HIVE-29447: Incorrect nullability inference when transforming SEARCH expression back to disjunctions

ql/src/java/org/apache/hadoop/hive/ql/optimizer/calcite/SearchTransformer.java

@@ -133,7 +133,8 @@ public Shuttle(RexBuilder rexBuilder, RexUnknownAs unknownContext) {
           return call;
         }
       case SEARCH:
-        return new SearchTransformer<>(rexBuilder, call, this.unknownContext).transform();
+        RexNode transformed = new SearchTransformer<>(rexBuilder, call, this.unknownContext).transform();
+        return rexBuilder.makeCast(call.getType(), transformed);
       default:
         // Some calls (e.g., IS [NOT] NULL, COALESCE) are sensitive to changes in the 3-valued
         // logic (notably nulls). In such cases, the nullability of the operands is important,

ql/src/test/queries/clientpositive/cbo_search.q

@@ -0,0 +1,25 @@
+CREATE TABLE main_table (
+    id_col STRING,
+    join_key STRING,
+    int_col_to_test INT,
+    misc_attr STRING
+);
+
+CREATE TABLE lookup_table (
+    lookup_key STRING,
+    replacement_val INT
+);
+
+-- Test search to disjuncts conversion when search is in a Project
+EXPLAIN CBO
+SELECT
+    CASE
+        WHEN base.int_col_to_test IS NULL
+             OR base.int_col_to_test = ''
+             OR base.int_col_to_test = 0
+        THEN COALESCE(lkup.replacement_val, 0)
+        ELSE base.int_col_to_test
+    END AS final_calculated_column
+FROM main_table base
+JOIN lookup_table lkup ON base.join_key = lkup.lookup_key
+;

ql/src/test/results/clientpositive/llap/cbo_search.q.out

@@ -0,0 +1,72 @@
+PREHOOK: query: CREATE TABLE main_table (
+    id_col STRING,
+    join_key STRING,
+    int_col_to_test INT,
+    misc_attr STRING
+)
+PREHOOK: type: CREATETABLE
+PREHOOK: Output: database:default
+PREHOOK: Output: default@main_table
+POSTHOOK: query: CREATE TABLE main_table (
+    id_col STRING,
+    join_key STRING,
+    int_col_to_test INT,
+    misc_attr STRING
+)
+POSTHOOK: type: CREATETABLE
+POSTHOOK: Output: database:default
+POSTHOOK: Output: default@main_table
+PREHOOK: query: CREATE TABLE lookup_table (
+    lookup_key STRING,
+    replacement_val INT
+)
+PREHOOK: type: CREATETABLE
+PREHOOK: Output: database:default
+PREHOOK: Output: default@lookup_table
+POSTHOOK: query: CREATE TABLE lookup_table (
+    lookup_key STRING,
+    replacement_val INT
+)
+POSTHOOK: type: CREATETABLE
+POSTHOOK: Output: database:default
+POSTHOOK: Output: default@lookup_table
+PREHOOK: query: EXPLAIN CBO
+SELECT
+    CASE
+        WHEN base.int_col_to_test IS NULL
+             OR base.int_col_to_test = ''
+             OR base.int_col_to_test = 0
+        THEN COALESCE(lkup.replacement_val, 0)
+        ELSE base.int_col_to_test
+    END AS final_calculated_column
+FROM main_table base
+JOIN lookup_table lkup ON base.join_key = lkup.lookup_key
+PREHOOK: type: QUERY
+PREHOOK: Input: default@lookup_table
+PREHOOK: Input: default@main_table
+#### A masked pattern was here ####
+POSTHOOK: query: EXPLAIN CBO
+SELECT
+    CASE
+        WHEN base.int_col_to_test IS NULL
+             OR base.int_col_to_test = ''
+             OR base.int_col_to_test = 0
+        THEN COALESCE(lkup.replacement_val, 0)
+        ELSE base.int_col_to_test
+    END AS final_calculated_column
+FROM main_table base
+JOIN lookup_table lkup ON base.join_key = lkup.lookup_key
+POSTHOOK: type: QUERY
+POSTHOOK: Input: default@lookup_table
+POSTHOOK: Input: default@main_table
+#### A masked pattern was here ####
+CBO PLAN:
+HiveProject(final_calculated_column=[CASE($2, $4, $1)])
+  HiveJoin(condition=[=($0, $3)], joinType=[inner], algorithm=[none], cost=[not available])
+    HiveProject(join_key=[$1], int_col_to_test=[$2], EXPR$0=[CAST(OR(IS NULL($2), =($2, 0))):BOOLEAN NOT NULL])
+      HiveFilter(condition=[IS NOT NULL($1)])
+        HiveTableScan(table=[[default, main_table]], table:alias=[base])
+    HiveProject(lookup_key=[$0], EXPR$0=[CASE(IS NOT NULL($1), $1, 0)])
+      HiveFilter(condition=[IS NOT NULL($0)])
+        HiveTableScan(table=[[default, lookup_table]], table:alias=[lkup])
+