aokolnychyi commented on a change in pull request #35395: URL: https://github.com/apache/spark/pull/35395#discussion_r840949905
########## File path: sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/v2/OptimizeMetadataOnlyDeleteFromTable.scala ########## @@ -0,0 +1,68 @@ +/* + * 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.sql.catalyst.expressions.{Expression, PredicateHelper, SubqueryExpression} +import org.apache.spark.sql.catalyst.expressions.Literal.TrueLiteral +import org.apache.spark.sql.catalyst.plans.logical.{DeleteFromTable, DeleteFromTableWithFilters, LogicalPlan} +import org.apache.spark.sql.catalyst.rules.Rule +import org.apache.spark.sql.connector.catalog.{SupportsDelete, TruncatableTable} +import org.apache.spark.sql.execution.datasources.DataSourceStrategy +import org.apache.spark.sql.sources + +/** + * Checks whether a delete using filters is possible and nullifies the rewrite plan + * if the source can handle this delete without executing the rewrite plan. + * + * Note this rule must be run after expression optimization but before scan planning. + */ +object OptimizeMetadataOnlyDeleteFromTable extends Rule[LogicalPlan] with PredicateHelper { + + override def apply(plan: LogicalPlan): LogicalPlan = plan transform { + case d @ DeleteFromTable(relation: DataSourceV2Relation, cond, Some(_)) => Review comment: > Next, the V2ScanRelationPushDown will just work and apply operator pushdown to GroupBasedReplaceData.query. I am afraid it is vice versa. The regular planning in `V2ScanRelationPushDown` can be applied to the `filter` subquery, not the `query` plan that contains remaining rows. The `query` plan contains a negated condition that cannot be pushed. That means the main scan must be fully built before `V2ScanRelationPushDown` kicks in. We cannot partially preconfigure the scan there. Parquet pushdown can be enabled in the `filter` subquery, not the main scan. > this rule can also evaluate the query and collect the affected groups If we execute the filter subquery in an optimizer rule to form an IN filter while building the main scan, we may not be able to benefit from reusing subqueries and other optimization as it will no longer be one plan. That’s why leveraging the existing framework for runtime filtering seemed so promising. Based on what we discussed, I removed nested commands by adding more information to `ReplaceData`. I think it simplifies the implementation a little bit. In the future, `ReplaceData` may also contain a filter subquery that will be assigned in the rewrite plan (just like you mentioned). However, let’s put the runtime filtering out of scope for now. We know we have multiple ways to do it: either evaluate a subquery in an optimizer rule or leverage the runtime filtering framework for DS V2. We want to support a custom scan builder anyway. I can work on that immediately after this PR. What do you think, @cloud-fan? -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
