3 SELECT evergreen.upgrade_deps_block_check('1340', :eg_version);
5 CREATE OR REPLACE FUNCTION search.disable_symspell_reification () RETURNS VOID AS $f$
6 INSERT INTO config.internal_flag (name,enabled)
7 VALUES ('ingest.disable_symspell_reification',TRUE)
8 ON CONFLICT (name) DO UPDATE SET enabled = TRUE;
11 CREATE OR REPLACE FUNCTION search.enable_symspell_reification () RETURNS VOID AS $f$
12 UPDATE config.internal_flag SET enabled = FALSE WHERE name = 'ingest.disable_symspell_reification';
15 CREATE OR REPLACE FUNCTION search.symspell_dictionary_full_reify () RETURNS SETOF search.symspell_dictionary AS $f$
17 DELETE FROM search.symspell_dictionary_updates RETURNING *
18 ), computed_rows AS ( -- this collapses the rows deleted into the format we need for UPSERT
19 SELECT SUM(keyword_count) AS keyword_count,
20 SUM(title_count) AS title_count,
21 SUM(author_count) AS author_count,
22 SUM(subject_count) AS subject_count,
23 SUM(series_count) AS series_count,
24 SUM(identifier_count) AS identifier_count,
28 ARRAY_REMOVE(ARRAY_AGG(DISTINCT keyword_suggestions[1]), NULL) AS keyword_suggestions,
29 ARRAY_REMOVE(ARRAY_AGG(DISTINCT title_suggestions[1]), NULL) AS title_suggestions,
30 ARRAY_REMOVE(ARRAY_AGG(DISTINCT author_suggestions[1]), NULL) AS author_suggestions,
31 ARRAY_REMOVE(ARRAY_AGG(DISTINCT subject_suggestions[1]), NULL) AS subject_suggestions,
32 ARRAY_REMOVE(ARRAY_AGG(DISTINCT series_suggestions[1]), NULL) AS series_suggestions,
33 ARRAY_REMOVE(ARRAY_AGG(DISTINCT identifier_suggestions[1]), NULL) AS identifier_suggestions
37 INSERT INTO search.symspell_dictionary AS d SELECT * FROM computed_rows
38 ON CONFLICT (prefix_key) DO UPDATE SET
39 keyword_count = GREATEST(0, d.keyword_count + EXCLUDED.keyword_count),
40 keyword_suggestions = evergreen.text_array_merge_unique(EXCLUDED.keyword_suggestions,d.keyword_suggestions),
42 title_count = GREATEST(0, d.title_count + EXCLUDED.title_count),
43 title_suggestions = evergreen.text_array_merge_unique(EXCLUDED.title_suggestions,d.title_suggestions),
45 author_count = GREATEST(0, d.author_count + EXCLUDED.author_count),
46 author_suggestions = evergreen.text_array_merge_unique(EXCLUDED.author_suggestions,d.author_suggestions),
48 subject_count = GREATEST(0, d.subject_count + EXCLUDED.subject_count),
49 subject_suggestions = evergreen.text_array_merge_unique(EXCLUDED.subject_suggestions,d.subject_suggestions),
51 series_count = GREATEST(0, d.series_count + EXCLUDED.series_count),
52 series_suggestions = evergreen.text_array_merge_unique(EXCLUDED.series_suggestions,d.series_suggestions),
54 identifier_count = GREATEST(0, d.identifier_count + EXCLUDED.identifier_count),
55 identifier_suggestions = evergreen.text_array_merge_unique(EXCLUDED.identifier_suggestions,d.identifier_suggestions)
59 -- Updated again to check for delayed symspell reification
60 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries(
62 skip_facet BOOL DEFAULT FALSE,
63 skip_display BOOL DEFAULT FALSE,
64 skip_browse BOOL DEFAULT FALSE,
65 skip_search BOOL DEFAULT FALSE,
66 only_fields INT[] DEFAULT '{}'::INT[]
67 ) RETURNS VOID AS $func$
70 ind_data metabib.field_entry_template%ROWTYPE;
71 mbe_row metabib.browse_entry%ROWTYPE;
78 field_list INT[] := only_fields;
79 field_types TEXT[] := '{}'::TEXT[];
82 IF field_list = '{}'::INT[] THEN
83 SELECT ARRAY_AGG(id) INTO field_list FROM config.metabib_field;
86 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;
87 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;
88 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;
89 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;
