record_xml( trgt_marc, tmpl_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule );
1357 $$ LANGUAGE PLPGSQL;
1359 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT) RETURNS BOOL AS $$
1360 SELECT vandelay.template_overlay_bib_record( $1, $2, NULL);
1363 CREATE OR REPLACE FUNCTION vandelay.overlay_bib_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1365 merge_profile vandelay.merge_profile%ROWTYPE;
1366 dyn_profile vandelay.compile_profile%ROWTYPE;
1376 SELECT q.marc INTO v_marc
1377 FROM vandelay.queued_record q
1378 JOIN vandelay.bib_match m ON (m.queued_record = q.id AND q.id = import_id)
1381 IF v_marc IS NULL THEN
1382 -- RAISE NOTICE 'no marc for vandelay or bib record';
1386 IF vandelay.template_overlay_bib_record( v_marc, eg_id, merge_profile_id) THEN
1387 UPDATE vandelay.queued_bib_record
1388 SET imported_as = eg_id,
1390 WHERE id = import_id;
1392 editor_string := (oils_xpath('//*[@tag="905"]/*[@code="u"]/text()',v_marc))[1];
1394 IF editor_string IS NOT NULL AND editor_string <> '' THEN
1395 SELECT usr INTO editor_id FROM actor.card WHERE barcode = editor_string;
1397 IF editor_id IS NULL THEN
1398 SELECT id INTO editor_id FROM actor.usr WHERE usrname = editor_string;
1401 IF editor_id IS NOT NULL THEN
1402 UPDATE biblio.record_entry SET editor = editor_id WHERE id = eg_id;
1409 -- RAISE NOTICE 'update of biblio.record_entry failed';
1414 $$ LANGUAGE PLPGSQL;
1416 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
1419 lwm_ratio_value NUMERIC;
1422 lwm_ratio_value := COALESCE(lwm_ratio_value_p, 0.0);
1424 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1427 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1431 SELECT m.eg_record INTO eg_id
1432 FROM vandelay.bib_match m
1433 JOIN vandelay.queued_bib_record qr ON (m.queued_record = qr.id)
1434 JOIN vandelay.bib_queue q ON (qr.queue = q.id)
1435 JOIN biblio.record_entry r ON (r.id = m.eg_record)
1436 WHERE m.queued_record = import_id
1437 AND qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC >= lwm_ratio_value
1438 ORDER BY m.match_score DESC, -- required match score
1439 qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC DESC, -- quality tie breaker
1440 m.id -- when in doubt, use the first match
1443 IF eg_id IS NULL THEN
1444 -- RAISE NOTICE 'incoming record is not of high enough quality';
1448 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1450 $$ LANGUAGE PLPGSQL;
1452 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1453 SELECT vandelay.auto_overlay_bib_record_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1456 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1462 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1465 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1469 SELECT COUNT(*) INTO match_count FROM vandelay.bib_match WHERE queued_record = import_id;
1471 IF match_count <> 1 THEN
1472 -- RAISE NOTICE 'not an exact match';
1476 -- Check that the one match is on the first 901c
1477 SELECT m.eg_record INTO eg_id
1478 FROM vandelay.queued_bib_record q
1479 JOIN vandelay.bib_match m ON (m.queued_record = q.id)
1480 WHERE q.id = import_id
1481 AND m.eg_record = oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]',marc)::BIGINT;
1484 -- RAISE NOTICE 'not a 901c match';
1488 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1490 $$ LANGUAGE PLPGSQL;
1492 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1494 queued_record vandelay.queued_bib_record%ROWTYPE;
1497 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1499 IF vandelay.auto_overlay_bib_record( queued_record.id, merge_profile_id ) THEN
1500 RETURN NEXT queued_record.id;
1508 $$ LANGUAGE PLPGSQL;
1510 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 $$
1512 queued_record vandelay.queued_bib_record%ROWTYPE;
1515 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1517 IF vandelay.auto_overlay_bib_record_with_best( queued_record.id, merge_profile_id, lwm_ratio_value ) THEN
1518 RETURN NEXT queued_record.id;
1526 $$ LANGUAGE PLPGSQL;
1528 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1529 SELECT vandelay.auto_overlay_bib_queue_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1532 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1533 SELECT * FROM vandelay.auto_overlay_bib_queue( $1, NULL );
1536 CREATE OR REPLACE FUNCTION vandelay.ingest_bib_marc ( ) RETURNS TRIGGER AS $$
1542 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1546 FOR adef IN SELECT * FROM vandelay.bib_attr_definition LOOP
1548 SELECT extract_marc_field('vandelay.queued_bib_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_bib_record WHERE id = NEW.id;
1549 IF (value IS NOT NULL AND value <> '') THEN
