[SPARK-19152][SQL]DataFrameWriter.saveAsTable support hive append

sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/rules.scala

@@ -112,12 +112,6 @@ case class AnalyzeCreateTable(sparkSession: SparkSession) extends Rule[LogicalPl
         throw new AnalysisException("Saving data into a view is not allowed.")
       }
 
-      if (DDLUtils.isHiveTable(existingTable)) {
-        throw new AnalysisException(s"Saving data in the Hive serde table $tableName is " +
-          "not supported yet. Please use the insertInto() API as an alternative.")
-      }
-
-      // Check if the specified data source match the data source of the existing table.
       val existingProvider = DataSource.lookupDataSource(existingTable.provider.get)
       val specifiedProvider = DataSource.lookupDataSource(tableDesc.provider.get)
       // TODO: Check that options from the resolved relation match the relation that we are

sql/hive/src/main/resources/META-INF/services/org.apache.spark.sql.sources.DataSourceRegister

@@ -1 +1,2 @@
 org.apache.spark.sql.hive.orc.OrcFileFormat
+org.apache.spark.sql.hive.execution.HiveFileFormat

sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveStrategies.scala

@@ -85,18 +85,11 @@ class HiveAnalysis(session: SparkSession) extends Rule[LogicalPlan] {
       InsertIntoHiveTable(table, partSpec, query, overwrite, ifNotExists)
 
     case CreateTable(tableDesc, mode, Some(query)) if DDLUtils.isHiveTable(tableDesc) =>
-      // Currently `DataFrameWriter.saveAsTable` doesn't support the Append mode of hive serde
-      // tables yet.
-      if (mode == SaveMode.Append) {
-        throw new AnalysisException(
-          "CTAS for hive serde tables does not support append semantics.")
-      }
-
       val dbName = tableDesc.identifier.database.getOrElse(session.catalog.currentDatabase)
       CreateHiveTableAsSelectCommand(
         tableDesc.copy(identifier = tableDesc.identifier.copy(database = Some(dbName))),
         query,
-        mode == SaveMode.Ignore)
+        mode)
   }
 
   /**

sql/hive/src/main/scala/org/apache/spark/sql/hive/execution/CreateHiveTableAsSelectCommand.scala

@@ -19,7 +19,7 @@ package org.apache.spark.sql.hive.execution
 
 import scala.util.control.NonFatal
 
-import org.apache.spark.sql.{AnalysisException, Row, SparkSession}
+import org.apache.spark.sql.{AnalysisException, Row, SaveMode, SparkSession}
 import org.apache.spark.sql.catalyst.catalog.{CatalogTable, SimpleCatalogRelation}
 import org.apache.spark.sql.catalyst.plans.logical.{InsertIntoTable, LogicalPlan, SubqueryAlias}
 import org.apache.spark.sql.execution.command.RunnableCommand
@@ -31,13 +31,12 @@ import org.apache.spark.sql.hive.MetastoreRelation
  *
  * @param tableDesc the Table Describe, which may contains serde, storage handler etc.
  * @param query the query whose result will be insert into the new relation
- * @param ignoreIfExists allow continue working if it's already exists, otherwise
- *                      raise exception
+ * @param mode SaveMode
  */
 case class CreateHiveTableAsSelectCommand(
     tableDesc: CatalogTable,
     query: LogicalPlan,
-    ignoreIfExists: Boolean)
+    mode: SaveMode)
   extends RunnableCommand {
 
   private val tableIdentifier = tableDesc.identifier
@@ -67,7 +66,7 @@ case class CreateHiveTableAsSelectCommand(
         withFormat
       }
 
-      sparkSession.sessionState.catalog.createTable(withSchema, ignoreIfExists = false)
+      sparkSession.sessionState.catalog.createTable(withSchema, ignoreIfExists = true)
 
       // Get the Metastore Relation
       sparkSession.sessionState.catalog.lookupRelation(tableIdentifier) match {
@@ -80,11 +79,18 @@ case class CreateHiveTableAsSelectCommand(
     // add the relation into catalog, just in case of failure occurs while data
     // processing.
     if (sparkSession.sessionState.catalog.tableExists(tableIdentifier)) {
-      if (ignoreIfExists) {
-        // table already exists, will do nothing, to keep consistent with Hive
-      } else {
+      assert(mode != SaveMode.Overwrite,
+        s"Expect the table $tableIdentifier has been dropped when the save mode is Overwrite")
+
+      if (mode == SaveMode.ErrorIfExists) {
         throw new AnalysisException(s"$tableIdentifier already exists.")
       }
+      if (mode == SaveMode.Ignore) {
+        // Since the table already exists and the save mode is Ignore, we will just return.
+        return Seq.empty
+      }
+      sparkSession.sessionState.executePlan(InsertIntoTable(
+        metastoreRelation, Map(), query, overwrite = false, ifNotExists = false)).toRdd
     } else {
       try {
         sparkSession.sessionState.executePlan(InsertIntoTable(

sql/hive/src/main/scala/org/apache/spark/sql/hive/execution/HiveFileFormat.scala

@@ -35,6 +35,7 @@ import org.apache.spark.sql.catalyst.InternalRow
 import org.apache.spark.sql.execution.datasources.{FileFormat, OutputWriter, OutputWriterFactory}
 import org.apache.spark.sql.hive.{HiveInspectors, HiveTableUtil}
 import org.apache.spark.sql.hive.HiveShim.{ShimFileSinkDesc => FileSinkDesc}
+import org.apache.spark.sql.sources.DataSourceRegister
 import org.apache.spark.sql.types.StructType
 import org.apache.spark.util.SerializableJobConf
 
