iodone commented on code in PR #3185: URL: https://github.com/apache/incubator-kyuubi/pull/3185#discussion_r941087333
########## externals/kyuubi-spark-sql-engine/src/main/scala/org/apache/spark/kyuubi/SparkSQLLineageParseHelper.scala: ########## @@ -0,0 +1,366 @@ +/* + * 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.kyuubi + +import scala.collection.immutable.ListMap +import scala.util.{Failure, Success, Try} + +import org.apache.spark.sql.SparkSession +import org.apache.spark.sql.catalyst.TableIdentifier +import org.apache.spark.sql.catalyst.catalog.{CatalogStorageFormat, CatalogTable, HiveTableRelation} +import org.apache.spark.sql.catalyst.expressions.{Alias, NamedExpression} +import org.apache.spark.sql.catalyst.plans.logical._ +import org.apache.spark.sql.connector.catalog.{CatalogPlugin, Identifier, TableCatalog} +import org.apache.spark.sql.execution.datasources.LogicalRelation +import org.apache.spark.sql.execution.datasources.v2.{DataSourceV2Relation, DataSourceV2ScanRelation} + +import org.apache.kyuubi.Logging +import org.apache.kyuubi.engine.spark.KyuubiSparkUtil.isSparkVersionAtMost +import org.apache.kyuubi.engine.spark.events.Lineage + +trait LineageParser { + def sparkSession: SparkSession + + def parse(plan: LogicalPlan): Lineage = { + val ret = { + extractColumnsLineage(plan, ListMap()).toList.collect { + case (k, columns) => + k.split('#').head -> columns.map(_.split('#').head) + } + } + val (inputTables, outputTables) = ret.foldLeft((List[String](), List[String]())) { + case ((inputs, outputs), (out, in)) => + val x = (inputs ++ in.map(_.split('.').init.mkString("."))).filter(_.nonEmpty) + val y = outputs ++ List(out.split('.').init.mkString(".")).filter(_.nonEmpty) + (x, y) + } + Lineage(inputTables.distinct, outputTables.distinct, ret) + } + + private def mergeColumnsLineage( + left: ListMap[String, List[String]], + right: ListMap[String, List[String]]): ListMap[String, List[String]] = { + if (left.isEmpty || right.isEmpty) left ++ right + else { + val head = + ListMap(left.head._1 -> (left.head._2 ++ right.getOrElse(left.head._1, List.empty))) + head ++ mergeColumnsLineage(left.tail, right - left.head._1) + } + } + + private def joinColumnsLineage( + parent: ListMap[String, List[String]], + child: ListMap[String, List[String]]): ListMap[String, List[String]] = { + if (parent.isEmpty) child + else { + val childTmp = child.map { case (k, v) => (k.split('#').last, v) } + parent.map { case (k, columns) => + k -> { + columns.flatMap { + case col if childTmp.contains(col.split('#').last) => childTmp(col.split('#').last) + case _ => List() + } + } + } + } + } + + private def extractColumnsLineage( + plan: LogicalPlan, + parentColumnsLineage: ListMap[String, List[String]]): ListMap[String, List[String]] = { + + def getPlanField[T](field: String): T = { + getFieldVal[T](plan, field) + } + + def getCurrentPlanField[T](curPlan: LogicalPlan, field: String): T = { + getFieldVal[T](curPlan, field) + } + + def getPlanMethod[T](name: String): T = { + getMethod[T](plan, name) + } + + def getQuery: LogicalPlan = { + getPlanField[LogicalPlan]("query") + } + + plan match { + // For command + case p if p.nodeName == "AlterViewAsCommand" => + val analyzed = sparkSession.sessionState.analyzer.execute(getQuery) + val view = getPlanField[TableIdentifier]("name").unquotedString + extractColumnsLineage(analyzed, parentColumnsLineage).map { case (k, v) => + (s"$view.