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 match_bucket INTEGER, -- REFERENCES container.biblio_record_entry_bucket(id);
122 CONSTRAINT vand_bib_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
123 ) INHERITS (vandelay.queue);
124 ALTER TABLE vandelay.bib_queue ADD PRIMARY KEY (id);
126 CREATE TABLE vandelay.queued_bib_record (
127 queue INT NOT NULL REFERENCES vandelay.bib_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
128 bib_source INT REFERENCES config.bib_source (id) DEFERRABLE INITIALLY DEFERRED,
129 imported_as BIGINT REFERENCES biblio.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
130 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
132 ) INHERITS (vandelay.queued_record);
133 ALTER TABLE vandelay.queued_bib_record ADD PRIMARY KEY (id);
134 CREATE INDEX queued_bib_record_queue_idx ON vandelay.queued_bib_record (queue);
136 CREATE TABLE vandelay.queued_bib_record_attr (
137 id BIGSERIAL PRIMARY KEY,
138 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) DEFERRABLE INITIALLY DEFERRED,
139 field INT NOT NULL REFERENCES vandelay.bib_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
140 attr_value TEXT NOT NULL
142 CREATE INDEX queued_bib_record_attr_record_idx ON vandelay.queued_bib_record_attr (record);
144 CREATE TABLE vandelay.bib_match (
145 id BIGSERIAL PRIMARY KEY,
146 queued_record BIGINT REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
147 eg_record BIGINT REFERENCES biblio.record_entry (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
148 quality INT NOT NULL DEFAULT 1,
149 match_score INT NOT NULL DEFAULT 0
152 CREATE TABLE vandelay.import_item (
153 id BIGSERIAL PRIMARY KEY,
154 record BIGINT NOT NULL REFERENCES vandelay.queued_bib_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
155 definition BIGINT NOT NULL REFERENCES vandelay.import_item_attr_definition (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
156 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
159 import_time TIMESTAMP WITH TIME ZONE,
168 deposit_amount NUMERIC(8,2),
179 internal_id BIGINT -- queue_type == 'acq' ? acq.lineitem_detail.id : asset.copy.id
182 CREATE TABLE vandelay.import_bib_trash_group(
183 id SERIAL PRIMARY KEY,
184 owner INTEGER NOT NULL REFERENCES actor.org_unit(id),
185 label TEXT NOT NULL, --i18n
186 always_apply BOOLEAN NOT NULL DEFAULT FALSE,
187 CONSTRAINT vand_import_bib_trash_grp_owner_label UNIQUE (owner, label)
190 CREATE TABLE vandelay.import_bib_trash_fields (
191 id BIGSERIAL PRIMARY KEY,
192 grp INTEGER NOT NULL REFERENCES vandelay.import_bib_trash_group,
194 CONSTRAINT vand_import_bib_trash_fields_once_per UNIQUE (grp, field)
197 CREATE TABLE vandelay.merge_profile (
198 id BIGSERIAL PRIMARY KEY,
199 owner INT NOT NULL REFERENCES actor.org_unit (id) ON UPDATE CASCADE ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
206 CONSTRAINT vand_merge_prof_owner_name_idx UNIQUE (owner,name),
207 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))
210 CREATE OR REPLACE FUNCTION vandelay.marc21_record_type( marc TEXT ) RETURNS config.marc21_rec_type_map AS $func$
217 retval config.marc21_rec_type_map%ROWTYPE;
219 ldr := oils_xpath_string( '//*[local-name()="leader"]', marc );
221 IF ldr IS NULL OR ldr = '' THEN
222 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
226 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
227 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
230 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
231 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
233 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
235 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
238 IF retval.code IS NULL THEN
239 SELECT * INTO retval FROM config.marc21_rec_type_map WHERE code = 'BKS';
244 $func$ LANGUAGE PLPGSQL;
246 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_fixed_field( marc TEXT, ff TEXT ) RETURNS TEXT AS $func$
253 rtype := (vandelay.marc21_record_type( marc )).code;
254 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = ff AND rec_type = rtype ORDER BY tag DESC LOOP
255 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(ff_pos.tag) || '"]/text()', marc ) ) x(value) LOOP
256 val := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
259 val := REPEAT( ff_pos.default_val, ff_pos.length );
265 $func$ LANGUAGE PLPGSQL;
267 CREATE TYPE biblio.record_ff_map AS (record BIGINT, ff_name TEXT, ff_value TEXT);
268 CREATE OR REPLACE FUNCTION vandelay.marc21_extract_all_fixed_fields( marc TEXT ) RETURNS SETOF biblio.record_ff_map AS $func$
273 output biblio.record_ff_map%ROWTYPE;
275 rtype := (vandelay.marc21_record_type( marc )).code;
277 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE rec_type = rtype ORDER BY tag DESC LOOP
278 output.ff_name := ff_pos.fixed_field;
279 output.ff_value := NULL;
281 FOR tag_data IN SELECT value FROM UNNEST( oils_xpath( '//*[@tag="' || UPPER(tag) || '"]/text()', marc ) ) x(value) LOOP
282 output.ff_value := SUBSTRING( tag_data.value, ff_pos.start_pos + 1, ff_pos.length );
283 IF output.ff_value IS NULL THEN output.ff_value := REPEAT( ff_pos.default_val, ff_pos.length ); END IF;
285 output.ff_value := NULL;
292 $func$ LANGUAGE PLPGSQL;
294 CREATE TYPE biblio.marc21_physical_characteristics AS ( id INT, record BIGINT, ptype TEXT, subfield INT, value INT );
295 CREATE OR REPLACE FUNCTION vandelay.marc21_physical_characteristics( marc TEXT) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
299 ptype config.marc21_physical_characteristic_type_map%ROWTYPE;
300 psf config.marc21_physical_characteristic_subfield_map%ROWTYPE;
301 pval config.marc21_physical_characteristic_value_map%ROWTYPE;
302 retval biblio.marc21_physical_characteristics%ROWTYPE;
305 _007 := oils_xpath_string( '//*[@tag="007"]', marc );
307 IF _007 IS NOT NULL AND _007 <> '' THEN
308 SELECT * INTO ptype FROM config.marc21_physical_characteristic_type_map WHERE ptype_key = SUBSTRING( _007, 1, 1 );