@@ -43,7 +44,13 @@ import org.apache.spark.util.SerializableJobConf
  *
  * TODO: implement the read logic.
  */
-class HiveFileFormat(fileSinkConf: FileSinkDesc) extends FileFormat with Logging {
+class HiveFileFormat(fileSinkConf: FileSinkDesc)
+  extends FileFormat with DataSourceRegister with Logging {
+
+  def this() = this(null)
+
+  override def shortName(): String = "hive"
+
   override def inferSchema(
       sparkSession: SparkSession,
       options: Map[String, String],

sql/hive/src/test/scala/org/apache/spark/sql/hive/MetastoreDataSourcesSuite.scala

@@ -419,12 +419,6 @@ class MetastoreDataSourcesSuite extends QueryTest with SQLTestUtils with TestHiv
       sql(s"CREATE TABLE $tableName STORED AS SEQUENCEFILE AS SELECT 1 AS key, 'abc' AS value")
 
       val df = sql(s"SELECT key, value FROM $tableName")
-      val e = intercept[AnalysisException] {
-        df.write.mode(SaveMode.Append).saveAsTable(tableName)
-      }.getMessage
-      assert(e.contains("Saving data in the Hive serde table default.tab1 is not supported " +
-        "yet. Please use the insertInto() API as an alternative."))
-
       df.write.insertInto(tableName)
       checkAnswer(
         sql(s"SELECT * FROM $tableName"),
@@ -1167,16 +1161,15 @@ class MetastoreDataSourcesSuite extends QueryTest with SQLTestUtils with TestHiv
 
   test("create a temp view using hive") {
     val tableName = "tab1"
-    withTable (tableName) {
-      val e = intercept[ClassNotFoundException] {
-        sql(
-          s"""
-             |CREATE TEMPORARY VIEW $tableName
-             |(col1 int)
-             |USING hive
-           """.stripMargin)
-      }.getMessage
-      assert(e.contains("Failed to find data source: hive"))
+    withTable(tableName) {
+      sql(
+        s"""
+           |CREATE TEMPORARY VIEW $tableName
+           |(col1 int)
+           |USING hive
+         """.stripMargin)
+      val tempView = spark.sessionState.catalog.getTempView(tableName)
+      assert(tempView.isDefined, "create a temp view using hive should success")
     }
   }
 

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

@@ -1319,7 +1319,7 @@ class HiveDDLSuite
   }
 
   test("create hive serde table with DataFrameWriter.saveAsTable") {
-    withTable("t", "t2") {
+    withTable("t", "t1") {
       Seq(1 -> "a").toDF("i", "j")
         .write.format("hive").option("fileFormat", "avro").saveAsTable("t")
       checkAnswer(spark.table("t"), Row(1, "a"))
@@ -1350,11 +1350,8 @@ class HiveDDLSuite
       assert(table.storage.serde ==
         Some("org.apache.hadoop.hive.serde2.avro.AvroSerDe"))
 
-      sql("INSERT INTO t SELECT 2, 'b'")
-      checkAnswer(spark.table("t"), Row(9, "x") :: Row(2, "b") :: Nil)
-
       val e2 = intercept[AnalysisException] {
-        Seq(1 -> "a").toDF("i", "j").write.format("hive").bucketBy(4, "i").saveAsTable("t2")
+        Seq(1 -> "a").toDF("i", "j").write.format("hive").bucketBy(4, "i").saveAsTable("t1")
       }
       assert(e2.message.contains("Creating bucketed Hive serde table is not supported yet"))
 
@@ -1365,6 +1362,35 @@ class HiveDDLSuite
     }
   }
 
+  test("append data to hive serde table") {
+    withTable("t", "t1") {
+      Seq(1 -> "a").toDF("i", "j")
+        .write.format("hive").option("fileFormat", "avro").saveAsTable("t")
+      checkAnswer(spark.table("t"), Row(1, "a"))
+
+      sql("INSERT INTO t SELECT 2, 'b'")
+      checkAnswer(spark.table("t"), Row(1, "a") :: Row(2, "b") :: Nil)
+
+      Seq(3 -> "c").toDF("i", "j")
+        .write.format("hive").mode("append").saveAsTable("t")
+      checkAnswer(spark.table("t"), Row(1, "a") :: Row(2, "b") :: Row(3, "c") :: Nil)
+
+      Seq("c" -> 3).toDF("i", "j")
+        .write.format("hive").mode("append").saveAsTable("t")
+      checkAnswer(spark.table("t"), Row(1, "a") :: Row(2, "b") :: Row(3, "c")
+        :: Row(null, "3") :: Nil)
+
+      Seq(4 -> "d").toDF("i", "j").write.saveAsTable("t1")
+
+      val e = intercept[AnalysisException] {
+        Seq(5 -> "e").toDF("i", "j")
+          .write.format("hive").mode("append").saveAsTable("t1")
+      }
+      assert(e.message.contains("The format of the existing table default.t1 is " +
+        "`ParquetFileFormat`. It doesn't match the specified format `HiveFileFormat`."))
+    }
+  }
+
   test("create partitioned hive serde table as select") {
     withTable("t", "t1") {
       withSQLConf("hive.exec.dynamic.partition.mode" -> "nonstrict") {