spark V2Writes 源码

  • 2022-10-20
  • 浏览 (156)

spark V2Writes 代码


 * 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
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * See the License for the specific language governing permissions and
 * limitations under the License.

package org.apache.spark.sql.execution.datasources.v2

import java.util.UUID

import org.apache.spark.sql.catalyst.expressions.PredicateHelper
import org.apache.spark.sql.catalyst.plans.logical.{AppendData, LogicalPlan, OverwriteByExpression, OverwritePartitionsDynamic, Project, ReplaceData}
import org.apache.spark.sql.catalyst.rules.Rule
import org.apache.spark.sql.catalyst.streaming.InternalOutputModes._
import org.apache.spark.sql.connector.catalog.{SupportsWrite, Table}
import org.apache.spark.sql.connector.expressions.filter.Predicate
import org.apache.spark.sql.connector.write.{LogicalWriteInfoImpl, SupportsDynamicOverwrite, SupportsOverwriteV2, SupportsTruncate, Write, WriteBuilder}
import org.apache.spark.sql.errors.{QueryCompilationErrors, QueryExecutionErrors}
import org.apache.spark.sql.execution.streaming.sources.{MicroBatchWrite, WriteToMicroBatchDataSource}
import org.apache.spark.sql.internal.connector.SupportsStreamingUpdateAsAppend
import org.apache.spark.sql.streaming.OutputMode
import org.apache.spark.sql.types.StructType

 * A rule that constructs logical writes.
object V2Writes extends Rule[LogicalPlan] with PredicateHelper {

  import DataSourceV2Implicits._

  override def apply(plan: LogicalPlan): LogicalPlan = plan transformDown {
    case a @ AppendData(r: DataSourceV2Relation, query, options, _, None) =>
      val writeBuilder = newWriteBuilder(r.table, options, query.schema)
      val write =
      val newQuery = DistributionAndOrderingUtils.prepareQuery(write, query, r.funCatalog)
      a.copy(write = Some(write), query = newQuery)

    case o @ OverwriteByExpression(r: DataSourceV2Relation, deleteExpr, query, options, _, None) =>
      // fail if any filter cannot be converted. correctness depends on removing all matching data.
      val predicates = splitConjunctivePredicates(deleteExpr).flatMap { pred =>
        val predicate = DataSourceV2Strategy.translateFilterV2(pred)
        if (predicate.isEmpty) {
          throw QueryCompilationErrors.cannotTranslateExpressionToSourceFilterError(pred)

      val table = r.table
      val writeBuilder = newWriteBuilder(table, options, query.schema)
      val write = writeBuilder match {
        case builder: SupportsTruncate if isTruncate(predicates) =>
        case builder: SupportsOverwriteV2 if builder.canOverwrite(predicates) =>
        case _ =>
          throw QueryExecutionErrors.overwriteTableByUnsupportedExpressionError(table)

      val newQuery = DistributionAndOrderingUtils.prepareQuery(write, query, r.funCatalog)
      o.copy(write = Some(write), query = newQuery)

    case o @ OverwritePartitionsDynamic(r: DataSourceV2Relation, query, options, _, None) =>
      val table = r.table
      val writeBuilder = newWriteBuilder(table, options, query.schema)
      val write = writeBuilder match {
        case builder: SupportsDynamicOverwrite =>
        case _ =>
          throw QueryExecutionErrors.dynamicPartitionOverwriteUnsupportedByTableError(table)
      val newQuery = DistributionAndOrderingUtils.prepareQuery(write, query, r.funCatalog)
      o.copy(write = Some(write), query = newQuery)

    case WriteToMicroBatchDataSource(
        relation, table, query, queryId, writeOptions, outputMode, Some(batchId)) =>

      val writeBuilder = newWriteBuilder(table, writeOptions, query.schema, queryId)
      val write = buildWriteForMicroBatch(table, writeBuilder, outputMode)
      val microBatchWrite = new MicroBatchWrite(batchId, write.toStreaming)
      val customMetrics = write.supportedCustomMetrics.toSeq
      val funCatalogOpt = relation.flatMap(_.funCatalog)
      val newQuery = DistributionAndOrderingUtils.prepareQuery(write, query, funCatalogOpt)
      WriteToDataSourceV2(relation, microBatchWrite, newQuery, customMetrics)

    case rd @ ReplaceData(r: DataSourceV2Relation, _, query, _, None) =>
      val rowSchema = StructType.fromAttributes(rd.dataInput)
      val writeBuilder = newWriteBuilder(r.table, Map.empty, rowSchema)
      val write =
      val newQuery = DistributionAndOrderingUtils.prepareQuery(write, query, r.funCatalog)
      // project away any metadata columns that could be used for distribution and ordering
      rd.copy(write = Some(write), query = Project(rd.dataInput, newQuery))


  private def buildWriteForMicroBatch(
      table: SupportsWrite,
      writeBuilder: WriteBuilder,
      outputMode: OutputMode): Write = {

    outputMode match {
      case Append =>
      case Complete =>
        // TODO: we should do this check earlier when we have capability API.
 + " does not support Complete mode.")
      case Update =>
 + " does not support Update mode.")

  private def isTruncate(predicates: Array[Predicate]): Boolean = {
    predicates.length == 1 && predicates(0).name().equals("ALWAYS_TRUE")

  private def newWriteBuilder(
      table: Table,
      writeOptions: Map[String, String],
      rowSchema: StructType,
      queryId: String = UUID.randomUUID().toString): WriteBuilder = {

    val info = LogicalWriteInfoImpl(queryId, rowSchema, writeOptions.asOptions)


spark 源码目录


spark AddPartitionExec 源码

spark AlterNamespaceSetPropertiesExec 源码

spark AlterTableExec 源码

spark BatchScanExec 源码

spark CacheTableExec 源码

spark ContinuousScanExec 源码

spark CreateIndexExec 源码

spark CreateNamespaceExec 源码

spark CreateTableExec 源码

spark DataSourceRDD 源码

0  赞