[GLUTEN-8598][CH] Fix diff for cast string to long (#8701)

apache · Feb 11, 2025 · cc7a355 · cc7a355
1 parent 39950cf
commit cc7a355
Show file tree

Hide file tree

Showing 2 changed files with 14 additions and 2 deletions.
diff --git a/...s-clickhouse/src/test/scala/org/apache/gluten/execution/GlutenFunctionValidateSuite.scala b/...s-clickhouse/src/test/scala/org/apache/gluten/execution/GlutenFunctionValidateSuite.scala
@@ -968,6 +968,17 @@ class GlutenFunctionValidateSuite extends GlutenClickHouseWholeStageTransformerS
     compareResultsAgainstVanillaSpark(sql, true, { _ => })
   }
 
+  test("GLUTEN-8598 Fix diff for cast string to long") {
+    withSQLConf(
+      SQLConf.OPTIMIZER_EXCLUDED_RULES.key ->
+        (ConstantFolding.ruleName + "," + NullPropagation.ruleName)) {
+      runQueryAndCompare(
+        "select cast(' \t2570852431\n' as long), cast('25708\t52431\n' as long)",
+        noFallBack = false
+      )(checkGlutenOperatorMatch[ProjectExecTransformer])
+    }
+  }
+
   test("Test transform_keys/transform_values") {
     val sql = """
                 |select id, sort_array(map_entries(m1)), sort_array(map_entries(m2)) from(

diff --git a/cpp-ch/local-engine/Parser/ExpressionParser.cpp b/cpp-ch/local-engine/Parser/ExpressionParser.cpp
@@ -368,8 +368,9 @@ ExpressionParser::NodeRawConstPtr ExpressionParser::parseExpression(ActionsDAG &
             else if (isString(denull_input_type) && isInt(denull_output_type))
             {
                 /// Spark cast(x as INT) if x is String -> CH cast(trim(x) as INT)
-                /// Refer to https://github.com/apache/incubator-gluten/issues/4956
-                args[0] = toFunctionNode(actions_dag, "trim", {args[0]});
+                /// Refer to https://github.com/apache/incubator-gluten/issues/4956 and https://github.com/apache/incubator-gluten/issues/8598
+                auto trim_str_arg = addConstColumn(actions_dag, std::make_shared<DataTypeString>(), " \t\n\r\f");
+                args[0] = toFunctionNode(actions_dag, "trimBothSpark", {args[0], trim_str_arg});
                 args.emplace_back(addConstColumn(actions_dag, std::make_shared<DataTypeString>(), output_type->getName()));
                 result_node = toFunctionNode(actions_dag, "CAST", args);
             }