1
0

[HUDI-3221] Support querying a table as of a savepoint (#4720)

This commit is contained in:
ForwardXu
2022-03-09 02:02:34 +08:00
committed by GitHub
parent 575bc63468
commit 08fd80c913
16 changed files with 5903 additions and 6 deletions

View File

@@ -20,14 +20,15 @@ package org.apache.spark.sql.hudi.analysis
import org.apache.hudi.DataSourceWriteOptions.MOR_TABLE_TYPE_OPT_VAL
import org.apache.hudi.common.model.HoodieRecord
import org.apache.hudi.common.util.ReflectionUtils
import org.apache.hudi.{HoodieSparkUtils, SparkAdapterSupport}
import org.apache.spark.sql.catalyst.analysis.{UnresolvedAttribute, UnresolvedStar}
import org.apache.hudi.{DataSourceReadOptions, HoodieSparkUtils, SparkAdapterSupport}
import org.apache.spark.sql.catalyst.analysis.{UnresolvedAttribute, UnresolvedRelation, UnresolvedStar}
import org.apache.spark.sql.catalyst.catalog.{CatalogUtils, HoodieCatalogTable}
import org.apache.spark.sql.catalyst.expressions.{Alias, Attribute, AttributeReference, Expression, GenericInternalRow, Literal, NamedExpression}
import org.apache.spark.sql.catalyst.plans.Inner
import org.apache.spark.sql.catalyst.plans.logical._
import org.apache.spark.sql.catalyst.rules.Rule
import org.apache.spark.sql.execution.command._
import org.apache.spark.sql.execution.datasources.{CreateTable, LogicalRelation}
import org.apache.spark.sql.execution.datasources.{CreateTable, DataSource, LogicalRelation}
import org.apache.spark.sql.hudi.HoodieSqlCommonUtils.{getTableIdentifier, removeMetaFields}
import org.apache.spark.sql.hudi.HoodieSqlUtils._
import org.apache.spark.sql.hudi.command._
@@ -113,6 +114,7 @@ case class HoodieAnalysis(sparkSession: SparkSession) extends Rule[LogicalPlan]
case _ =>
l
}
// Convert to CreateHoodieTableAsSelectCommand
case CreateTable(table, mode, Some(query))
if query.resolved && sparkAdapter.isHoodieTable(table) =>
@@ -396,6 +398,37 @@ case class HoodieResolveReferences(sparkSession: SparkSession) extends Rule[Logi
l
}
case l if sparkAdapter.isRelationTimeTravel(l) =>
val (plan: UnresolvedRelation, timestamp, version) =
sparkAdapter.getRelationTimeTravel(l).get
if (timestamp.isEmpty && version.nonEmpty) {
throw new AnalysisException(
"version expression is not supported for time travel")
}
val tableIdentifier = sparkAdapter.toTableIdentifier(plan)
if (sparkAdapter.isHoodieTable(tableIdentifier, sparkSession)) {
val hoodieCatalogTable = HoodieCatalogTable(sparkSession, tableIdentifier)
val table = hoodieCatalogTable.table
val pathOption = table.storage.locationUri.map("path" -> CatalogUtils.URIToString(_))
val instantOption = Map(
DataSourceReadOptions.TIME_TRAVEL_AS_OF_INSTANT.key -> timestamp.get.toString())
val dataSource =
DataSource(
sparkSession,
userSpecifiedSchema = if (table.schema.isEmpty) None else Some(table.schema),
partitionColumns = table.partitionColumnNames,
bucketSpec = table.bucketSpec,
className = table.provider.get,
options = table.storage.properties ++ pathOption ++ instantOption,
catalogTable = Some(table))
LogicalRelation(dataSource.resolveRelation(checkFilesExist = false), table)
} else {
l
}
case p => p
}

