4 ALTER TABLE config.metabib_field
5 ADD COLUMN display_xpath TEXT,
6 ADD COLUMN display_field BOOL NOT NULL DEFAULT FALSE;
8 CREATE TABLE config.display_field_map (
10 field INTEGER REFERENCES config.metabib_field (id),
11 multi BOOLEAN DEFAULT FALSE
14 CREATE TABLE metabib.display_entry (
15 id BIGSERIAL PRIMARY KEY,
16 source BIGINT NOT NULL REFERENCES biblio.record_entry (id),
17 field INT NOT NULL REFERENCES config.metabib_field (id),
21 CREATE INDEX metabib_display_entry_field_idx ON metabib.display_entry (field);
22 CREATE INDEX metabib_display_entry_source_idx ON metabib.display_entry (source);
24 -- one row per display entry fleshed with field info
25 CREATE VIEW metabib.flat_display_entry AS
33 FROM metabib.display_entry mde
34 JOIN config.metabib_field cmf ON (cmf.id = mde.field)
35 JOIN config.display_field_map cdfm ON (cdfm.field = mde.field)
38 -- like flat_display_entry except values are compressed
39 -- into one row per display_field_map and JSON-ified.
40 CREATE VIEW metabib.compressed_display_entry AS
48 TO_JSON(ARRAY_AGG(value))
52 FROM metabib.flat_display_entry
53 GROUP BY 1, 2, 3, 4, 5
56 -- TODO: expand to encompass all well-known fields
57 CREATE VIEW metabib.wide_display_entry AS
60 COALESCE(mcde_title.value, 'null') AS title,
61 COALESCE(mcde_author.value, 'null') AS author,
62 COALESCE(mcde_subject.value, 'null') AS subject,
63 COALESCE(mcde_topic_subject.value, 'null') AS topic_subject,
64 COALESCE(mcde_isbn.value, 'null') AS isbn
65 -- ensure one row per bre regardless of any display fields
66 FROM biblio.record_entry bre
67 LEFT JOIN metabib.compressed_display_entry mcde_title
68 ON (bre.id = mcde_title.source AND mcde_title.name = 'title')
69 LEFT JOIN metabib.compressed_display_entry mcde_author
70 ON (bre.id = mcde_author.source AND mcde_author.name = 'author')
71 LEFT JOIN metabib.compressed_display_entry mcde_subject
72 ON (bre.id = mcde_subject.source AND mcde_subject.name = 'subject')
73 LEFT JOIN metabib.compressed_display_entry mcde_topic_subject
74 ON (bre.id = mcde_topic_subject.source AND mcde_topic_subject.name = 'topic_subject')
75 LEFT JOIN metabib.compressed_display_entry mcde_isbn
76 ON (bre.id = mcde_isbn.source AND mcde_isbn.name = 'isbn')
80 CREATE OR REPLACE FUNCTION metabib.display_field_normalize_trigger ()
84 display_field_text TEXT;
86 display_field_text := NEW.value;
89 SELECT n.func AS func,
90 n.param_count AS param_count,
92 FROM config.index_normalizer n
93 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
94 WHERE m.field = NEW.field AND m.pos < 0
97 EXECUTE 'SELECT ' || normalizer.func || '(' ||
98 quote_literal( display_field_text ) ||
100 WHEN normalizer.param_count > 0
101 THEN ',' || REPLACE(REPLACE(BTRIM(
102 normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
105 ')' INTO display_field_text;
109 NEW.value = display_field_text;
115 CREATE TRIGGER display_field_normalize_tgr
116 BEFORE UPDATE OR INSERT ON metabib.display_entry
117 FOR EACH ROW EXECUTE PROCEDURE metabib.display_field_normalize_trigger();
119 CREATE OR REPLACE FUNCTION evergreen.display_field_force_nfc()
120 RETURNS TRIGGER AS $$
122 NEW.value := force_unicode_normal_form(NEW.value,'NFC');
127 CREATE TRIGGER display_field_force_nfc_tgr
128 BEFORE UPDATE OR INSERT ON metabib.display_entry
129 FOR EACH ROW EXECUTE PROCEDURE evergreen.display_field_force_nfc();
131 ALTER TYPE metabib.field_entry_template ADD ATTRIBUTE display_field BOOL;
133 DROP FUNCTION metabib.reingest_metabib_field_entries(BIGINT, BOOL, BOOL, BOOL);
134 DROP FUNCTION biblio.extract_metabib_field_entry(BIGINT);
135 DROP FUNCTION biblio.extract_metabib_field_entry(BIGINT, TEXT);
137 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry (
142 ) RETURNS SETOF metabib.field_entry_template AS $func$
144 bib biblio.record_entry%ROWTYPE;
145 idx config.metabib_field%ROWTYPE;
146 xfrm config.xml_transform%ROWTYPE;
148 transformed_xml TEXT;
150 xml_node_list TEXT[];
157 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
159 authority_link BIGINT;
