2 * Copyright (C) 2004-2008 Georgia Public Library Service
3 * Copyright (C) 2007-2008 Equinox Software, Inc.
4 * Mike Rylander <miker@esilibrary.com>
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public License
8 * as published by the Free Software Foundation; either version 2
9 * of the License, or (at your option) any later version.
11 * This program is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
18 DROP SCHEMA IF EXISTS metabib CASCADE;
21 CREATE SCHEMA metabib;
23 CREATE TABLE metabib.metarecord (
24 id BIGSERIAL PRIMARY KEY,
25 fingerprint TEXT NOT NULL,
29 CREATE INDEX metabib_metarecord_master_record_idx ON metabib.metarecord (master_record);
30 CREATE INDEX metabib_metarecord_fingerprint_idx ON metabib.metarecord (fingerprint);
32 CREATE TABLE metabib.identifier_field_entry (
33 id BIGSERIAL PRIMARY KEY,
34 source BIGINT NOT NULL,
37 index_vector tsvector NOT NULL
39 CREATE TRIGGER metabib_identifier_field_entry_fti_trigger
40 BEFORE UPDATE OR INSERT ON metabib.identifier_field_entry
41 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('identifier');
43 CREATE INDEX metabib_identifier_field_entry_index_vector_idx ON metabib.identifier_field_entry USING GIST (index_vector);
44 CREATE INDEX metabib_identifier_field_entry_value_idx ON metabib.identifier_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
45 CREATE INDEX metabib_identifier_field_entry_source_idx ON metabib.identifier_field_entry (source);
47 CREATE TABLE metabib.combined_identifier_field_entry (
48 record BIGINT NOT NULL,
49 metabib_field INT NULL,
50 index_vector tsvector NOT NULL
52 CREATE UNIQUE INDEX metabib_combined_identifier_field_entry_fakepk_idx ON metabib.combined_identifier_field_entry (record, COALESCE(metabib_field::TEXT,''));
53 CREATE INDEX metabib_combined_identifier_field_entry_index_vector_idx ON metabib.combined_identifier_field_entry USING GIST (index_vector);
54 CREATE INDEX metabib_combined_identifier_field_source_idx ON metabib.combined_identifier_field_entry (metabib_field);
56 CREATE TABLE metabib.title_field_entry (
57 id BIGSERIAL PRIMARY KEY,
58 source BIGINT NOT NULL,
61 index_vector tsvector NOT NULL
63 CREATE TRIGGER metabib_title_field_entry_fti_trigger
64 BEFORE UPDATE OR INSERT ON metabib.title_field_entry
65 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('title');
67 CREATE INDEX metabib_title_field_entry_index_vector_idx ON metabib.title_field_entry USING GIST (index_vector);
68 CREATE INDEX metabib_title_field_entry_value_idx ON metabib.title_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
69 CREATE INDEX metabib_title_field_entry_source_idx ON metabib.title_field_entry (source);
71 CREATE TABLE metabib.combined_title_field_entry (
72 record BIGINT NOT NULL,
73 metabib_field INT NULL,
74 index_vector tsvector NOT NULL
76 CREATE UNIQUE INDEX metabib_combined_title_field_entry_fakepk_idx ON metabib.combined_title_field_entry (record, COALESCE(metabib_field::TEXT,''));
77 CREATE INDEX metabib_combined_title_field_entry_index_vector_idx ON metabib.combined_title_field_entry USING GIST (index_vector);
78 CREATE INDEX metabib_combined_title_field_source_idx ON metabib.combined_title_field_entry (metabib_field);
80 CREATE TABLE metabib.author_field_entry (
81 id BIGSERIAL PRIMARY KEY,
82 source BIGINT NOT NULL,
85 index_vector tsvector NOT NULL
87 CREATE TRIGGER metabib_author_field_entry_fti_trigger
88 BEFORE UPDATE OR INSERT ON metabib.author_field_entry
89 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('author');
91 CREATE INDEX metabib_author_field_entry_index_vector_idx ON metabib.author_field_entry USING GIST (index_vector);
92 CREATE INDEX metabib_author_field_entry_value_idx ON metabib.author_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
93 CREATE INDEX metabib_author_field_entry_source_idx ON metabib.author_field_entry (source);
95 CREATE TABLE metabib.combined_author_field_entry (
96 record BIGINT NOT NULL,
97 metabib_field INT NULL,
98 index_vector tsvector NOT NULL
100 CREATE UNIQUE INDEX metabib_combined_author_field_entry_fakepk_idx ON metabib.combined_author_field_entry (record, COALESCE(metabib_field::TEXT,''));
101 CREATE INDEX metabib_combined_author_field_entry_index_vector_idx ON metabib.combined_author_field_entry USING GIST (index_vector);
102 CREATE INDEX metabib_combined_author_field_source_idx ON metabib.combined_author_field_entry (metabib_field);
104 CREATE TABLE metabib.subject_field_entry (
105 id BIGSERIAL PRIMARY KEY,
106 source BIGINT NOT NULL,
109 index_vector tsvector NOT NULL
111 CREATE TRIGGER metabib_subject_field_entry_fti_trigger
112 BEFORE UPDATE OR INSERT ON metabib.subject_field_entry
113 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('subject');
115 CREATE INDEX metabib_subject_field_entry_index_vector_idx ON metabib.subject_field_entry USING GIST (index_vector);
116 CREATE INDEX metabib_subject_field_entry_value_idx ON metabib.subject_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
117 CREATE INDEX metabib_subject_field_entry_source_idx ON metabib.subject_field_entry (source);
119 CREATE TABLE metabib.combined_subject_field_entry (
120 record BIGINT NOT NULL,
121 metabib_field INT NULL,
122 index_vector tsvector NOT NULL
124 CREATE UNIQUE INDEX metabib_combined_subject_field_entry_fakepk_idx ON metabib.combined_subject_field_entry (record, COALESCE(metabib_field::TEXT,''));
125 CREATE INDEX metabib_combined_subject_field_entry_index_vector_idx ON metabib.combined_subject_field_entry USING GIST (index_vector);
126 CREATE INDEX metabib_combined_subject_field_source_idx ON metabib.combined_subject_field_entry (metabib_field);
128 CREATE TABLE metabib.keyword_field_entry (
129 id BIGSERIAL PRIMARY KEY,
130 source BIGINT NOT NULL,
133 index_vector tsvector NOT NULL
135 CREATE TRIGGER metabib_keyword_field_entry_fti_trigger
136 BEFORE UPDATE OR INSERT ON metabib.keyword_field_entry
137 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('keyword');
139 CREATE INDEX metabib_keyword_field_entry_index_vector_idx ON metabib.keyword_field_entry USING GIST (index_vector);
140 CREATE INDEX metabib_keyword_field_entry_value_idx ON metabib.keyword_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
141 CREATE INDEX metabib_keyword_field_entry_source_idx ON metabib.keyword_field_entry (source);
143 CREATE TABLE metabib.combined_keyword_field_entry (
144 record BIGINT NOT NULL,
145 metabib_field INT NULL,
146 index_vector tsvector NOT NULL
148 CREATE UNIQUE INDEX metabib_combined_keyword_field_entry_fakepk_idx ON metabib.combined_keyword_field_entry (record, COALESCE(metabib_field::TEXT,''));
149 CREATE INDEX metabib_combined_keyword_field_entry_index_vector_idx ON metabib.combined_keyword_field_entry USING GIST (index_vector);
150 CREATE INDEX metabib_combined_keyword_field_source_idx ON metabib.combined_keyword_field_entry (metabib_field);
152 CREATE TABLE metabib.series_field_entry (
153 id BIGSERIAL PRIMARY KEY,
154 source BIGINT NOT NULL,
157 index_vector tsvector NOT NULL
159 CREATE TRIGGER metabib_series_field_entry_fti_trigger
160 BEFORE UPDATE OR INSERT ON metabib.series_field_entry
161 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('series');
163 CREATE INDEX metabib_series_field_entry_index_vector_idx ON metabib.series_field_entry USING GIST (index_vector);
164 CREATE INDEX metabib_series_field_entry_value_idx ON metabib.series_field_entry (SUBSTRING(value,1,1024)) WHERE index_vector = ''::TSVECTOR;
165 CREATE INDEX metabib_series_field_entry_source_idx ON metabib.series_field_entry (source);