91 IF NOT b_skip_facet THEN field_types := field_types || '{facet}'; END IF;
92 IF NOT b_skip_display THEN field_types := field_types || '{display}'; END IF;
93 IF NOT b_skip_browse THEN field_types := field_types || '{browse}'; END IF;
94 IF NOT b_skip_search THEN field_types := field_types || '{search}'; END IF;
96 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
98 IF NOT b_skip_search THEN
99 FOR fclass IN SELECT * FROM config.metabib_class LOOP
100 -- RAISE NOTICE 'Emptying out %', fclass.name;
101 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
104 IF NOT b_skip_facet THEN
105 DELETE FROM metabib.facet_entry WHERE source = bib_id;
107 IF NOT b_skip_display THEN
108 DELETE FROM metabib.display_entry WHERE source = bib_id;
110 IF NOT b_skip_browse THEN
111 DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
115 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id, ' ', field_types, field_list ) LOOP
117 -- don't store what has been normalized away
118 CONTINUE WHEN ind_data.value IS NULL;
120 IF ind_data.field < 0 THEN
121 ind_data.field = -1 * ind_data.field;
124 IF ind_data.facet_field AND NOT b_skip_facet THEN
125 INSERT INTO metabib.facet_entry (field, source, value)
126 VALUES (ind_data.field, ind_data.source, ind_data.value);
129 IF ind_data.display_field AND NOT b_skip_display THEN
130 INSERT INTO metabib.display_entry (field, source, value)
131 VALUES (ind_data.field, ind_data.source, ind_data.value);
135 IF ind_data.browse_field AND NOT b_skip_browse THEN
136 -- A caveat about this SELECT: this should take care of replacing
137 -- old mbe rows when data changes, but not if normalization (by
138 -- which I mean specifically the output of
139 -- evergreen.oils_tsearch2()) changes. It may or may not be
140 -- expensive to add a comparison of index_vector to index_vector
141 -- to the WHERE clause below.
143 CONTINUE WHEN ind_data.sort_value IS NULL;
145 value_prepped := metabib.browse_normalize(ind_data.value, ind_data.field);
146 IF ind_data.browse_nocase THEN
147 SELECT INTO mbe_row * FROM metabib.browse_entry
148 WHERE evergreen.lowercase(value) = evergreen.lowercase(value_prepped) AND sort_value = ind_data.sort_value
149 ORDER BY sort_value, value LIMIT 1; -- gotta pick something, I guess
151 SELECT INTO mbe_row * FROM metabib.browse_entry
152 WHERE value = value_prepped AND sort_value = ind_data.sort_value;
156 mbe_id := mbe_row.id;
158 INSERT INTO metabib.browse_entry
159 ( value, sort_value ) VALUES
160 ( value_prepped, ind_data.sort_value );
162 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
165 INSERT INTO metabib.browse_entry_def_map (entry, def, source, authority)
166 VALUES (mbe_id, ind_data.field, ind_data.source, ind_data.authority);
169 IF ind_data.search_field AND NOT b_skip_search THEN
170 -- Avoid inserting duplicate rows
171 EXECUTE 'SELECT 1 FROM metabib.' || ind_data.field_class ||
172 '_field_entry WHERE field = $1 AND source = $2 AND value = $3'
173 INTO mbe_id USING ind_data.field, ind_data.source, ind_data.value;
174 -- RAISE NOTICE 'Search for an already matching row returned %', mbe_id;
175 IF mbe_id IS NULL THEN
177 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
179 quote_literal(ind_data.field) || $$, $$ ||
180 quote_literal(ind_data.source) || $$, $$ ||
181 quote_literal(ind_data.value) ||
188 IF NOT b_skip_search THEN
189 PERFORM metabib.update_combined_index_vectors(bib_id);
190 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_symspell_reification' AND enabled;
192 PERFORM search.symspell_dictionary_reify();
198 $func$ LANGUAGE PLPGSQL;