1550 INSERT INTO vandelay.queued_bib_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1557 $$ LANGUAGE PLPGSQL;
1559 CREATE OR REPLACE FUNCTION vandelay.cleanup_bib_marc ( ) RETURNS TRIGGER AS $$
1561 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1565 DELETE FROM vandelay.queued_bib_record_attr WHERE record = OLD.id;
1566 DELETE FROM vandelay.import_item WHERE record = OLD.id;
1568 IF TG_OP = 'UPDATE' THEN
1573 $$ LANGUAGE PLPGSQL;
1575 CREATE TRIGGER cleanup_bib_trigger
1576 BEFORE UPDATE OR DELETE ON vandelay.queued_bib_record
1577 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_bib_marc();
1579 CREATE TRIGGER ingest_bib_trigger
1580 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1581 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_bib_marc();
1583 CREATE TRIGGER zz_match_bibs_trigger
1584 BEFORE INSERT OR UPDATE ON vandelay.queued_bib_record
1585 FOR EACH ROW EXECUTE PROCEDURE vandelay.match_bib_record();
1588 /* Authority stuff down here */
1589 ---------------------------------------
1590 CREATE TABLE vandelay.authority_attr_definition (
1591 id SERIAL PRIMARY KEY,
1592 code TEXT UNIQUE NOT NULL,
1594 xpath TEXT NOT NULL,
1595 remove TEXT NOT NULL DEFAULT ''
1598 CREATE TYPE vandelay.authority_queue_queue_type AS ENUM ('authority');
1599 CREATE TABLE vandelay.authority_queue (
1600 queue_type vandelay.authority_queue_queue_type NOT NULL DEFAULT 'authority',
1601 CONSTRAINT vand_authority_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
1602 ) INHERITS (vandelay.queue);
1603 ALTER TABLE vandelay.authority_queue ADD PRIMARY KEY (id);
1605 CREATE TABLE vandelay.queued_authority_record (
1606 queue INT NOT NULL REFERENCES vandelay.authority_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1607 imported_as INT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1608 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
1610 ) INHERITS (vandelay.queued_record);
1611 ALTER TABLE vandelay.queued_authority_record ADD PRIMARY KEY (id);
1612 CREATE INDEX queued_authority_record_queue_idx ON vandelay.queued_authority_record (queue);
1614 CREATE TABLE vandelay.queued_authority_record_attr (
1615 id BIGSERIAL PRIMARY KEY,
1616 record BIGINT NOT NULL REFERENCES vandelay.queued_authority_record (id) DEFERRABLE INITIALLY DEFERRED,
1617 field INT NOT NULL REFERENCES vandelay.authority_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
1618 attr_value TEXT NOT NULL
1620 CREATE INDEX queued_authority_record_attr_record_idx ON vandelay.queued_authority_record_attr (record);
1622 CREATE TABLE vandelay.authority_match (
1623 id BIGSERIAL PRIMARY KEY,
1624 queued_record BIGINT REFERENCES vandelay.queued_authority_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1625 eg_record BIGINT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1626 quality INT NOT NULL DEFAULT 0
1629 CREATE OR REPLACE FUNCTION vandelay.ingest_authority_marc ( ) RETURNS TRIGGER AS $$
1635 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1639 FOR adef IN SELECT * FROM vandelay.authority_attr_definition LOOP
1641 SELECT extract_marc_field('vandelay.queued_authority_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_authority_record WHERE id = NEW.id;
1642 IF (value IS NOT NULL AND value <> '') THEN
1643 INSERT INTO vandelay.queued_authority_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1650 $$ LANGUAGE PLPGSQL;
1652 CREATE OR REPLACE FUNCTION vandelay.cleanup_authority_marc ( ) RETURNS TRIGGER AS $$
1654 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1658 DELETE FROM vandelay.queued_authority_record_attr WHERE record = OLD.id;
1659 IF TG_OP = 'UPDATE' THEN
1664 $$ LANGUAGE PLPGSQL;
1666 CREATE TRIGGER cleanup_authority_trigger
1667 BEFORE UPDATE OR DELETE ON vandelay.queued_authority_record
1668 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_authority_marc();
1670 CREATE TRIGGER ingest_authority_trigger
1671 AFTER INSERT OR UPDATE ON vandelay.queued_authority_record
1672 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_authority_marc();
1674 CREATE OR REPLACE FUNCTION vandelay.overlay_authority_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1676 merge_profile vandelay.merge_profile%ROWTYPE;
1677 dyn_profile vandelay.compile_profile%ROWTYPE;
1686 SELECT b.marc INTO eg_marc
1687 FROM authority.record_entry b
1688 JOIN vandelay.authority_match m ON (m.eg_record = b.id AND m.queued_record = import_id)
1691 SELECT q.marc INTO v_marc
1692 FROM vandelay.queued_record q
1693 JOIN vandelay.authority_match m ON (m.queued_record = q.id AND q.id = import_id)
1696 IF eg_marc IS NULL OR v_marc IS NULL THEN
1697 -- RAISE NOTICE 'no marc for vandelay or authority record';
1701 dyn_profile := vandelay.compile_profile( v_marc );
