Avoid collisions of normalized values going into metabib.browse_entry

This fixes a bug that prevents successful reingestion of bib records in
some cases.  Reported by George Duimovich.  Key part here:

> ERROR: duplicate key value violates unique constraint
> "browse_entry_value_key"
> DETAIL: Key (value)=(545575) already exists.
> CONTEXT: SQL statement "INSERT INTO metabib.browse_entry (value) VALUES
>                     (metabib.browse_normalize(ind_data.value,

Signed-off-by: Lebbeous Fogle-Weekley <lebbeous@esilibrary.com>

diff --git a/Open-ILS/src/sql/Pg/upgrade/XXXX.schema.reingest_avoid_collision_better.sql b/Open-ILS/src/sql/Pg/upgrade/XXXX.schema.reingest_avoid_collision_better.sql
new file mode 100644 (file)
index 0000000..a6d5961
--- /dev/null
+++ b/Open-ILS/src/sql/Pg/upgrade/XXXX.schema.reingest_avoid_collision_better.sql
@@ -0,0 +1,78 @@
+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;
+    normalized_value    TEXT;
+BEGIN
+    PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
+    IF NOT FOUND THEN
+        IF NOT 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 skip_facet THEN
+            DELETE FROM metabib.facet_entry WHERE source = bib_id;
+        END IF;
+        IF NOT 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 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 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.
+            normalized_value := metabib.browse_normalize(
+                ind_data.value, ind_data.field
+            );
+
+            SELECT INTO mbe_row * FROM metabib.browse_entry WHERE value = normalized_value;
+            IF FOUND THEN
+                mbe_id := mbe_row.id;
+            ELSE
+                INSERT INTO metabib.browse_entry (value) VALUES (normalized_value);
+                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;
+
+        IF ind_data.search_field AND NOT skip_search 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 LOOP;
+
+    RETURN;
+END;
+$func$ LANGUAGE PLPGSQL;
+
+COMMIT;

diff --git a/Open-ILS/src/sql/Pg/version-upgrade/2.1-2.2-upgrade-db.sql b/Open-ILS/src/sql/Pg/version-upgrade/2.1-2.2-upgrade-db.sql
index 6d1a138..46b68fc 100644 (file)
--- a/Open-ILS/src/sql/Pg/version-upgrade/2.1-2.2-upgrade-db.sql
+++ b/Open-ILS/src/sql/Pg/version-upgrade/2.1-2.2-upgrade-db.sql
@@ -10091,6 +10091,7 @@ DECLARE
     ind_data        metabib.field_entry_template%ROWTYPE;
     mbe_row         metabib.browse_entry%ROWTYPE;
     mbe_id          BIGINT;
+    normalized_value    TEXT;
 BEGIN
     PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
     IF NOT FOUND THEN
@@ -10125,12 +10126,15 @@ 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;
+            normalized_value := metabib.browse_normalize(
+                ind_data.value, ind_data.field
+            );
+
+            SELECT INTO mbe_row * FROM metabib.browse_entry WHERE value = normalized_value;
             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 (normalized_value);
                 mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
             END IF;