View File

@@ -0,0 +1,241 @@
/*
* 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
import org.apache.hudi.HoodieSparkUtils
import org.apache.hudi.common.table.HoodieTableMetaClient
class TestTimeTravelTable extends TestHoodieSqlBase {
test("Test Insert and Update Record with time travel") {
if (HoodieSparkUtils.gteqSpark3_2) {
withTempDir { tmp =>
val tableName1 = generateTableName
spark.sql(
s"""
|create table $tableName1 (
| id int,
| name string,
| price double,
| ts long
|) using hudi
| tblproperties (
| type = 'cow',
| primaryKey = 'id',
| preCombineField = 'ts'
| )
| location '${tmp.getCanonicalPath}/$tableName1'
""".stripMargin)
spark.sql(s"insert into $tableName1 values(1, 'a1', 10, 1000)")
val metaClient1 = HoodieTableMetaClient.builder()
.setBasePath(s"${tmp.getCanonicalPath}/$tableName1")
.setConf(spark.sessionState.newHadoopConf())
.build()
val instant1 = metaClient1.getActiveTimeline.getAllCommitsTimeline
.lastInstant().get().getTimestamp
spark.sql(s"insert into $tableName1 values(1, 'a2', 20, 2000)")
checkAnswer(s"select id, name, price, ts from $tableName1")(
Seq(1, "a2", 20.0, 2000)
)
// time travel from instant1
checkAnswer(
s"select id, name, price, ts from $tableName1 TIMESTAMP AS OF '$instant1'")(
Seq(1, "a1", 10.0, 1000)
)
}
}
}
test("Test Insert Into Records with time travel To new Table") {
if (HoodieSparkUtils.gteqSpark3_2) {
withTempDir { tmp =>
// Create Non-Partitioned table
val tableName1 = generateTableName
spark.sql(
s"""
|create table $tableName1 (
| id int,
| name string,
| price double,
| ts long
|) using hudi
| tblproperties (
| type = 'cow',
| primaryKey = 'id',
| preCombineField = 'ts'
| )
| location '${tmp.getCanonicalPath}/$tableName1'
""".stripMargin)
spark.sql(s"insert into $tableName1 values(1, 'a1', 10, 1000)")
val metaClient1 = HoodieTableMetaClient.builder()
.setBasePath(s"${tmp.getCanonicalPath}/$tableName1")
.setConf(spark.sessionState.newHadoopConf())
.build()
val instant1 = metaClient1.getActiveTimeline.getAllCommitsTimeline
.lastInstant().get().getTimestamp
val tableName2 = generateTableName
// Create a partitioned table
spark.sql(
s"""
|create table $tableName2 (
| id int,
| name string,
| price double,
| ts long,
| dt string
|) using hudi
| tblproperties (primaryKey = 'id')
| partitioned by (dt)
| location '${tmp.getCanonicalPath}/$tableName2'
""".stripMargin)
// Insert into dynamic partition
spark.sql(
s"""
| insert into $tableName2
| select id, name, price, ts, '2022-02-14' as dt
| from $tableName1 TIMESTAMP AS OF '$instant1'
""".stripMargin)
checkAnswer(s"select id, name, price, ts, dt from $tableName2")(
Seq(1, "a1", 10.0, 1000, "2022-02-14")
)
// Insert into static partition
spark.sql(
s"""
| insert into $tableName2 partition(dt = '2022-02-15')
| select 2 as id, 'a2' as name, price, ts
| from $tableName1 TIMESTAMP AS OF '$instant1'
""".stripMargin)
checkAnswer(
s"select id, name, price, ts, dt from $tableName2")(
Seq(1, "a1", 10.0, 1000, "2022-02-14"),
Seq(2, "a2", 10.0, 1000, "2022-02-15")
)
}
}
}
test("Test Two Table's Union Join with time travel") {
if (HoodieSparkUtils.gteqSpark3_2) {
withTempDir { tmp =>
Seq("cow", "mor").foreach { tableType =>
val tableName = generateTableName
val basePath = tmp.getCanonicalPath
val tableName1 = tableName + "_1"
val tableName2 = tableName + "_2"
val path1 = s"$basePath/$tableName1"
val path2 = s"$basePath/$tableName2"
spark.sql(
s"""
|create table $tableName1 (
| id int,
| name string,
| price double,
| ts long
|) using hudi
| tblproperties (
| type = '$tableType',
| primaryKey = 'id',
| preCombineField = 'ts'
| )
| location '$path1'
""".stripMargin)
spark.sql(
s"""
|create table $tableName2 (
| id int,
| name string,
| price double,
| ts long
|) using hudi
| tblproperties (
| type = '$tableType',
| primaryKey = 'id',
| preCombineField = 'ts'
| )
| location '$path2'
""".stripMargin)
spark.sql(s"insert into $tableName1 values(1, 'a1', 10, 1000)")
spark.sql(s"insert into $tableName1 values(2, 'a2', 20, 1000)")
checkAnswer(s"select id, name, price, ts from $tableName1")(
Seq(1, "a1", 10.0, 1000),
Seq(2, "a2", 20.0, 1000)
)
checkAnswer(s"select id, name, price, ts from $tableName1")(
Seq(1, "a1", 10.0, 1000),
Seq(2, "a2", 20.0, 1000)
)
spark.sql(s"insert into $tableName2 values(3, 'a3', 10, 1000)")
spark.sql(s"insert into $tableName2 values(4, 'a4', 20, 1000)")
checkAnswer(s"select id, name, price, ts from $tableName2")(
Seq(3, "a3", 10.0, 1000),
Seq(4, "a4", 20.0, 1000)
)
val metaClient1 = HoodieTableMetaClient.builder()
.setBasePath(path1)
.setConf(spark.sessionState.newHadoopConf())
.build()
val metaClient2 = HoodieTableMetaClient.builder()
.setBasePath(path2)
.setConf(spark.sessionState.newHadoopConf())
.build()
val instant1 = metaClient1.getActiveTimeline.getAllCommitsTimeline
.lastInstant().get().getTimestamp
val instant2 = metaClient2.getActiveTimeline.getAllCommitsTimeline
.lastInstant().get().getTimestamp
val sql =
s"""
|select id, name, price, ts from $tableName1 TIMESTAMP AS OF '$instant1' where id=1
|union
|select id, name, price, ts from $tableName2 TIMESTAMP AS OF '$instant2' where id>1
|""".stripMargin
checkAnswer(sql)(
Seq(1, "a1", 10.0, 1000),
Seq(3, "a3", 10.0, 1000),
Seq(4, "a4", 20.0, 1000)
)
}
}
}
}
}