4 SELECT evergreen.upgrade_deps_block_check('1073', :eg_version);
6 ALTER TABLE config.metabib_field
7 ADD COLUMN display_xpath TEXT,
8 ADD COLUMN display_field BOOL NOT NULL DEFAULT FALSE;
10 CREATE TABLE config.display_field_map (
11 name TEXT PRIMARY KEY,
12 field INTEGER REFERENCES config.metabib_field (id),
13 multi BOOLEAN DEFAULT FALSE
16 CREATE TABLE metabib.display_entry (
17 id BIGSERIAL PRIMARY KEY,
18 source BIGINT NOT NULL REFERENCES biblio.record_entry (id),
19 field INT NOT NULL REFERENCES config.metabib_field (id),
23 CREATE INDEX metabib_display_entry_field_idx ON metabib.display_entry (field);
24 CREATE INDEX metabib_display_entry_source_idx ON metabib.display_entry (source);
26 -- one row per display entry fleshed with field info
27 CREATE VIEW metabib.flat_display_entry AS
35 FROM metabib.display_entry mde
36 JOIN config.metabib_field cmf ON (cmf.id = mde.field)
37 JOIN config.display_field_map cdfm ON (cdfm.field = mde.field)
40 -- like flat_display_entry except values are compressed
41 -- into one row per display_field_map and JSON-ified.
42 CREATE VIEW metabib.compressed_display_entry AS
50 TO_JSON(ARRAY_AGG(value))
54 FROM metabib.flat_display_entry
55 GROUP BY 1, 2, 3, 4, 5
58 -- TODO: expand to encompass all well-known fields
59 CREATE VIEW metabib.wide_display_entry AS
62 COALESCE(mcde_title.value, 'null') AS title,
63 COALESCE(mcde_author.value, 'null') AS author,
64 COALESCE(mcde_subject.value, 'null') AS subject,
65 COALESCE(mcde_creators.value, 'null') AS creators,
66 COALESCE(mcde_isbn.value, 'null') AS isbn
67 -- ensure one row per bre regardless of any display fields
68 FROM biblio.record_entry bre
69 LEFT JOIN metabib.compressed_display_entry mcde_title
70 ON (bre.id = mcde_title.source AND mcde_title.name = 'title')
71 LEFT JOIN metabib.compressed_display_entry mcde_author
72 ON (bre.id = mcde_author.source AND mcde_author.name = 'author')
73 LEFT JOIN metabib.compressed_display_entry mcde_subject
74 ON (bre.id = mcde_subject.source AND mcde_subject.name = 'subject')
75 LEFT JOIN metabib.compressed_display_entry mcde_creators
76 ON (bre.id = mcde_creators.source AND mcde_creators.name = 'creators')
77 LEFT JOIN metabib.compressed_display_entry mcde_isbn
78 ON (bre.id = mcde_isbn.source AND mcde_isbn.name = 'isbn')
82 CREATE OR REPLACE FUNCTION metabib.display_field_normalize_trigger ()
86 display_field_text TEXT;
88 display_field_text := NEW.value;
91 SELECT n.func AS func,
92 n.param_count AS param_count,
94 FROM config.index_normalizer n
95 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
96 WHERE m.field = NEW.field AND m.pos < 0
99 EXECUTE 'SELECT ' || normalizer.func || '(' ||
100 quote_literal( display_field_text ) ||
102 WHEN normalizer.param_count > 0
103 THEN ',' || REPLACE(REPLACE(BTRIM(
104 normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
107 ')' INTO display_field_text;
111 NEW.value = display_field_text;
117 CREATE TRIGGER display_field_normalize_tgr
118 BEFORE UPDATE OR INSERT ON metabib.display_entry
119 FOR EACH ROW EXECUTE PROCEDURE metabib.display_field_normalize_trigger();
121 CREATE OR REPLACE FUNCTION evergreen.display_field_force_nfc()
122 RETURNS TRIGGER AS $$
124 NEW.value := force_unicode_normal_form(NEW.value,'NFC');
129 CREATE TRIGGER display_field_force_nfc_tgr
130 BEFORE UPDATE OR INSERT ON metabib.display_entry
131 FOR EACH ROW EXECUTE PROCEDURE evergreen.display_field_force_nfc();
133 ALTER TYPE metabib.field_entry_template ADD ATTRIBUTE display_field BOOL;
135 DROP FUNCTION metabib.reingest_metabib_field_entries(BIGINT, BOOL, BOOL, BOOL);
136 DROP FUNCTION biblio.extract_metabib_field_entry(BIGINT);
137 DROP FUNCTION biblio.extract_metabib_field_entry(BIGINT, TEXT);
139 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry (
144 ) RETURNS SETOF metabib.field_entry_template AS $func$
146 bib biblio.record_entry%ROWTYPE;
147 idx config.metabib_field%ROWTYPE;
148 xfrm config.xml_transform%ROWTYPE;
150 transformed_xml TEXT;
152 xml_node_list TEXT[];