1703 IF merge_profile_id IS NOT NULL THEN
1704 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1706 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1707 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1708 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1709 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1713 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1714 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1718 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1719 --Since we have nothing to do, just return a NOOP "we did it"
1721 ELSIF dyn_profile.replace_rule <> '' THEN
1722 source_marc = v_marc;
1723 target_marc = eg_marc;
1724 replace_rule = dyn_profile.replace_rule;
1726 source_marc = eg_marc;
1727 target_marc = v_marc;
1728 replace_rule = dyn_profile.preserve_rule;
1731 UPDATE authority.record_entry
1732 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1736 UPDATE vandelay.queued_authority_record
1737 SET imported_as = eg_id,
1739 WHERE id = import_id;
1743 -- RAISE NOTICE 'update of authority.record_entry failed';
1748 $$ LANGUAGE PLPGSQL;
1750 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1755 SELECT COUNT(*) INTO match_count FROM vandelay.authority_match WHERE queued_record = import_id;
1757 IF match_count <> 1 THEN
1758 -- RAISE NOTICE 'not an exact match';
1762 SELECT m.eg_record INTO eg_id
1763 FROM vandelay.authority_match m
1764 WHERE m.queued_record = import_id
1767 IF eg_id IS NULL THEN
1771 RETURN vandelay.overlay_authority_record( import_id, eg_id, merge_profile_id );
1773 $$ LANGUAGE PLPGSQL;
1775 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1777 queued_record vandelay.queued_authority_record%ROWTYPE;
1780 FOR queued_record IN SELECT * FROM vandelay.queued_authority_record WHERE queue = queue_id AND import_time IS NULL LOOP
1782 IF vandelay.auto_overlay_authority_record( queued_record.id, merge_profile_id ) THEN
1783 RETURN NEXT queued_record.id;
1791 $$ LANGUAGE PLPGSQL;
1793 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1794 SELECT * FROM vandelay.auto_overlay_authority_queue( $1, NULL );
1798 -- Vandelay (for importing and exporting records) 012.schema.vandelay.sql
1799 --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)]');
1800 --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)]');
1801 --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]');
1802 --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]');
1803 --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$);
1804 --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$);
1805 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (7, 'price',oils_i18n_gettext(7, 'vqbrad', 'Price', 'description'),'//*[@tag="020" or @tag="022"]/*[@code="c"][1]');
1806 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (8, 'rec_identifier',oils_i18n_gettext(8, 'vqbrad', 'Accession Number', 'description'),'//*[@tag="001"]', TRUE);
1807 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (9, 'eg_tcn',oils_i18n_gettext(9, 'vqbrad', 'TCN Value', 'description'),'//*[@tag="901"]/*[@code="a"]', TRUE);
1808 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (10, 'eg_tcn_source',oils_i18n_gettext(10, 'vqbrad', 'TCN Source', 'description'),'//*[@tag="901"]/*[@code="b"]', TRUE);
1809 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (11, 'eg_identifier',oils_i18n_gettext(11, 'vqbrad', 'Internal ID', 'description'),'//*[@tag="901"]/*[@code="c"]', TRUE);
1810 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (12, 'publisher',oils_i18n_gettext(12, 'vqbrad', 'Publisher', 'description'),'//*[@tag="260"]/*[@code="b"][1]');
1811 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, remove ) VALUES (13, 'pubdate',oils_i18n_gettext(13, 'vqbrad', 'Publication Date', 'description'),'//*[@tag="260"]/*[@code="c"][1]',$r$\D$r$);
1812 --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]');
1814 --INSERT INTO vandelay.import_item_attr_definition (
1815 -- owner, name, tag, owning_lib, circ_lib, location,
1816 -- call_number, circ_modifier, barcode, price, copy_number,
1817 -- circulate, ref, holdable, opac_visible, status
1820 -- 'Evergreen 852 export format',
1822 -- '[@code = "b"][1]',
1823 -- '[@code = "b"][2]',
1830 -- '[@code = "x" and text() = "circulating"]',
1831 -- '[@code = "x" and text() = "reference"]',
1832 -- '[@code = "x" and text() = "holdable"]',
1833 -- '[@code = "x" and text() = "visible"]',
1837 --INSERT INTO vandelay.import_item_attr_definition (
1850 -- 'Unicorn Import format -- 999',
1861 --INSERT INTO vandelay.authority_attr_definition ( code, description, xpath, ident ) VALUES ('rec_identifier','Identifier','//*[@tag="001"]', TRUE);