310 IF ptype.ptype_key IS NOT NULL THEN
311 FOR psf IN SELECT * FROM config.marc21_physical_characteristic_subfield_map WHERE ptype_key = ptype.ptype_key LOOP
312 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 );
314 IF pval.id IS NOT NULL THEN
317 retval.ptype := ptype.ptype_key;
318 retval.subfield := psf.id;
319 retval.value := pval.id;
329 $func$ LANGUAGE PLPGSQL;
331 CREATE TYPE vandelay.flat_marc AS ( tag CHAR(3), ind1 TEXT, ind2 TEXT, subfield TEXT, value TEXT );
332 CREATE OR REPLACE FUNCTION vandelay.flay_marc ( TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
335 use MARC::File::XML (BinaryEncoding => 'UTF-8');
339 MARC::Charset->assume_unicode(1);
342 my $r = MARC::Record->new_from_xml( $xml );
344 return_next( { tag => 'LDR', value => $r->leader } );
346 for my $f ( $r->fields ) {
347 if ($f->is_control_field) {
348 return_next({ tag => $f->tag, value => $f->data });
350 for my $s ($f->subfields) {
353 ind1 => $f->indicator(1),
354 ind2 => $f->indicator(2),
359 if ( $f->tag eq '245' and $s->[0] eq 'a' ) {
360 my $trim = $f->indicator(2) || 0;
363 ind1 => $f->indicator(1),
364 ind2 => $f->indicator(2),
366 value => substr( $s->[1], $trim )
375 $func$ LANGUAGE PLPERLU;
377 CREATE OR REPLACE FUNCTION vandelay.flatten_marc ( marc TEXT ) RETURNS SETOF vandelay.flat_marc AS $func$
379 output vandelay.flat_marc%ROWTYPE;
382 FOR field IN SELECT * FROM vandelay.flay_marc( marc ) LOOP
383 output.ind1 := field.ind1;
384 output.ind2 := field.ind2;
385 output.tag := field.tag;
386 output.subfield := field.subfield;
387 IF field.subfield IS NOT NULL AND field.tag NOT IN ('020','022','024') THEN -- exclude standard numbers and control fields
388 output.value := naco_normalize(field.value, field.subfield);
390 output.value := field.value;
393 CONTINUE WHEN output.value IS NULL;
398 $func$ LANGUAGE PLPGSQL;
400 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT, attr_defs TEXT[]) RETURNS hstore AS $_$
402 transformed_xml TEXT;
405 xfrm config.xml_transform%ROWTYPE;
407 new_attrs HSTORE := ''::HSTORE;
408 attr_def config.record_attr_definition%ROWTYPE;
411 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE name IN (SELECT * FROM UNNEST(attr_defs)) ORDER BY format LOOP
413 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
414 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(x.value), COALESCE(attr_def.joiner,' ')) INTO attr_value
415 FROM vandelay.flatten_marc(xml) AS x
416 WHERE x.tag LIKE attr_def.tag
418 WHEN attr_def.sf_list IS NOT NULL
419 THEN POSITION(x.subfield IN attr_def.sf_list) > 0
426 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
427 attr_value := vandelay.marc21_extract_fixed_field(xml, attr_def.fixed_field);
429 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
431 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
433 -- See if we can skip the XSLT ... it's expensive
434 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
435 -- Can't skip the transform
436 IF xfrm.xslt <> '---' THEN
437 transformed_xml := oils_xslt_process(xml,xfrm.xslt);
439 transformed_xml := xml;
442 prev_xfrm := xfrm.name;
445 IF xfrm.name IS NULL THEN
446 -- just grab the marcxml (empty) transform
447 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
448 prev_xfrm := xfrm.name;
451 attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]);
453 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
454 SELECT m.value::TEXT INTO attr_value
455 FROM vandelay.marc21_physical_characteristics(xml) v
456 JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
457 WHERE v.subfield = attr_def.phys_char_sf
458 LIMIT 1; -- Just in case ...
462 -- apply index normalizers to attr_value
464 SELECT n.func AS func,
465 n.param_count AS param_count,
467 FROM config.index_normalizer n
468 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
469 WHERE attr = attr_def.name
471 EXECUTE 'SELECT ' || normalizer.func || '(' ||
472 quote_nullable( attr_value ) ||
474 WHEN normalizer.param_count > 0
475 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
482 -- Add the new value to the hstore
483 new_attrs := new_attrs || hstore( attr_def.name, attr_value );
489 $_$ LANGUAGE PLPGSQL;
491 CREATE OR REPLACE FUNCTION vandelay.extract_rec_attrs ( xml TEXT ) RETURNS hstore AS $_$
492 SELECT vandelay.extract_rec_attrs( $1, (SELECT ARRAY_ACCUM(name) FROM config.record_attr_definition));
495 -- Everything between this comment and the beginning of the definition of
496 -- vandelay.match_bib_record() is strictly in service of that function.
497 CREATE TYPE vandelay.match_set_test_result AS (record BIGINT, quality INTEGER);
499 CREATE OR REPLACE FUNCTION vandelay.match_set_test_marcxml(
500 match_set_id INTEGER, record_xml TEXT, bucket_id INTEGER
501 ) RETURNS SETOF vandelay.match_set_test_result AS $$
512 tags_rstore := vandelay.flatten_marc_hstore(record_xml);
513 svf_rstore := vandelay.extract_rec_attrs(record_xml);
515 CREATE TEMPORARY TABLE _vandelay_tmp_qrows (q INTEGER);
516 CREATE TEMPORARY TABLE _vandelay_tmp_jrows (j TEXT);
518 -- generate the where clause and return that directly (into wq), and as
519 -- a side-effect, populate the _vandelay_tmp_[qj]rows tables.
520 wq := vandelay.get_expr_from_match_set(match_set_id, tags_rstore);
522 query_ := 'SELECT DISTINCT(record), ';
524 -- qrows table is for the quality bits we add to the SELECT clause
525 SELECT ARRAY_TO_STRING(
526 ARRAY_ACCUM('COALESCE(n' || q::TEXT || '.quality, 0)'), ' + '
527 ) INTO coal FROM _vandelay_tmp_qrows;
529 -- our query string so far is the SELECT clause and the inital FROM.
530 -- no JOINs yet nor the WHERE clause
531 query_ := query_ || coal || ' AS quality ' || E'\n';
533 -- jrows table is for the joins we must make (and the real text conditions)
534 SELECT ARRAY_TO_STRING(ARRAY_ACCUM(j), E'\n') INTO joins
535 FROM _vandelay_tmp_jrows;
537 -- add those joins and the where clause to our query.
