Github user budde commented on a diff in the pull request:
https://github.com/apache/spark/pull/16944#discussion_r101908105
--- Diff:
sql/hive/src/test/scala/org/apache/spark/sql/hive/HiveSchemaInferenceSuite.scala
---
@@ -0,0 +1,192 @@
+/*
+ * 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.hive
+
+import java.io.File
+import java.util.concurrent.{Executors, TimeUnit}
+
+import org.scalatest.BeforeAndAfterEach
+
+import org.apache.spark.metrics.source.HiveCatalogMetrics
+import org.apache.spark.sql.catalyst.TableIdentifier
+import org.apache.spark.sql.catalyst.catalog._
+import org.apache.spark.sql.execution.datasources.FileStatusCache
+import org.apache.spark.sql.QueryTest
+import org.apache.spark.sql.hive.client.HiveClient
+import org.apache.spark.sql.hive.test.TestHiveSingleton
+import org.apache.spark.sql.internal.SQLConf
+import org.apache.spark.sql.internal.SQLConf.HiveCaseSensitiveInferenceMode
+import org.apache.spark.sql.test.SQLTestUtils
+import org.apache.spark.sql.types._
+
+class HiveSchemaInferenceSuite
+ extends QueryTest with TestHiveSingleton with SQLTestUtils with
BeforeAndAfterEach {
+
+ import HiveSchemaInferenceSuite._
+
+ override def beforeEach(): Unit = {
+ super.beforeEach()
+ FileStatusCache.resetForTesting()
+ }
+
+ override def afterEach(): Unit = {
+ super.afterEach()
+ FileStatusCache.resetForTesting()
+ }
+
+ private val externalCatalog =
spark.sharedState.externalCatalog.asInstanceOf[HiveExternalCatalog]
+ private val lowercaseSchema = StructType(Seq(
+ StructField("fieldone", LongType),
+ StructField("partcol1", IntegerType),
+ StructField("partcol2", IntegerType)))
+ private val caseSensitiveSchema = StructType(Seq(
+ StructField("fieldOne", LongType),
+ // Partition columns remain case-insensitive
+ StructField("partcol1", IntegerType),
+ StructField("partcol2", IntegerType)))
+
+ // Create a CatalogTable instance modeling an external Hive Metastore
table backed by
+ // Parquet data files.
+ private def hiveExternalCatalogTable(
+ tableName: String,
+ location: String,
+ schema: StructType,
+ partitionColumns: Seq[String],
+ properties: Map[String, String] = Map.empty): CatalogTable = {
+ CatalogTable(
+ identifier = TableIdentifier(table = tableName, database =
Option(DATABASE)),
+ tableType = CatalogTableType.EXTERNAL,
+ storage = CatalogStorageFormat(
+ locationUri = Option(location),
+ inputFormat =
Option("org.apache.hadoop.hive.ql.io.parquet.MapredParquetInputFormat"),
+ outputFormat =
Option("org.apache.hadoop.hive.ql.io.parquet.MapredParquetOutputFormat"),
+ serde =
Option("org.apache.hadoop.hive.ql.io.parquet.serde.ParquetHiveSerDe"),
+ compressed = false,
+ properties = Map("serialization.format" -> "1")),
+ schema = schema,
+ provider = Option("hive"),
+ partitionColumnNames = partitionColumns,
+ properties = properties)
+ }
+
+ // Creates CatalogTablePartition instances for adding partitions of data
to our test table.
+ private def hiveCatalogPartition(location: String, index: Int):
CatalogTablePartition
+ = CatalogTablePartition(
+ spec = Map("partcol1" -> index.toString, "partcol2" ->
index.toString),
+ storage = CatalogStorageFormat(
+ locationUri =
Option(s"${location}/partCol1=$index/partCol2=$index/"),
+ inputFormat =
Option("org.apache.hadoop.hive.ql.io.parquet.MapredParquetInputFormat"),
+ outputFormat =
Option("org.apache.hadoop.hive.ql.io.parquet.MapredParquetOutputFormat"),
+ serde =
Option("org.apache.hadoop.hive.ql.io.parquet.serde.ParquetHiveSerDe"),
+ compressed = false,
+ properties = Map("serialization.format" -> "1")))
+
+ // Creates a case-sensitive external Hive table for testing schema
inference options. Table
+ // will not have Spark-specific table properties set.
