rdblue commented on a change in pull request #1947:
URL: https://github.com/apache/iceberg/pull/1947#discussion_r559053410



##########
File path: 
spark3-extensions/src/main/scala/org/apache/spark/sql/execution/datasources/v2/MergeIntoExec.scala
##########
@@ -0,0 +1,111 @@
+/*
+ * 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.datasources.v2
+
+import org.apache.spark.rdd.RDD
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Attribute, BasePredicate, 
Expression, UnsafeProjection}
+import org.apache.spark.sql.catalyst.expressions.codegen.GeneratePredicate
+import org.apache.spark.sql.catalyst.plans.logical.MergeIntoParams
+import org.apache.spark.sql.execution.{SparkPlan, UnaryExecNode}
+
+case class MergeIntoExec(mergeIntoProcessor: MergeIntoParams,
+                         @transient targetRelation: DataSourceV2Relation,
+                         override val child: SparkPlan) extends UnaryExecNode {
+
+  override def output: Seq[Attribute] = targetRelation.output
+
+  protected override def doExecute(): RDD[InternalRow] = {
+    child.execute().mapPartitions {
+      processPartition(mergeIntoProcessor, _)
+    }
+  }
+
+  private def generateProjection(exprs: Seq[Expression], attrs: 
Seq[Attribute]): UnsafeProjection = {
+    UnsafeProjection.create(exprs, attrs)
+  }
+
+  private def generatePredicate(expr: Expression, attrs: Seq[Attribute]): 
BasePredicate = {
+    GeneratePredicate.generate(expr, attrs)
+  }
+
+  def applyProjection(predicates: Seq[BasePredicate],
+                      projections: Seq[UnsafeProjection],
+                      projectTargetCols: UnsafeProjection,
+                      projectDeleteRow: UnsafeProjection,
+                      inputRow: InternalRow,
+                      targetRowNotPresent: Boolean): InternalRow = {
+    // Find the first combination where the predicate evaluates to true
+    val pair = (predicates zip projections).find {
+      case (predicate, _) => predicate.eval(inputRow)
+    }
+
+    // Now apply the appropriate projection to either :
+    // - Insert a row into target
+    // - Update a row of target
+    // - Delete a row in target. The projected row will have the deleted bit 
set.
+    pair match {
+      case Some((_, projection)) =>
+        projection.apply(inputRow)
+      case None =>
+        if (targetRowNotPresent) {
+          projectDeleteRow.apply(inputRow)

Review comment:
       What if we just returned `null` instead?
   
   I think the problem is that this is trying to create one output for each 
input row, then filtering happens afterward. An extra column is added to signal 
that the row should be kept or not. But we don't need to copy the row if it is 
going to be removed. We also don't need to copy incoming target rows just to 
add a `true` at the end if they are going to be kept.
   
   So what if we changed all of the delete cases to produce `null` instead?
   
   Let's not worry about this for now, but I'll open a PR after this is merged 
to simplify and avoid some of the copies.




----------------------------------------------------------------
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:
[email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to