Add max_by.

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

@@ -289,6 +289,7 @@ object FunctionRegistry {
     expression[Last]("last"),
     expression[Last]("last_value"),
     expression[Max]("max"),
+    expression[MaxBy]("max_by"),
     expression[Average]("mean"),
     expression[Min]("min"),
     expression[Percentile]("percentile"),

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/MaxBy.scala

@@ -0,0 +1,81 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql.catalyst.expressions.aggregate
+
+import org.apache.spark.sql.catalyst.analysis.TypeCheckResult
+import org.apache.spark.sql.catalyst.dsl.expressions._
+import org.apache.spark.sql.catalyst.expressions._
+import org.apache.spark.sql.catalyst.util.TypeUtils
+import org.apache.spark.sql.types._
+
+@ExpressionDescription(
+  usage = "_FUNC_(x, y) - Returns the value of `x` associated with the maximum value of `y`.",
+  examples = """
+    Examples:
+      > SELECT _FUNC_(x, y) FROM VALUES (('a', 10)), (('b', 50)), (('c', 20)) AS tab(x, y);
+       b
+  """,
+  since = "3.0")
+case class MaxBy(valueExpr: Expression, maxExpr: Expression) extends DeclarativeAggregate {
+
+  override def children: Seq[Expression] = valueExpr :: maxExpr :: Nil
+
+  override def nullable: Boolean = true
+
+  // Return data type.
+  override def dataType: DataType = valueExpr.dataType
+
+  override def checkInputDataTypes(): TypeCheckResult =
+    TypeUtils.checkForOrderingExpr(maxExpr.dataType, "function max_by")
+
+  private lazy val max = AttributeReference("max", maxExpr.dataType)()
+  private lazy val value = AttributeReference("value", valueExpr.dataType)()
+
+  override lazy val aggBufferAttributes: Seq[AttributeReference] = value :: max :: Nil
+
+  override lazy val initialValues: Seq[Literal] = Seq(
+    /* value = */ Literal.create(null, valueExpr.dataType),
+    /* max = */ Literal.create(null, maxExpr.dataType)
+  )
+
+  override lazy val updateExpressions: Seq[Expression] = Seq(
+    /* value = */
+    CaseWhen(
+      (And(IsNull(max), IsNull(maxExpr)), Literal.create(null, valueExpr.dataType)) ::
+        (IsNull(max), valueExpr) ::
+        (IsNull(maxExpr), value) :: Nil,
+      If(GreaterThan(max, maxExpr), value, valueExpr)
+    ),
+    /* max = */ greatest(max, maxExpr)
+  )
+
+  override lazy val mergeExpressions: Seq[Expression] = {
+    Seq(
+      /* value = */
+      CaseWhen(
+        (And(IsNull(max.left), IsNull(max.right)), Literal.create(null, valueExpr.dataType)) ::
+          (IsNull(max.left), value.right) ::
+          (IsNull(max.right), value.left) :: Nil,
+        If(GreaterThan(max.left, max.right), value.left, value.right)
+      ),
+      /* max = */ greatest(max.left, max.right)
+    )
+  }
+
+  override lazy val evaluateExpression: AttributeReference = value
+}

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

@@ -782,4 +782,47 @@ class DataFrameAggregateSuite extends QueryTest with SharedSQLContext {
     val countAndDistinct = df.select(count("*"), countDistinct("*"))
     checkAnswer(countAndDistinct, Row(100000, 100))
   }
+
+  test("max_by") {
+    val yearOfMaxEarnings =
+      sql("SELECT course, max_by(year, earnings) FROM courseSales GROUP BY course")
+    checkAnswer(yearOfMaxEarnings, Row("dotNET", 2013) :: Row("Java", 2013) :: Nil)
+
+    checkAnswer(
+      sql("SELECT max_by(x, y) FROM VALUES (('a', 10)), (('b', 50)), (('c', 20)) AS tab(x, y)"),
+      Row("b") :: Nil
+    )
+
+    checkAnswer(
+      sql("SELECT max_by(x, y) FROM VALUES (('a', 10)), (('b', null)), (('c', 20)) AS tab(x, y)"),
+      Row("c") :: Nil
+    )
+
+    checkAnswer(
+      sql("SELECT max_by(x, y) FROM VALUES (('a', null)), (('b', null)), (('c', 20)) AS tab(x, y)"),
+      Row("c") :: Nil
+    )
+
+    checkAnswer(
+      sql("SELECT max_by(x, y) FROM VALUES (('a', 10)), (('b', 50)), (('c', null)) AS tab(x, y)"),
+      Row("b") :: Nil
+    )
+
+    checkAnswer(
+      sql("SELECT max_by(x, y) FROM VALUES (('a', null)), (('b', null)) AS tab(x, y)"),
+      Row(null) :: Nil
+    )
+
+    withTempView("tempView") {
+      val dfWithMap = Seq((0, "a"), (1, "b"), (2, "c"))
+        .toDF("x", "y")
+        .select($"x", map($"x", $"y").as("y"))
+        .createOrReplaceTempView("tempView")
+      val error = intercept[AnalysisException] {
+        sql("SELECT max_by(x, y) FROM tempView").show
+      }
+      assert(
+        error.message.contains("function max_by does not support ordering on type map<int,string>"))
+    }
+  }
 }