167 CREATE TABLE metabib.combined_series_field_entry (
168 record BIGINT NOT NULL,
169 metabib_field INT NULL,
170 index_vector tsvector NOT NULL
172 CREATE UNIQUE INDEX metabib_combined_series_field_entry_fakepk_idx ON metabib.combined_series_field_entry (record, COALESCE(metabib_field::TEXT,''));
173 CREATE INDEX metabib_combined_series_field_entry_index_vector_idx ON metabib.combined_series_field_entry USING GIST (index_vector);
174 CREATE INDEX metabib_combined_series_field_source_idx ON metabib.combined_series_field_entry (metabib_field);
176 CREATE TABLE metabib.facet_entry (
177 id BIGSERIAL PRIMARY KEY,
178 source BIGINT NOT NULL,
182 CREATE INDEX metabib_facet_entry_field_idx ON metabib.facet_entry (field);
183 CREATE INDEX metabib_facet_entry_value_idx ON metabib.facet_entry (SUBSTRING(value,1,1024));
184 CREATE INDEX metabib_facet_entry_source_idx ON metabib.facet_entry (source);
186 CREATE TABLE metabib.browse_entry (
187 id BIGSERIAL PRIMARY KEY,
189 index_vector tsvector,
190 sort_value TEXT NOT NULL,
191 UNIQUE(sort_value, value)
195 CREATE INDEX browse_entry_sort_value_idx
196 ON metabib.browse_entry USING BTREE (sort_value);
198 CREATE INDEX metabib_browse_entry_index_vector_idx ON metabib.browse_entry USING GIN (index_vector);
199 CREATE TRIGGER metabib_browse_entry_fti_trigger
200 BEFORE INSERT OR UPDATE ON metabib.browse_entry
201 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('keyword');
204 CREATE TABLE metabib.browse_entry_def_map (
205 id BIGSERIAL PRIMARY KEY,
206 entry BIGINT REFERENCES metabib.browse_entry (id),
207 def INT REFERENCES config.metabib_field (id) ON UPDATE CASCADE DEFERRABLE INITIALLY DEFERRED,
208 source BIGINT REFERENCES biblio.record_entry (id),
209 authority BIGINT REFERENCES authority.record_entry (id) ON DELETE SET NULL
211 CREATE INDEX browse_entry_def_map_def_idx ON metabib.browse_entry_def_map (def);
212 CREATE INDEX browse_entry_def_map_entry_idx ON metabib.browse_entry_def_map (entry);
213 CREATE INDEX browse_entry_def_map_source_idx ON metabib.browse_entry_def_map (source);
215 CREATE TABLE metabib.browse_entry_simple_heading_map (
216 id BIGSERIAL PRIMARY KEY,
217 entry BIGINT REFERENCES metabib.browse_entry (id),
218 simple_heading BIGINT REFERENCES authority.simple_heading (id) ON DELETE CASCADE
220 CREATE INDEX browse_entry_sh_map_entry_idx ON metabib.browse_entry_simple_heading_map (entry);
221 CREATE INDEX browse_entry_sh_map_sh_idx ON metabib.browse_entry_simple_heading_map (simple_heading);
223 CREATE OR REPLACE FUNCTION metabib.facet_normalize_trigger () RETURNS TRIGGER AS $$
228 facet_text := NEW.value;
231 SELECT n.func AS func,
232 n.param_count AS param_count,
234 FROM config.index_normalizer n
235 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
236 WHERE m.field = NEW.field AND m.pos < 0
239 EXECUTE 'SELECT ' || normalizer.func || '(' ||
240 quote_literal( facet_text ) ||
242 WHEN normalizer.param_count > 0
243 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
250 NEW.value = facet_text;
256 CREATE TRIGGER facet_normalize_tgr
257 BEFORE UPDATE OR INSERT ON metabib.facet_entry
258 FOR EACH ROW EXECUTE PROCEDURE metabib.facet_normalize_trigger();
260 CREATE OR REPLACE FUNCTION evergreen.facet_force_nfc() RETURNS TRIGGER AS $$
262 NEW.value := force_unicode_normal_form(NEW.value,'NFC');
267 CREATE TRIGGER facet_force_nfc_tgr
268 BEFORE UPDATE OR INSERT ON metabib.facet_entry
269 FOR EACH ROW EXECUTE PROCEDURE evergreen.facet_force_nfc();
271 -- DECREMENTING serial starts at -1
272 CREATE SEQUENCE metabib.uncontrolled_record_attr_value_id_seq INCREMENT BY -1;
274 CREATE TABLE metabib.uncontrolled_record_attr_value (
275 id BIGINT PRIMARY KEY DEFAULT nextval('metabib.uncontrolled_record_attr_value_id_seq'),
276 attr TEXT NOT NULL REFERENCES config.record_attr_definition (name),
279 CREATE UNIQUE INDEX muv_once_idx ON metabib.uncontrolled_record_attr_value (attr,value);
281 CREATE VIEW metabib.record_attr_id_map AS
282 SELECT id, attr, value FROM metabib.uncontrolled_record_attr_value
284 SELECT c.id, c.ctype AS attr, c.code AS value
285 FROM config.coded_value_map c
286 JOIN config.record_attr_definition d ON (d.name = c.ctype AND NOT d.composite);
288 CREATE VIEW metabib.composite_attr_id_map AS
289 SELECT c.id, c.ctype AS attr, c.code AS value
290 FROM config.coded_value_map c
291 JOIN config.record_attr_definition d ON (d.name = c.ctype AND d.composite);
293 CREATE VIEW metabib.full_attr_id_map AS
294 SELECT id, attr, value FROM metabib.record_attr_id_map
296 SELECT id, attr, value FROM metabib.composite_attr_id_map;
299 CREATE OR REPLACE FUNCTION metabib.compile_composite_attr_cache_init () RETURNS BOOL AS $f$
300 $_SHARED{metabib_compile_composite_attr_cache} = {}
301 if ! exists $_SHARED{metabib_compile_composite_attr_cache};
302 return exists $_SHARED{metabib_compile_composite_attr_cache};
303 $f$ LANGUAGE PLPERLU;
305 CREATE OR REPLACE FUNCTION metabib.compile_composite_attr_cache_disable () RETURNS BOOL AS $f$
306 delete $_SHARED{metabib_compile_composite_attr_cache};
307 return ! exists $_SHARED{metabib_compile_composite_attr_cache};
308 $f$ LANGUAGE PLPERLU;
310 CREATE OR REPLACE FUNCTION metabib.compile_composite_attr_cache_invalidate () RETURNS BOOL AS $f$
311 SELECT metabib.compile_composite_attr_cache_disable() AND metabib.compile_composite_attr_cache_init();
314 CREATE OR REPLACE FUNCTION metabib.composite_attr_def_cache_inval_tgr () RETURNS TRIGGER AS $f$
316 PERFORM metabib.compile_composite_attr_cache_invalidate();
319 $f$ LANGUAGE PLPGSQL;
321 CREATE TRIGGER ccraed_cache_inval_tgr AFTER INSERT OR UPDATE OR DELETE ON config.composite_attr_entry_definition FOR EACH STATEMENT EXECUTE PROCEDURE metabib.composite_attr_def_cache_inval_tgr();
323 CREATE OR REPLACE FUNCTION metabib.compile_composite_attr ( cattr_def TEXT ) RETURNS query_int AS $func$
328 my $def = decode_json($json);
330 die("Composite attribute definition not supplied") unless $def;
332 my $_cache = (exists $_SHARED{metabib_compile_composite_attr_cache}) ? 1 : 0;
334 return $_SHARED{metabib_compile_composite_attr_cache}{$json}
335 if ($_cache && $_SHARED{metabib_compile_composite_attr_cache}{$json});
342 if (ref $d eq 'HASH') { # node or AND
343 if (exists $d->{_attr}) { # it is a node
344 my $plan = spi_prepare('SELECT * FROM metabib.full_attr_id_map WHERE attr = $1 AND value = $2', qw/TEXT TEXT/);
345 my $id = spi_exec_prepared(
346 $plan, {limit => 1}, $d->{_attr}, $d->{_val}
350 } elsif (exists $d->{_not} && scalar(keys(%$d)) == 1) { # it is a NOT
351 return '!' . recurse($$d{_not});
352 } else { # an AND list
353 @list = map { recurse($$d{$_}) } sort keys %$d;
355 } elsif (ref $d eq 'ARRAY') {
357 @list = map { recurse($_) } @$d;
360 @list = grep { defined && $_ ne '' } @list;
362 return '(' . join($j,@list) . ')' if @list;
366 my $val = recurse($def) || undef;
367 $_SHARED{metabib_compile_composite_attr_cache}{$json} = $val if $_cache;
370 $func$ IMMUTABLE LANGUAGE plperlu;
372 CREATE OR REPLACE FUNCTION metabib.compile_composite_attr ( cattr_id INT ) RETURNS query_int AS $func$
373 SELECT metabib.compile_composite_attr(definition) FROM config.composite_attr_entry_definition WHERE coded_value = $1;
374 $func$ STRICT IMMUTABLE LANGUAGE SQL;
376 CREATE TABLE metabib.record_attr_vector_list (
377 source BIGINT PRIMARY KEY REFERENCES biblio.record_entry (id),
378 vlist INT[] NOT NULL -- stores id from ccvm AND murav
380 CREATE INDEX mrca_vlist_idx ON metabib.record_attr_vector_list USING gin ( vlist gin__int_ops );
382 /* This becomes a view, and we do sorters differently ...
