Search in sources :

Example 1 with GroupByHashPageIndexerFactory

use of io.trino.operator.GroupByHashPageIndexerFactory in project trino by trinodb.

the class AbstractTestHive method setup.

protected final void setup(String databaseName, HiveConfig hiveConfig, HiveMetastore hiveMetastore, HdfsEnvironment hdfsConfiguration) {
    setupHive(databaseName);
    metastoreClient = hiveMetastore;
    hdfsEnvironment = hdfsConfiguration;
    HivePartitionManager partitionManager = new HivePartitionManager(hiveConfig);
    locationService = new HiveLocationService(hdfsEnvironment);
    JsonCodec<PartitionUpdate> partitionUpdateCodec = JsonCodec.jsonCodec(PartitionUpdate.class);
    metadataFactory = new HiveMetadataFactory(new CatalogName("hive"), HiveMetastoreFactory.ofInstance(metastoreClient), hdfsEnvironment, partitionManager, 10, 10, 10, false, false, false, true, true, false, 1000, Optional.empty(), true, TESTING_TYPE_MANAGER, NOOP_METADATA_PROVIDER, locationService, partitionUpdateCodec, newFixedThreadPool(2), heartbeatService, TEST_SERVER_VERSION, (session, tableHandle) -> {
        if (!tableHandle.getTableName().contains("apply_redirection_tester")) {
            return Optional.empty();
        }
        return Optional.of(new TableScanRedirectApplicationResult(new CatalogSchemaTableName("hive", databaseName, "mock_redirection_target"), ImmutableMap.of(), TupleDomain.all()));
    }, ImmutableSet.of(new PartitionsSystemTableProvider(partitionManager, TESTING_TYPE_MANAGER), new PropertiesSystemTableProvider()), metastore -> new NoneHiveMaterializedViewMetadata() {

        @Override
        public Optional<ConnectorMaterializedViewDefinition> getMaterializedView(ConnectorSession session, SchemaTableName viewName) {
            if (!viewName.getTableName().contains("materialized_view_tester")) {
                return Optional.empty();
            }
            return Optional.of(new ConnectorMaterializedViewDefinition("dummy_view_sql", Optional.empty(), Optional.empty(), Optional.empty(), ImmutableList.of(new ConnectorMaterializedViewDefinition.Column("abc", TypeId.of("type"))), Optional.empty(), Optional.of("alice"), ImmutableMap.of()));
        }
    }, SqlStandardAccessControlMetadata::new, NO_REDIRECTIONS, TableInvalidationCallback.NOOP);
    transactionManager = new HiveTransactionManager(metadataFactory);
    splitManager = new HiveSplitManager(transactionManager, partitionManager, new NamenodeStats(), hdfsEnvironment, new CachingDirectoryLister(hiveConfig), directExecutor(), new CounterStat(), 100, hiveConfig.getMaxOutstandingSplitsSize(), hiveConfig.getMinPartitionBatchSize(), hiveConfig.getMaxPartitionBatchSize(), hiveConfig.getMaxInitialSplits(), hiveConfig.getSplitLoaderConcurrency(), hiveConfig.getMaxSplitsPerSecond(), false, TESTING_TYPE_MANAGER);
    pageSinkProvider = new HivePageSinkProvider(getDefaultHiveFileWriterFactories(hiveConfig, hdfsEnvironment), hdfsEnvironment, PAGE_SORTER, HiveMetastoreFactory.ofInstance(metastoreClient), new GroupByHashPageIndexerFactory(JOIN_COMPILER, BLOCK_TYPE_OPERATORS), TESTING_TYPE_MANAGER, getHiveConfig(), locationService, partitionUpdateCodec, new TestingNodeManager("fake-environment"), new HiveEventClient(), getHiveSessionProperties(hiveConfig), new HiveWriterStats());
