cloud-fan commented on code in PR #45739: URL: https://github.com/apache/spark/pull/45739#discussion_r1557699328
########## sql/core/src/test/scala/org/apache/spark/sql/TPCDSCollationQueryTestSuite.scala: ########## @@ -0,0 +1,270 @@ +/* + * 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 + +import java.nio.file.{Files, Paths} +import java.util.Locale + +import org.apache.spark.{SparkConf, SparkContext} +import org.apache.spark.sql.catalyst.util.resourceToString +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.test.TestSparkSession +import org.apache.spark.tags.ExtendedSQLTest +import org.apache.spark.util.Utils + +/** + * End-to-end tests to validate TPC-DS query results against collation-aware + * modified data and queries. + * + * For each collation, table schemas are replicated into two databases in such way that in first + * DB all table columns are collated with specified collation, while the second DB collates table + * columns with case-insensitive version of the collation. Tables from first DB are then populated + * with lowercase-converted data from tpc-ds kit and tables from second DB are populated with + * randomized-case data. + * + * When running arbitrary SQL query, we convert the query to lowercase for execution + * against first DB; second DB receives original query. Results should compare equal + * ignoring case. We use this method to validate collations are working with arbitrary standard + * SQL constructs. + * + * Additionally, we perform trims on string data to properly convert it from CharType + * to StringType and do sanity checks to verify that results are non-empty as expected. + * + * To run this test suite: + * {{{ + * SPARK_TPCDS_DATA=<path of TPCDS SF=1 data> + * build/sbt "sql/testOnly *TPCDSCollationQueryTestSuite" + * }}} + * + * To run a single test file upon change: + * {{{ + * SPARK_TPCDS_DATA=<path of TPCDS SF=1 data> + * build/sbt "sql/testOnly *TPCDSCollationQueryTestSuite -- -z q79" + * }}} + */ +@ExtendedSQLTest +class TPCDSCollationQueryTestSuite extends QueryTest with TPCDSBase with SQLQueryTestHelper { + + private val tpcdsDataPath = sys.env.get("SPARK_TPCDS_DATA") + + // To make output results deterministic + override protected def sparkConf: SparkConf = super.sparkConf + .set(SQLConf.SHUFFLE_PARTITIONS.key, "1") + + protected override def createSparkSession: TestSparkSession = { + new TestSparkSession(new SparkContext("local[1]", this.getClass.getSimpleName, sparkConf)) + } + + if (tpcdsDataPath.nonEmpty) { + val nonExistentTables = tableColumns.keys.filterNot { tableName => + Files.exists(Paths.get(s"${tpcdsDataPath.get}/$tableName")) + } + if (nonExistentTables.nonEmpty) { + fail(s"Non-existent TPCDS table paths found in ${tpcdsDataPath.get}: " + + nonExistentTables.mkString(", ")) + } + } + + private def withDB[T](dbName: String)(fun: => T): T = { + Utils.tryWithSafeFinally({ + spark.sql(s"USE `$dbName`") + fun + }) { + spark.sql("USE DEFAULT") + } + } + + abstract class CollationCheck( + val dbName: String, + val collation: String, + val columnTransform: String) { + + def queryTransform: String => String + } + + case class CaseInsensitiveCollationCheck( + override val dbName: String, + override val collation: String, + override val columnTransform: String) + extends CollationCheck(dbName, collation, columnTransform) { + + override def queryTransform: String => String = identity + } + + case class CaseSensitiveCollationCheck( + override val dbName: String, + override val collation: String, + override val columnTransform: String) + extends CollationCheck(dbName, collation, columnTransform) { + + override def queryTransform: String => String = _.toLowerCase(Locale.ROOT) + } + + val collateNormalize = "COLLATE_NORMALIZE" + val randomizeCase = "RANDOMIZE_CASE" + + // List of batches of runs which should yield the same result when run on a query + val checks: Seq[Seq[CollationCheck]] = Seq( + Seq( + CaseSensitiveCollationCheck("tpcds_utf8", "UTF8_BINARY", collateNormalize), + CaseInsensitiveCollationCheck("tpcds_utf8_random", "UTF8_BINARY_LCASE", randomizeCase) + ), + Seq( + CaseSensitiveCollationCheck("tpcds_unicode", "UNICODE", collateNormalize), + CaseInsensitiveCollationCheck("tpcds_unicode_random", "UNICODE_CI", randomizeCase) + ) + ) + + override def createTables(): Unit = { + // trim collated strings with udf Review Comment: there is a `trim/rtrim` function in Spark, and `lower` -- 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]