159 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
161 authority_link BIGINT;
162 output_row metabib.field_entry_template%ROWTYPE;
166 -- Start out with no field-use bools set
167 output_row.browse_field = FALSE;
168 output_row.facet_field = FALSE;
169 output_row.display_field = FALSE;
170 output_row.search_field = FALSE;
173 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
175 -- Loop over the indexing entries
176 FOR idx IN SELECT * FROM config.metabib_field WHERE id = ANY (only_fields) ORDER BY format LOOP
178 process_idx := FALSE;
179 IF idx.display_field AND 'display' = ANY (field_types) THEN process_idx = TRUE; END IF;
180 IF idx.browse_field AND 'browse' = ANY (field_types) THEN process_idx = TRUE; END IF;
181 IF idx.search_field AND 'search' = ANY (field_types) THEN process_idx = TRUE; END IF;
182 IF idx.facet_field AND 'facet' = ANY (field_types) THEN process_idx = TRUE; END IF;
183 CONTINUE WHEN process_idx = FALSE;
185 joiner := COALESCE(idx.joiner, default_joiner);
187 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
189 -- See if we can skip the XSLT ... it's expensive
190 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
191 -- Can't skip the transform
192 IF xfrm.xslt <> '---' THEN
193 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
195 transformed_xml := bib.marc;
198 prev_xfrm := xfrm.name;
201 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
204 FOR xml_node IN SELECT x FROM unnest(xml_node_list) AS x LOOP
205 CONTINUE WHEN xml_node !~ E'^\\s*<';
207 -- XXX much of this should be moved into oils_xpath_string...
208 curr_text := ARRAY_TO_STRING(evergreen.array_remove_item_by_value(evergreen.array_remove_item_by_value(
209 oils_xpath( '//text()', -- get the content of all the nodes within the main selected node
210 REGEXP_REPLACE( xml_node, E'\\s+', ' ', 'g' ) -- Translate adjacent whitespace to a single space
211 ), ' '), ''), -- throw away morally empty (bankrupt?) strings
215 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
217 IF raw_text IS NOT NULL THEN
218 raw_text := raw_text || joiner;
221 raw_text := COALESCE(raw_text,'') || curr_text;
223 -- autosuggest/metabib.browse_entry
224 IF idx.browse_field THEN
226 IF idx.browse_xpath IS NOT NULL AND idx.browse_xpath <> '' THEN
227 browse_text := oils_xpath_string( idx.browse_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
229 browse_text := curr_text;
232 IF idx.browse_sort_xpath IS NOT NULL AND
233 idx.browse_sort_xpath <> '' THEN
235 sort_value := oils_xpath_string(
236 idx.browse_sort_xpath, xml_node, joiner,
237 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
240 sort_value := browse_text;
243 output_row.field_class = idx.field_class;
244 output_row.field = idx.id;
245 output_row.source = rid;
246 output_row.value = BTRIM(REGEXP_REPLACE(browse_text, E'\\s+', ' ', 'g'));
247 output_row.sort_value :=
248 public.naco_normalize(sort_value);
250 output_row.authority := NULL;
252 IF idx.authority_xpath IS NOT NULL AND idx.authority_xpath <> '' THEN
253 authority_text := oils_xpath_string(
254 idx.authority_xpath, xml_node, joiner,
256 ARRAY[xfrm.prefix, xfrm.namespace_uri],
257 ARRAY['xlink','http://www.w3.org/1999/xlink']
261 IF authority_text ~ '^\d+$' THEN
262 authority_link := authority_text::BIGINT;
263 PERFORM * FROM authority.record_entry WHERE id = authority_link;
265 output_row.authority := authority_link;
271 output_row.browse_field = TRUE;
272 -- Returning browse rows with search_field = true for search+browse
273 -- configs allows us to retain granularity of being able to search
274 -- browse fields with "starts with" type operators (for example, for
275 -- titles of songs in music albums)
276 IF idx.search_field THEN
277 output_row.search_field = TRUE;
279 RETURN NEXT output_row;
280 output_row.browse_field = FALSE;
281 output_row.search_field = FALSE;
282 output_row.sort_value := NULL;
285 -- insert raw node text for faceting
286 IF idx.facet_field THEN
288 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
