[SPARK-20680][SQL][FOLLOW-UP] Add HiveVoidType in HiveClientImpl

### What changes were proposed in this pull request?

Discussion with [comment](https://github.com/apache/spark/pull/29244#issuecomment-671746329).

Add `HiveVoidType` class in `HiveClientImpl` then we can replace `NullType` to `HiveVoidType` before we call hive client.

### Why are the changes needed?

Better compatible with hive.

More details in [#29244](https://github.com/apache/spark/pull/29244).

### Does this PR introduce _any_ user-facing change?

Yes, user can create view with null type in Hive.

### How was this patch tested?

New test.

Closes #29423 from ulysses-you/add-HiveVoidType.

Authored-by: ulysses <youxiduo@weidian.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

sql/core/src/test/scala/org/apache/spark/sql/execution/SQLViewSuite.scala

@@ -723,7 +723,7 @@ abstract class SQLViewSuite extends QueryTest with SQLTestUtils {
 
   test("sparkSession API view resolution with different default database") {
     withDatabase("db2") {
-      withView("v1") {
+      withView("default.v1") {
         withTable("t1") {
           sql("USE default")
           sql("CREATE TABLE t1 USING parquet AS SELECT 1 AS c0")

sql/hive/src/main/scala/org/apache/spark/sql/hive/client/HiveClientImpl.scala

@@ -986,7 +986,8 @@ private[hive] object HiveClientImpl extends Logging {
     val typeString = if (c.metadata.contains(HIVE_TYPE_STRING)) {
       c.metadata.getString(HIVE_TYPE_STRING)
     } else {
-      c.dataType.catalogString
+      // replace NullType to HiveVoidType since Hive parse void not null.
+      HiveVoidType.replaceVoidType(c.dataType).catalogString
     }
     new FieldSchema(c.name, typeString, c.getComment().orNull)
   }
@@ -1005,7 +1006,8 @@ private[hive] object HiveClientImpl extends Logging {
   /** Builds the native StructField from Hive's FieldSchema. */
   def fromHiveColumn(hc: FieldSchema): StructField = {
     val columnType = getSparkSQLDataType(hc)
-    val metadata = if (hc.getType != columnType.catalogString) {
+    val replacedVoidType = HiveVoidType.replaceVoidType(columnType)
+    val metadata = if (hc.getType != replacedVoidType.catalogString) {
       new MetadataBuilder().putString(HIVE_TYPE_STRING, hc.getType).build()
     } else {
       Metadata.empty
@@ -1273,3 +1275,22 @@ private[hive] object HiveClientImpl extends Logging {
     hiveConf
   }
 }
+
+case object HiveVoidType extends DataType {
+  override def defaultSize: Int = 1
+  override def asNullable: DataType = HiveVoidType
+  override def simpleString: String = "void"
+
+  def replaceVoidType(dt: DataType): DataType = dt match {
+    case ArrayType(et, nullable) =>
+      ArrayType(replaceVoidType(et), nullable)
+    case MapType(kt, vt, nullable) =>
+      MapType(replaceVoidType(kt), replaceVoidType(vt), nullable)
+    case StructType(fields) =>
+      StructType(fields.map { field =>
+        field.copy(dataType = replaceVoidType(field.dataType))
+      })
+    case _: NullType => HiveVoidType
+    case _ => dt
+  }
+}

sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveSQLViewSuite.scala

@@ -22,7 +22,7 @@ import org.apache.spark.sql.catalyst.TableIdentifier
 import org.apache.spark.sql.catalyst.catalog.{CatalogStorageFormat, CatalogTable, CatalogTableType}
 import org.apache.spark.sql.execution.SQLViewSuite
 import org.apache.spark.sql.hive.test.{TestHive, TestHiveSingleton}
-import org.apache.spark.sql.types.StructType
+import org.apache.spark.sql.types.{NullType, StructType}
 
 /**
  * A test suite for Hive view related functionality.
@@ -137,4 +137,24 @@ class HiveSQLViewSuite extends SQLViewSuite with TestHiveSingleton {
       }
     }
   }
+
+  test("SPARK-20680: Add HiveVoidType to compatible with Hive void type") {
+    withView("v1") {
+      sql("create view v1 as select null as c")
+      val df = sql("select * from v1")
+      assert(df.schema.fields.head.dataType == NullType)
+      checkAnswer(
+        df,
+        Row(null)
+      )
+
+      sql("alter view v1 as select null as c1, 1 as c2")
+      val df2 = sql("select * from v1")
+      assert(df2.schema.fields.head.dataType == NullType)
+      checkAnswer(
+        df2,
+        Row(null, 1)
+      )
+    }
+  }
 }