[SPARK-16278][SPARK-16279][SQL] Implement map_keys/map_values SQL fun…

…ctions

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/FunctionRegistry.scala

@@ -171,6 +171,8 @@ object FunctionRegistry {
     expression[IsNotNull]("isnotnull"),
     expression[Least]("least"),
     expression[CreateMap]("map"),
+    expression[MapKeys]("map_keys"),
+    expression[MapValues]("map_values"),
     expression[CreateNamedStruct]("named_struct"),
     expression[NaNvl]("nanvl"),
     expression[NullIf]("nullif"),

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/collectionOperations.scala

@@ -43,6 +43,56 @@ case class Size(child: Expression) extends UnaryExpression with ExpectsInputType
   }
 }
 
+/**
+ * Returns an unordered array containing the keys of the map.
+ */
+@ExpressionDescription(
+  usage = "_FUNC_(map) - Returns an unordered array containing the keys of the map.")
+case class MapKeys(child: Expression)
+  extends UnaryExpression with ExpectsInputTypes {
+
+  override def inputTypes: Seq[AbstractDataType] = Seq(MapType)
+
+  override def dataType: DataType = ArrayType(child.dataType.asInstanceOf[MapType].keyType)
+
+  override def foldable: Boolean = child.foldable
+
+  override def nullSafeEval(map: Any): Any = {
+    map.asInstanceOf[MapData].keyArray().copy()
+  }
+
+  override def doGenCode(ctx: CodegenContext, ev: ExprCode): ExprCode = {
+    nullSafeCodeGen(ctx, ev, c => s"${ev.value} = ($c).keyArray().copy();")
+  }
+
+  override def prettyName: String = "map_keys"
+}
+
+/**
+ * Returns an unordered array containing the values of the map.
+ */
+@ExpressionDescription(
+  usage = "_FUNC_(map) - Returns an unordered array containing the values of the map.")
+case class MapValues(child: Expression)
+  extends UnaryExpression with ExpectsInputTypes {
+
+  override def inputTypes: Seq[AbstractDataType] = Seq(MapType)
+
+  override def dataType: DataType = ArrayType(child.dataType.asInstanceOf[MapType].valueType)
+
+  override def foldable: Boolean = child.foldable
+
+  override def nullSafeEval(map: Any): Any = {
+    map.asInstanceOf[MapData].valueArray().copy()
+  }
+
+  override def doGenCode(ctx: CodegenContext, ev: ExprCode): ExprCode = {
+    nullSafeCodeGen(ctx, ev, c => s"${ev.value} = ($c).valueArray().copy();")
+  }
+
+  override def prettyName: String = "map_values"
+}
+
 /**
  * Sorts the input array in ascending / descending order according to the natural ordering of
  * the array elements and returns it.

sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/expressions/CollectionFunctionsSuite.scala

@@ -44,6 +44,16 @@ class CollectionFunctionsSuite extends SparkFunSuite with ExpressionEvalHelper {
     checkEvaluation(Literal.create(null, ArrayType(StringType)), null)
   }
 
+  test("MapKeys/MapValues") {
+    val m0 = Literal.create(Map("a" -> "1", "b" -> "2"), MapType(StringType, StringType))
+    val m1 = Literal.create(Map[String, String](), MapType(StringType, StringType))
+
+    checkEvaluation(MapKeys(m0), Seq("a", "b"))
+    checkEvaluation(MapValues(m0), Seq("1", "2"))
+    checkEvaluation(MapKeys(m1), Seq())
+    checkEvaluation(MapValues(m1), Seq())
+  }
+
   test("Sort Array") {
     val a0 = Literal.create(Seq(2, 1, 3), ArrayType(IntegerType))
     val a1 = Literal.create(Seq[Integer](), ArrayType(IntegerType))

sql/core/src/main/scala/org/apache/spark/sql/functions.scala

@@ -2760,6 +2760,22 @@ object functions {
    */
   def size(e: Column): Column = withExpr { Size(e.expr) }
 
+  /**
+   * Returns the key array of the map.
+   *
+   * @group collection_funcs
+   * @since 2.1.0
+   */
+  def map_keys(e: Column): Column = withExpr { MapKeys(e.expr) }
+
+  /**
+   * Returns the key array of the map.
+   *
+   * @group collection_funcs
+   * @since 2.1.0
+   */
+  def map_values(e: Column): Column = withExpr { MapValues(e.expr) }
+
   /**
    * Sorts the input array for the given column in ascending order,
    * according to the natural ordering of the array elements.

sql/core/src/test/scala/org/apache/spark/sql/DataFrameFunctionsSuite.scala

@@ -352,6 +352,30 @@ class DataFrameFunctionsSuite extends QueryTest with SharedSQLContext {
     )
   }
 
+  test("map_keys/map_values function") {
+    val df = Seq(
+      (Map[Int, Int](1 -> 100, 2 -> 200), "x"),
+      (Map[Int, Int](), "y"),
+      (Map[Int, Int](1 -> 100, 2 -> 200, 3 -> 300), "z")
+    ).toDF("a", "b")
+    checkAnswer(
+      df.select(map_keys($"a")),
+      Seq(Row(Seq(1, 2)), Row(Seq.empty), Row(Seq(1, 2, 3)))
+    )
+    checkAnswer(
+      df.selectExpr("map_keys(a)"),
+      Seq(Row(Seq(1, 2)), Row(Seq.empty), Row(Seq(1, 2, 3)))
+    )
+    checkAnswer(
+      df.select(map_values($"a")),
+      Seq(Row(Seq(100, 200)), Row(Seq.empty), Row(Seq(100, 200, 300)))
+    )
+    checkAnswer(
+      df.selectExpr("map_values(a)"),
+      Seq(Row(Seq(100, 200)), Row(Seq.empty), Row(Seq(100, 200, 300)))
+    )
+  }
+
   test("array contains function") {
     val df = Seq(
       (Seq[Int](1, 2), "x"),

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

@@ -239,7 +239,6 @@ private[sql] class HiveSessionCatalog(
   // str_to_map, windowingtablefunction.
   private val hiveFunctions = Seq(
     "hash", "java_method", "histogram_numeric",
-    "map_keys", "map_values",
     "parse_url", "percentile", "percentile_approx", "reflect", "sentences", "stack", "str_to_map",
     "xpath", "xpath_double", "xpath_float", "xpath_int", "xpath_long",
     "xpath_number", "xpath_short", "xpath_string",