Browse normalization timing fix

When indexing browse entries, we need to normalize the value we want to use
before we go looking for it in the table, for uniqueness.  We do in master,
we need to in 2.4 (and before) as well.

Signed-off-by: Mike Rylander <mrylander@gmail.com>
Signed-off-by: Ben Shum <bshum@biblio.org>

diff --git a/Open-ILS/src/sql/Pg/030.schema.metabib.sql b/Open-ILS/src/sql/Pg/030.schema.metabib.sql
index 68d44fd..fc6e09d 100644 (file)
--- a/Open-ILS/src/sql/Pg/030.schema.metabib.sql
+++ b/Open-ILS/src/sql/Pg/030.schema.metabib.sql
@@ -586,6 +586,7 @@ DECLARE
     ind_data        metabib.field_entry_template%ROWTYPE;
     mbe_row         metabib.browse_entry%ROWTYPE;
     mbe_id          BIGINT;
+    mbe_txt         TEXT;
     b_skip_facet    BOOL;
     b_skip_browse   BOOL;
     b_skip_search   BOOL;
@@ -628,12 +629,12 @@ BEGIN
             -- evergreen.oils_tsearch2()) changes.  It may or may not be
             -- expensive to add a comparison of index_vector to index_vector
             -- to the WHERE clause below.
-            SELECT INTO mbe_row * FROM metabib.browse_entry WHERE value = ind_data.value;
+            mbe_txt := metabib.browse_normalize(ind_data.value, ind_data.field);
+            SELECT INTO mbe_row * FROM metabib.browse_entry WHERE value = mbe_txt;
             IF FOUND THEN
                 mbe_id := mbe_row.id;
             ELSE
-                INSERT INTO metabib.browse_entry (value) VALUES
-                    (metabib.browse_normalize(ind_data.value, ind_data.field));
+                INSERT INTO metabib.browse_entry (value) VALUES (mbe_txt);
                 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
             END IF;
 

diff --git a/Open-ILS/src/sql/Pg/upgrade/XXXX.function.browse-normalize-timing.sql b/Open-ILS/src/sql/Pg/upgrade/XXXX.function.browse-normalize-timing.sql
new file mode 100644 (file)
index 0000000..5b59127
--- /dev/null
+++ b/Open-ILS/src/sql/Pg/upgrade/XXXX.function.browse-normalize-timing.sql
@@ -0,0 +1,97 @@
+BEGIN;
+
+SELECT evergreen.upgrade_deps_block_check('XXXX', :eg_version);
+
+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 $func$
+DECLARE
+    fclass          RECORD;
+    ind_data        metabib.field_entry_template%ROWTYPE;
+    mbe_row         metabib.browse_entry%ROWTYPE;
+    mbe_id          BIGINT;
+    mbe_txt         TEXT;
+    b_skip_facet    BOOL;
+    b_skip_browse   BOOL;
+    b_skip_search   BOOL;
+BEGIN
+
+    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;
+    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;
+    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;
+
+    PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
+    IF NOT FOUND THEN
+        IF NOT b_skip_search THEN
+            FOR fclass IN SELECT * FROM config.metabib_class LOOP
+                -- RAISE NOTICE 'Emptying out %', fclass.name;
+                EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
+            END LOOP;
+        END IF;
+        IF NOT b_skip_facet THEN
+            DELETE FROM metabib.facet_entry WHERE source = bib_id;
+        END IF;
+        IF NOT b_skip_browse THEN
+            DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
+        END IF;
+    END IF;
+
+    FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
+        IF ind_data.field < 0 THEN
+            ind_data.field = -1 * ind_data.field;
+        END IF;
+
+        IF ind_data.facet_field AND NOT b_skip_facet THEN
+            INSERT INTO metabib.facet_entry (field, source, value)
+                VALUES (ind_data.field, ind_data.source, ind_data.value);
+        END IF;
+
+        IF ind_data.browse_field AND NOT b_skip_browse THEN
+            -- A caveat about this SELECT: this should take care of replacing
+            -- old mbe rows when data changes, but not if normalization (by
+            -- which I mean specifically the output of
+            -- evergreen.oils_tsearch2()) changes.  It may or may not be
+            -- expensive to add a comparison of index_vector to index_vector
+            -- to the WHERE clause below.
+            mbe_txt := metabib.browse_normalize(ind_data.value, ind_data.field);
+            SELECT INTO mbe_row * FROM metabib.browse_entry WHERE value = mbe_txt;
+            IF FOUND THEN
+                mbe_id := mbe_row.id;
+            ELSE
+                INSERT INTO metabib.browse_entry (value) VALUES (mbe_txt);
+                mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
+            END IF;
+
+            INSERT INTO metabib.browse_entry_def_map (entry, def, source)
+                VALUES (mbe_id, ind_data.field, ind_data.source);
+        END IF;
+
+        -- Avoid inserting duplicate rows, but retain granularity of being
+        -- able to search browse fields with "starts with" type operators
+        -- (for example, for titles of songs in music albums)
+        IF (ind_data.search_field OR ind_data.browse_field) AND NOT b_skip_search THEN
+            EXECUTE 'SELECT 1 FROM metabib.' || ind_data.field_class ||
+                '_field_entry WHERE field = $1 AND source = $2 AND value = $3'
+                INTO mbe_id USING ind_data.field, ind_data.source, ind_data.value;
+                -- RAISE NOTICE 'Search for an already matching row returned %', mbe_id;
+            IF mbe_id IS NULL THEN
+                EXECUTE $$
+                INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
+                    VALUES ($$ ||
+                        quote_literal(ind_data.field) || $$, $$ ||
+                        quote_literal(ind_data.source) || $$, $$ ||
+                        quote_literal(ind_data.value) ||
+                    $$);$$;
+            END IF;
+        END IF;
+
+    END LOOP;
+
+    IF NOT b_skip_search THEN
+        PERFORM metabib.update_combined_index_vectors(bib_id);
+    END IF;
+
+    RETURN;
+END;
+$func$ LANGUAGE PLPGSQL;
+
+COMMIT;
+