$k", v) + } + + case p if p.nodeName == "CreateViewCommand" => + val view = getPlanField[TableIdentifier]("name").unquotedString + val outputCols = + getPlanField[Seq[(String, Option[String])]]("userSpecifiedColumns").map(_._1) + val query = + if (isSparkVersionAtMost("3.1")) { + getPlanField[LogicalPlan]("child") + } else { + getPlanField[LogicalPlan]("plan") + } + val analyzed = sparkSession.sessionState.analyzer.execute(query) + extractColumnsLineage(analyzed, parentColumnsLineage).zipWithIndex.map { + case (x, i) if outputCols.nonEmpty => s"$view.${outputCols(i)}" -> x._2 + case ((k, v), _) => s"$view.$k" -> v + } + + case p if p.nodeName == "CreateDataSourceTableAsSelectCommand" => + val table = getPlanField[CatalogTable]("table").qualifiedName + val analyzed = sparkSession.sessionState.analyzer.execute(getQuery) + extractColumnsLineage(analyzed, parentColumnsLineage).map { case (k, v) => + s"$table.$k" -> v + } + + case p + if p.nodeName == "CreateHiveTableAsSelectCommand" || + p.nodeName == "OptimizedCreateHiveTableAsSelectCommand" => + val table = getPlanField[CatalogTable]("tableDesc").qualifiedName + extractColumnsLineage(getQuery, parentColumnsLineage).map { case (k, v) => + s"$table.$k" -> v + } + + case p + if p.nodeName == "CreateTableAsSelect" || + p.nodeName == "ReplaceTableAsSelect" => + val (table, namespace, catalog) = + if (isSparkVersionAtMost("3.2")) { + ( + getPlanField[Identifier]("tableName").name, + getPlanField[Identifier]("tableName").namespace.mkString("."), + getPlanField[TableCatalog]("catalog").name()) + } else { + ( + getPlanMethod[Identifier]("tableName").name(), + getPlanMethod[Identifier]("tableName").namespace().mkString("."), + getCurrentPlanField[CatalogPlugin]( + getPlanMethod[LogicalPlan]("left"), + "catalog").name()) + } + extractColumnsLineage(getQuery, parentColumnsLineage).map { case (k, v) => + s"$catalog.$namespace.$table.$k" -> v + } + + case p if p.nodeName == "InsertIntoDataSourceCommand" => + val logicalRelation = getPlanField[LogicalRelation]("logicalRelation") + val table = logicalRelation.catalogTable.map(_.qualifiedName).getOrElse("") + extractColumnsLineage(getQuery, parentColumnsLineage).map { case (k, v) => + s"$table.$k" -> v + } + + case p if p.nodeName == "InsertIntoHadoopFsRelationCommand" => + val table = + getPlanField[Option[CatalogTable]]("catalogTable").map(_.qualifiedName).getOrElse("") + extractColumnsLineage(getQuery, parentColumnsLineage).map { case (k, v) => + s"$table.$k" -> v + } + + case p + if p.nodeName == "InsertIntoDataSourceDirCommand" || + p.nodeName == "InsertIntoHiveDirCommand" => + val dir = + getPlanField[CatalogStorageFormat]("storage").locationUri.map(_.toString).getOrElse("") + extractColumnsLineage(getQuery, parentColumnsLineage).map { case (k, v) => + s"`$dir`.$k" -> v + } + + case p if p.nodeName == "InsertIntoHiveTable" => + val table = getPlanField[CatalogTable]("table").qualifiedName + extractColumnsLineage(getQuery, parentColumnsLineage).map { case (k, v) => + s"$table.$k" -> v + } + + case p if p.nodeName == "SaveIntoDataSourceCommand" => + // TODO: Should get the table via datasource options? + extractColumnsLineage(getQuery, parentColumnsLineage) + + // For query + case p: Project => + val nextColumnsLineage = + joinColumnsLineage(parentColumnsLineage, getSelectColumnLineage(p.projectList)) + p.children.map(extractColumnsLineage(_, nextColumnsLineage)).reduce(mergeColumnsLineage) + + case p: Aggregate => + val nextColumnsLineage = + joinColumnsLineage(parentColumnsLineage, getSelectColumnLineage(p.aggregateExpressions)) + p.children.map(extractColumnsLineage(_, nextColumnsLineage)).reduce(mergeColumnsLineage) + + case