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 metabib.display_entry (
9 id BIGSERIAL PRIMARY KEY,
10 source BIGINT NOT NULL REFERENCES biblio.record_entry (id),
11 field INT NOT NULL REFERENCES config.metabib_field (id),
15 CREATE TABLE config.display_field_map (
16 name TEXT PRIMARY KEY,
17 field INTEGER REFERENCES config.metabib_field (id),
18 multi BOOLEAN DEFAULT FALSE
21 CREATE VIEW metabib.flat_display_entry AS
25 CASE WHEN cdfm.multi THEN
26 TO_JSON(ARRAY_AGG(value))
30 FROM metabib.display_entry mde
31 JOIN config.display_field_map cdfm ON (cdfm.field = mde.field)
34 CREATE INDEX metabib_display_entry_field_idx
35 ON metabib.display_entry (field);
36 CREATE INDEX metabib_display_entry_source_idx
37 ON metabib.display_entry (source);
39 CREATE OR REPLACE FUNCTION metabib.display_field_normalize_trigger ()
43 display_field_text TEXT;
45 display_field_text := NEW.value;
48 SELECT n.func AS func,
49 n.param_count AS param_count,
51 FROM config.index_normalizer n
52 JOIN config.metabib_field_index_norm_map m ON (m.norm = n.id)
53 WHERE m.field = NEW.field AND m.pos < 0
56 EXECUTE 'SELECT ' || normalizer.func || '(' ||
57 quote_literal( display_field_text ) ||
59 WHEN normalizer.param_count > 0
60 THEN ',' || REPLACE(REPLACE(BTRIM(
61 normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
64 ')' INTO display_field_text;
68 NEW.value = display_field_text;
74 CREATE TRIGGER display_field_normalize_tgr
75 BEFORE UPDATE OR INSERT ON metabib.display_entry
76 FOR EACH ROW EXECUTE PROCEDURE metabib.display_field_normalize_trigger();
78 CREATE OR REPLACE FUNCTION evergreen.display_field_force_nfc()
81 NEW.value := force_unicode_normal_form(NEW.value,'NFC');
86 CREATE TRIGGER display_field_force_nfc_tgr
87 BEFORE UPDATE OR INSERT ON metabib.display_entry
88 FOR EACH ROW EXECUTE PROCEDURE evergreen.display_field_force_nfc();
90 ALTER TYPE metabib.field_entry_template ADD ATTRIBUTE display_field BOOL;
92 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
94 bib biblio.record_entry%ROWTYPE;
95 idx config.metabib_field%ROWTYPE;
96 xfrm config.xml_transform%ROWTYPE;
100 xml_node_list TEXT[];
107 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
109 authority_link BIGINT;
110 output_row metabib.field_entry_template%ROWTYPE;
113 -- Start out with no field-use bools set
114 output_row.browse_field = FALSE;
115 output_row.facet_field = FALSE;
116 output_row.display_field = FALSE;
117 output_row.search_field = FALSE;
120 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
122 -- Loop over the indexing entries
123 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
125 joiner := COALESCE(idx.joiner, default_joiner);
127 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
129 -- See if we can skip the XSLT ... it's expensive
130 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
131 -- Can't skip the transform
132 IF xfrm.xslt <> '---' THEN
133 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
135 transformed_xml := bib.marc;
138 prev_xfrm := xfrm.name;
141 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
144 FOR xml_node IN SELECT x FROM unnest(xml_node_list) AS x LOOP
145 CONTINUE WHEN xml_node !~ E'^\\s*<';
147 -- XXX much of this should be moved into oils_xpath_string...
148 curr_text := ARRAY_TO_STRING(evergreen.array_remove_item_by_value(evergreen.array_remove_item_by_value(
149 oils_xpath( '//text()', -- get the content of all the nodes within the main selected node
150 REGEXP_REPLACE( xml_node, E'\\s+', ' ', 'g' ) -- Translate adjacent whitespace to a single space
151 ), ' '), ''), -- throw away morally empty (bankrupt?) strings
155 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
157 IF raw_text IS NOT NULL THEN
158 raw_text := raw_text || joiner;
161 raw_text := COALESCE(raw_text,'') || curr_text;
163 -- autosuggest/metabib.browse_entry
164 IF idx.browse_field THEN
166 IF idx.browse_xpath IS NOT NULL AND idx.browse_xpath <> '' THEN
167 browse_text := oils_xpath_string( idx.browse_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
169 browse_text := curr_text;
172 IF idx.browse_sort_xpath IS NOT NULL AND
173 idx.browse_sort_xpath <> '' THEN
175 sort_value := oils_xpath_string(
176 idx.browse_sort_xpath, xml_node, joiner,
177 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
180 sort_value := browse_text;
183 output_row.field_class = idx.field_class;
184 output_row.field = idx.id;
185 output_row.source = rid;
186 output_row.value = BTRIM(REGEXP_REPLACE(browse_text, E'\\s+', ' ', 'g'));
187 output_row.sort_value :=
188 public.naco_normalize(sort_value);
190 output_row.authority := NULL;
192 IF idx.authority_xpath IS NOT NULL AND idx.authority_xpath <> '' THEN
193 authority_text := oils_xpath_string(
194 idx.authority_xpath, xml_node, joiner,
196 ARRAY[xfrm.prefix, xfrm.namespace_uri],
197 ARRAY['xlink','http://www.w3.org/1999/xlink']
201 IF authority_text ~ '^\d+$' THEN
202 authority_link := authority_text::BIGINT;
203 PERFORM * FROM authority.record_entry WHERE id = authority_link;
205 output_row.authority := authority_link;
211 output_row.browse_field = TRUE;
212 -- Returning browse rows with search_field = true for search+browse
213 -- configs allows us to retain granularity of being able to search