538 query_ := query_ || joins || E'\n';
540 -- join the record bucket
541 IF bucket_id IS NOT NULL THEN
542 query_ := query_ || 'JOIN container.biblio_record_entry_bucket_item ' ||
543 'brebi ON (brebi.target_biblio_record_entry = record ' ||
544 'AND brebi.bucket = ' || bucket_id || E')\n';
547 query_ := query_ || 'JOIN biblio.record_entry bre ON (bre.id = record) ' || 'WHERE ' || wq || ' AND not bre.deleted';
549 -- this will return rows of record,quality
550 FOR rec IN EXECUTE query_ USING tags_rstore, svf_rstore LOOP
554 DROP TABLE _vandelay_tmp_qrows;
555 DROP TABLE _vandelay_tmp_jrows;
561 CREATE OR REPLACE FUNCTION vandelay.flatten_marc_hstore(
563 ) RETURNS HSTORE AS $func$
567 ARRAY_ACCUM(tag || (COALESCE(subfield, ''))),
571 SELECT tag, subfield, ARRAY_ACCUM(value)::TEXT AS value
574 CASE WHEN tag = '020' THEN -- caseless -- isbn
575 LOWER((REGEXP_MATCHES(value,$$^(\S{10,17})$$))[1] || '%')
576 WHEN tag = '022' THEN -- caseless -- issn
577 LOWER((REGEXP_MATCHES(value,$$^(\S{4}[- ]?\S{4})$$))[1] || '%')
578 WHEN tag = '024' THEN -- caseless -- upc (other)
583 FROM vandelay.flatten_marc(record_xml)) x
584 GROUP BY tag, subfield ORDER BY tag, subfield
588 $func$ LANGUAGE PLPGSQL;
590 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set(
591 match_set_id INTEGER,
595 root vandelay.match_set_point;
597 SELECT * INTO root FROM vandelay.match_set_point
598 WHERE parent IS NULL AND match_set = match_set_id;
600 RETURN vandelay.get_expr_from_match_set_point(root, tags_rstore);
604 CREATE OR REPLACE FUNCTION vandelay.get_expr_from_match_set_point(
605 node vandelay.match_set_point,
613 child vandelay.match_set_point;
615 SELECT ARRAY_ACCUM(id) INTO children FROM vandelay.match_set_point
616 WHERE parent = node.id;
618 IF ARRAY_LENGTH(children, 1) > 0 THEN
619 this_op := vandelay._get_expr_render_one(node);
622 WHILE children[i] IS NOT NULL LOOP
623 SELECT * INTO child FROM vandelay.match_set_point
624 WHERE id = children[i];
626 q := q || ' ' || this_op || ' ';
629 q := q || vandelay.get_expr_from_match_set_point(child, tags_rstore);
633 ELSIF node.bool_op IS NULL THEN
634 PERFORM vandelay._get_expr_push_qrow(node);
635 PERFORM vandelay._get_expr_push_jrow(node, tags_rstore);
636 RETURN vandelay._get_expr_render_one(node);
643 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_qrow(
644 node vandelay.match_set_point
648 INSERT INTO _vandelay_tmp_qrows (q) VALUES (node.id);
652 CREATE OR REPLACE FUNCTION vandelay._get_expr_push_jrow(
653 node vandelay.match_set_point,
666 -- remember $1 is tags_rstore, and $2 is svf_rstore
669 SELECT COUNT(*) INTO jrow_count FROM _vandelay_tmp_jrows;
670 IF jrow_count > 0 THEN
671 my_using := ' USING (record)';
672 my_join := 'FULL OUTER JOIN';
678 IF node.tag IS NOT NULL THEN
679 caseless := (node.tag IN ('020', '022', '024'));
681 IF node.subfield IS NOT NULL THEN
682 tagkey := tagkey || node.subfield;
700 my_alias := 'n' || node.id::TEXT;
702 jrow := my_join || ' (SELECT *, ';
703 IF node.tag IS NOT NULL THEN
704 jrow := jrow || node.quality ||
705 ' AS quality FROM metabib.full_rec mfr WHERE mfr.tag = ''' ||
707 IF node.subfield IS NOT NULL THEN
708 jrow := jrow || ' AND mfr.subfield = ''' ||
709 node.subfield || '''';
711 jrow := jrow || ' AND (';
712 jrow := jrow || vandelay._node_tag_comparisons(caseless, op, tags_rstore, tagkey);
713 jrow := jrow || ')) ' || my_alias || my_using || E'\n';
715 jrow := jrow || 'id AS record, ' || node.quality ||
716 ' AS quality FROM metabib.record_attr mra WHERE mra.attrs->''' ||
717 node.svf || ''' ' || op || ' $2->''' || node.svf || ''') ' ||
718 my_alias || my_using || E'\n';
720 INSERT INTO _vandelay_tmp_jrows (j) VALUES (jrow);
724 CREATE OR REPLACE FUNCTION vandelay._node_tag_comparisons(
736 vals := tags_rstore->tagkey;
741 IF vals[i] IS NULL THEN
744 result := result || ' OR ';
749 result := result || 'LOWER(mfr.value) ' || op;
751 result := result || 'mfr.value ' || op;
754 result := result || ' ' || COALESCE('''' || vals[i] || '''', 'NULL');
756 IF vals[i] IS NULL THEN
767 CREATE OR REPLACE FUNCTION vandelay._get_expr_render_one(
768 node vandelay.match_set_point
773 IF node.bool_op IS NOT NULL THEN
776 RETURN '(n' || node.id::TEXT || '.id IS NOT NULL)';
781 CREATE OR REPLACE FUNCTION vandelay.match_bib_record() RETURNS TRIGGER AS $func$
783 incoming_existing_id TEXT;
784 test_result vandelay.match_set_test_result%ROWTYPE;
789 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
793 DELETE FROM vandelay.bib_match WHERE queued_record = NEW.id;
795 SELECT q.match_set INTO match_set FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
797 IF match_set IS NOT NULL THEN
798 NEW.quality := vandelay.measure_record_quality( NEW.marc, match_set );
801 -- Perfect matches on 901$c exit early with a match with high quality.