    pageSourceProvider = new HivePageSourceProvider(TESTING_TYPE_MANAGER, hdfsEnvironment, hiveConfig, getDefaultHivePageSourceFactories(hdfsEnvironment, hiveConfig), getDefaultHiveRecordCursorProviders(hiveConfig, hdfsEnvironment), new GenericHiveRecordCursorProvider(hdfsEnvironment, hiveConfig), Optional.empty());
    nodePartitioningProvider = new HiveNodePartitioningProvider(new TestingNodeManager("fake-environment"), TESTING_TYPE_MANAGER);
}
Also used : Assertions.assertInstanceOf(io.airlift.testing.Assertions.assertInstanceOf) FileSystem(org.apache.hadoop.fs.FileSystem) Test(org.testng.annotations.Test) FileStatus(org.apache.hadoop.fs.FileStatus) NOT_SUPPORTED(io.trino.spi.StandardErrorCode.NOT_SUPPORTED) TableNotFoundException(io.trino.spi.connector.TableNotFoundException) HiveColumnStatistics.createDateColumnStatistics(io.trino.plugin.hive.metastore.HiveColumnStatistics.createDateColumnStatistics) Files.createTempDirectory(java.nio.file.Files.createTempDirectory) Map(java.util.Map) PRESTO_QUERY_ID_NAME(io.trino.plugin.hive.HiveMetadata.PRESTO_QUERY_ID_NAME) ConnectorPageSource(io.trino.spi.connector.ConnectorPageSource) ViewNotFoundException(io.trino.spi.connector.ViewNotFoundException) MaterializedRow(io.trino.testing.MaterializedRow) ROLLBACK_AFTER_FINISH_INSERT(io.trino.plugin.hive.AbstractTestHive.TransactionDeleteInsertTestTag.ROLLBACK_AFTER_FINISH_INSERT) ENGLISH(java.util.Locale.ENGLISH) Assert.assertFalse(org.testng.Assert.assertFalse) HiveIdentity(io.trino.plugin.hive.authentication.HiveIdentity) Domain(io.trino.spi.predicate.Domain) MANAGED_TABLE(org.apache.hadoop.hive.metastore.TableType.MANAGED_TABLE) ASCENDING(io.trino.plugin.hive.metastore.SortingColumn.Order.ASCENDING) ValueSet(io.trino.spi.predicate.ValueSet) MoreExecutors.directExecutor(com.google.common.util.concurrent.MoreExecutors.directExecutor) COMMIT(io.trino.plugin.hive.AbstractTestHive.TransactionDeleteInsertTestTag.COMMIT) NOT_PARTITIONED(io.trino.spi.connector.NotPartitionedPartitionHandle.NOT_PARTITIONED) Lists.newArrayList(com.google.common.collect.Lists.newArrayList) DESCENDING(io.trino.plugin.hive.metastore.SortingColumn.Order.DESCENDING) ConnectorPartitioningHandle(io.trino.spi.connector.ConnectorPartitioningHandle) TrinoS3ConfigurationInitializer(io.trino.plugin.hive.s3.TrinoS3ConfigurationInitializer) CatalogSchemaTableName(io.trino.spi.connector.CatalogSchemaTableName) MetastoreLocator(io.trino.plugin.hive.metastore.thrift.MetastoreLocator) ROLLBACK_AFTER_SINK_FINISH(io.trino.plugin.hive.AbstractTestHive.TransactionDeleteInsertTestTag.ROLLBACK_AFTER_SINK_FINISH) TableScanRedirectApplicationResult(io.trino.spi.connector.TableScanRedirectApplicationResult) TableColumnsMetadata(io.trino.spi.connector.TableColumnsMetadata) REAL(io.trino.spi.type.RealType.REAL) Partition(io.trino.plugin.hive.metastore.Partition) BUCKETED_BY_PROPERTY(io.trino.plugin.hive.HiveTableProperties.BUCKETED_BY_PROPERTY) ColumnMetadata(io.trino.spi.connector.ColumnMetadata) TIMESTAMP_MILLIS(io.trino