3 CREATE OR REPLACE FUNCTION search.disable_symspell_reification () RETURNS VOID AS $f$
4 INSERT INTO config.internal_flag (name,enabled)
5 VALUES ('ingest.disable_symspell_reification',TRUE)
6 ON CONFLICT (name) DO UPDATE SET enabled = TRUE;
9 CREATE OR REPLACE FUNCTION search.enable_symspell_reification () RETURNS VOID AS $f$
10 UPDATE config.internal_flag SET enabled = FALSE WHERE name = 'ingest.disable_symspell_reification';
13 CREATE OR REPLACE FUNCTION search.symspell_dictionary_full_reify () RETURNS SETOF search.symspell_dictionary AS $f$
15 DELETE FROM search.symspell_dictionary_updates RETURNING *
16 ), computed_rows AS ( -- this collapses the rows deleted into the format we need for UPSERT
17 SELECT SUM(keyword_count) AS keyword_count,
18 SUM(title_count) AS title_count,
19 SUM(author_count) AS author_count,
20 SUM(subject_count) AS subject_count,
21 SUM(series_count) AS series_count,
22 SUM(identifier_count) AS identifier_count,
26 ARRAY_REMOVE(ARRAY_AGG(DISTINCT keyword_suggestions[1]), NULL) AS keyword_suggestions,
27 ARRAY_REMOVE(ARRAY_AGG(DISTINCT title_suggestions[1]), NULL) AS title_suggestions,
28 ARRAY_REMOVE(ARRAY_AGG(DISTINCT author_suggestions[1]), NULL) AS author_suggestions,
29 ARRAY_REMOVE(ARRAY_AGG(DISTINCT subject_suggestions[1]), NULL) AS subject_suggestions,
30 ARRAY_REMOVE(ARRAY_AGG(DISTINCT series_suggestions[1]), NULL) AS series_suggestions,
31 ARRAY_REMOVE(ARRAY_AGG(DISTINCT identifier_suggestions[1]), NULL) AS identifier_suggestions
35 INSERT INTO search.symspell_dictionary AS d SELECT * FROM computed_rows
36 ON CONFLICT (prefix_key) DO UPDATE SET
37 keyword_count = GREATEST(0, d.keyword_count + EXCLUDED.keyword_count),
38 keyword_suggestions = evergreen.text_array_merge_unique(EXCLUDED.keyword_suggestions,d.keyword_suggestions),
40 title_count = GREATEST(0, d.title_count + EXCLUDED.title_count),
41 title_suggestions = evergreen.text_array_merge_unique(EXCLUDED.title_suggestions,d.title_suggestions),
43 author_count = GREATEST(0, d.author_count + EXCLUDED.author_count),
44 author_suggestions = evergreen.text_array_merge_unique(EXCLUDED.author_suggestions,d.author_suggestions),
46 subject_count = GREATEST(0, d.subject_count + EXCLUDED.subject_count),
47 subject_suggestions = evergreen.text_array_merge_unique(EXCLUDED.subject_suggestions,d.subject_suggestions),
49 series_count = GREATEST(0, d.series_count + EXCLUDED.series_count),
50 series_suggestions = evergreen.text_array_merge_unique(EXCLUDED.series_suggestions,d.series_suggestions),
52 identifier_count = GREATEST(0, d.identifier_count + EXCLUDED.identifier_count),
53 identifier_suggestions = evergreen.text_array_merge_unique(EXCLUDED.identifier_suggestions,d.identifier_suggestions)
57 -- Updated again to check for delayed symspell reification
58 CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries(
60 skip_facet BOOL DEFAULT FALSE,
61 skip_display BOOL DEFAULT FALSE,
62 skip_browse BOOL DEFAULT FALSE,
63 skip_search BOOL DEFAULT FALSE,
64 only_fields INT[] DEFAULT '{}'::INT[]
65 ) RETURNS VOID AS $func$
68 ind_data metabib.field_entry_template%ROWTYPE;
69 mbe_row metabib.browse_entry%ROWTYPE;
76 field_list INT[] := only_fields;
77 field_types TEXT[] := '{}'::TEXT[];
80 IF field_list = '{}'::INT[] THEN
81 SELECT ARRAY_AGG(id) INTO field_list FROM config.metabib_field;
84 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;
85 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;
86 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;
87 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;