383 CREATE TABLE metabib.record_attr (
384 id BIGINT PRIMARY KEY REFERENCES biblio.record_entry (id) ON DELETE CASCADE,
385 attrs HSTORE NOT NULL DEFAULT ''::HSTORE
387 CREATE INDEX metabib_svf_attrs_idx ON metabib.record_attr USING GIST (attrs);
388 CREATE INDEX metabib_svf_date1_idx ON metabib.record_attr ((attrs->'date1'));
389 CREATE INDEX metabib_svf_dates_idx ON metabib.record_attr ((attrs->'date1'),(attrs->'date2'));
393 CREATE TABLE metabib.record_sorter (
394 id BIGSERIAL PRIMARY KEY,
395 source BIGINT NOT NULL REFERENCES biblio.record_entry (id) ON DELETE CASCADE,
396 attr TEXT NOT NULL REFERENCES config.record_attr_definition (name) ON DELETE CASCADE,
399 CREATE INDEX metabib_sorter_source_idx ON metabib.record_sorter (source); -- we may not need one of this or the next ... stats will tell
400 CREATE INDEX metabib_sorter_s_a_idx ON metabib.record_sorter (source, attr);
401 CREATE INDEX metabib_sorter_a_v_idx ON metabib.record_sorter (attr, value);
404 CREATE TYPE metabib.record_attr_type AS (
409 -- Back-compat view ... we're moving to an INTARRAY world
410 CREATE VIEW metabib.record_attr_flat AS
411 SELECT v.source AS id,
414 FROM metabib.record_attr_vector_list v
415 LEFT JOIN metabib.uncontrolled_record_attr_value m ON ( m.id = ANY( v.vlist ) )
417 SELECT v.source AS id,
420 FROM metabib.record_attr_vector_list v
421 LEFT JOIN config.coded_value_map c ON ( c.id = ANY( v.vlist ) );
423 CREATE VIEW metabib.record_attr AS
424 SELECT id, HSTORE( ARRAY_AGG( attr ), ARRAY_AGG( value ) ) AS attrs
425 FROM metabib.record_attr_flat
426 WHERE attr IS NOT NULL
429 -- Back-back-compat view ... we use to live in an HSTORE world
430 CREATE TYPE metabib.rec_desc_type AS (
448 CREATE VIEW metabib.rec_descriptor AS
451 (populate_record(NULL::metabib.rec_desc_type, attrs)).*
452 FROM metabib.record_attr;
454 -- Use a sequence that matches previous version, for easier upgrading.
455 CREATE SEQUENCE metabib.full_rec_id_seq;
457 CREATE TABLE metabib.real_full_rec (
458 id BIGINT NOT NULL DEFAULT NEXTVAL('metabib.full_rec_id_seq'::REGCLASS),
459 record BIGINT NOT NULL,
460 tag CHAR(3) NOT NULL,
465 index_vector tsvector NOT NULL
467 ALTER TABLE metabib.real_full_rec ADD PRIMARY KEY (id);
469 CREATE INDEX metabib_full_rec_tag_subfield_idx ON metabib.real_full_rec (tag,subfield);
470 CREATE INDEX metabib_full_rec_value_idx ON metabib.real_full_rec (substring(value,1,1024));
471 /* Enable LIKE to use an index for database clusters with locales other than C or POSIX */
472 CREATE INDEX metabib_full_rec_value_tpo_index ON metabib.real_full_rec (substring(value,1,1024) text_pattern_ops);
473 CREATE INDEX metabib_full_rec_record_idx ON metabib.real_full_rec (record);
474 CREATE INDEX metabib_full_rec_index_vector_idx ON metabib.real_full_rec USING GIST (index_vector);
475 CREATE INDEX metabib_full_rec_isxn_caseless_idx
476 ON metabib.real_full_rec (LOWER(value))
477 WHERE tag IN ('020', '022', '024');
478 -- This next index might fully supplant the one above, but leaving both for now.
479 -- (they are not too large)
480 -- The reason we need this index is to ensure that the query parser always
481 -- prefers this index over the simpler tag/subfield index, as this greatly
482 -- increases Vandelay overlay speed for these identifiers, especially when
483 -- a record has many of these fields (around > 4-6 seems like the cutoff
484 -- on at least one PG9.1 system)
485 -- A similar index could be added for other fields (e.g. 010), but one should
486 -- leave out the LOWER() in all other cases.
487 -- TODO: verify whether we can discard the non tag/subfield/substring version
488 -- above (metabib_full_rec_isxn_caseless_idx)
489 CREATE INDEX metabib_full_rec_02x_tag_subfield_lower_substring
490 ON metabib.real_full_rec (tag, subfield, LOWER(substring(value, 1, 1024)))
491 WHERE tag IN ('020', '022', '024');
494 CREATE TRIGGER metabib_full_rec_fti_trigger
495 BEFORE UPDATE OR INSERT ON metabib.real_full_rec
496 FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('default');
498 CREATE OR REPLACE VIEW metabib.full_rec AS
505 SUBSTRING(value,1,1024) AS value,
507 FROM metabib.real_full_rec;
509 CREATE OR REPLACE RULE metabib_full_rec_insert_rule
510 AS ON INSERT TO metabib.full_rec
512 INSERT INTO metabib.real_full_rec VALUES (
513 COALESCE(NEW.id, NEXTVAL('metabib.full_rec_id_seq'::REGCLASS)),
523 CREATE OR REPLACE RULE metabib_full_rec_update_rule
524 AS ON UPDATE TO metabib.full_rec
526 UPDATE metabib.real_full_rec SET
532 subfield = NEW.subfield,
534 index_vector = NEW.index_vector
537 CREATE OR REPLACE RULE metabib_full_rec_delete_rule
538 AS ON DELETE TO metabib.full_rec
540 DELETE FROM metabib.real_full_rec WHERE id = OLD.id;
542 CREATE TABLE metabib.metarecord_source_map (
543 id BIGSERIAL PRIMARY KEY,
544 metarecord BIGINT NOT NULL,
545 source BIGINT NOT NULL
547 CREATE INDEX metabib_metarecord_source_map_metarecord_idx ON metabib.metarecord_source_map (metarecord);
548 CREATE INDEX metabib_metarecord_source_map_source_record_idx ON metabib.metarecord_source_map (source);
550 CREATE TYPE metabib.field_entry_template AS (
563 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
565 bib biblio.record_entry%ROWTYPE;
566 idx config.metabib_field%ROWTYPE;
567 xfrm config.xml_transform%ROWTYPE;
569 transformed_xml TEXT;
571 xml_node_list TEXT[];
577 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
579 authority_link BIGINT;
580 output_row metabib.field_entry_template%ROWTYPE;
583 -- Start out with no field-use bools set
584 output_row.browse_field = FALSE;
585 output_row.facet_field = FALSE;
586 output_row.search_field = FALSE;
589 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
591 -- Loop over the indexing entries
592 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
594 joiner := COALESCE(idx.joiner, default_joiner);
596 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
598 -- See if we can skip the XSLT ... it's expensive
599 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
600 -- Can't skip the transform
601 IF xfrm.xslt <> '---' THEN
602 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
604 transformed_xml := bib.marc;
607 prev_xfrm := xfrm.name;
610 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
613 FOR xml_node IN SELECT x FROM unnest(xml_node_list) AS x LOOP
614 CONTINUE WHEN xml_node !~ E'^\\s*<';
616 -- XXX much of this should be moved into oils_xpath_string...
