1 -- XXX REVERT FILE -- DELETE THIS BEFORE MERGING XXX --
3 DELETE FROM metabib.display_entry;
4 DELETE FROM config.display_field_map;
5 DELETE FROM config.metabib_field WHERE display_field; -- ASSUMES ALL NEW FIELDS
7 DELETE FROM config.internal_flag WHERE name = 'ingest.skip_display_indexing';
11 DROP FUNCTION metabib.reingest_metabib_field_entries(BIGINT, BOOL, BOOL, BOOL, BOOL);
13 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 $func$
16 ind_data metabib.field_entry_template%ROWTYPE;
17 mbe_row metabib.browse_entry%ROWTYPE;
25 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;
26 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;
27 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;
29 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
31 IF NOT b_skip_search THEN
32 FOR fclass IN SELECT * FROM config.metabib_class LOOP
33 -- RAISE NOTICE 'Emptying out %', fclass.name;
34 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
37 IF NOT b_skip_facet THEN
38 DELETE FROM metabib.facet_entry WHERE source = bib_id;
40 IF NOT b_skip_browse THEN
41 DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
45 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
47 -- don't store what has been normalized away
48 CONTINUE WHEN ind_data.value IS NULL;
50 IF ind_data.field < 0 THEN
51 ind_data.field = -1 * ind_data.field;
54 IF ind_data.facet_field AND NOT b_skip_facet THEN
55 INSERT INTO metabib.facet_entry (field, source, value)
56 VALUES (ind_data.field, ind_data.source, ind_data.value);
59 IF ind_data.browse_field AND NOT b_skip_browse THEN
60 -- A caveat about this SELECT: this should take care of replacing
61 -- old mbe rows when data changes, but not if normalization (by
62 -- which I mean specifically the output of
63 -- evergreen.oils_tsearch2()) changes. It may or may not be
64 -- expensive to add a comparison of index_vector to index_vector
65 -- to the WHERE clause below.
67 CONTINUE WHEN ind_data.sort_value IS NULL;
69 value_prepped := metabib.browse_normalize(ind_data.value, ind_data.field);
70 SELECT INTO mbe_row * FROM metabib.browse_entry
71 WHERE value = value_prepped AND sort_value = ind_data.sort_value;
76 INSERT INTO metabib.browse_entry
77 ( value, sort_value ) VALUES
78 ( value_prepped, ind_data.sort_value );
80 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
83 INSERT INTO metabib.browse_entry_def_map (entry, def, source, authority)
84 VALUES (mbe_id, ind_data.field, ind_data.source, ind_data.authority);
87 IF ind_data.search_field AND NOT b_skip_search THEN
88 -- Avoid inserting duplicate rows
89 EXECUTE 'SELECT 1 FROM metabib.' || ind_data.field_class ||
90 '_field_entry WHERE field = $1 AND source = $2 AND value = $3'
91 INTO mbe_id USING ind_data.field, ind_data.source, ind_data.value;
92 -- RAISE NOTICE 'Search for an already matching row returned %', mbe_id;
93 IF mbe_id IS NULL THEN
95 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
97 quote_literal(ind_data.field) || $$, $$ ||
98 quote_literal(ind_data.source) || $$, $$ ||
99 quote_literal(ind_data.value) ||
106 IF NOT b_skip_search THEN
107 PERFORM metabib.update_combined_index_vectors(bib_id);
112 $func$ LANGUAGE PLPGSQL;
115 CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
117 bib biblio.record_entry%ROWTYPE;
118 idx config.metabib_field%ROWTYPE;
119 xfrm config.xml_transform%ROWTYPE;
121 transformed_xml TEXT;
123 xml_node_list TEXT[];
129 joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
131 authority_link BIGINT;
132 output_row metabib.field_entry_template%ROWTYPE;
135 -- Start out with no field-use bools set
136 output_row.browse_field = FALSE;
137 output_row.facet_field = FALSE;
138 output_row.search_field = FALSE;
141 SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
143 -- Loop over the indexing entries
144 FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
146 joiner := COALESCE(idx.joiner, default_joiner);
148 SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
150 -- See if we can skip the XSLT ... it's expensive
151 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
152 -- Can't skip the transform
153 IF xfrm.xslt <> '---' THEN
154 transformed_xml := oils_xslt_process(bib.marc,xfrm.xslt);
156 transformed_xml := bib.marc;
159 prev_xfrm := xfrm.name;
162 xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
165 FOR xml_node IN SELECT x FROM unnest(xml_node_list) AS x LOOP
166 CONTINUE WHEN xml_node !~ E'^\\s*<';
168 -- XXX much of this should be moved into oils_xpath_string...
