4 ALTER TABLE config.metabib_field ADD COLUMN display_xpath TEXT, display_field BOOL NOT NULL DEFAULT TRUE;
5 UPDATE config.metabib_field SET display_field = FALSE WHERE field_class = 'keyword' OR name = 'complete';
7 CREATE TABLE metabib.display_entry (
8 id BIGSERIAL PRIMARY KEY,
9 source BIGINT NOT NULL,
14 CREATE INDEX metabib_display_entry_field_idx
15 ON metabib.display_entry (field);
16 CREATE INDEX metabib_display_entry_source_idx
17 ON metabib.display_entry (source);
19 CREATE OR REPLACE FUNCTION metabib.display_field_normalize_trigger ()
23 display_field_text TEXT;
25 display_field_text := NEW.value;
28 SELECT n.func AS func,
29 n.param_count AS param_count,
31 FROM config.index_normalizer n
32 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
33 WHERE m.field = NEW.field AND m.pos < 0
36 EXECUTE 'SELECT ' || normalizer.func || '(' ||
37 quote_literal( display_field_text ) ||
39 WHEN normalizer.param_count > 0
40 THEN ',' || REPLACE(REPLACE(BTRIM(
41 normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
44 ')' INTO display_field_text;
48 NEW.value = display_field_text;
54 CREATE TRIGGER display_field_normalize_tgr
55 BEFORE UPDATE OR INSERT ON metabib.display_entry
56 FOR EACH ROW EXECUTE PROCEDURE metabib.display_field_normalize_trigger();
58 CREATE OR REPLACE FUNCTION evergreen.display_field_force_nfc()
61 NEW.value := force_unicode_normal_form(NEW.value,'NFC');
66 CREATE TRIGGER display_field_force_nfc_tgr
67 BEFORE UPDATE OR INSERT ON metabib.display_entry
68 FOR EACH ROW EXECUTE PROCEDURE evergreen.display_field_force_nfc();
70 ALTER TABLE config.metabib_field
71 ADD COLUMN display_field BOOL NOT NULL DEFAULT TRUE;
73 ALTER TYPE metabib.field_entry_template ADD ATTRIBUTE display_field BOOL;
75 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
77 bib biblio.record_entry%ROWTYPE;
78 idx config.metabib_field%ROWTYPE;
79 xfrm config.xml_transform%ROWTYPE;
90 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
92 authority_link BIGINT;
93 output_row metabib.field_entry_template%ROWTYPE;
96 -- Start out with no field-use bools set
97 output_row.browse_field = FALSE;
98 output_row.facet_field = FALSE;
99 output_row.display_field = FALSE;
100 output_row.search_field = FALSE;
103 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
105 -- Loop over the indexing entries
106 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
108 joiner := COALESCE(idx.joiner, default_joiner);
110 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
112 -- See if we can skip the XSLT ... it's expensive
113 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
114 -- Can't skip the transform
115 IF xfrm.xslt <> '---' THEN
116 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
118 transformed_xml := bib.marc;
121 prev_xfrm := xfrm.name;
124 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
127 FOR xml_node IN SELECT x FROM unnest(xml_node_list) AS x LOOP
128 CONTINUE WHEN xml_node !~ E'^\\s*<';
130 -- XXX much of this should be moved into oils_xpath_string...
131 curr_text := ARRAY_TO_STRING(evergreen.array_remove_item_by_value(evergreen.array_remove_item_by_value(
132 oils_xpath( '//text()',
134 REGEXP_REPLACE( -- This escapes all &s not followed by "amp;". Data ise returned from oils_xpath (above) in UTF-8, not entity encoded
135 REGEXP_REPLACE( -- This escapes embeded <s
137 $re$(>[^<]+)(<)([^>]+<)$re$,
153 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
155 IF raw_text IS NOT NULL THEN
156 raw_text := raw_text || joiner;
159 raw_text := COALESCE(raw_text,'') || curr_text;
161 -- autosuggest/metabib.browse_entry
162 IF idx.browse_field THEN
164 IF idx.browse_xpath IS NOT NULL AND idx.browse_xpath <> '' THEN
165 browse_text := oils_xpath_string( idx.browse_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
167 browse_text := curr_text;
170 IF idx.browse_sort_xpath IS NOT NULL AND
171 idx.browse_sort_xpath <> '' THEN
173 sort_value := oils_xpath_string(
174 idx.browse_sort_xpath, xml_node, joiner,
175 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
178 sort_value := browse_text;
181 output_row.field_class = idx.field_class;
182 output_row.field = idx.id;
183 output_row.source = rid;
184 output_row.value = BTRIM(REGEXP_REPLACE(browse_text, E'\\s+', ' ', 'g'));
185 output_row.sort_value :=
186 public.naco_normalize(sort_value);
188 output_row.authority := NULL;
190 IF idx.authority_xpath IS NOT NULL AND idx.authority_xpath <> '' THEN
191 authority_text := oils_xpath_string(
192 idx.authority_xpath, xml_node, joiner,
194 ARRAY[xfrm.prefix, xfrm.namespace_uri],
195 ARRAY['xlink','http://www.w3.org/1999/xlink']
199 IF authority_text ~ '^\d+$' THEN
200 authority_link := authority_text::BIGINT;
201 PERFORM * FROM authority.record_entry WHERE id = authority_link;
203 output_row.authority := authority_link;
209 output_row.browse_field = TRUE;
210 -- Returning browse rows with search_field = true for search+browse
211 -- configs allows us to retain granularity of being able to search
212 -- browse fields with "starts with" type operators (for example, for
213 -- titles of songs in music albums)
214 IF idx.search_field THEN
215 output_row.search_field = TRUE;
217 RETURN NEXT output_row;
218 output_row.browse_field = FALSE;
219 output_row.search_field = FALSE;
220 output_row.sort_value := NULL;
223 -- insert raw node text for faceting