617 curr_text := ARRAY_TO_STRING(evergreen.array_remove_item_by_value(evergreen.array_remove_item_by_value(
618 oils_xpath( '//text()', -- get the content of all the nodes within the main selected node
619 REGEXP_REPLACE( xml_node, E'\\s+', ' ', 'g' ) -- Translate adjacent whitespace to a single space
620 ), ' '), ''), -- throw away morally empty (bankrupt?) strings
624 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
626 IF raw_text IS NOT NULL THEN
627 raw_text := raw_text || joiner;
630 raw_text := COALESCE(raw_text,'') || curr_text;
632 -- autosuggest/metabib.browse_entry
633 IF idx.browse_field THEN
635 IF idx.browse_xpath IS NOT NULL AND idx.browse_xpath <> '' THEN
636 browse_text := oils_xpath_string( idx.browse_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
638 browse_text := curr_text;
641 IF idx.browse_sort_xpath IS NOT NULL AND
642 idx.browse_sort_xpath <> '' THEN
644 sort_value := oils_xpath_string(
645 idx.browse_sort_xpath, xml_node, joiner,
646 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
649 sort_value := browse_text;
652 output_row.field_class = idx.field_class;
653 output_row.field = idx.id;
654 output_row.source = rid;
655 output_row.value = BTRIM(REGEXP_REPLACE(browse_text, E'\\s+', ' ', 'g'));
656 output_row.sort_value :=
657 public.naco_normalize(sort_value);
659 output_row.authority := NULL;
661 IF idx.authority_xpath IS NOT NULL AND idx.authority_xpath <> '' THEN
662 authority_text := oils_xpath_string(
663 idx.authority_xpath, xml_node, joiner,
665 ARRAY[xfrm.prefix, xfrm.namespace_uri],
666 ARRAY['xlink','http://www.w3.org/1999/xlink']
670 IF authority_text ~ '^\d+$' THEN
671 authority_link := authority_text::BIGINT;
672 PERFORM * FROM authority.record_entry WHERE id = authority_link;
674 output_row.authority := authority_link;
680 output_row.browse_field = TRUE;
681 -- Returning browse rows with search_field = true for search+browse
682 -- configs allows us to retain granularity of being able to search
683 -- browse fields with "starts with" type operators (for example, for
684 -- titles of songs in music albums)
685 IF idx.search_field THEN
686 output_row.search_field = TRUE;
688 RETURN NEXT output_row;
689 output_row.browse_field = FALSE;
690 output_row.search_field = FALSE;
691 output_row.sort_value := NULL;
694 -- insert raw node text for faceting
695 IF idx.facet_field THEN
697 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
698 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
700 facet_text := curr_text;
703 output_row.field_class = idx.field_class;
704 output_row.field = -1 * idx.id;
705 output_row.source = rid;
706 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
708 output_row.facet_field = TRUE;
709 RETURN NEXT output_row;
710 output_row.facet_field = FALSE;
715 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
717 -- insert combined node text for searching
718 IF idx.search_field THEN
719 output_row.field_class = idx.field_class;
720 output_row.field = idx.id;
721 output_row.source = rid;
722 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
724 output_row.search_field = TRUE;
725 RETURN NEXT output_row;
726 output_row.search_field = FALSE;
733 $func$ LANGUAGE PLPGSQL;
735 CREATE OR REPLACE FUNCTION metabib.update_combined_index_vectors(bib_id BIGINT) RETURNS VOID AS $func$
737 DELETE FROM metabib.combined_keyword_field_entry WHERE record = bib_id;
738 INSERT INTO metabib.combined_keyword_field_entry(record, metabib_field, index_vector)
739 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
740 FROM metabib.keyword_field_entry WHERE source = bib_id GROUP BY field;
741 INSERT INTO metabib.combined_keyword_field_entry(record, metabib_field, index_vector)
742 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
743 FROM metabib.keyword_field_entry WHERE source = bib_id;
745 DELETE FROM metabib.combined_title_field_entry WHERE record = bib_id;
746 INSERT INTO metabib.combined_title_field_entry(record, metabib_field, index_vector)
747 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
748 FROM metabib.title_field_entry WHERE source = bib_id GROUP BY field;
749 INSERT INTO metabib.combined_title_field_entry(record, metabib_field, index_vector)
750 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
751 FROM metabib.title_field_entry WHERE source = bib_id;
753 DELETE FROM metabib.combined_author_field_entry WHERE record = bib_id;
754 INSERT INTO metabib.combined_author_field_entry(record, metabib_field, index_vector)
755 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
756 FROM metabib.author_field_entry WHERE source = bib_id GROUP BY field;
757 INSERT INTO metabib.combined_author_field_entry(record, metabib_field, index_vector)
758 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
759 FROM metabib.author_field_entry WHERE source = bib_id;
761 DELETE FROM metabib.combined_subject_field_entry WHERE record = bib_id;
762 INSERT INTO metabib.combined_subject_field_entry(record, metabib_field, index_vector)
763 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
764 FROM metabib.subject_field_entry WHERE source = bib_id GROUP BY field;
765 INSERT INTO metabib.combined_subject_field_entry(record, metabib_field, index_vector)
766 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
767 FROM metabib.subject_field_entry WHERE source = bib_id;
769 DELETE FROM metabib.combined_series_field_entry WHERE record = bib_id;
770 INSERT INTO metabib.combined_series_field_entry(record, metabib_field, index_vector)
771 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
772 FROM metabib.series_field_entry WHERE source = bib_id GROUP BY field;
773 INSERT INTO metabib.combined_series_field_entry(record, metabib_field, index_vector)
774 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
775 FROM metabib.series_field_entry WHERE source = bib_id;
777 DELETE FROM metabib.combined_identifier_field_entry WHERE record = bib_id;
778 INSERT INTO metabib.combined_identifier_field_entry(record, metabib_field, index_vector)
779 SELECT bib_id, field, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
780 FROM metabib.identifier_field_entry WHERE source = bib_id GROUP BY field;
781 INSERT INTO metabib.combined_identifier_field_entry(record, metabib_field, index_vector)
782 SELECT bib_id, NULL, strip(COALESCE(string_agg(index_vector::TEXT,' '),'')::tsvector)
783 FROM metabib.identifier_field_entry WHERE source = bib_id;
786 $func$ LANGUAGE PLPGSQL;
788 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries( bib_id BIGINT, skip_facet BOOL DEFAULT FALSE, skip_browse BOOL DEFAULT FALSE, skip_search BOOL DEFAULT FALSE ) RETURNS VOID AS $func$
791 ind_data metabib.field_entry_template%ROWTYPE;
792 mbe_row metabib.browse_entry%ROWTYPE;
800 SELECT COALESCE(NULLIF(skip_facet, FALSE), EXISTS (SELECT enabled FROM config.internal_flag WHERE name = 'ingest.skip_facet_indexing' AND enabled)) INTO b_skip_facet;
801 SELECT COALESCE(NULLIF(skip_browse, FALSE), EXISTS (SELECT enabled FROM config.internal_flag WHERE name = 'ingest.skip_browse_indexing' AND enabled)) INTO b_skip_browse;
802 SELECT COALESCE(NULLIF(skip_search, FALSE), EXISTS (SELECT enabled FROM config.internal_flag WHERE name = 'ingest.skip_search_indexing' AND enabled)) INTO b_skip_search;
804 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
806 IF NOT b_skip_search THEN
807 FOR fclass IN SELECT * FROM config.metabib_class LOOP
808 -- RAISE NOTICE 'Emptying out %', fclass.name;
809 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
812 IF NOT b_skip_facet THEN
813 DELETE FROM metabib.facet_entry WHERE source = bib_id;
815 IF NOT b_skip_browse THEN
816 DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
820 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
822 -- don't store what has been normalized away
823 CONTINUE WHEN ind_data.value IS NULL;
825 IF ind_data.field < 0 THEN
826 ind_data.field = -1 * ind_data.field;
829 IF ind_data.facet_field AND NOT b_skip_facet THEN
830 INSERT INTO metabib.facet_entry (field, source, value)
831 VALUES (ind_data.field, ind_data.source, ind_data.value);
834 IF ind_data.browse_field AND NOT b_skip_browse THEN
835 -- A caveat about this SELECT: this should take care of replacing
836 -- old mbe rows when data changes, but not if normalization (by
837 -- which I mean specifically the output of
838 -- evergreen.oils_tsearch2()) changes. It may or may not be
839 -- expensive to add a comparison of index_vector to index_vector
840 -- to the WHERE clause below.