.spi.type.TimestampType.TIMESTAMP_MILLIS) LocalDateTime(java.time.LocalDateTime) ConnectorTableMetadata(io.trino.spi.connector.ConnectorTableMetadata) HiveBasicStatistics.createEmptyStatistics(io.trino.plugin.hive.HiveBasicStatistics.createEmptyStatistics) Variable(io.trino.spi.expression.Variable) StorageFormat.fromHiveStorageFormat(io.trino.plugin.hive.metastore.StorageFormat.fromHiveStorageFormat) OptionalLong(java.util.OptionalLong) VARCHAR(io.trino.spi.type.VarcharType.VARCHAR) HiveMetastore(io.trino.plugin.hive.metastore.HiveMetastore) OrcPageSource(io.trino.plugin.hive.orc.OrcPageSource) SEQUENCEFILE(io.trino.plugin.hive.HiveStorageFormat.SEQUENCEFILE) ScheduledExecutorService(java.util.concurrent.ScheduledExecutorService) HIVE_INVALID_PARTITION_VALUE(io.trino.plugin.hive.HiveErrorCode.HIVE_INVALID_PARTITION_VALUE) ImmutableSet.toImmutableSet(com.google.common.collect.ImmutableSet.toImmutableSet) Assertions.assertGreaterThanOrEqual(io.airlift.testing.Assertions.assertGreaterThanOrEqual) ImmutableMultimap(com.google.common.collect.ImmutableMultimap) HiveSessionProperties.isTemporaryStagingDirectoryEnabled(io.trino.plugin.hive.HiveSessionProperties.isTemporaryStagingDirectoryEnabled) AfterClass(org.testng.annotations.AfterClass) HiveAzureConfig(io.trino.plugin.hive.azure.HiveAzureConfig) FileUtils.makePartName(org.apache.hadoop.hive.common.FileUtils.makePartName) MapType(io.trino.spi.type.MapType) ConnectorSplit(io.trino.spi.connector.ConnectorSplit) TRANSACTIONAL(io.trino.plugin.hive.HiveTableProperties.TRANSACTIONAL) SPARK_TABLE_PROVIDER_KEY(io.trino.plugin.hive.util.HiveUtil.SPARK_TABLE_PROVIDER_KEY) ConnectorSplitSource(io.trino.spi.connector.ConnectorSplitSource) IOException(java.io.IOException) Iterables.getOnlyElement(com.google.common.collect.Iterables.getOnlyElement) STAGE_AND_MOVE_TO_TARGET_DIRECTORY(io.trino.plugin.hive.LocationHandle.WriteMode.STAGE_AND_MOVE_TO_TARGET_DIRECTORY) ROLLBACK_AFTER_BEGIN_INSERT(io.trino.plugin.hive.AbstractTestHive.TransactionDeleteInsertTestTag.ROLLBACK_AFTER_BEGIN_INSERT) HostAndPort(com.google.common.net.HostAndPort) CatalogName(io.trino.plugin.base.CatalogName) DOUBLE(io.trino.spi.type.DoubleType.DOUBLE) HIVE_INT(io.trino.plugin.hive.HiveType.HIVE_INT) ConnectorTableProperties(io.trino.spi.connector.ConnectorTableProperties) ConnectorExpression(io.trino.spi.expression.ConnectorExpression) HiveTestUtils.mapType(io.trino.plugin.hive.HiveTestUtils.mapType) ParquetPageSource(io.trino.plugin.hive.parquet.ParquetPageSource) ThriftMetastoreConfig(io.trino.plugin.hive.metastore.thrift.ThriftMetastoreConfig) RCTEXT(io.trino.plugin.hive.HiveStorageFormat.RCTEXT) VarcharType.createVarcharType(io.trino.spi.type.VarcharType.createVarcharType) WriteInfo(io.trino.plugin.hive.LocationService.WriteInfo) HivePrivilege(io.trino.plugin.hive.metastore.HivePrivilegeInfo.HivePrivilege) PARTITION_KEY(io.trino.plugin.hive.HiveColumnHandle.ColumnType.PARTITION_KEY) HiveColumnStatistics.createStringColumnStatistics(io.trino.plugin.hive.metastore.HiveColumnStatistics.createStringColumnStatistics) MoreFiles.deleteRecursively(com.google.common.io.MoreFiles.deleteRecursively