289 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
291 facet_text := curr_text;
294 output_row.field_class = idx.field_class;
295 output_row.field = -1 * idx.id;
296 output_row.source = rid;
297 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
299 output_row.facet_field = TRUE;
300 RETURN NEXT output_row;
301 output_row.facet_field = FALSE;
304 -- insert raw node text for display
305 IF idx.display_field THEN
307 IF idx.display_xpath IS NOT NULL AND idx.display_xpath <> '' THEN
308 display_text := oils_xpath_string( idx.display_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
310 display_text := curr_text;
313 output_row.field_class = idx.field_class;
314 output_row.field = -1 * idx.id;
315 output_row.source = rid;
316 output_row.value = BTRIM(REGEXP_REPLACE(display_text, E'\\s+', ' ', 'g'));
318 output_row.display_field = TRUE;
319 RETURN NEXT output_row;
320 output_row.display_field = FALSE;
325 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
327 -- insert combined node text for searching
328 IF idx.search_field THEN
329 output_row.field_class = idx.field_class;
330 output_row.field = idx.id;
331 output_row.source = rid;
332 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
334 output_row.search_field = TRUE;
335 RETURN NEXT output_row;
336 output_row.search_field = FALSE;
343 $func$ LANGUAGE PLPGSQL;
345 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries(
347 skip_facet BOOL DEFAULT FALSE,
348 skip_display BOOL DEFAULT FALSE,
349 skip_browse BOOL DEFAULT FALSE,
350 skip_search BOOL DEFAULT FALSE,
351 only_fields INT[] DEFAULT '{}'::INT[]
352 ) RETURNS VOID AS $func$
355 ind_data metabib.field_entry_template%ROWTYPE;
356 mbe_row metabib.browse_entry%ROWTYPE;
363 field_list INT[] := only_fields;
364 field_types TEXT[] := '{}'::TEXT[];
367 IF field_list = '{}'::INT[] THEN
368 SELECT ARRAY_AGG(id) INTO field_list FROM config.metabib_field;
371 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;
372 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;
373 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;
374 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;
376 IF NOT b_skip_facet THEN field_types := field_types || '{facet}'; END IF;
377 IF NOT b_skip_display THEN field_types := field_types || '{display}'; END IF;
378 IF NOT b_skip_browse THEN field_types := field_types || '{browse}'; END IF;
379 IF NOT b_skip_search THEN field_types := field_types || '{search}'; END IF;
381 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
383 IF NOT b_skip_search THEN
384 FOR fclass IN SELECT * FROM config.metabib_class LOOP
385 -- RAISE NOTICE 'Emptying out %', fclass.name;
386 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
389 IF NOT b_skip_facet THEN
390 DELETE FROM metabib.facet_entry WHERE source = bib_id;
392 IF NOT b_skip_display THEN
393 DELETE FROM metabib.display_entry WHERE source = bib_id;
395 IF NOT b_skip_browse THEN
396 DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
400 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id, ' ', field_types, field_list ) LOOP
402 -- don't store what has been normalized away
403 CONTINUE WHEN ind_data.value IS NULL;
405 IF ind_data.field < 0 THEN
406 ind_data.field = -1 * ind_data.field;
409 IF ind_data.facet_field AND NOT b_skip_facet THEN
410 INSERT INTO metabib.facet_entry (field, source, value)
411 VALUES (ind_data.field, ind_data.source, ind_data.value);
414 IF ind_data.display_field AND NOT b_skip_display THEN
415 INSERT INTO metabib.display_entry (field, source, value)
416 VALUES (ind_data.field, ind_data.source, ind_data.value);
420 IF ind_data.browse_field AND NOT b_skip_browse THEN
421 -- A caveat about this SELECT: this should take care of replacing
422 -- old mbe rows when data changes, but not if normalization (by
423 -- which I mean specifically the output of
424 -- evergreen.oils_tsearch2()) changes. It may or may not be
425 -- expensive to add a comparison of index_vector to index_vector
426 -- to the WHERE clause below.