842 CONTINUE WHEN ind_data.sort_value IS NULL;
844 value_prepped := metabib.browse_normalize(ind_data.value, ind_data.field);
845 SELECT INTO mbe_row * FROM metabib.browse_entry
846 WHERE value = value_prepped AND sort_value = ind_data.sort_value;
849 mbe_id := mbe_row.id;
851 INSERT INTO metabib.browse_entry
852 ( value, sort_value ) VALUES
853 ( value_prepped, ind_data.sort_value );
855 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
858 INSERT INTO metabib.browse_entry_def_map (entry, def, source, authority)
859 VALUES (mbe_id, ind_data.field, ind_data.source, ind_data.authority);
862 IF ind_data.search_field AND NOT b_skip_search THEN
863 -- Avoid inserting duplicate rows
864 EXECUTE 'SELECT 1 FROM metabib.' || ind_data.field_class ||
865 '_field_entry WHERE field = $1 AND source = $2 AND value = $3'
866 INTO mbe_id USING ind_data.field, ind_data.source, ind_data.value;
867 -- RAISE NOTICE 'Search for an already matching row returned %', mbe_id;
868 IF mbe_id IS NULL THEN
870 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
872 quote_literal(ind_data.field) || $$, $$ ||
873 quote_literal(ind_data.source) || $$, $$ ||
874 quote_literal(ind_data.value) ||
881 IF NOT b_skip_search THEN
882 PERFORM metabib.update_combined_index_vectors(bib_id);
887 $func$ LANGUAGE PLPGSQL;
889 -- default to a space joiner
890 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( BIGINT ) RETURNS SETOF metabib.field_entry_template AS $func$
891 SELECT * FROM biblio.extract_metabib_field_entry($1, ' ');
894 CREATE OR REPLACE FUNCTION authority.flatten_marc ( rid BIGINT ) RETURNS SETOF authority.full_rec AS $func$
896 auth authority.record_entry%ROWTYPE;
897 output authority.full_rec%ROWTYPE;
900 SELECT INTO auth * FROM authority.record_entry WHERE id = rid;
902 FOR field IN SELECT * FROM vandelay.flatten_marc( auth.marc ) LOOP
903 output.record := rid;
904 output.ind1 := field.ind1;
905 output.ind2 := field.ind2;
906 output.tag := field.tag;
907 output.subfield := field.subfield;
908 output.value := field.value;
913 $func$ LANGUAGE PLPGSQL;
915 CREATE OR REPLACE FUNCTION biblio.flatten_marc ( rid BIGINT ) RETURNS SETOF metabib.full_rec AS $func$
917 bib biblio.record_entry%ROWTYPE;
918 output metabib.full_rec%ROWTYPE;
921 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
923 FOR field IN SELECT * FROM vandelay.flatten_marc( bib.marc ) LOOP
924 output.record := rid;
925 output.ind1 := field.ind1;
926 output.ind2 := field.ind2;
927 output.tag := field.tag;
928 output.subfield := field.subfield;
929 output.value := field.value;
934 $func$ LANGUAGE PLPGSQL;
936 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field_list( rid BIGINT, ff TEXT ) RETURNS TEXT[] AS $func$
937 SELECT * FROM vandelay.marc21_extract_fixed_field_list( (SELECT marc FROM biblio.record_entry WHERE id = $1), $2, TRUE );
940 CREATE OR REPLACE FUNCTION biblio.marc21_extract_fixed_field( rid BIGINT, ff TEXT ) RETURNS TEXT AS $func$
941 SELECT * FROM vandelay.marc21_extract_fixed_field( (SELECT marc FROM biblio.record_entry WHERE id = $1), $2, TRUE );
944 CREATE OR REPLACE FUNCTION biblio.marc21_extract_all_fixed_fields( rid BIGINT ) RETURNS SETOF biblio.record_ff_map AS $func$
945 SELECT $1 AS record, ff_name, ff_value FROM vandelay.marc21_extract_all_fixed_fields( (SELECT marc FROM biblio.record_entry WHERE id = $1), TRUE );
948 CREATE OR REPLACE FUNCTION biblio.marc21_physical_characteristics( rid BIGINT ) RETURNS SETOF biblio.marc21_physical_characteristics AS $func$
949 SELECT id, $1 AS record, ptype, subfield, value FROM vandelay.marc21_physical_characteristics( (SELECT marc FROM biblio.record_entry WHERE id = $1) );
952 CREATE OR REPLACE FUNCTION biblio.extract_quality ( marc TEXT, best_lang TEXT, best_type TEXT ) RETURNS INT AS $func$
965 IF marc IS NULL OR marc = '' THEN
969 -- First, the count of tags
970 qual := ARRAY_UPPER(oils_xpath('*[local-name()="datafield"]', marc), 1);
972 -- now go through a bunch of pain to get the record type
973 IF best_type IS NOT NULL THEN
974 ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];
976 IF ldr IS NOT NULL THEN
977 SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
978 SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same
981 tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
982 bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );
984 -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;
986 SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';
988 IF type_map.code IS NOT NULL THEN
989 IF best_type = type_map.code THEN
990 qual := qual + qual / 2;
993 FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = 'Lang' AND rec_type = type_map.code ORDER BY tag DESC LOOP
994 ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
995 IF ff_tag_data = best_lang THEN
1003 -- Now look for some quality metrics
1005 IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
1010 IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
1017 $func$ LANGUAGE PLPGSQL;
1019 CREATE OR REPLACE FUNCTION biblio.extract_fingerprint ( marc text ) RETURNS TEXT AS $func$
1021 idx config.biblio_fingerprint%ROWTYPE;
1022 xfrm config.xml_transform%ROWTYPE;
1024 transformed_xml TEXT;
1026 xml_node_list TEXT[];
1028 output_text TEXT := '';
1031 IF marc IS NULL OR marc = '' THEN
1035 -- Loop over the indexing entries
1036 FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP
1038 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
1040 -- See if we can skip the XSLT ... it's expensive
1041 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
1042 -- Can't skip the transform
1043 IF xfrm.xslt <> '---' THEN
1044 transformed_xml := oils_xslt_process(marc,xfrm.xslt);
1046 transformed_xml := marc;
1049 prev_xfrm := xfrm.name;
1052 raw_text := COALESCE(
1060 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
1069 raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
1070 raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!
1072 IF idx.first_word IS TRUE THEN
1073 raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
1076 output_text := output_text || idx.name || ':' ||
1077 REGEXP_REPLACE(raw_text, E'\\s+', '', 'g') || ' ';
1081 RETURN BTRIM(output_text);
1084 $func$ LANGUAGE PLPGSQL;
1086 -- BEFORE UPDATE OR INSERT trigger for biblio.record_entry
1087 CREATE OR REPLACE FUNCTION biblio.fingerprint_trigger () RETURNS TRIGGER AS $func$
1090 -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')
1092 IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
1096 NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
1097 NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);
1102 $func$ LANGUAGE PLPGSQL;
1104 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_full_rec( bib_id BIGINT ) RETURNS VOID AS $func$
1106 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
1108 DELETE FROM metabib.real_full_rec WHERE record = bib_id;
1110 INSERT INTO metabib.real_full_rec (record, tag, ind1, ind2, subfield, value)
1111 SELECT record, tag, ind1, ind2, subfield, value FROM biblio.flatten_marc( bib_id );
1115 $func$ LANGUAGE PLPGSQL;
1117 CREATE OR REPLACE FUNCTION biblio.extract_located_uris( bib_id BIGINT, marcxml TEXT, editor_id INT ) RETURNS VOID AS $func$
1124 uri_owner_list TEXT[];
1132 -- Clear any URI mappings and call numbers for this bib.
1133 -- This leads to acn / auricnm inflation, but also enables
1134 -- old acn/auricnm's to go away and for bibs to be deleted.
1135 FOR uri_cn_id IN SELECT id FROM asset.call_number WHERE record = bib_id AND label = '##URI##' AND NOT deleted LOOP
1136 DELETE FROM asset.uri_call_number_map WHERE call_number = uri_cn_id;
1137 DELETE FROM asset.call_number WHERE id = uri_cn_id;
1140 uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',marcxml);
1141 IF ARRAY_UPPER(uris,1) > 0 THEN
1142 FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
1143 -- First we pull info out of the 856
1146 uri_href := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
1147 uri_label := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()',uri_xml))[1];
1148 uri_use := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()',uri_xml))[1];
1150 IF uri_label IS NULL THEN
1151 uri_label := uri_href;
1153 CONTINUE WHEN uri_href IS NULL;
1155 -- Get the distinct list of libraries wanting to use
1157 DISTINCT REGEXP_REPLACE(
1159 $re$^.*?\((\w+)\).*$$re$,
1162 ) INTO uri_owner_list
1165 '//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',
1170 IF ARRAY_UPPER(uri_owner_list,1) > 0 THEN
1172 -- look for a matching uri
1173 IF uri_use IS NULL THEN
1174 SELECT id INTO uri_id
1176 WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active
1177 ORDER BY id LIMIT 1;
1178 IF NOT FOUND THEN -- create one
1179 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
1180 SELECT id INTO uri_id
1182 WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active;
1185 SELECT id INTO uri_id
1187 WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active
1188 ORDER BY id LIMIT 1;
1189 IF NOT FOUND THEN -- create one
1190 INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
1191 SELECT id INTO uri_id
1193 WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
1197 FOR j IN 1 .. ARRAY_UPPER(uri_owner_list, 1) LOOP
1198 uri_owner := uri_owner_list[j];
1200 SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = uri_owner;
1201 CONTINUE WHEN NOT FOUND;
1203 -- we need a call number to link through
1204 SELECT id INTO uri_cn_id FROM asset.call_number WHERE owning_lib = uri_owner_id AND record = bib_id AND label = '##URI##' AND NOT deleted;
1206 INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
1207 VALUES (uri_owner_id, bib_id, 'now', 'now', editor_id, editor_id, '##URI##');
1208 SELECT id INTO uri_cn_id FROM asset.call_number WHERE owning_lib = uri_owner_id AND record = bib_id AND label = '##URI##' AND NOT deleted;
1211 -- now, link them if they're not already
1212 SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
1214 INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
1226 $func$ LANGUAGE PLPGSQL;
1228 CREATE OR REPLACE FUNCTION metabib.remap_metarecord_for_bib( bib_id BIGINT, fp TEXT, bib_is_deleted BOOL DEFAULT FALSE, retain_deleted BOOL DEFAULT FALSE ) RETURNS BIGINT AS $func$
1230 new_mapping BOOL := TRUE;
1233 tmp_mr metabib.metarecord%ROWTYPE;
1234 deleted_mrs BIGINT[];
1237 -- We need to make sure we're not a deleted master record of an MR
1238 IF bib_is_deleted THEN
1239 FOR old_mr IN SELECT id FROM metabib.metarecord WHERE master_record = bib_id LOOP
1241 IF NOT retain_deleted THEN -- Go away for any MR that we're master of, unless retained
1242 DELETE FROM metabib.metarecord_source_map WHERE source = bib_id;
1245 -- Now, are there any more sources on this MR?