160 output_row metabib.field_entry_template%ROWTYPE;
164 -- Start out with no field-use bools set
165 output_row.browse_field = FALSE;
166 output_row.facet_field = FALSE;
167 output_row.display_field = FALSE;
168 output_row.search_field = FALSE;
171 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
173 -- Loop over the indexing entries
174 FOR idx IN SELECT * FROM config.metabib_field WHERE id = ANY (only_fields) ORDER BY format LOOP
176 process_idx := FALSE;
177 IF idx.display_field AND 'display' = ANY (field_types) THEN process_idx = TRUE; END IF;
178 IF idx.browse_field AND 'browse' = ANY (field_types) THEN process_idx = TRUE; END IF;
179 IF idx.search_field AND 'search' = ANY (field_types) THEN process_idx = TRUE; END IF;
180 IF idx.facet_field AND 'facet' = ANY (field_types) THEN process_idx = TRUE; END IF;
181 CONTINUE WHEN process_idx = FALSE;
183 joiner := COALESCE(idx.joiner, default_joiner);
185 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
187 -- See if we can skip the XSLT ... it's expensive
188 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
189 -- Can't skip the transform
190 IF xfrm.xslt <> '---' THEN
191 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
193 transformed_xml := bib.marc;
196 prev_xfrm := xfrm.name;
199 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
202 FOR xml_node IN SELECT x FROM unnest(xml_node_list) AS x LOOP
203 CONTINUE WHEN xml_node !~ E'^\\s*<';
205 -- XXX much of this should be moved into oils_xpath_string...
206 curr_text := ARRAY_TO_STRING(evergreen.array_remove_item_by_value(evergreen.array_remove_item_by_value(
207 oils_xpath( '//text()', -- get the content of all the nodes within the main selected node
208 REGEXP_REPLACE( xml_node, E'\\s+', ' ', 'g' ) -- Translate adjacent whitespace to a single space
209 ), ' '), ''), -- throw away morally empty (bankrupt?) strings
213 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
215 IF raw_text IS NOT NULL THEN
216 raw_text := raw_text || joiner;
219 raw_text := COALESCE(raw_text,'') || curr_text;
221 -- autosuggest/metabib.browse_entry
222 IF idx.browse_field THEN
224 IF idx.browse_xpath IS NOT NULL AND idx.browse_xpath <> '' THEN
225 browse_text := oils_xpath_string( idx.browse_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
227 browse_text := curr_text;
230 IF idx.browse_sort_xpath IS NOT NULL AND
231 idx.browse_sort_xpath <> '' THEN
233 sort_value := oils_xpath_string(
234 idx.browse_sort_xpath, xml_node, joiner,
235 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
238 sort_value := browse_text;
241 output_row.field_class = idx.field_class;
242 output_row.field = idx.id;
243 output_row.source = rid;
244 output_row.value = BTRIM(REGEXP_REPLACE(browse_text, E'\\s+', ' ', 'g'));
245 output_row.sort_value :=
246 public.naco_normalize(sort_value);
248 output_row.authority := NULL;
250 IF idx.authority_xpath IS NOT NULL AND idx.authority_xpath <> '' THEN
251 authority_text := oils_xpath_string(
252 idx.authority_xpath, xml_node, joiner,
254 ARRAY[xfrm.prefix, xfrm.namespace_uri],
255 ARRAY['xlink','http://www.w3.org/1999/xlink']
259 IF authority_text ~ '^\d+$' THEN
260 authority_link := authority_text::BIGINT;
261 PERFORM * FROM authority.record_entry WHERE id = authority_link;
263 output_row.authority := authority_link;
269 output_row.browse_field = TRUE;
270 -- Returning browse rows with search_field = true for search+browse
271 -- configs allows us to retain granularity of being able to search
272 -- browse fields with "starts with" type operators (for example, for
273 -- titles of songs in music albums)
274 IF idx.search_field THEN
275 output_row.search_field = TRUE;
277 RETURN NEXT output_row;
278 output_row.browse_field = FALSE;
279 output_row.search_field = FALSE;
280 output_row.sort_value := NULL;
283 -- insert raw node text for faceting
284 IF idx.facet_field THEN
286 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
287 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
289 facet_text := curr_text;
