[GLUTEN-9240][VL] Write NULL value into relation in gluten unit tests

backends-velox/src/test/scala/org/apache/gluten/execution/JsonFunctionsValidateSuite.scala

@@ -30,10 +30,20 @@ class JsonFunctionsValidateSuite extends FunctionsValidateSuite {
       checkGlutenOperatorMatch[ProjectExecTransformer]
     }
 
-    runQueryAndCompare(
-      "SELECT l_orderkey, get_json_object('{\"a\":\"b\"}', '$.a') " +
-        "from lineitem limit 1;") {
-      checkGlutenOperatorMatch[ProjectExecTransformer]
+    withTempPath {
+      path =>
+        Seq[(String)](
+          ("""{"a":"b"}""")
+        )
+          .toDF("txt")
+          .write
+          .parquet(path.getCanonicalPath)
+
+        spark.read.parquet(path.getCanonicalPath).createOrReplaceTempView("tbl")
+
+        runQueryAndCompare("select get_json_object(txt, '$.a') from tbl") {
+          checkGlutenOperatorMatch[ProjectExecTransformer]
+        }
     }
 
     // Invalid UTF-8 encoding.
@@ -51,12 +61,22 @@ class JsonFunctionsValidateSuite extends FunctionsValidateSuite {
     runQueryAndCompare(
       s"select *, json_array_length(string_field1) " +
         s"from datatab limit 5")(checkGlutenOperatorMatch[ProjectExecTransformer])
-    runQueryAndCompare(
-      s"select l_orderkey, json_array_length('[1,2,3,4]') " +
-        s"from lineitem limit 5")(checkGlutenOperatorMatch[ProjectExecTransformer])
-    runQueryAndCompare(
-      s"select l_orderkey, json_array_length(null) " +
-        s"from lineitem limit 5")(checkGlutenOperatorMatch[ProjectExecTransformer])
+    withTempPath {
+      path =>
+        Seq[(String)](
+          ("[1,2,3,4]"),
+          (null.asInstanceOf[String])
+        )
+          .toDF("txt")
+          .write
+          .parquet(path.getCanonicalPath)
+
+        spark.read.parquet(path.getCanonicalPath).createOrReplaceTempView("tbl")
+
+        runQueryAndCompare("select json_array_length(txt) from tbl") {
+          checkGlutenOperatorMatch[ProjectExecTransformer]
+        }
+    }
   }
 
   testWithSpecifiedSparkVersion("from_json function bool", Some("3.4")) {

backends-velox/src/test/scala/org/apache/gluten/execution/ScalarFunctionsValidateSuite.scala

@@ -187,8 +187,20 @@ abstract class ScalarFunctionsValidateSuite extends FunctionsValidateSuite {
   }
 
   testWithSpecifiedSparkVersion("null input for array_size", Some("3.3")) {
-    runQueryAndCompare("SELECT array_size(null)") {
-      checkGlutenOperatorMatch[ProjectExecTransformer]
+    withTempPath {
+      path =>
+        Seq[(Array[Int])](
+          (null.asInstanceOf[Array[Int]])
+        )
+          .toDF("txt")
+          .write
+          .parquet(path.getCanonicalPath)
+
+        spark.read.parquet(path.getCanonicalPath).createOrReplaceTempView("tbl")
+
+        runQueryAndCompare("select array_size(txt) from tbl") {
+          checkGlutenOperatorMatch[ProjectExecTransformer]
+        }
     }
   }
 
@@ -801,8 +813,20 @@ abstract class ScalarFunctionsValidateSuite extends FunctionsValidateSuite {
   }
 
   test("Test sum/count function") {
-    runQueryAndCompare("""SELECT sum(2),count(2) from lineitem""".stripMargin) {
-      checkGlutenOperatorMatch[BatchScanExecTransformer]
+    withTempPath {
+      path =>
+        Seq[(Integer, Integer)](
+          (2, 2)
+        )
+          .toDF("val1", "val2")
+          .write
+          .parquet(path.getCanonicalPath)
+
+        spark.read.parquet(path.getCanonicalPath).createOrReplaceTempView("tbl")
+
+        runQueryAndCompare("SELECT sum(val1),count(val2) from tbl") {
+          checkGlutenOperatorMatch[BatchScanExecTransformer]
+        }
     }
   }
 
@@ -818,9 +842,20 @@ abstract class ScalarFunctionsValidateSuite extends FunctionsValidateSuite {
   }
 
   testWithSpecifiedSparkVersion("Test width_bucket function", Some("3.4")) {
-    runQueryAndCompare("""SELECT width_bucket(2, 0, 4, 3), l_orderkey
-                         | from lineitem limit 100""".stripMargin) {
-      checkGlutenOperatorMatch[ProjectExecTransformer]
+    withTempPath {
+      path =>
+        Seq[(Integer, Integer, Integer, Integer)](
+          (2, 0, 4, 3)
+        )
+          .toDF("val1", "val2", "val3", "val4")
+          .write
+          .parquet(path.getCanonicalPath)
+
+        spark.read.parquet(path.getCanonicalPath).createOrReplaceTempView("tbl")
+
+        runQueryAndCompare("SELECT width_bucket(val1, val2, val3, val4) from tbl") {
+          checkGlutenOperatorMatch[BatchScanExecTransformer]
+        }
     }
   }