802 incoming_existing_id :=
803 oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]', NEW.marc);
805 IF incoming_existing_id IS NOT NULL AND incoming_existing_id != '' THEN
806 SELECT id INTO tmp_rec FROM biblio.record_entry WHERE id = incoming_existing_id::bigint;
807 IF tmp_rec IS NOT NULL THEN
808 INSERT INTO vandelay.bib_match (queued_record, eg_record, match_score, quality)
813 -- note: no match_set means quality==0
814 vandelay.measure_record_quality( b.marc, match_set )
815 FROM biblio.record_entry b
816 WHERE id = incoming_existing_id::bigint;
820 IF match_set IS NULL THEN
824 SELECT q.match_bucket INTO match_bucket FROM vandelay.bib_queue q WHERE q.id = NEW.queue;
826 FOR test_result IN SELECT * FROM
827 vandelay.match_set_test_marcxml(match_set, NEW.marc, match_bucket) LOOP
829 INSERT INTO vandelay.bib_match ( queued_record, eg_record, match_score, quality )
834 vandelay.measure_record_quality( b.marc, match_set )
835 FROM biblio.record_entry b
836 WHERE id = test_result.record;
842 $func$ LANGUAGE PLPGSQL;
844 CREATE OR REPLACE FUNCTION vandelay.measure_record_quality ( xml TEXT, match_set_id INT ) RETURNS INT AS $_$
848 test vandelay.match_set_quality%ROWTYPE;
851 FOR test IN SELECT * FROM vandelay.match_set_quality WHERE match_set = match_set_id LOOP
852 IF test.tag IS NOT NULL THEN
853 FOR rvalue IN SELECT value FROM vandelay.flatten_marc( xml ) WHERE tag = test.tag AND subfield = test.subfield LOOP
854 IF test.value = rvalue THEN
855 out_q := out_q + test.quality;
859 IF test.value = vandelay.extract_rec_attrs(xml, ARRAY[test.svf]) -> test.svf THEN
860 out_q := out_q + test.quality;
867 $_$ LANGUAGE PLPGSQL;
869 CREATE TYPE vandelay.tcn_data AS (tcn TEXT, tcn_source TEXT, used BOOL);
870 CREATE OR REPLACE FUNCTION vandelay.find_bib_tcn_data ( xml TEXT ) RETURNS SETOF vandelay.tcn_data AS $_$
874 output vandelay.tcn_data%ROWTYPE;
878 eg_tcn := BTRIM((oils_xpath('//*[@tag="001"]/text()',xml))[1]);
879 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
881 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="003"]/text()',xml))[1]);
882 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
883 eg_tcn_source := 'System Local';
886 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
889 output.used := FALSE;
894 output.tcn := eg_tcn;
895 output.tcn_source := eg_tcn_source;
901 eg_tcn := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="a"]/text()',xml))[1]);
902 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
904 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="901"]/*[@code="b"]/text()',xml))[1]);
905 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
906 eg_tcn_source := 'System Local';
909 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
912 output.used := FALSE;
917 output.tcn := eg_tcn;
918 output.tcn_source := eg_tcn_source;
924 eg_tcn := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="a"]/text()',xml))[1]);
925 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
927 eg_tcn_source := BTRIM((oils_xpath('//*[@tag="039"]/*[@code="b"]/text()',xml))[1]);
928 IF eg_tcn_source IS NULL OR eg_tcn_source = '' THEN
929 eg_tcn_source := 'System Local';
932 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
935 output.used := FALSE;
940 output.tcn := eg_tcn;
941 output.tcn_source := eg_tcn_source;
947 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="020"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
948 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
950 eg_tcn_source := 'ISBN';
952 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
955 output.used := FALSE;
960 output.tcn := eg_tcn;
961 output.tcn_source := eg_tcn_source;
967 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="022"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
968 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
970 eg_tcn_source := 'ISSN';
972 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
975 output.used := FALSE;
980 output.tcn := eg_tcn;
981 output.tcn_source := eg_tcn_source;
987 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="010"]/*[@code="a"]/text()',xml))[1], $re$^(\w+).*?$$re$, $re$\1$re$);
988 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
990 eg_tcn_source := 'LCCN';
992 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
995 output.used := FALSE;
1000 output.tcn := eg_tcn;
1001 output.tcn_source := eg_tcn_source;
1007 eg_tcn := REGEXP_REPLACE((oils_xpath('//*[@tag="035"]/*[@code="a"]/text()',xml))[1], $re$^.*?(\w+)$$re$, $re$\1$re$);
1008 IF eg_tcn IS NOT NULL AND eg_tcn <> '' THEN
1010 eg_tcn_source := 'System Legacy';
1012 PERFORM id FROM biblio.record_entry WHERE tcn_value = eg_tcn AND NOT deleted;
1015 output.used := FALSE;
1017 output.used := TRUE;
1020 output.tcn := eg_tcn;
1021 output.tcn_source := eg_tcn_source;
1028 $_$ LANGUAGE PLPGSQL;
1030 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT, force_add INT ) RETURNS TEXT AS $_$
1033 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1037 MARC::Charset->assume_unicode(1);
1039 my $target_xml = shift;
1040 my $source_xml = shift;
1041 my $field_spec = shift;
1042 my $force_add = shift || 0;
1044 my $target_r = MARC::Record->new_from_xml( $target_xml );
1045 my $source_r = MARC::Record->new_from_xml( $source_xml );
1047 return $target_xml unless ($target_r && $source_r);
1049 my @field_list = split(',', $field_spec);
1052 for my $f (@field_list) {
1053 $f =~ s/^\s*//; $f =~ s/\s*$//;
1054 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1060 $match =~ s/^\s*//; $match =~ s/\s*$//;
1061 $fields{$field} = { sf => [ split('', $sf) ] };
1063 my ($msf,$mre) = split('~', $match);
1064 if (length($msf) > 0 and length($mre) > 0) {
1065 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1066 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1067 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1073 for my $f ( keys %fields) {
1074 if ( @{$fields{$f}{sf}} ) {
1075 for my $from_field ($source_r->field( $f )) {
1076 my @tos = $target_r->field( $f );
1078 next if (exists($fields{$f}{match}) and !$force_add);
1079 my @new_fields = map { $_->clone } $source_r->field( $f );
1080 $target_r->insert_fields_ordered( @new_fields );
1082 for my $to_field (@tos) {
1083 if (exists($fields{$f}{match})) {
1084 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1086 my @new_sf = map { ($_ => $from_field->subfield($_)) } grep { defined($from_field->subfield($_)) } @{$fields{$f}{sf}};
1087 $to_field->add_subfields( @new_sf );
1092 my @new_fields = map { $_->clone } $source_r->field( $f );
1093 $target_r->insert_fields_ordered( @new_fields );
1097 $target_xml = $target_r->as_xml_record;