) MaterializedResult(io.trino.testing.MaterializedResult) HiveUtil.toPartitionValues(io.trino.plugin.hive.util.HiveUtil.toPartitionValues) NO_RETRIES(io.trino.spi.connector.RetryMode.NO_RETRIES) ConnectorMaterializedViewDefinition(io.trino.spi.connector.ConnectorMaterializedViewDefinition) ICEBERG_TABLE_TYPE_VALUE(io.trino.plugin.hive.util.HiveUtil.ICEBERG_TABLE_TYPE_VALUE) Duration(io.airlift.units.Duration) BUCKETING_V1(io.trino.plugin.hive.util.HiveBucketing.BucketingVersion.BUCKETING_V1) SqlTimestamp(io.trino.spi.type.SqlTimestamp) ICEBERG_TABLE_TYPE_NAME(io.trino.plugin.hive.util.HiveUtil.ICEBERG_TABLE_TYPE_NAME) NOOP_METADATA_PROVIDER(io.trino.spi.connector.MetadataProvider.NOOP_METADATA_PROVIDER) HiveMetastoreFactory(io.trino.plugin.hive.metastore.HiveMetastoreFactory) Preconditions.checkArgument(com.google.common.base.Preconditions.checkArgument) HiveTestUtils.arrayType(io.trino.plugin.hive.HiveTestUtils.arrayType) Block(io.trino.spi.block.Block) HIVE_PARTITION_SCHEMA_MISMATCH(io.trino.plugin.hive.HiveErrorCode.HIVE_PARTITION_SCHEMA_MISMATCH) ConnectorViewDefinition(io.trino.spi.connector.ConnectorViewDefinition) INTEGER(io.trino.spi.type.IntegerType.INTEGER) ROLLBACK_RIGHT_AWAY(io.trino.plugin.hive.AbstractTestHive.TransactionDeleteInsertTestTag.ROLLBACK_RIGHT_AWAY) ImmutableSet(com.google.common.collect.ImmutableSet) BeforeClass(org.testng.annotations.BeforeClass) Collection(java.util.Collection) UUID(java.util.UUID) Assert.assertNotNull(org.testng.Assert.assertNotNull) TrinoAzureConfigurationInitializer(io.trino.plugin.hive.azure.TrinoAzureConfigurationInitializer) HiveColumnStatistics.createDoubleColumnStatistics(io.trino.plugin.hive.metastore.HiveColumnStatistics.createDoubleColumnStatistics) BUCKET_COLUMN_NAME(io.trino.plugin.hive.HiveColumnHandle.BUCKET_COLUMN_NAME) BIGINT(io.trino.spi.type.BigintType.BIGINT) SORTED_BY_PROPERTY(io.trino.plugin.hive.HiveTableProperties.SORTED_BY_PROPERTY) LocalDate(java.time.LocalDate) JsonCodec(io.airlift.json.JsonCodec) IntStream(java.util.stream.IntStream) HiveTestUtils.getDefaultHivePageSourceFactories(io.trino.plugin.hive.HiveTestUtils.getDefaultHivePageSourceFactories) Constraint(io.trino.spi.connector.Constraint) Assert.assertNull(org.testng.Assert.assertNull) OptionalDouble(java.util.OptionalDouble) Assert.assertEquals(org.testng.Assert.assertEquals) OptionalInt(java.util.OptionalInt) Function(java.util.function.Function) HashSet(java.util.HashSet) HYPER_LOG_LOG(io.trino.spi.type.HyperLogLogType.HYPER_LOG_LOG) ImmutableList(com.google.common.collect.ImmutableList) BridgingHiveMetastore(io.trino.plugin.hive.metastore.thrift.BridgingHiveMetastore) TableStatistics(io.trino.spi.statistics.TableStatistics) HiveColumnHandle.createBaseColumn(io.trino.plugin.hive.HiveColumnHandle.createBaseColumn) Math.toIntExact(java.lang.Math.toIntExact) ExecutorService(java.util.concurrent.ExecutorService) ConnectorPageSink(io.trino.spi.connector.ConnectorPageSink) DESC_NULLS_LAST(io.trino.spi.connector.SortOrder.DESC_NULLS_LAST) ORC(io.trino.plugin.hive.HiveStorageFormat.ORC) HiveColumnStatistics.createIntegerColumnStatistics(io.trino.plugin.hive.metastore.