214 -- browse fields with "starts with" type operators (for example, for
215 -- titles of songs in music albums)
216 IF idx.search_field THEN
217 output_row.search_field = TRUE;
219 RETURN NEXT output_row;
220 output_row.browse_field = FALSE;
221 output_row.search_field = FALSE;
222 output_row.sort_value := NULL;
225 -- insert raw node text for faceting
226 IF idx.facet_field THEN
228 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
229 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
231 facet_text := curr_text;
234 output_row.field_class = idx.field_class;
235 output_row.field = -1 * idx.id;
236 output_row.source = rid;
237 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
239 output_row.facet_field = TRUE;
240 RETURN NEXT output_row;
241 output_row.facet_field = FALSE;
244 -- insert raw node text for display
245 IF idx.display_field THEN
247 IF idx.display_xpath IS NOT NULL AND idx.display_xpath <> '' THEN
248 display_text := oils_xpath_string( idx.display_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
250 display_text := curr_text;
253 output_row.field_class = idx.field_class;
254 output_row.field = -1 * idx.id;
255 output_row.source = rid;
256 output_row.value = BTRIM(REGEXP_REPLACE(display_text, E'\\s+', ' ', 'g'));
258 output_row.display_field = TRUE;
259 RETURN NEXT output_row;
260 output_row.display_field = FALSE;
265 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
267 -- insert combined node text for searching
268 IF idx.search_field THEN
269 output_row.field_class = idx.field_class;
270 output_row.field = idx.id;
271 output_row.source = rid;
272 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
274 output_row.search_field = TRUE;
275 RETURN NEXT output_row;
276 output_row.search_field = FALSE;
283 $func$ LANGUAGE PLPGSQL;
285 DROP FUNCTION metabib.reingest_metabib_field_entries(BIGINT, BOOL, BOOL, BOOL);
287 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries(
288 bib_id BIGINT, skip_facet BOOL DEFAULT FALSE,
289 skip_display BOOL DEFAULT FALSE, skip_browse BOOL DEFAULT FALSE,
290 skip_search BOOL DEFAULT FALSE ) RETURNS VOID AS $func$
293 ind_data metabib.field_entry_template%ROWTYPE;
294 mbe_row metabib.browse_entry%ROWTYPE;
303 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;
304 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;
305 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;
306 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;
308 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
310 IF NOT b_skip_search THEN
311 FOR fclass IN SELECT * FROM config.metabib_class LOOP
312 -- RAISE NOTICE 'Emptying out %', fclass.name;
313 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
316 IF NOT b_skip_facet THEN
317 DELETE FROM metabib.facet_entry WHERE source = bib_id;
319 IF NOT b_skip_display THEN
320 DELETE FROM metabib.display_entry WHERE source = bib_id;
322 IF NOT b_skip_browse THEN
323 DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
327 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
329 -- don't store what has been normalized away
330 CONTINUE WHEN ind_data.value IS NULL;
332 IF ind_data.field < 0 THEN
333 ind_data.field = -1 * ind_data.field;
336 IF ind_data.facet_field AND NOT b_skip_facet THEN
337 INSERT INTO metabib.facet_entry (field, source, value)
338 VALUES (ind_data.field, ind_data.source, ind_data.value);
341 IF ind_data.display_field AND NOT b_skip_display THEN
342 INSERT INTO metabib.display_entry (field, source, value)
343 VALUES (ind_data.field, ind_data.source, ind_data.value);
347 IF ind_data.browse_field AND NOT b_skip_browse THEN
348 -- A caveat about this SELECT: this should take care of replacing
349 -- old mbe rows when data changes, but not if normalization (by
350 -- which I mean specifically the output of
351 -- evergreen.oils_tsearch2()) changes. It may or may not be
352 -- expensive to add a comparison of index_vector to index_vector
353 -- to the WHERE clause below.
355 CONTINUE WHEN ind_data.sort_value IS NULL;
357 value_prepped := metabib.browse_normalize(ind_data.value, ind_data.field);
358 SELECT INTO mbe_row * FROM metabib.browse_entry
359 WHERE value = value_prepped AND sort_value = ind_data.sort_value;
362 mbe_id := mbe_row.id;
364 INSERT INTO metabib.browse_entry
365 ( value, sort_value ) VALUES
366 ( value_prepped, ind_data.sort_value );
368 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
371 INSERT INTO metabib.browse_entry_def_map (entry, def, source, authority)
372 VALUES (mbe_id, ind_data.field, ind_data.source, ind_data.authority);
375 IF ind_data.search_field AND NOT b_skip_search THEN
376 -- Avoid inserting duplicate rows
377 EXECUTE 'SELECT 1 FROM metabib.' || ind_data.field_class ||
378 '_field_entry WHERE field = $1 AND source = $2 AND value = $3'
379 INTO mbe_id USING ind_data.field, ind_data.source, ind_data.value;
380 -- RAISE NOTICE 'Search for an already matching row returned %', mbe_id;
381 IF mbe_id IS NULL THEN
383 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
385 quote_literal(ind_data.field) || $$, $$ ||
386 quote_literal(ind_data.source) || $$, $$ ||
387 quote_literal(ind_data.value) ||
394 IF NOT b_skip_search THEN
395 PERFORM metabib.update_combined_index_vectors(bib_id);
400 $func$ LANGUAGE PLPGSQL;