+ private def setupCaseSensitiveTable(
+ tableName: String,
+ dir: File): Unit = {
+ spark.range(NUM_RECORDS)
+ .selectExpr("id as fieldOne", "id as partCol1", "id as partCol2")
+ .write
+ .partitionBy("partCol1", "partCol2")
+ .mode("overwrite")
+ .parquet(dir.getAbsolutePath)
+
+
+ val client = externalCatalog.client
+
+ val catalogTable = hiveExternalCatalogTable(
+ tableName,
+ dir.getAbsolutePath,
+ lowercaseSchema,
+ Seq("partcol1", "partcol2"))
+ client.createTable(catalogTable, true)
+
+ val partitions = (0 until
NUM_RECORDS).map(hiveCatalogPartition(dir.getAbsolutePath, _)).toSeq
+ client.createPartitions(DATABASE, tableName, partitions, true)
+ }
+
+ // Create a test table used for a single unit test, with data stored in
the specified directory.
+ private def withTestTable(dir: File)(f: File => Unit): Unit = {
+ setupCaseSensitiveTable(TEST_TABLE_NAME, dir)
+ try f(dir) finally spark.sql(s"DROP TABLE IF EXISTS $TEST_TABLE_NAME")
+ }
+
+ private val inferenceKey = SQLConf.HIVE_CASE_SENSITIVE_INFERENCE.key
+
+ test("Schema should be inferred and written to table properties when
INFER_AND_SAVE is " +
+ "specified") {
+ withSQLConf(inferenceKey ->
HiveCaseSensitiveInferenceMode.INFER_AND_SAVE.toString) {
+ withTempDir { dir =>
+ withTestTable(dir) { dir =>
+ assert(spark.sql(FIELD_QUERY).count == NUM_RECORDS)
+ assert(spark.sql(PARTITION_COLUMN_QUERY).count == NUM_RECORDS)
+ // Test that the case-sensitive schema was storied as a table
property after inference
+ assert(spark.sql(SELECT_ALL_QUERY).schema == caseSensitiveSchema)
+
+ // Verify the catalog table now contains the udpated schema and
properties
+ val catalogTable = externalCatalog.getTable(DATABASE,
TEST_TABLE_NAME)
+ assert(catalogTable.schemaPreservesCase == true)
+ assert(catalogTable.schema == caseSensitiveSchema)
+ assert(catalogTable.partitionColumnNames == Seq("partcol1",
"partcol2"))
+ val rawTable = externalCatalog.client.getTable(DATABASE,
TEST_TABLE_NAME)
+
assert(rawTable.properties.contains(HiveExternalCatalog.DATASOURCE_SCHEMA_NUMPARTS))
+ }
+ }
+ }
+ }
+
+ test("Schema should be inferred but not stored when INFER_ONLY is
specified") {
+ withSQLConf(inferenceKey ->
HiveCaseSensitiveInferenceMode.INFER_ONLY.toString) {
+ withTempDir { dir =>
+ withTestTable(dir) { dir =>
+ val originalTable = externalCatalog.getTable(DATABASE,
TEST_TABLE_NAME)
+ val existingSchema = spark.sql(SELECT_ALL_QUERY).schema
+ assert(spark.sql(FIELD_QUERY).count == NUM_RECORDS)
+ assert(spark.sql(PARTITION_COLUMN_QUERY).count == NUM_RECORDS)
+ assert(spark.sql(SELECT_ALL_QUERY).schema == existingSchema)
+ // Catalog table shouldn't be altered
+ assert(externalCatalog.getTable(DATABASE, TEST_TABLE_NAME) ==
originalTable)
+ }
+ }
+ }
+ }
+
+ test("Schema should not be inferred when NEVER_INFER is specified") {
+ withSQLConf(inferenceKey ->
HiveCaseSensitiveInferenceMode.NEVER_INFER.toString) {
+ withTempDir { dir =>
+ withTestTable(dir) { dir =>
+ val originalTable = externalCatalog.getTable(DATABASE,
TEST_TABLE_NAME)
+ // Only check the schema returned by SELECT * as other queries
will break
--- End diff --
As mentioned in #16797 this issue actually won't cause exceptions, at least
for Parquet data. The queries will simply return 0 results due to
ParquetReadSupport using case-sensitive field resolution. If enabled, any
pushed-down filter containing a case-sensitive field will also return 0 results
since the lowercase filter field name won't match the case-sensitive Parquet
column name.
I'll put some thought towards whether this test can be made more robust in
other ways.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