1246 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = old_mr;
1248 IF source_count = 0 AND NOT retain_deleted THEN -- No other records
1249 deleted_mrs := ARRAY_APPEND(deleted_mrs, old_mr); -- Just in case...
1250 DELETE FROM metabib.metarecord WHERE id = old_mr;
1252 ELSE -- indeed there are. Update it with a null cache and recalcualated master record
1253 UPDATE metabib.metarecord
1255 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp AND NOT deleted ORDER BY quality DESC LIMIT 1)
1260 ELSE -- insert or update
1262 FOR tmp_mr IN SELECT m.* FROM metabib.metarecord m JOIN metabib.metarecord_source_map s ON (s.metarecord = m.id) WHERE s.source = bib_id LOOP
1264 -- Find the first fingerprint-matching
1265 IF old_mr IS NULL AND fp = tmp_mr.fingerprint THEN
1266 old_mr := tmp_mr.id;
1267 new_mapping := FALSE;
1269 ELSE -- Our fingerprint changed ... maybe remove the old MR
1270 DELETE FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id AND source = bib_id; -- remove the old source mapping
1271 SELECT COUNT(*) INTO source_count FROM metabib.metarecord_source_map WHERE metarecord = tmp_mr.id;
1272 IF source_count = 0 THEN -- No other records
1273 deleted_mrs := ARRAY_APPEND(deleted_mrs, tmp_mr.id);
1274 DELETE FROM metabib.metarecord WHERE id = tmp_mr.id;
1280 -- we found no suitable, preexisting MR based on old source maps
1281 IF old_mr IS NULL THEN
1282 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp; -- is there one for our current fingerprint?
1284 IF old_mr IS NULL THEN -- nope, create one and grab its id
1285 INSERT INTO metabib.metarecord ( fingerprint, master_record ) VALUES ( fp, bib_id );
1286 SELECT id INTO old_mr FROM metabib.metarecord WHERE fingerprint = fp;
1288 ELSE -- indeed there is. update it with a null cache and recalcualated master record
1289 UPDATE metabib.metarecord
1291 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp AND NOT deleted ORDER BY quality DESC LIMIT 1)
1295 ELSE -- there was one we already attached to, update its mods cache and master_record
1296 UPDATE metabib.metarecord
1298 master_record = ( SELECT id FROM biblio.record_entry WHERE fingerprint = fp AND NOT deleted ORDER BY quality DESC LIMIT 1)
1303 INSERT INTO metabib.metarecord_source_map (metarecord, source) VALUES (old_mr, bib_id); -- new source mapping
1308 IF ARRAY_UPPER(deleted_mrs,1) > 0 THEN
1309 UPDATE action.hold_request SET target = old_mr WHERE target IN ( SELECT unnest(deleted_mrs) ) AND hold_type = 'M'; -- if we had to delete any MRs above, make sure their holds are moved
1315 $func$ LANGUAGE PLPGSQL;
1318 CREATE OR REPLACE FUNCTION biblio.map_authority_linking (bibid BIGINT, marc TEXT) RETURNS BIGINT AS $func$
1319 DELETE FROM authority.bib_linking WHERE bib = $1;
1320 INSERT INTO authority.bib_linking (bib, authority)
1323 FROM ( SELECT DISTINCT $1 AS bib,
1324 BTRIM(remove_paren_substring(txt))::BIGINT AS authority
1325 FROM unnest(oils_xpath('//*[@code="0"]/text()',$2)) x(txt)
1326 WHERE BTRIM(remove_paren_substring(txt)) ~ $re$^\d+$$re$
1327 ) y JOIN authority.record_entry r ON r.id = y.authority;
1329 $func$ LANGUAGE SQL;
1331 CREATE OR REPLACE FUNCTION metabib.reingest_record_attributes (rid BIGINT, pattr_list TEXT[] DEFAULT NULL, prmarc TEXT DEFAULT NULL, rdeleted BOOL DEFAULT TRUE) RETURNS VOID AS $func$
1333 transformed_xml TEXT;
1334 rmarc TEXT := prmarc;
1338 xfrm config.xml_transform%ROWTYPE;
1339 attr_vector INT[] := '{}'::INT[];
1340 attr_vector_tmp INT[];
1341 attr_list TEXT[] := pattr_list;
1343 norm_attr_value TEXT[];
1345 attr_def config.record_attr_definition%ROWTYPE;
1346 ccvm_row config.coded_value_map%ROWTYPE;
1349 IF attr_list IS NULL OR rdeleted THEN -- need to do the full dance on INSERT or undelete
1350 SELECT ARRAY_AGG(name) INTO attr_list FROM config.record_attr_definition
1353 fixed_field IS NOT NULL OR
1354 xpath IS NOT NULL OR
1355 phys_char_sf IS NOT NULL OR
1362 IF rmarc IS NULL THEN
1363 SELECT marc INTO rmarc FROM biblio.record_entry WHERE id = rid;
1366 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE NOT composite AND name = ANY( attr_list ) ORDER BY format LOOP
1368 attr_value := '{}'::TEXT[];
1369 norm_attr_value := '{}'::TEXT[];
1370 attr_vector_tmp := '{}'::INT[];
1372 SELECT * INTO ccvm_row FROM config.coded_value_map c WHERE c.ctype = attr_def.name LIMIT 1;
1374 -- tag+sf attrs only support SVF
1375 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
1376 SELECT ARRAY[ARRAY_TO_STRING(ARRAY_AGG(value), COALESCE(attr_def.joiner,' '))] INTO attr_value
1377 FROM (SELECT * FROM metabib.full_rec ORDER BY tag, subfield) AS x
1379 AND tag LIKE attr_def.tag
1381 WHEN attr_def.sf_list IS NOT NULL
1382 THEN POSITION(subfield IN attr_def.sf_list) > 0
1389 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
1390 attr_value := vandelay.marc21_extract_fixed_field_list(rmarc, attr_def.fixed_field);
1392 IF NOT attr_def.multi THEN
1393 attr_value := ARRAY[attr_value[1]];
1396 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
1398 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
1400 -- See if we can skip the XSLT ... it's expensive
1401 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
1402 -- Can't skip the transform
1403 IF xfrm.xslt <> '---' THEN
1404 transformed_xml := oils_xslt_process(rmarc,xfrm.xslt);
1406 transformed_xml := rmarc;
1409 prev_xfrm := xfrm.name;
1412 IF xfrm.name IS NULL THEN
1413 -- just grab the marcxml (empty) transform
1414 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
1415 prev_xfrm := xfrm.name;
1418 FOR tmp_xml IN SELECT UNNEST(oils_xpath(attr_def.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]])) LOOP
1419 tmp_val := oils_xpath_string(
1422 COALESCE(attr_def.joiner,' '),
1423 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
1425 IF tmp_val IS NOT NULL AND BTRIM(tmp_val) <> '' THEN
1426 attr_value := attr_value || tmp_val;
1427 EXIT WHEN NOT attr_def.multi;
1431 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
1432 SELECT ARRAY_AGG(m.value) INTO attr_value
1433 FROM vandelay.marc21_physical_characteristics(rmarc) v
1434 LEFT JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
1435 WHERE v.subfield = attr_def.phys_char_sf AND (m.value IS NOT NULL AND BTRIM(m.value) <> '')
1436 AND ( ccvm_row.id IS NULL OR ( ccvm_row.id IS NOT NULL AND v.id IS NOT NULL) );
1438 IF NOT attr_def.multi THEN
1439 attr_value := ARRAY[attr_value[1]];
1444 -- apply index normalizers to attr_value
1445 FOR tmp_val IN SELECT value FROM UNNEST(attr_value) x(value) LOOP
1447 SELECT n.func AS func,
1448 n.param_count AS param_count,
1450 FROM config.index_normalizer n
1451 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
1452 WHERE attr = attr_def.name
1454 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1455 COALESCE( quote_literal( tmp_val ), 'NULL' ) ||
1457 WHEN normalizer.param_count > 0
1458 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1464 IF tmp_val IS NOT NULL AND tmp_val <> '' THEN
1465 -- note that a string that contains only blanks
1466 -- is a valid value for some attributes
1467 norm_attr_value := norm_attr_value || tmp_val;
1471 IF attr_def.filter THEN
1472 -- Create unknown uncontrolled values and find the IDs of the values
1473 IF ccvm_row.id IS NULL THEN
1474 FOR tmp_val IN SELECT value FROM UNNEST(norm_attr_value) x(value) LOOP
1475 IF tmp_val IS NOT NULL AND BTRIM(tmp_val) <> '' THEN
1476 BEGIN -- use subtransaction to isolate unique constraint violations
1477 INSERT INTO metabib.uncontrolled_record_attr_value ( attr, value ) VALUES ( attr_def.name, tmp_val );
1478 EXCEPTION WHEN unique_violation THEN END;
1482 SELECT ARRAY_AGG(id) INTO attr_vector_tmp FROM metabib.uncontrolled_record_attr_value WHERE attr = attr_def.name AND value = ANY( norm_attr_value );
1484 SELECT ARRAY_AGG(id) INTO attr_vector_tmp FROM config.coded_value_map WHERE ctype = attr_def.name AND code = ANY( norm_attr_value );
1487 -- Add the new value to the vector
1488 attr_vector := attr_vector || attr_vector_tmp;
1491 IF attr_def.sorter THEN
1492 DELETE FROM metabib.record_sorter WHERE source = rid AND attr = attr_def.name;
1493 IF norm_attr_value[1] IS NOT NULL THEN
1494 INSERT INTO metabib.record_sorter (source, attr, value) VALUES (rid, attr_def.name, norm_attr_value[1]);
1500 /* We may need to rewrite the vlist to contain
1501 the intersection of new values for requested
1502 attrs and old values for ignored attrs. To
1503 do this, we take the old attr vlist and
1504 subtract any values that are valid for the
1505 requested attrs, and then add back the new
1506 set of attr values. */
1508 IF ARRAY_LENGTH(pattr_list, 1) > 0 THEN
1509 SELECT vlist INTO attr_vector_tmp FROM metabib.record_attr_vector_list WHERE source = rid;
1510 SELECT attr_vector_tmp - ARRAY_AGG(id::INT) INTO attr_vector_tmp FROM metabib.full_attr_id_map WHERE attr = ANY (pattr_list);
1511 attr_vector := attr_vector || attr_vector_tmp;
1514 -- On to composite attributes, now that the record attrs have been pulled. Processed in name order, so later composite
1515 -- attributes can depend on earlier ones.