169 curr_text := ARRAY_TO_STRING(evergreen.array_remove_item_by_value(evergreen.array_remove_item_by_value(
170 oils_xpath( '//text()', -- get the content of all the nodes within the main selected node
171 REGEXP_REPLACE( xml_node, E'\\s+', ' ', 'g' ) -- Translate adjacent whitespace to a single space
172 ), ' '), ''), -- throw away morally empty (bankrupt?) strings
176 CONTINUE WHEN curr_text IS NULL OR curr_text = '';
178 IF raw_text IS NOT NULL THEN
179 raw_text := raw_text || joiner;
182 raw_text := COALESCE(raw_text,'') || curr_text;
184 -- autosuggest/metabib.browse_entry
185 IF idx.browse_field THEN
187 IF idx.browse_xpath IS NOT NULL AND idx.browse_xpath <> '' THEN
188 browse_text := oils_xpath_string( idx.browse_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
190 browse_text := curr_text;
193 IF idx.browse_sort_xpath IS NOT NULL AND
194 idx.browse_sort_xpath <> '' THEN
196 sort_value := oils_xpath_string(
197 idx.browse_sort_xpath, xml_node, joiner,
198 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
201 sort_value := browse_text;
204 output_row.field_class = idx.field_class;
205 output_row.field = idx.id;
206 output_row.source = rid;
207 output_row.value = BTRIM(REGEXP_REPLACE(browse_text, E'\\s+', ' ', 'g'));
208 output_row.sort_value :=
209 public.naco_normalize(sort_value);
211 output_row.authority := NULL;
213 IF idx.authority_xpath IS NOT NULL AND idx.authority_xpath <> '' THEN
214 authority_text := oils_xpath_string(
215 idx.authority_xpath, xml_node, joiner,
217 ARRAY[xfrm.prefix, xfrm.namespace_uri],
218 ARRAY['xlink','http://www.w3.org/1999/xlink']
222 IF authority_text ~ '^\d+$' THEN
223 authority_link := authority_text::BIGINT;
224 PERFORM * FROM authority.record_entry WHERE id = authority_link;
226 output_row.authority := authority_link;
232 output_row.browse_field = TRUE;
233 -- Returning browse rows with search_field = true for search+browse
234 -- configs allows us to retain granularity of being able to search
235 -- browse fields with "starts with" type operators (for example, for
236 -- titles of songs in music albums)
237 IF idx.search_field THEN
238 output_row.search_field = TRUE;
240 RETURN NEXT output_row;
241 output_row.browse_field = FALSE;
242 output_row.search_field = FALSE;
243 output_row.sort_value := NULL;
246 -- insert raw node text for faceting
247 IF idx.facet_field THEN
249 IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
250 facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
252 facet_text := curr_text;
255 output_row.field_class = idx.field_class;
256 output_row.field = -1 * idx.id;
257 output_row.source = rid;
258 output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
260 output_row.facet_field = TRUE;
261 RETURN NEXT output_row;
262 output_row.facet_field = FALSE;
267 CONTINUE WHEN raw_text IS NULL OR raw_text = '';
269 -- insert combined node text for searching
270 IF idx.search_field THEN
271 output_row.field_class = idx.field_class;
272 output_row.field = idx.id;
273 output_row.source = rid;
274 output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
276 output_row.search_field = TRUE;
277 RETURN NEXT output_row;
278 output_row.search_field = FALSE;
285 $func$ LANGUAGE PLPGSQL;
288 ALTER TYPE metabib.field_entry_template DROP ATTRIBUTE display_field;
289 DROP TRIGGER display_field_force_nfc_tgr ON metabib.display_entry;
290 DROP FUNCTION evergreen.display_field_force_nfc();
291 DROP TRIGGER display_field_normalize_tgr ON metabib.display_entry;
292 DROP FUNCTION metabib.display_field_normalize_trigger();
293 DROP INDEX metabib.metabib_display_entry_source_idx;
294 DROP INDEX metabib.metabib_display_entry_field_idx;
295 DROP VIEW metabib.flat_display_entry;
296 DROP TABLE config.display_field_map;
297 DROP TABLE metabib.display_entry;
299 ALTER TABLE config.metabib_field
300 DROP COLUMN display_xpath,
301 DROP COLUMN display_field;