292 output_row.field_class = idx.field_class;
293 output_row.field = -1 * idx.id;
294 output_row.source = rid;
295 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
297 output_row.facet_field = TRUE;
298 RETURN NEXT output_row;
299 output_row.facet_field = FALSE;
302 -- insert raw node text for display
303 IF idx.display_field THEN
305 IF idx.display_xpath IS NOT NULL AND idx.display_xpath <> '' THEN
306 display_text := oils_xpath_string( idx.display_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
308 display_text := curr_text;
311 output_row.field_class = idx.field_class;
312 output_row.field = -1 * idx.id;
313 output_row.source = rid;
314 output_row.value = BTRIM(REGEXP_REPLACE(display_text, E'\\s+', ' ', 'g'));
316 output_row.display_field = TRUE;
317 RETURN NEXT output_row;
318 output_row.display_field = FALSE;
323 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
325 -- insert combined node text for searching
326 IF idx.search_field THEN
327 output_row.field_class = idx.field_class;
328 output_row.field = idx.id;
329 output_row.source = rid;
330 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
332 output_row.search_field = TRUE;
333 RETURN NEXT output_row;
334 output_row.search_field = FALSE;
341 $func$ LANGUAGE PLPGSQL;
343 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries(
345 skip_facet BOOL DEFAULT FALSE,
346 skip_display BOOL DEFAULT FALSE,
347 skip_browse BOOL DEFAULT FALSE,
348 skip_search BOOL DEFAULT FALSE,
349 only_fields INT[] DEFAULT '{}'::INT[]
350 ) RETURNS VOID AS $func$
353 ind_data metabib.field_entry_template%ROWTYPE;
354 mbe_row metabib.browse_entry%ROWTYPE;
361 field_list INT[] := only_fields;
362 field_types TEXT[] := '{}'::TEXT[];
365 IF field_list = '{}'::INT[] THEN
366 SELECT ARRAY_AGG(id) INTO field_list FROM config.metabib_field;
369 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;
370 SELECT COALESCE(NULLIF(skip_display, FALSE), EXISTS (SELECT enabled FROM config.internal_flag WHERE name = 'ingest.skip_display_indexing' AND enabled)) INTO b_skip_display;
371 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;
372 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;
374 IF NOT b_skip_facet THEN field_types := field_types || '{facet}'; END IF;
375 IF NOT b_skip_display THEN field_types := field_types || '{display}'; END IF;
376 IF NOT b_skip_browse THEN field_types := field_types || '{browse}'; END IF;
377 IF NOT b_skip_search THEN field_types := field_types || '{search}'; END IF;
379 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
381 IF NOT b_skip_search THEN
382 FOR fclass IN SELECT * FROM config.metabib_class LOOP
383 -- RAISE NOTICE 'Emptying out %', fclass.name;
384 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
387 IF NOT b_skip_facet THEN
388 DELETE FROM metabib.facet_entry WHERE source = bib_id;
390 IF NOT b_skip_display THEN
391 DELETE FROM metabib.display_entry WHERE source = bib_id;
393 IF NOT b_skip_browse THEN
394 DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
398 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id, ' ', field_types, field_list ) LOOP
400 -- don't store what has been normalized away
401 CONTINUE WHEN ind_data.value IS NULL;
403 IF ind_data.field < 0 THEN
404 ind_data.field = -1 * ind_data.field;
407 IF ind_data.facet_field AND NOT b_skip_facet THEN
408 INSERT INTO metabib.facet_entry (field, source, value)
409 VALUES (ind_data.field, ind_data.source, ind_data.value);
412 IF ind_data.display_field AND NOT b_skip_display THEN
413 INSERT INTO metabib.display_entry (field, source, value)
414 VALUES (ind_data.field, ind_data.source, ind_data.value);
418 IF ind_data.browse_field AND NOT b_skip_browse THEN
419 -- A caveat about this SELECT: this should take care of replacing
420 -- old mbe rows when data changes, but not if normalization (by
421 -- which I mean specifically the output of
422 -- evergreen.oils_tsearch2()) changes. It may or may not be
423 -- expensive to add a comparison of index_vector to index_vector
424 -- to the WHERE clause below.