p: Union => + // merge all children in to one derivedColumns + val childrenUnion = + p.children.map(extractColumnsLineage(_, ListMap())).map(_.values).reduce { + (left, right) => + left.zip(right).map(col => (col._1 ++ col._2).distinct) + } + val childrenColumnsLineage = + ListMap(p.output.map(a => s"${a.name}#${a.exprId.id}").toList.zip(childrenUnion): _*) + joinColumnsLineage(parentColumnsLineage, childrenColumnsLineage) + + case p: SubqueryAlias => + extractColumnsLineage(p.child, parentColumnsLineage) + + case p: LogicalRelation if p.catalogTable.nonEmpty => + val tableName = p.catalogTable.get.qualifiedName + val dataColIds = p.output.map(_.exprId.id.toString).toSet + val dataColNames = p.output.map(_.name) + if (parentColumnsLineage.nonEmpty) { + parentColumnsLineage.map { + case (k, columns) => + k -> columns.collect { + case col if dataColIds.contains(col.split('#').last) => s"$tableName.$col" + } + } + } else { + val colPair = dataColNames.map { col => + col -> List(s"$tableName.$col") + } + ListMap(colPair: _*) + } + + case p: HiveTableRelation => + val tableName = p.tableMeta.qualifiedName + val dataCols = p.dataCols.map(_.exprId.id.toString).toSet + val dataColNames = p.output.map(_.name) + if (parentColumnsLineage.nonEmpty) { + parentColumnsLineage.map { case (k, columns) => + k -> columns.collect { + case col if dataCols.contains(col.split('#').last) => s"$tableName.$col" + } + } + } else { + val colPair = dataColNames.map { col => + col -> List(s"$tableName.$col") + } + ListMap(colPair: _*) + } + + case p: DataSourceV2Relation => + val tableName = p.name + val dataCols = p.output.map(_.exprId.id.toString).toSet + val dataColNames = p.output.map(_.name) + if (parentColumnsLineage.nonEmpty) { + parentColumnsLineage.map { case (k, columns) => + k -> columns.collect { + case col if dataCols.contains(col.split('#').last) => s"$tableName.$col" + } + } + } else { + val colPair = dataColNames.map { col => + col -> List(s"$tableName.$col") + } + ListMap(colPair: _*) + } + + case p: DataSourceV2ScanRelation => + val tableName = p.name + val dataCols = p.output.map(_.exprId.id.toString).toSet + val dataColNames = p.output.map(_.name) + if (parentColumnsLineage.nonEmpty) { + parentColumnsLineage.map { case (k, columns) => + k -> columns.collect { + case col if dataCols.contains(col.split('#').last) => s"$tableName.$col" + } + } + } else { + val colPair = dataColNames.map { col => + col -> List(s"$tableName.$col") + } + ListMap(colPair: _*) + } + + case p: LocalRelation => + val dataCols = p.output.map(_.exprId.id.toString).toSet + val dataColNames = p.output.map(_.name) + if (parentColumnsLineage.nonEmpty) { + parentColumnsLineage.map { case (k, columns) => + k -> columns.collect { + case col if dataCols.contains(col.split('#').last) => col + } + } + } else { + val colPair = dataColNames.map { col => + col -> List(col) + } + ListMap(colPair: _*) + } + + case p if p.children.isEmpty => ListMap.empty + case p => + p.children.map(extractColumnsLineage(_, parentColumnsLineage)).reduce(mergeColumnsLineage) + } + } + + private def getSelectColumnLineage(exps: Seq[NamedExpression]): ListMap[String, List[String]] = { + val result = exps.map { exp => + val derivedColumns = exp.references.map(e => s"${e.name}#${e.exprId.id}").toList + exp match { + case e: Alias => + (s"${e.name}#${e.exprId.id}" -> derivedColumns) + case _ => + (exp.toString -> derivedColumns) + } Review Comment: Thanks for the very good advice, I am ready to change it -- 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]