1098 $target_xml =~ s/^<\?.+?\?>$//mo;
1099 $target_xml =~ s/\n//sgo;
1100 $target_xml =~ s/>\s+</></sgo;
1104 $_$ LANGUAGE PLPERLU;
1106 CREATE OR REPLACE FUNCTION vandelay.add_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1107 SELECT vandelay.add_field( $1, $2, $3, 0 );
1110 CREATE OR REPLACE FUNCTION vandelay.strip_field ( xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1113 use MARC::File::XML (BinaryEncoding => 'UTF-8');
1117 MARC::Charset->assume_unicode(1);
1120 my $r = MARC::Record->new_from_xml( $xml );
1122 return $xml unless ($r);
1124 my $field_spec = shift;
1125 my @field_list = split(',', $field_spec);
1128 for my $f (@field_list) {
1129 $f =~ s/^\s*//; $f =~ s/\s*$//;
1130 if ($f =~ /^(.{3})(\w*)(?:\[([^]]*)\])?$/) {
1136 $match =~ s/^\s*//; $match =~ s/\s*$//;
1137 $fields{$field} = { sf => [ split('', $sf) ] };
1139 my ($msf,$mre) = split('~', $match);
1140 if (length($msf) > 0 and length($mre) > 0) {
1141 $msf =~ s/^\s*//; $msf =~ s/\s*$//;
1142 $mre =~ s/^\s*//; $mre =~ s/\s*$//;
1143 $fields{$field}{match} = { sf => $msf, re => qr/$mre/ };
1149 for my $f ( keys %fields) {
1150 for my $to_field ($r->field( $f )) {
1151 if (exists($fields{$f}{match})) {
1152 next unless (grep { $_ =~ $fields{$f}{match}{re} } $to_field->subfield($fields{$f}{match}{sf}));
1155 if ( @{$fields{$f}{sf}} ) {
1156 $to_field->delete_subfield(code => $fields{$f}{sf});
1158 $r->delete_field( $to_field );
1163 $xml = $r->as_xml_record;
1164 $xml =~ s/^<\?.+?\?>$//mo;
1166 $xml =~ s/>\s+</></sgo;
1170 $_$ LANGUAGE PLPERLU;
1172 CREATE OR REPLACE FUNCTION vandelay.replace_field ( target_xml TEXT, source_xml TEXT, field TEXT ) RETURNS TEXT AS $_$
1179 parsed_target := vandelay.strip_field( target_xml, ''); -- this dance normalizes the format of the xml for the IF below
1180 xml_output := parsed_target; -- if there are no replace rules, just return the input
1182 FOR curr_field IN SELECT UNNEST( STRING_TO_ARRAY(field, ',') ) LOOP -- naive split, but it's the same we use in the perl
1184 xml_output := vandelay.strip_field( parsed_target, curr_field);
1186 IF xml_output <> parsed_target AND curr_field ~ E'~' THEN
1187 -- we removed something, and there was a regexp restriction in the curr_field definition, so proceed
1188 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 1 );
1189 ELSIF curr_field !~ E'~' THEN
1190 -- No regexp restriction, add the curr_field
1191 xml_output := vandelay.add_field( xml_output, source_xml, curr_field, 0 );
1194 parsed_target := xml_output; -- in prep for any following loop iterations
1200 $_$ LANGUAGE PLPGSQL;
1202 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 $_$
1203 SELECT vandelay.replace_field( vandelay.add_field( vandelay.strip_field( $1, $5) , $2, $3 ), $2, $4);
1206 CREATE TYPE vandelay.compile_profile AS (add_rule TEXT, replace_rule TEXT, preserve_rule TEXT, strip_rule TEXT);
1207 CREATE OR REPLACE FUNCTION vandelay.compile_profile ( incoming_xml TEXT ) RETURNS vandelay.compile_profile AS $_$
1209 output vandelay.compile_profile%ROWTYPE;
1210 profile vandelay.merge_profile%ROWTYPE;
1212 profile_tmpl_owner TEXT;
1213 add_rule TEXT := '';
1214 strip_rule TEXT := '';
1215 replace_rule TEXT := '';
1216 preserve_rule TEXT := '';
1220 profile_tmpl := (oils_xpath('//*[@tag="905"]/*[@code="t"]/text()',incoming_xml))[1];
1221 profile_tmpl_owner := (oils_xpath('//*[@tag="905"]/*[@code="o"]/text()',incoming_xml))[1];
1223 IF profile_tmpl IS NOT NULL AND profile_tmpl <> '' AND profile_tmpl_owner IS NOT NULL AND profile_tmpl_owner <> '' THEN
1224 SELECT p.* INTO profile
1225 FROM vandelay.merge_profile p
1226 JOIN actor.org_unit u ON (u.id = p.owner)
1227 WHERE p.name = profile_tmpl
1228 AND u.shortname = profile_tmpl_owner;
1230 IF profile.id IS NOT NULL THEN
1231 add_rule := COALESCE(profile.add_spec,'');
1232 strip_rule := COALESCE(profile.strip_spec,'');
1233 replace_rule := COALESCE(profile.replace_spec,'');
1234 preserve_rule := COALESCE(profile.preserve_spec,'');
1238 add_rule := add_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="a"]/text()',incoming_xml),','),'');
1239 strip_rule := strip_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="d"]/text()',incoming_xml),','),'');
1240 replace_rule := replace_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="r"]/text()',incoming_xml),','),'');
1241 preserve_rule := preserve_rule || ',' || COALESCE(ARRAY_TO_STRING(oils_xpath('//*[@tag="905"]/*[@code="p"]/text()',incoming_xml),','),'');
1243 output.add_rule := BTRIM(add_rule,',');
1244 output.replace_rule := BTRIM(replace_rule,',');
1245 output.strip_rule := BTRIM(strip_rule,',');
1246 output.preserve_rule := BTRIM(preserve_rule,',');
1250 $_$ LANGUAGE PLPGSQL;
1252 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1254 merge_profile vandelay.merge_profile%ROWTYPE;
1255 dyn_profile vandelay.compile_profile%ROWTYPE;
1265 SELECT b.marc INTO eg_marc
1266 FROM biblio.record_entry b
1270 IF eg_marc IS NULL OR v_marc IS NULL THEN
1271 -- RAISE NOTICE 'no marc for template or bib record';
1275 dyn_profile := vandelay.compile_profile( v_marc );
1277 IF merge_profile_id IS NOT NULL THEN
1278 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1280 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1281 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1282 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1283 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1287 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1288 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1292 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1293 --Since we have nothing to do, just return a NOOP "we did it"
1295 ELSIF dyn_profile.replace_rule <> '' THEN
1296 source_marc = v_marc;
1297 target_marc = eg_marc;
1298 replace_rule = dyn_profile.replace_rule;
1300 source_marc = eg_marc;
1301 target_marc = v_marc;
1302 replace_rule = dyn_profile.preserve_rule;
1305 UPDATE biblio.record_entry
1306 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1310 -- RAISE NOTICE 'update of biblio.record_entry failed';
1317 $$ LANGUAGE PLPGSQL;
1319 CREATE OR REPLACE FUNCTION vandelay.merge_record_xml ( target_marc TEXT, template_marc TEXT ) RETURNS TEXT AS $$
1321 dyn_profile vandelay.compile_profile%ROWTYPE;
1329 IF target_marc IS NULL OR template_marc IS NULL THEN
1330 -- RAISE NOTICE 'no marc for target or template record';
1334 dyn_profile := vandelay.compile_profile( template_marc );
1336 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1337 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1341 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1342 --Since we have nothing to do, just return what we were given.