426 CONTINUE WHEN ind_data.sort_value IS NULL;
428 value_prepped := metabib.browse_normalize(ind_data.value, ind_data.field);
429 SELECT INTO mbe_row * FROM metabib.browse_entry
430 WHERE value = value_prepped AND sort_value = ind_data.sort_value;
433 mbe_id := mbe_row.id;
435 INSERT INTO metabib.browse_entry
436 ( value, sort_value ) VALUES
437 ( value_prepped, ind_data.sort_value );
439 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
442 INSERT INTO metabib.browse_entry_def_map (entry, def, source, authority)
443 VALUES (mbe_id, ind_data.field, ind_data.source, ind_data.authority);
446 IF ind_data.search_field AND NOT b_skip_search THEN
447 -- Avoid inserting duplicate rows
448 EXECUTE 'SELECT 1 FROM metabib.' || ind_data.field_class ||
449 '_field_entry WHERE field = $1 AND source = $2 AND value = $3'
450 INTO mbe_id USING ind_data.field, ind_data.source, ind_data.value;
451 -- RAISE NOTICE 'Search for an already matching row returned %', mbe_id;
452 IF mbe_id IS NULL THEN
454 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
456 quote_literal(ind_data.field) || $$, $$ ||
457 quote_literal(ind_data.source) || $$, $$ ||
458 quote_literal(ind_data.value) ||
465 IF NOT b_skip_search THEN
466 PERFORM metabib.update_combined_index_vectors(bib_id);
471 $func$ LANGUAGE PLPGSQL;
473 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
474 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
479 IF NEW.deleted THEN -- If this bib is deleted
481 PERFORM * FROM config.internal_flag WHERE
482 name = 'ingest.metarecord_mapping.preserve_on_delete' AND enabled;
484 tmp_bool := FOUND; -- Just in case this is changed by some other statement
486 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint, TRUE, tmp_bool );
489 -- One needs to keep these around to support searches
490 -- with the #deleted modifier, so one should turn on the named
491 -- internal flag for that functionality.
492 DELETE FROM metabib.record_attr_vector_list WHERE source = NEW.id;
495 DELETE FROM authority.bib_linking WHERE bib = NEW.id; -- Avoid updating fields in bibs that are no longer visible
496 DELETE FROM biblio.peer_bib_copy_map WHERE peer_record = NEW.id; -- Separate any multi-homed items
497 DELETE FROM metabib.browse_entry_def_map WHERE source = NEW.id; -- Don't auto-suggest deleted bibs
498 RETURN NEW; -- and we're done
501 IF TG_OP = 'UPDATE' THEN -- re-ingest?
502 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
504 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
509 -- Record authority linking
510 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_linking' AND enabled;
512 PERFORM biblio.map_authority_linking( NEW.id, NEW.marc );
515 -- Flatten and insert the mfr data
516 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_full_rec' AND enabled;
518 PERFORM metabib.reingest_metabib_full_rec(NEW.id);
520 -- Now we pull out attribute data, which is dependent on the mfr for all but XPath-based fields
521 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_rec_descriptor' AND enabled;
523 PERFORM metabib.reingest_record_attributes(NEW.id, NULL, NEW.marc, TG_OP = 'INSERT' OR OLD.deleted);
527 -- Gather and insert the field entry data
528 PERFORM metabib.reingest_metabib_field_entries(NEW.id);
531 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
532 IF NOT FOUND THEN PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor ); END IF;
534 -- (re)map metarecord-bib linking
535 IF TG_OP = 'INSERT' THEN -- if not deleted and performing an insert, check for the flag
536 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
538 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
540 ELSE -- we're doing an update, and we're not deleted, remap
541 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_update' AND enabled;
543 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
549 $func$ LANGUAGE PLPGSQL;