gengliangwang commented on code in PR #55508: URL: https://github.com/apache/spark/pull/55508#discussion_r3133997864
########## sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/ResolveChangelogTable.scala: ########## @@ -0,0 +1,315 @@ +/* + * 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.analysis + +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.{Count, Max, Min} +import org.apache.spark.sql.catalyst.plans.logical._ +import org.apache.spark.sql.catalyst.rules.Rule +import org.apache.spark.sql.catalyst.streaming.StreamingRelationV2 +import org.apache.spark.sql.catalyst.trees.TreeNodeTag +import org.apache.spark.sql.connector.catalog.{Changelog, ChangelogInfo} +import org.apache.spark.sql.errors.QueryCompilationErrors +import org.apache.spark.sql.execution.datasources.v2.{ChangelogTable, DataSourceV2Relation} +import org.apache.spark.sql.types.{IntegerType, StringType} + +/** + * Post-processes a resolved [[ChangelogTable]] read to apply CDC option semantics + * (carry-over removal, update detection) and to enforce supported option combinations. + * + * Fires after [[ResolveRelations]] has wrapped the connector's [[Changelog]] in a + * [[ChangelogTable]]. Both batch ([[DataSourceV2Relation]]) and streaming + * ([[StreamingRelationV2]]) reads are handled: + * - Batch: the requested post-processing passes are injected as logical operators on top + * of the relation. Carry-over removal and update detection are fused into a single + * pass over a (rowId, _commit_version)-partitioned Window: the Filter drops CoW + * carry-over pairs (same rowVersion on both sides) and the subsequent Project relabels + * real delete+insert pairs as update_preimage / update_postimage. + * - Streaming: post-processing is not yet supported. If the requested options would + * require any post-processing, the rule throws an explicit [[AnalysisException]] to + * prevent silent wrong results. Streams that don't require post-processing pass + * through unchanged. + * + * Net change computation (`deduplicationMode = netChanges`) is not yet implemented and + * is rejected up-front for both batch and streaming. + */ +object ResolveChangelogTable extends Rule[LogicalPlan] { + + private val CHANGELOG_TRANSFORMED_TAG = + TreeNodeTag[Boolean]("changelog_transformed") + + private object HelperColumn { + final val DelCnt = "_del_cnt" + final val InsCnt = "_ins_cnt" + final val MinRv = "_min_rv" + final val MaxRv = "_max_rv" + + val all: Set[String] = Set(DelCnt, InsCnt, MinRv, MaxRv) + } + + override def apply(plan: LogicalPlan): LogicalPlan = { + if (isAlreadyTransformed(plan)) return plan + var updatedPlan = plan + updatedPlan = plan.resolveOperatorsUp { + case rel @ DataSourceV2Relation(table: ChangelogTable, _, _, _, _, _) => + val changelog = table.changelog + val req = evaluateRequirements(changelog, table.changelogInfo) + + var updatedRel: LogicalPlan = rel + if (req.requiresCarryOverRemoval || req.requiresUpdateDetection) { + updatedRel = addRowLevelPostProcessing( + rel, changelog, req.requiresCarryOverRemoval, req.requiresUpdateDetection) + } + if (req.requiresNetChanges) { + updatedRel = injectNetChangeComputation(updatedRel, changelog) + } + updatedRel + + case rel @ StreamingRelationV2(_, _, table: ChangelogTable, _, _, _, _, _, _) => + // Streaming CDC reads do not yet apply post-processing. Run the same option / + // capability validation as the batch path so silent wrong results are impossible: + // either no post-processing would be required (fall through, return raw stream), + // or we throw an explicit AnalysisException. + val changelog = table.changelog + val req = evaluateRequirements(changelog, table.changelogInfo) + if (req.needsAny) { + throw QueryCompilationErrors.cdcStreamingPostProcessingNotSupported(changelog.name()) + } + rel + } + if (updatedPlan ne plan) { + updatedPlan.setTagValue(CHANGELOG_TRANSFORMED_TAG, true) Review Comment: The top-level plan node may change. Instead of setting tag on the top-level plan node, shall we marking `ChangelogTable` as resolved? For example, we can have a field `resolved:Boolean = false` in `ChangelogTable` -- 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]
