YannByron commented on a change in pull request #4611: URL: https://github.com/apache/hudi/pull/4611#discussion_r790287042
########## File path: hudi-spark-datasource/hudi-spark3/src/main/scala/org/apache/spark/sql/hudi/analysis/HoodieSpark3Analysis.scala ########## @@ -0,0 +1,187 @@ +/* + * 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.hudi.analysis + +import org.apache.hudi.common.table.HoodieTableMetaClient +import org.apache.hudi.{DefaultSource, SparkAdapterSupport} +import org.apache.spark.sql.catalyst.TableIdentifier +import org.apache.spark.sql.catalyst.analysis.{ResolvedTable, UnresolvedPartitionSpec} +import org.apache.spark.sql.catalyst.expressions.{Alias, Attribute} +import org.apache.spark.sql.catalyst.plans.logical._ +import org.apache.spark.sql.catalyst.rules.Rule +import org.apache.spark.sql.connector.catalog.CatalogV2Implicits.IdentifierHelper +import org.apache.spark.sql.execution.datasources.LogicalRelation +import org.apache.spark.sql.execution.datasources.v2.{DataSourceV2Relation, V2SessionCatalog} +import org.apache.spark.sql.hudi.HoodieSqlCommonUtils.{castIfNeeded, getTableLocation, isHoodieTable, removeMetaFields, tableExistsInPath} +import org.apache.spark.sql.hudi.catalog.{HoodieCatalog, HoodieConfigHelper, HoodieInternalV2Table} +import org.apache.spark.sql.hudi.command.{AlterHoodieTableDropPartitionCommand, ShowHoodieTablePartitionsCommand, TruncateHoodieTableCommand} +import org.apache.spark.sql.hudi.HoodieSqlCommonUtils +import org.apache.spark.sql.types.StructType +import org.apache.spark.sql.{AnalysisException, SQLContext, SparkSession} + +/** + * Rule for convert the logical plan to command. + * @param sparkSession + */ +case class HoodieSpark3Analysis(sparkSession: SparkSession) extends Rule[LogicalPlan] + with SparkAdapterSupport with HoodieConfigHelper { + + override def apply(plan: LogicalPlan): LogicalPlan = plan.resolveOperatorsDown { + case dsv2 @ DataSourceV2Relation(d: HoodieInternalV2Table, _, _, _, _) => + val output = dsv2.output + val catalogTable = if (d.catalogTable.isDefined) { + Some(d.v1Table) + } else { + None + } + val relation = new DefaultSource().createRelation(new SQLContext(sparkSession), + buildHoodieConfig(d.hoodieCatalogTable)) + LogicalRelation(relation, output, catalogTable, isStreaming = false) + case a @ InsertIntoStatement(r: DataSourceV2Relation, partitionSpec, _, _, _, _) if a.query.resolved && + r.table.isInstanceOf[HoodieInternalV2Table] && + needsSchemaAdjustment(a.query, r.table.asInstanceOf[HoodieInternalV2Table], partitionSpec, r.schema) => + val projection = resolveQueryColumnsByOrdinal(a.query, r.output) + if (projection != a.query) { + a.copy(query = projection) + } else { + a + } + } + + private def needsSchemaAdjustment(query: LogicalPlan, + hoodieTable: HoodieInternalV2Table, + partitionSpec: Map[String, Option[String]], + schema: StructType): Boolean = { + val output = query.output + val queryOutputWithoutMetaFields = removeMetaFields(output) + val partitionFields = hoodieTable.hoodieCatalogTable.partitionFields + val partitionSchema = hoodieTable.hoodieCatalogTable.partitionSchema + val staticPartitionValues = partitionSpec.filter(p => p._2.isDefined).mapValues(_.get) + + assert(staticPartitionValues.isEmpty || + staticPartitionValues.size == partitionSchema.size, + s"Required partition columns is: ${partitionSchema.json}, Current static partitions " + + s"is: ${staticPartitionValues.mkString("," + "")}") + + assert(staticPartitionValues.size + queryOutputWithoutMetaFields.size + == hoodieTable.hoodieCatalogTable.tableSchemaWithoutMetaFields.size, + s"Required select columns count: ${hoodieTable.hoodieCatalogTable.tableSchemaWithoutMetaFields.size}, " + + s"Current select columns(including static partition column) count: " + + s"${staticPartitionValues.size + queryOutputWithoutMetaFields.size}，columns: " + + s"(${(queryOutputWithoutMetaFields.map(_.name) ++ staticPartitionValues.keys).mkString(",")})") + + // static partition insert. + if (staticPartitionValues.nonEmpty) { + // drop partition fields in origin schema to align fields. + schema.dropWhile(p => partitionFields.contains(p.name)) + } + + val existingSchemaOutput = output.take(schema.length) + existingSchemaOutput.map(_.name) != schema.map(_.name) || + existingSchemaOutput.map(_.dataType) != schema.map(_.dataType) + } + + private def resolveQueryColumnsByOrdinal(query: LogicalPlan, + targetAttrs: Seq[Attribute]): LogicalPlan = { + // always add a Cast. it will be removed in the optimizer if it is unnecessary. + val project = query.output.zipWithIndex.map { case (attr, i) => + if (i < targetAttrs.length) { + val targetAttr = targetAttrs(i) + val castAttr = castIfNeeded(attr.withNullability(targetAttr.nullable), targetAttr.dataType, conf) + Alias(castAttr, targetAttr.name)() + } else { + attr + } + } + Project(project, query) + } +} + +/** + * Rule for resolve hoodie's extended syntax or rewrite some logical plan. + * @param sparkSession + */ +case class HoodieSpark3ResolveReferences(sparkSession: SparkSession) extends Rule[LogicalPlan] + with SparkAdapterSupport with HoodieConfigHelper { + + def apply(plan: LogicalPlan): LogicalPlan = plan resolveOperatorsUp { + // Fill schema for Create Table without specify schema info + case c @ CreateV2Table(tableCatalog, tableName, schema, _, properties, _) + if isHoodieTable(properties) => + + val hoodieCatalog = tableCatalog match { + case catalog: HoodieCatalog => catalog + case _ => tableCatalog.asInstanceOf[V2SessionCatalog] + } + val tablePath = getTableLocation(properties, + TableIdentifier(tableName.name(), tableName.namespace().lastOption), sparkSession) + + val tableExistInCatalog = hoodieCatalog.tableExists(tableName) + // Only when the table has not exist in catalog, we need to fill the schema info for creating table. + if (!tableExistInCatalog && tableExistsInPath(tablePath, sparkSession.sessionState.newHadoopConf())) { + val metaClient = HoodieTableMetaClient.builder() + .setBasePath(tablePath) + .setConf(sparkSession.sessionState.newHadoopConf()) + .build() + val tableSchema = HoodieSqlCommonUtils.getTableSqlSchema(metaClient) + if (tableSchema.isDefined && schema.isEmpty) { + // Fill the schema with the schema from the table + c.copy(tableSchema = tableSchema.get) Review comment: i think we don't need to fill the table schema in here. We can move this process to `HoodieCatalog.createHoodieTable`. -- 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]