428 CONTINUE WHEN ind_data.sort_value IS NULL;
430 value_prepped := metabib.browse_normalize(ind_data.value, ind_data.field);
431 SELECT INTO mbe_row * FROM metabib.browse_entry
432 WHERE value = value_prepped AND sort_value = ind_data.sort_value;
435 mbe_id := mbe_row.id;
437 INSERT INTO metabib.browse_entry
438 ( value, sort_value ) VALUES
439 ( value_prepped, ind_data.sort_value );
441 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
444 INSERT INTO metabib.browse_entry_def_map (entry, def, source, authority)
445 VALUES (mbe_id, ind_data.field, ind_data.source, ind_data.authority);
448 IF ind_data.search_field AND NOT b_skip_search THEN
449 -- Avoid inserting duplicate rows
450 EXECUTE 'SELECT 1 FROM metabib.' || ind_data.field_class ||
451 '_field_entry WHERE field = $1 AND source = $2 AND value = $3'
452 INTO mbe_id USING ind_data.field, ind_data.source, ind_data.value;
453 -- RAISE NOTICE 'Search for an already matching row returned %', mbe_id;
454 IF mbe_id IS NULL THEN
456 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
458 quote_literal(ind_data.field) || $$, $$ ||
459 quote_literal(ind_data.source) || $$, $$ ||
460 quote_literal(ind_data.value) ||
467 IF NOT b_skip_search THEN
468 PERFORM metabib.update_combined_index_vectors(bib_id);
473 $func$ LANGUAGE PLPGSQL;
475 -- AFTER UPDATE OR INSERT trigger for biblio.record_entry
476 CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$
481 IF NEW.deleted THEN -- If this bib is deleted
483 PERFORM * FROM config.internal_flag WHERE
484 name = 'ingest.metarecord_mapping.preserve_on_delete' AND enabled;
486 tmp_bool := FOUND; -- Just in case this is changed by some other statement
488 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint, TRUE, tmp_bool );
491 -- One needs to keep these around to support searches
492 -- with the #deleted modifier, so one should turn on the named
493 -- internal flag for that functionality.
494 DELETE FROM metabib.record_attr_vector_list WHERE source = NEW.id;
497 DELETE FROM authority.bib_linking WHERE bib = NEW.id; -- Avoid updating fields in bibs that are no longer visible
498 DELETE FROM biblio.peer_bib_copy_map WHERE peer_record = NEW.id; -- Separate any multi-homed items
499 DELETE FROM metabib.browse_entry_def_map WHERE source = NEW.id; -- Don't auto-suggest deleted bibs
500 RETURN NEW; -- and we're done
503 IF TG_OP = 'UPDATE' THEN -- re-ingest?
504 PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled;
506 IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change
511 -- Record authority linking
512 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_linking' AND enabled;
514 PERFORM biblio.map_authority_linking( NEW.id, NEW.marc );
517 -- Flatten and insert the mfr data
518 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_full_rec' AND enabled;
520 PERFORM metabib.reingest_metabib_full_rec(NEW.id);
522 -- Now we pull out attribute data, which is dependent on the mfr for all but XPath-based fields
523 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_rec_descriptor' AND enabled;
525 PERFORM metabib.reingest_record_attributes(NEW.id, NULL, NEW.marc, TG_OP = 'INSERT' OR OLD.deleted);
529 -- Gather and insert the field entry data
530 PERFORM metabib.reingest_metabib_field_entries(NEW.id);
533 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled;
534 IF NOT FOUND THEN PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor ); END IF;
536 -- (re)map metarecord-bib linking
537 IF TG_OP = 'INSERT' THEN -- if not deleted and performing an insert, check for the flag
538 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled;
540 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
542 ELSE -- we're doing an update, and we're not deleted, remap
543 PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_update' AND enabled;
545 PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint );
551 $func$ LANGUAGE PLPGSQL;