1344 ELSIF dyn_profile.replace_rule <> '' THEN
1345 trgt_marc = target_marc;
1346 tmpl_marc = template_marc;
1347 replace_rule = dyn_profile.replace_rule;
1349 tmp_marc = target_marc;
1350 trgt_marc = template_marc;
1351 tmpl_marc = tmp_marc;
1352 replace_rule = dyn_profile.preserve_rule;
1355 RETURN vandelay.merge_record_xml( trgt_marc, tmpl_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule );
1358 $$ LANGUAGE PLPGSQL;
1360 CREATE OR REPLACE FUNCTION vandelay.template_overlay_bib_record ( v_marc TEXT, eg_id BIGINT) RETURNS BOOL AS $$
1361 SELECT vandelay.template_overlay_bib_record( $1, $2, NULL);
1364 CREATE OR REPLACE FUNCTION vandelay.overlay_bib_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1366 merge_profile vandelay.merge_profile%ROWTYPE;
1367 dyn_profile vandelay.compile_profile%ROWTYPE;
1377 SELECT q.marc INTO v_marc
1378 FROM vandelay.queued_record q
1379 JOIN vandelay.bib_match m ON (m.queued_record = q.id AND q.id = import_id)
1382 IF v_marc IS NULL THEN
1383 -- RAISE NOTICE 'no marc for vandelay or bib record';
1387 IF vandelay.template_overlay_bib_record( v_marc, eg_id, merge_profile_id) THEN
1388 UPDATE vandelay.queued_bib_record
1389 SET imported_as = eg_id,
1391 WHERE id = import_id;
1393 editor_string := (oils_xpath('//*[@tag="905"]/*[@code="u"]/text()',v_marc))[1];
1395 IF editor_string IS NOT NULL AND editor_string <> '' THEN
1396 SELECT usr INTO editor_id FROM actor.card WHERE barcode = editor_string;
1398 IF editor_id IS NULL THEN
1399 SELECT id INTO editor_id FROM actor.usr WHERE usrname = editor_string;
1402 IF editor_id IS NOT NULL THEN
1403 UPDATE biblio.record_entry SET editor = editor_id WHERE id = eg_id;
1410 -- RAISE NOTICE 'update of biblio.record_entry failed';
1415 $$ LANGUAGE PLPGSQL;
1417 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT, lwm_ratio_value_p NUMERIC ) RETURNS BOOL AS $$
1420 lwm_ratio_value NUMERIC;
1423 lwm_ratio_value := COALESCE(lwm_ratio_value_p, 0.0);
1425 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1428 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1432 SELECT m.eg_record INTO eg_id
1433 FROM vandelay.bib_match m
1434 JOIN vandelay.queued_bib_record qr ON (m.queued_record = qr.id)
1435 JOIN vandelay.bib_queue q ON (qr.queue = q.id)
1436 JOIN biblio.record_entry r ON (r.id = m.eg_record)
1437 WHERE m.queued_record = import_id
1438 AND qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC >= lwm_ratio_value
1439 ORDER BY m.match_score DESC, -- required match score
1440 qr.quality::NUMERIC / COALESCE(NULLIF(m.quality,0),1)::NUMERIC DESC, -- quality tie breaker
1441 m.id -- when in doubt, use the first match
1444 IF eg_id IS NULL THEN
1445 -- RAISE NOTICE 'incoming record is not of high enough quality';
1449 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1451 $$ LANGUAGE PLPGSQL;
1453 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1454 SELECT vandelay.auto_overlay_bib_record_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1457 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1463 PERFORM * FROM vandelay.queued_bib_record WHERE import_time IS NOT NULL AND id = import_id;
1466 -- RAISE NOTICE 'already imported, cannot auto-overlay'
1470 SELECT COUNT(*) INTO match_count FROM vandelay.bib_match WHERE queued_record = import_id;
1472 IF match_count <> 1 THEN
1473 -- RAISE NOTICE 'not an exact match';
1477 -- Check that the one match is on the first 901c
1478 SELECT m.eg_record INTO eg_id
1479 FROM vandelay.queued_bib_record q
1480 JOIN vandelay.bib_match m ON (m.queued_record = q.id)
1481 WHERE q.id = import_id
1482 AND m.eg_record = oils_xpath_string('//*[@tag="901"]/*[@code="c"][1]',marc)::BIGINT;
1485 -- RAISE NOTICE 'not a 901c match';
1489 RETURN vandelay.overlay_bib_record( import_id, eg_id, merge_profile_id );
1491 $$ LANGUAGE PLPGSQL;
1493 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1495 queued_record vandelay.queued_bib_record%ROWTYPE;
1498 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1500 IF vandelay.auto_overlay_bib_record( queued_record.id, merge_profile_id ) THEN
1501 RETURN NEXT queued_record.id;
1509 $$ LANGUAGE PLPGSQL;
1511 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 $$
1513 queued_record vandelay.queued_bib_record%ROWTYPE;
1516 FOR queued_record IN SELECT * FROM vandelay.queued_bib_record WHERE queue = queue_id AND import_time IS NULL LOOP
1518 IF vandelay.auto_overlay_bib_record_with_best( queued_record.id, merge_profile_id, lwm_ratio_value ) THEN
1519 RETURN NEXT queued_record.id;
1527 $$ LANGUAGE PLPGSQL;
1529 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue_with_best ( import_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1530 SELECT vandelay.auto_overlay_bib_queue_with_best( $1, $2, p.lwm_ratio ) FROM vandelay.merge_profile p WHERE id = $2;
1533 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_bib_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1534 SELECT * FROM vandelay.auto_overlay_bib_queue( $1, NULL );
1537 CREATE OR REPLACE FUNCTION vandelay.ingest_bib_marc ( ) RETURNS TRIGGER AS $$
1543 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1547 FOR adef IN SELECT * FROM vandelay.bib_attr_definition LOOP