89 IF NOT b_skip_facet THEN field_types := field_types || '{facet}'; END IF;
90 IF NOT b_skip_display THEN field_types := field_types || '{display}'; END IF;
91 IF NOT b_skip_browse THEN field_types := field_types || '{browse}'; END IF;
92 IF NOT b_skip_search THEN field_types := field_types || '{search}'; END IF;
94 PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
96 IF NOT b_skip_search THEN
97 FOR fclass IN SELECT * FROM config.metabib_class LOOP
98 -- RAISE NOTICE 'Emptying out %', fclass.name;
99 EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
102 IF NOT b_skip_facet THEN
103 DELETE FROM metabib.facet_entry WHERE source = bib_id;
105 IF NOT b_skip_display THEN
106 DELETE FROM metabib.display_entry WHERE source = bib_id;
108 IF NOT b_skip_browse THEN
109 DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
113 FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id, ' ', field_types, field_list ) LOOP
115 -- don't store what has been normalized away
116 CONTINUE WHEN ind_data.value IS NULL;
118 IF ind_data.field < 0 THEN
119 ind_data.field = -1 * ind_data.field;
122 IF ind_data.facet_field AND NOT b_skip_facet THEN
123 INSERT INTO metabib.facet_entry (field, source, value)
124 VALUES (ind_data.field, ind_data.source, ind_data.value);
127 IF ind_data.display_field AND NOT b_skip_display THEN
128 INSERT INTO metabib.display_entry (field, source, value)
129 VALUES (ind_data.field, ind_data.source, ind_data.value);
133 IF ind_data.browse_field AND NOT b_skip_browse THEN
134 -- A caveat about this SELECT: this should take care of replacing
135 -- old mbe rows when data changes, but not if normalization (by
136 -- which I mean specifically the output of
137 -- evergreen.oils_tsearch2()) changes. It may or may not be
138 -- expensive to add a comparison of index_vector to index_vector
139 -- to the WHERE clause below.
141 CONTINUE WHEN ind_data.sort_value IS NULL;
143 value_prepped := metabib.browse_normalize(ind_data.value, ind_data.field);
144 IF ind_data.browse_nocase THEN
145 SELECT INTO mbe_row * FROM metabib.browse_entry
146 WHERE evergreen.lowercase(value) = evergreen.lowercase(value_prepped) AND sort_value = ind_data.sort_value
147 ORDER BY sort_value, value LIMIT 1; -- gotta pick something, I guess
149 SELECT INTO mbe_row * FROM metabib.browse_entry
150 WHERE value = value_prepped AND sort_value = ind_data.sort_value;
154 mbe_id := mbe_row.id;
156 INSERT INTO metabib.browse_entry
157 ( value, sort_value ) VALUES
158 ( value_prepped, ind_data.sort_value );
160 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
163 INSERT INTO metabib.browse_entry_def_map (entry, def, source, authority)
164 VALUES (mbe_id, ind_data.field, ind_data.source, ind_data.authority);
167 IF ind_data.search_field AND NOT b_skip_search THEN
168 -- Avoid inserting duplicate rows
169 EXECUTE 'SELECT 1 FROM metabib.' || ind_data.field_class ||
170 '_field_entry WHERE field = $1 AND source = $2 AND value = $3'
171 INTO mbe_id USING ind_data.field, ind_data.source, ind_data.value;
172 -- RAISE NOTICE 'Search for an already matching row returned %', mbe_id;
173 IF mbe_id IS NULL THEN
175 INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
177 quote_literal(ind_data.field) || $$, $$ ||
178 quote_literal(ind_data.source) || $$, $$ ||
179 quote_literal(ind_data.value) ||
186 IF NOT b_skip_search THEN
187 PERFORM metabib.update_combined_index_vectors(bib_id);
188 PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_symspell_reification' AND enabled;
190 PERFORM search.symspell_dictionary_reify();
196 $func$ LANGUAGE PLPGSQL;