1516 PERFORM metabib.compile_composite_attr_cache_init();
1517 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE composite AND name = ANY( attr_list ) ORDER BY name LOOP
1519 FOR ccvm_row IN SELECT * FROM config.coded_value_map c WHERE c.ctype = attr_def.name ORDER BY value LOOP
1521 tmp_val := metabib.compile_composite_attr( ccvm_row.id );
1522 CONTINUE WHEN tmp_val IS NULL OR tmp_val = ''; -- nothing to do
1524 IF attr_def.filter THEN
1525 IF attr_vector @@ tmp_val::query_int THEN
1526 attr_vector = attr_vector + intset(ccvm_row.id);
1527 EXIT WHEN NOT attr_def.multi;
1531 IF attr_def.sorter THEN
1532 IF attr_vector @@ tmp_val THEN
1533 DELETE FROM metabib.record_sorter WHERE source = rid AND attr = attr_def.name;
1534 INSERT INTO metabib.record_sorter (source, attr, value) VALUES (rid, attr_def.name, ccvm_row.code);
1542 IF ARRAY_LENGTH(attr_vector, 1) > 0 THEN
1543 IF rdeleted THEN -- initial insert OR revivication
1544 DELETE FROM metabib.record_attr_vector_list WHERE source = rid;
1545 INSERT INTO metabib.record_attr_vector_list (source, vlist) VALUES (rid, attr_vector);
1547 UPDATE metabib.record_attr_vector_list SET vlist = attr_vector WHERE source = rid;
1553 $func$ LANGUAGE PLPGSQL;
1556 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
1557 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
1562 IF NEW.deleted THEN -- If this bib is deleted
1564 PERFORM * FROM config.internal_flag WHERE
1565 name = 'ingest.metarecord_mapping.preserve_on_delete' AND enabled;
1567 tmp_bool := FOUND; -- Just in case this is changed by some other statement
1569 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint, TRUE, tmp_bool );
1571 IF NOT tmp_bool THEN
1572 -- One needs to keep these around to support searches
1573 -- with the #deleted modifier, so one should turn on the named
1574 -- internal flag for that functionality.
1575 DELETE FROM metabib.record_attr_vector_list WHERE source = NEW.id;
1578 DELETE FROM authority.bib_linking WHERE bib = NEW.id; -- Avoid updating fields in bibs that are no longer visible
1579 DELETE FROM biblio.peer_bib_copy_map WHERE peer_record = NEW.id; -- Separate any multi-homed items
1580 DELETE FROM metabib.browse_entry_def_map WHERE source = NEW.id; -- Don't auto-suggest deleted bibs
1581 RETURN NEW; -- and we're done
1584 IF TG_OP = 'UPDATE' THEN -- re-ingest?
1585 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
1587 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
1592 -- Record authority linking
1593 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_linking' AND enabled;
1595 PERFORM biblio.map_authority_linking( NEW.id, NEW.marc );
1598 -- Flatten and insert the mfr data
1599 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_full_rec' AND enabled;
1601 PERFORM metabib.reingest_metabib_full_rec(NEW.id);
1603 -- Now we pull out attribute data, which is dependent on the mfr for all but XPath-based fields
1604 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_rec_descriptor' AND enabled;
1606 PERFORM metabib.reingest_record_attributes(NEW.id, NULL, NEW.marc, TG_OP = 'INSERT' OR OLD.deleted);
1610 -- Gather and insert the field entry data
1611 PERFORM metabib.reingest_metabib_field_entries(NEW.id);
1613 -- Located URI magic
1614 IF TG_OP = 'INSERT' THEN
1615 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1617 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1620 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
1622 PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor );
1626 -- (re)map metarecord-bib linking
1627 IF TG_OP = 'INSERT' THEN -- if not deleted and performing an insert, check for the flag
1628 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
1630 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1632 ELSE -- we're doing an update, and we're not deleted, remap
1633 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_update' AND enabled;
1635 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
1641 $func$ LANGUAGE PLPGSQL;
1643 CREATE OR REPLACE FUNCTION metabib.browse_normalize(facet_text TEXT, mapped_field INT) RETURNS TEXT AS $$
1649 SELECT n.func AS func,
1650 n.param_count AS param_count,
1652 FROM config.index_normalizer n
1653 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
1654 WHERE m.field = mapped_field AND m.pos < 0
1657 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1658 quote_literal( facet_text ) ||
1660 WHEN normalizer.param_count > 0
1661 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1664 ')' INTO facet_text;
1671 $$ LANGUAGE PLPGSQL;
1674 -- This mimics a specific part of QueryParser, turning the first part of a
1675 -- classed search (search_class) into a set of classes and possibly fields.
1676 -- search_class might look like "author" or "title|proper" or "ti|uniform"
1677 -- or "au" or "au|corporate|personal" or anything like that, where the first
1678 -- element of the list you get by separating on the "|" character is either
1679 -- a registered class (config.metabib_class) or an alias
1680 -- (config.metabib_search_alias), and the rest of any such elements are
1681 -- fields (config.metabib_field).
1683 FUNCTION metabib.search_class_to_registered_components(search_class TEXT)
1684 RETURNS SETOF RECORD AS $func$
1686 search_parts TEXT[];
1688 search_part_count INTEGER;
1690 registered_class config.metabib_class%ROWTYPE;
1691 registered_alias config.metabib_search_alias%ROWTYPE;
1692 registered_field config.metabib_field%ROWTYPE;
1694 search_parts := REGEXP_SPLIT_TO_ARRAY(search_class, E'\\|');
1696 search_part_count := ARRAY_LENGTH(search_parts, 1);
1697 IF search_part_count = 0 THEN
1700 SELECT INTO registered_class
1701 * FROM config.metabib_class WHERE name = search_parts[1];
1703 IF search_part_count < 2 THEN -- all fields
1704 rec := (registered_class.name, NULL::INTEGER);
1708 FOR field_name IN SELECT *
1709 FROM UNNEST(search_parts[2:search_part_count]) LOOP
1710 SELECT INTO registered_field
1711 * FROM config.metabib_field
1712 WHERE name = field_name AND
1713 field_class = registered_class.name;
1715 rec := (registered_class.name, registered_field.id);
1720 -- maybe we have an alias?
1721 SELECT INTO registered_alias
1722 * FROM config.metabib_search_alias WHERE alias=search_parts[1];
1726 IF search_part_count < 2 THEN -- return w/e the alias says
1728 registered_alias.field_class, registered_alias.field
1733 FOR field_name IN SELECT *
1734 FROM UNNEST(search_parts[2:search_part_count]) LOOP
1735 SELECT INTO registered_field
1736 * FROM config.metabib_field
1737 WHERE name = field_name AND
1738 field_class = registered_alias.field_class;
1741 registered_alias.field_class,
1752 $func$ LANGUAGE PLPGSQL ROWS 1;
1755 -- Given a string such as a user might type into a search box, prepare
1756 -- two changed variants for TO_TSQUERY(). See
1757 -- http://www.postgresql.org/docs/9.0/static/textsearch-controls.html
1758 -- The first variant is normalized to match indexed documents regardless
1759 -- of diacritics. The second variant keeps its diacritics for proper
1760 -- highlighting via TS_HEADLINE().