1549 SELECT extract_marc_field('vandelay.queued_bib_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_bib_record WHERE id = NEW.id;
1550 IF (value IS NOT NULL AND value <> '') THEN
1551 INSERT INTO vandelay.queued_bib_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1558 $$ LANGUAGE PLPGSQL;
1560 CREATE OR REPLACE FUNCTION vandelay.cleanup_bib_marc ( ) RETURNS TRIGGER AS $$
1562 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1566 DELETE FROM vandelay.queued_bib_record_attr WHERE record = OLD.id;
1567 DELETE FROM vandelay.import_item WHERE record = OLD.id;
1569 IF TG_OP = 'UPDATE' THEN
1574 $$ LANGUAGE PLPGSQL;
1576 CREATE TRIGGER cleanup_bib_trigger
1577 BEFORE UPDATE OR DELETE ON vandelay.queued_bib_record
1578 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_bib_marc();
1580 CREATE TRIGGER ingest_bib_trigger
1581 AFTER INSERT OR UPDATE ON vandelay.queued_bib_record
1582 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_bib_marc();
1584 CREATE TRIGGER zz_match_bibs_trigger
1585 BEFORE INSERT OR UPDATE ON vandelay.queued_bib_record
1586 FOR EACH ROW EXECUTE PROCEDURE vandelay.match_bib_record();
1589 /* Authority stuff down here */
1590 ---------------------------------------
1591 CREATE TABLE vandelay.authority_attr_definition (
1592 id SERIAL PRIMARY KEY,
1593 code TEXT UNIQUE NOT NULL,
1595 xpath TEXT NOT NULL,
1596 remove TEXT NOT NULL DEFAULT ''
1599 CREATE TYPE vandelay.authority_queue_queue_type AS ENUM ('authority');
1600 CREATE TABLE vandelay.authority_queue (
1601 queue_type vandelay.authority_queue_queue_type NOT NULL DEFAULT 'authority',
1602 CONSTRAINT vand_authority_queue_name_once_per_owner_const UNIQUE (owner,name,queue_type)
1603 ) INHERITS (vandelay.queue);
1604 ALTER TABLE vandelay.authority_queue ADD PRIMARY KEY (id);
1606 CREATE TABLE vandelay.queued_authority_record (
1607 queue INT NOT NULL REFERENCES vandelay.authority_queue (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1608 imported_as INT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1609 import_error TEXT REFERENCES vandelay.import_error (code) ON DELETE SET NULL ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
1611 ) INHERITS (vandelay.queued_record);
1612 ALTER TABLE vandelay.queued_authority_record ADD PRIMARY KEY (id);
1613 CREATE INDEX queued_authority_record_queue_idx ON vandelay.queued_authority_record (queue);
1615 CREATE TABLE vandelay.queued_authority_record_attr (
1616 id BIGSERIAL PRIMARY KEY,
1617 record BIGINT NOT NULL REFERENCES vandelay.queued_authority_record (id) DEFERRABLE INITIALLY DEFERRED,
1618 field INT NOT NULL REFERENCES vandelay.authority_attr_definition (id) DEFERRABLE INITIALLY DEFERRED,
1619 attr_value TEXT NOT NULL
1621 CREATE INDEX queued_authority_record_attr_record_idx ON vandelay.queued_authority_record_attr (record);
1623 CREATE TABLE vandelay.authority_match (
1624 id BIGSERIAL PRIMARY KEY,
1625 queued_record BIGINT REFERENCES vandelay.queued_authority_record (id) ON DELETE CASCADE DEFERRABLE INITIALLY DEFERRED,
1626 eg_record BIGINT REFERENCES authority.record_entry (id) DEFERRABLE INITIALLY DEFERRED,
1627 quality INT NOT NULL DEFAULT 0
1630 CREATE OR REPLACE FUNCTION vandelay.ingest_authority_marc ( ) RETURNS TRIGGER AS $$
1636 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1640 FOR adef IN SELECT * FROM vandelay.authority_attr_definition LOOP
1642 SELECT extract_marc_field('vandelay.queued_authority_record', id, adef.xpath, adef.remove) INTO value FROM vandelay.queued_authority_record WHERE id = NEW.id;
1643 IF (value IS NOT NULL AND value <> '') THEN
1644 INSERT INTO vandelay.queued_authority_record_attr (record, field, attr_value) VALUES (NEW.id, adef.id, value);
1651 $$ LANGUAGE PLPGSQL;
1653 CREATE OR REPLACE FUNCTION vandelay.cleanup_authority_marc ( ) RETURNS TRIGGER AS $$
1655 IF TG_OP IN ('INSERT','UPDATE') AND NEW.imported_as IS NOT NULL THEN
1659 DELETE FROM vandelay.queued_authority_record_attr WHERE record = OLD.id;
1660 IF TG_OP = 'UPDATE' THEN
1665 $$ LANGUAGE PLPGSQL;
1667 CREATE TRIGGER cleanup_authority_trigger
1668 BEFORE UPDATE OR DELETE ON vandelay.queued_authority_record
1669 FOR EACH ROW EXECUTE PROCEDURE vandelay.cleanup_authority_marc();
1671 CREATE TRIGGER ingest_authority_trigger
1672 AFTER INSERT OR UPDATE ON vandelay.queued_authority_record
1673 FOR EACH ROW EXECUTE PROCEDURE vandelay.ingest_authority_marc();
1675 CREATE OR REPLACE FUNCTION vandelay.overlay_authority_record ( import_id BIGINT, eg_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1677 merge_profile vandelay.merge_profile%ROWTYPE;
1678 dyn_profile vandelay.compile_profile%ROWTYPE;
1687 SELECT b.marc INTO eg_marc
1688 FROM authority.record_entry b
1689 JOIN vandelay.authority_match m ON (m.eg_record = b.id AND m.queued_record = import_id)
1692 SELECT q.marc INTO v_marc
1693 FROM vandelay.queued_record q
1694 JOIN vandelay.authority_match m ON (m.queued_record = q.id AND q.id = import_id)
1697 IF eg_marc IS NULL OR v_marc IS NULL THEN
1698 -- RAISE NOTICE 'no marc for vandelay or authority record';
1702 dyn_profile := vandelay.compile_profile( v_marc );
1704 IF merge_profile_id IS NOT NULL THEN
1705 SELECT * INTO merge_profile FROM vandelay.merge_profile WHERE id = merge_profile_id;
