[GitHub] [spark] cloud-fan commented on a change in pull request #26012: [SPARK-29346][SQL] Add Aggregating Accumulator

[GitBox <gi...@apache.org>]

cloud-fan commented on a change in pull request #26012: [SPARK-29346][SQL] Add Aggregating Accumulator
URL: https://github.com/apache/spark/pull/26012#discussion_r331480687
 
 

 ##########
 File path: sql/core/src/main/scala/org/apache/spark/sql/execution/AggregatingAccumulator.scala
 ##########
 @@ -0,0 +1,264 @@
+/*
+ * 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.execution
+
+import scala.collection.mutable
+
+import org.apache.spark.TaskContext
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Attribute, AttributeReference, AttributeSeq, BindReferences, Expression, InterpretedMutableProjection, InterpretedUnsafeProjection, JoinedRow, MutableProjection, NamedExpression, Projection, SpecificInternalRow}
+import org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression, DeclarativeAggregate, ImperativeAggregate, NoOp, TypedImperativeAggregate}
+import org.apache.spark.sql.internal.SQLConf
+import org.apache.spark.sql.types.{DataType, StructField, StructType}
+import org.apache.spark.util.AccumulatorV2
+
+/**
+ * Accumulator that computes a global aggregate.
+ */
+class AggregatingAccumulator private(
+    bufferSchema: Seq[DataType],
+    initialValues: Seq[Expression],
+    updateExpressions: Seq[Expression],
+    @transient private val mergeExpressions: Seq[Expression],
+    @transient private val resultExpressions: Seq[Expression],
+    imperatives: Array[ImperativeAggregate],
+    typedImperatives: Array[TypedImperativeAggregate[_]],
+    @transient private val conf: SQLConf)
+  extends AccumulatorV2[InternalRow, InternalRow] {
+  assert(bufferSchema.size == initialValues.size)
+  assert(bufferSchema.size == updateExpressions.size)
+  assert(mergeExpressions == null || bufferSchema.size == mergeExpressions.size)
+
+  private[this] var joinedRow: JoinedRow = _
+
+  private var buffer: SpecificInternalRow = _
+
+  private def createBuffer(): SpecificInternalRow = {
+    val buffer = new SpecificInternalRow(bufferSchema)
+
+    // Initialize the buffer. Note that we do not use a code generated projection here because
+    // generating and compiling a projection is probably more expensive than using an interpreted
+    // projection.
+    InterpretedMutableProjection.createProjection(initialValues)
+      .target(buffer)
+      .apply(InternalRow.empty)
+    imperatives.foreach(_.initialize(buffer))
+    typedImperatives.foreach(_.initialize(buffer))
+    buffer
+  }
+
+  private def getOrCreateBuffer(): SpecificInternalRow = {
+    if (buffer == null) {
+      buffer = createBuffer()
+
+      // Create the joined row and set the buffer as its 'left' row.
+      joinedRow = new JoinedRow()
+      joinedRow.withLeft(buffer)
+    }
+    buffer
+  }
+
+  private def initializeProjection[T <: Projection](projection: T): T = {
+    projection.initialize(TaskContext.getPartitionId())
+    projection
+  }
+
+  @transient
+  private[this] lazy val updateProjection = initializeProjection {
+    MutableProjection.create(updateExpressions)
+  }
+
+  @transient
+  private[this] lazy val mergeProjection = initializeProjection {
+    InterpretedMutableProjection.createProjection(mergeExpressions)
+  }
+
+  @transient
+  private[this] lazy val resultProjection = initializeProjection {
+    InterpretedUnsafeProjection.createProjection(resultExpressions)
+  }
+
+  /**
+   * Driver side operations like `merge` and `value` are executed in the DAGScheduler thread. This
+   * thread does not have a SQL configuration so we attach our own here. Note that we can't (and
+   * shouldn't) call `merge` or `value` on an accumulator originating from an executor so we just
 
 Review comment:
   What I mean is, we don't need the default value. We can simply `assert(conf != null)` in the method.

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org