1762 FUNCTION metabib.autosuggest_prepare_tsquery(orig TEXT) RETURNS TEXT[] AS
1765 orig_ended_in_space BOOLEAN;
1770 orig_ended_in_space := orig ~ E'\\s$';
1772 orig := ARRAY_TO_STRING(
1773 evergreen.regexp_split_to_array(orig, E'\\W+'), ' '
1776 normalized := public.naco_normalize(orig); -- also trim()s
1777 plain := trim(orig);
1779 IF NOT orig_ended_in_space THEN
1780 plain := plain || ':*';
1781 normalized := normalized || ':*';
1784 plain := ARRAY_TO_STRING(
1785 evergreen.regexp_split_to_array(plain, E'\\s+'), ' & '
1787 normalized := ARRAY_TO_STRING(
1788 evergreen.regexp_split_to_array(normalized, E'\\s+'), ' & '
1791 RETURN ARRAY[normalized, plain];
1793 $$ LANGUAGE PLPGSQL;
1795 -- Functions metabib.browse, metabib.staged_browse, and metabib.suggest_browse_entries
1796 -- will be created later, after internal dependencies are resolved.
1798 CREATE OR REPLACE FUNCTION public.oils_tsearch2 () RETURNS TRIGGER AS $$
1802 temp_vector TEXT := '';
1808 NEW.index_vector = ''::tsvector;
1810 IF TG_TABLE_NAME::TEXT ~ 'field_entry$' THEN
1812 SELECT n.func AS func,
1813 n.param_count AS param_count,
1815 FROM config.index_normalizer n
1816 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
1817 WHERE field = NEW.field AND m.pos < 0
1819 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1820 quote_literal( value ) ||
1822 WHEN normalizer.param_count > 0
1823 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1833 SELECT n.func AS func,
1834 n.param_count AS param_count,
1836 FROM config.index_normalizer n
1837 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
1838 WHERE field = NEW.field AND m.pos >= 0
1840 EXECUTE 'SELECT ' || normalizer.func || '(' ||
1841 quote_literal( value ) ||
1843 WHEN normalizer.param_count > 0
1844 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
1852 IF TG_TABLE_NAME::TEXT ~ 'browse_entry$' THEN
1854 value := ARRAY_TO_STRING(
1855 evergreen.regexp_split_to_array(value, E'\\W+'), ' '
1857 value := public.search_normalize(value);
1858 NEW.index_vector = to_tsvector(TG_ARGV[0]::regconfig, value);
1860 ELSIF TG_TABLE_NAME::TEXT ~ 'field_entry$' THEN
1863 SELECT DISTINCT m.ts_config, m.index_weight
1864 FROM config.metabib_class_ts_map m
1865 LEFT JOIN metabib.record_attr_vector_list r ON (r.source = NEW.source)
1866 LEFT JOIN config.coded_value_map ccvm ON (
1867 ccvm.ctype IN ('item_lang', 'language') AND
1868 ccvm.code = m.index_lang AND
1869 r.vlist @> intset(ccvm.id)
1871 WHERE m.field_class = TG_ARGV[0]
1873 AND (m.always OR NOT EXISTS (SELECT 1 FROM config.metabib_field_ts_map WHERE metabib_field = NEW.field))
1874 AND (m.index_lang IS NULL OR ccvm.id IS NOT NULL)
1876 SELECT DISTINCT m.ts_config, m.index_weight
1877 FROM config.metabib_field_ts_map m
1878 LEFT JOIN metabib.record_attr_vector_list r ON (r.source = NEW.source)
1879 LEFT JOIN config.coded_value_map ccvm ON (
1880 ccvm.ctype IN ('item_lang', 'language') AND
1881 ccvm.code = m.index_lang AND
1882 r.vlist @> intset(ccvm.id)
1884 WHERE m.metabib_field = NEW.field
1886 AND (m.index_lang IS NULL OR ccvm.id IS NOT NULL)
1887 ORDER BY index_weight ASC
1891 IF cur_weight IS NOT NULL AND cur_weight != ts_rec.index_weight THEN
1892 NEW.index_vector = NEW.index_vector || setweight(temp_vector::tsvector,cur_weight);
1896 cur_weight = ts_rec.index_weight;
1897 SELECT INTO temp_vector temp_vector || ' ' || to_tsvector(ts_rec.ts_config::regconfig, value)::TEXT;
1900 NEW.index_vector = NEW.index_vector || setweight(temp_vector::tsvector,cur_weight);
1902 NEW.index_vector = to_tsvector(TG_ARGV[0]::regconfig, value);
1907 $$ LANGUAGE PLPGSQL;
1910 CREATE TYPE metabib.flat_browse_entry_appearance AS (
1911 browse_entry BIGINT,
1916 sources INT, -- visible ones, that is
1917 asources INT, -- visible ones, that is
1918 row_number INT, -- internal use, sort of
1919 accurate BOOL, -- Count in sources field is accurate? Not
1920 -- if we had more than a browse superpage
1921 -- of records to look at.
1922 aaccurate BOOL, -- See previous comment...
1927 CREATE OR REPLACE FUNCTION metabib.browse_bib_pivot(
1930 ) RETURNS BIGINT AS $p$
1932 FROM metabib.browse_entry mbe
1933 JOIN metabib.browse_entry_def_map mbedm ON (
1934 mbedm.entry = mbe.id
1935 AND mbedm.def = ANY($1)
1937 WHERE mbe.sort_value >= public.naco_normalize($2)
1938 ORDER BY mbe.sort_value, mbe.value LIMIT 1;
1939 $p$ LANGUAGE SQL STABLE;
1941 CREATE OR REPLACE FUNCTION metabib.browse_authority_pivot(
1944 ) RETURNS BIGINT AS $p$
1946 FROM metabib.browse_entry mbe
1947 JOIN metabib.browse_entry_simple_heading_map mbeshm ON ( mbeshm.entry = mbe.id )
1948 JOIN authority.simple_heading ash ON ( mbeshm.simple_heading = ash.id )
1949 JOIN authority.control_set_auth_field_metabib_field_map_refs map ON (
1950 ash.atag = map.authority_field
1951 AND map.metabib_field = ANY($1)
1953 WHERE mbe.sort_value >= public.naco_normalize($2)
1954 ORDER BY mbe.sort_value, mbe.value LIMIT 1;
1955 $p$ LANGUAGE SQL STABLE;
1957 CREATE OR REPLACE FUNCTION metabib.browse_authority_refs_pivot(
1960 ) RETURNS BIGINT AS $p$
1962 FROM metabib.browse_entry mbe
1963 JOIN metabib.browse_entry_simple_heading_map mbeshm ON ( mbeshm.entry = mbe.id )
1964 JOIN authority.simple_heading ash ON ( mbeshm.simple_heading = ash.id )
1965 JOIN authority.control_set_auth_field_metabib_field_map_refs_only map ON (
1966 ash.atag = map.authority_field
1967 AND map.metabib_field = ANY($1)
1969 WHERE mbe.sort_value >= public.naco_normalize($2)
1970 ORDER BY mbe.sort_value, mbe.value LIMIT 1;
1971 $p$ LANGUAGE SQL STABLE;
1973 CREATE OR REPLACE FUNCTION metabib.browse_pivot(
1976 ) RETURNS BIGINT AS $p$
1977 SELECT id FROM metabib.browse_entry
1979 metabib.browse_bib_pivot($1, $2),
1980 metabib.browse_authority_refs_pivot($1,$2) -- only look in 4xx, 5xx, 7xx of authority
1982 ORDER BY sort_value, value LIMIT 1;
1983 $p$ LANGUAGE SQL STABLE;
1986 -- This function is used to help clean up facet labels. Due to quirks in
1987 -- MARC parsing, some facet labels may be generated with periods or commas
1988 -- at the end. This will strip a trailing commas off all the time, and
1989 -- periods when they don't look like they are part of initials.
1990 -- Smith, John => no change
1991 -- Smith, John, => Smith, John
1992 -- Smith, John. => Smith, John
1993 -- Public, John Q. => no change
1994 CREATE OR REPLACE FUNCTION metabib.trim_trailing_punctuation ( TEXT ) RETURNS TEXT AS $$
2000 last_char = substring(result from '.$');
2002 IF last_char = ',' THEN
2003 result := substring(result from '^(.*),$');
2005 ELSIF last_char = '.' THEN
2006 IF substring(result from ' \w\.$') IS NULL THEN
2007 result := substring(result from '^(.*)\.$');
2014 $$ language 'plpgsql';