1707 dyn_profile.add_rule := BTRIM( dyn_profile.add_rule || ',' || COALESCE(merge_profile.add_spec,''), ',');
1708 dyn_profile.strip_rule := BTRIM( dyn_profile.strip_rule || ',' || COALESCE(merge_profile.strip_spec,''), ',');
1709 dyn_profile.replace_rule := BTRIM( dyn_profile.replace_rule || ',' || COALESCE(merge_profile.replace_spec,''), ',');
1710 dyn_profile.preserve_rule := BTRIM( dyn_profile.preserve_rule || ',' || COALESCE(merge_profile.preserve_spec,''), ',');
1714 IF dyn_profile.replace_rule <> '' AND dyn_profile.preserve_rule <> '' THEN
1715 -- RAISE NOTICE 'both replace [%] and preserve [%] specified', dyn_profile.replace_rule, dyn_profile.preserve_rule;
1719 IF dyn_profile.replace_rule = '' AND dyn_profile.preserve_rule = '' AND dyn_profile.add_rule = '' AND dyn_profile.strip_rule = '' THEN
1720 --Since we have nothing to do, just return a NOOP "we did it"
1722 ELSIF dyn_profile.replace_rule <> '' THEN
1723 source_marc = v_marc;
1724 target_marc = eg_marc;
1725 replace_rule = dyn_profile.replace_rule;
1727 source_marc = eg_marc;
1728 target_marc = v_marc;
1729 replace_rule = dyn_profile.preserve_rule;
1732 UPDATE authority.record_entry
1733 SET marc = vandelay.merge_record_xml( target_marc, source_marc, dyn_profile.add_rule, replace_rule, dyn_profile.strip_rule )
1737 UPDATE vandelay.queued_authority_record
1738 SET imported_as = eg_id,
1740 WHERE id = import_id;
1744 -- RAISE NOTICE 'update of authority.record_entry failed';
1749 $$ LANGUAGE PLPGSQL;
1751 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_record ( import_id BIGINT, merge_profile_id INT ) RETURNS BOOL AS $$
1756 SELECT COUNT(*) INTO match_count FROM vandelay.authority_match WHERE queued_record = import_id;
1758 IF match_count <> 1 THEN
1759 -- RAISE NOTICE 'not an exact match';
1763 SELECT m.eg_record INTO eg_id
1764 FROM vandelay.authority_match m
1765 WHERE m.queued_record = import_id
1768 IF eg_id IS NULL THEN
1772 RETURN vandelay.overlay_authority_record( import_id, eg_id, merge_profile_id );
1774 $$ LANGUAGE PLPGSQL;
1776 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT, merge_profile_id INT ) RETURNS SETOF BIGINT AS $$
1778 queued_record vandelay.queued_authority_record%ROWTYPE;
1781 FOR queued_record IN SELECT * FROM vandelay.queued_authority_record WHERE queue = queue_id AND import_time IS NULL LOOP
1783 IF vandelay.auto_overlay_authority_record( queued_record.id, merge_profile_id ) THEN
1784 RETURN NEXT queued_record.id;
1792 $$ LANGUAGE PLPGSQL;
1794 CREATE OR REPLACE FUNCTION vandelay.auto_overlay_authority_queue ( queue_id BIGINT ) RETURNS SETOF BIGINT AS $$
1795 SELECT * FROM vandelay.auto_overlay_authority_queue( $1, NULL );
1799 -- Vandelay (for importing and exporting records) 012.schema.vandelay.sql
1800 --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)]');
1801 --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)]');
1802 --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]');
1803 --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]');
1804 --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$);
1805 --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$);
1806 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (7, 'price',oils_i18n_gettext(7, 'vqbrad', 'Price', 'description'),'//*[@tag="020" or @tag="022"]/*[@code="c"][1]');
1807 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (8, 'rec_identifier',oils_i18n_gettext(8, 'vqbrad', 'Accession Number', 'description'),'//*[@tag="001"]', TRUE);
1808 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (9, 'eg_tcn',oils_i18n_gettext(9, 'vqbrad', 'TCN Value', 'description'),'//*[@tag="901"]/*[@code="a"]', TRUE);
1809 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (10, 'eg_tcn_source',oils_i18n_gettext(10, 'vqbrad', 'TCN Source', 'description'),'//*[@tag="901"]/*[@code="b"]', TRUE);
1810 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, ident ) VALUES (11, 'eg_identifier',oils_i18n_gettext(11, 'vqbrad', 'Internal ID', 'description'),'//*[@tag="901"]/*[@code="c"]', TRUE);
1811 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath ) VALUES (12, 'publisher',oils_i18n_gettext(12, 'vqbrad', 'Publisher', 'description'),'//*[@tag="260"]/*[@code="b"][1]');
1812 --INSERT INTO vandelay.bib_attr_definition ( id, code, description, xpath, remove ) VALUES (13, 'pubdate',oils_i18n_gettext(13, 'vqbrad', 'Publication Date', 'description'),'//*[@tag="260"]/*[@code="c"][1]',$r$\D$r$);
1813 --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]');
1815 --INSERT INTO vandelay.import_item_attr_definition (
1816 -- owner, name, tag, owning_lib, circ_lib, location,
1817 -- call_number, circ_modifier, barcode, price, copy_number,
1818 -- circulate, ref, holdable, opac_visible, status
1821 -- 'Evergreen 852 export format',
1823 -- '[@code = "b"][1]',
1824 -- '[@code = "b"][2]',
1831 -- '[@code = "x" and text() = "circulating"]',
1832 -- '[@code = "x" and text() = "reference"]',
1833 -- '[@code = "x" and text() = "holdable"]',
1834 -- '[@code = "x" and text() = "visible"]',
1838 --INSERT INTO vandelay.import_item_attr_definition (
1851 -- 'Unicorn Import format -- 999',
1862 --INSERT INTO vandelay.authority_attr_definition ( code, description, xpath, ident ) VALUES ('rec_identifier','Identifier','//*[@tag="001"]', TRUE);