224 IF idx.facet_field THEN
226 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
227 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
229 facet_text := curr_text;
232 output_row.field_class = idx.field_class;
233 output_row.field = -1 * idx.id;
234 output_row.source = rid;
235 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
237 output_row.facet_field = TRUE;
238 RETURN NEXT output_row;
239 output_row.facet_field = FALSE;
242 -- insert raw node text for display
243 IF idx.display_field THEN
245 IF idx.display_xpath IS NOT NULL AND idx.display_xpath <> '' THEN
246 display_text := oils_xpath_string( idx.display_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
248 display_text := curr_text;
251 output_row.field_class = idx.field_class;
252 output_row.field = -1 * idx.id;
253 output_row.source = rid;
254 output_row.value = BTRIM(REGEXP_REPLACE(display_text, E'\\s+', ' ', 'g'));
256 output_row.display_field = TRUE;
257 RETURN NEXT output_row;
258 output_row.display_field = FALSE;
263 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
265 -- insert combined node text for searching
266 IF idx.search_field THEN
267 output_row.field_class = idx.field_class;
268 output_row.field = idx.id;
269 output_row.source = rid;
270 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
272 output_row.search_field = TRUE;
273 RETURN NEXT output_row;
274 output_row.search_field = FALSE;
281 $func$ LANGUAGE PLPGSQL;
283 DROP FUNCTION metabib.reingest_metabib_field_entries(BIGINT, BOOL, BOOL, BOOL);
285 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries(
286 bib_id BIGINT, skip_facet BOOL DEFAULT FALSE,
287 skip_display BOOL DEFAULT FALSE, skip_browse BOOL DEFAULT FALSE,
288 skip_search BOOL DEFAULT FALSE ) RETURNS VOID AS $func$
291 ind_data metabib.field_entry_template%ROWTYPE;
292 mbe_row metabib.browse_entry%ROWTYPE;
301 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;
302 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;
303 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;
304 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;
306 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
308 IF NOT b_skip_search THEN
309 FOR fclass IN SELECT * FROM config.metabib_class LOOP
310 -- RAISE NOTICE 'Emptying out %', fclass.name;
311 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
314 IF NOT b_skip_facet THEN
315 DELETE FROM metabib.facet_entry WHERE source = bib_id;
317 IF NOT b_skip_display THEN
318 DELETE FROM metabib.display_entry WHERE source = bib_id;
320 IF NOT b_skip_browse THEN
321 DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
325 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
326 IF ind_data.field < 0 THEN
327 ind_data.field = -1 * ind_data.field;
330 IF ind_data.facet_field AND NOT b_skip_facet THEN
331 INSERT INTO metabib.facet_entry (field, source, value)
332 VALUES (ind_data.field, ind_data.source, ind_data.value);
335 IF ind_data.display_field AND NOT b_skip_display THEN
336 INSERT INTO metabib.display_entry (field, source, value)
337 VALUES (ind_data.field, ind_data.source, ind_data.value);
341 IF ind_data.browse_field AND NOT b_skip_browse THEN
342 -- A caveat about this SELECT: this should take care of replacing
343 -- old mbe rows when data changes, but not if normalization (by
344 -- which I mean specifically the output of
345 -- evergreen.oils_tsearch2()) changes. It may or may not be
346 -- expensive to add a comparison of index_vector to index_vector
347 -- to the WHERE clause below.
349 value_prepped := metabib.browse_normalize(ind_data.value, ind_data.field);
350 SELECT INTO mbe_row * FROM metabib.browse_entry
351 WHERE value = value_prepped AND sort_value = ind_data.sort_value;
354 mbe_id := mbe_row.id;
356 INSERT INTO metabib.browse_entry
357 ( value, sort_value ) VALUES
358 ( value_prepped, ind_data.sort_value );
360 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
363 INSERT INTO metabib.browse_entry_def_map (entry, def, source, authority)
364 VALUES (mbe_id, ind_data.field, ind_data.source, ind_data.authority);
367 IF ind_data.search_field AND NOT b_skip_search THEN
368 -- Avoid inserting duplicate rows
369 EXECUTE 'SELECT 1 FROM metabib.' || ind_data.field_class ||
370 '_field_entry WHERE field = $1 AND source = $2 AND value = $3'
371 INTO mbe_id USING ind_data.field, ind_data.source, ind_data.value;
372 -- RAISE NOTICE 'Search for an already matching row returned %', mbe_id;
373 IF mbe_id IS NULL THEN
375 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
377 quote_literal(ind_data.field) || $$, $$ ||
378 quote_literal(ind_data.source) || $$, $$ ||
379 quote_literal(ind_data.value) ||
386 IF NOT b_skip_search THEN
387 PERFORM metabib.update_combined_index_vectors(bib_id);
392 $func$ LANGUAGE PLPGSQL;
395 -- DATA -------------------
397 -- "General Keywords" and "All Subjects"
399 UPDATE config.metabib_field SET display_field = FALSE WHERE id IN (15, 16);
401 INSERT INTO config.internal_flag (name, enabled)
402 VALUES ('ingest.skip_display_indexing', FALSE);
404 -- TODO: targeted ingest?
407 --UPDATE config.internal_flag SET enabled = TRUE
408 -- WHERE name = 'ingest.reingest.force_on_same_marc';
409 --UPDATE biblio.record_entry SET marc = marc;
410 --UPDATE config.internal_flag SET enabled = FALSE
411 -- WHERE name = 'ingest.reingest.force_on_same_marc';