HiveColumnStatistics.createIntegerColumnStatistics) UTF_8(java.nio.charset.StandardCharsets.UTF_8) Assert.fail(org.testng.Assert.fail) ConnectorPageSourceProvider(io.trino.spi.connector.ConnectorPageSourceProvider) DateTime(org.joda.time.DateTime) PAGE_SORTER(io.trino.plugin.hive.HiveTestUtils.PAGE_SORTER) Executors.newFixedThreadPool(java.util.concurrent.Executors.newFixedThreadPool) HIVE_STRING(io.trino.plugin.hive.HiveType.HIVE_STRING) Hashing.sha256(com.google.common.hash.Hashing.sha256) HiveTestUtils.getHiveSession(io.trino.plugin.hive.HiveTestUtils.getHiveSession) Collectors.toList(java.util.stream.Collectors.toList) PRESTO_VERSION_NAME(io.trino.plugin.hive.HiveMetadata.PRESTO_VERSION_NAME) Assert.assertTrue(org.testng.Assert.assertTrue) PrincipalPrivileges(io.trino.plugin.hive.metastore.PrincipalPrivileges) ConnectorPageSinkProvider(io.trino.spi.connector.ConnectorPageSinkProvider) ConnectorTransactionHandle(io.trino.spi.connector.ConnectorTransactionHandle) ConnectorSplitManager(io.trino.spi.connector.ConnectorSplitManager) Arrays(java.util.Arrays) NamedTypeSignature(io.trino.spi.type.NamedTypeSignature) USER(io.trino.spi.security.PrincipalType.USER) Maps.uniqueIndex(com.google.common.collect.Maps.uniqueIndex) NO_ACID_TRANSACTION(io.trino.plugin.hive.acid.AcidTransaction.NO_ACID_TRANSACTION) HiveColumnStatistics.createDecimalColumnStatistics(io.trino.plugin.hive.metastore.HiveColumnStatistics.createDecimalColumnStatistics) TypeOperators(io.trino.spi.type.TypeOperators) ROLLBACK_AFTER_APPEND_PAGE(io.trino.plugin.hive.AbstractTestHive.TransactionDeleteInsertTestTag.ROLLBACK_AFTER_APPEND_PAGE) HiveTestUtils.rowType(io.trino.plugin.hive.HiveTestUtils.rowType) TrinoExceptionAssert.assertTrinoExceptionThrownBy(io.trino.testing.assertions.TrinoExceptionAssert.assertTrinoExceptionThrownBy) CharType.createCharType(io.trino.spi.type.CharType.createCharType) BigDecimal(java.math.BigDecimal) TypeId(io.trino.spi.type.TypeId) Sets.difference(com.google.common.collect.Sets.difference) PARQUET(io.trino.plugin.hive.HiveStorageFormat.PARQUET) Column(io.trino.plugin.hive.metastore.Column) Executors.newScheduledThreadPool(java.util.concurrent.Executors.newScheduledThreadPool) ConnectorOutputTableHandle(io.trino.spi.connector.ConnectorOutputTableHandle) ConnectorTableHandle(io.trino.spi.connector.ConnectorTableHandle) ViewColumn(io.trino.spi.connector.ConnectorViewDefinition.ViewColumn) ProjectionApplicationResult(io.trino.spi.connector.ProjectionApplicationResult) Slices.utf8Slice(io.airlift.slice.Slices.utf8Slice) PartitionWithStatistics(io.trino.plugin.hive.metastore.PartitionWithStatistics) SMALLINT(io.trino.spi.type.SmallintType.SMALLINT) HiveTestUtils.getDefaultHiveRecordCursorProviders(io.trino.plugin.hive.HiveTestUtils.getDefaultHiveRecordCursorProviders) ConnectorNodePartitioningProvider(io.trino.spi.connector.ConnectorNodePartitioningProvider) Table(io.trino.plugin.hive.metastore.Table) TestingNodeManager(io.trino.testing.TestingNodeManager) Range(io.trino.spi.predicate.Range) ImmutableList.toImmutableList(com.google.common.collect.ImmutableList.toImmutableList) PARTITIONED_BY_PROPERTY(io.trino.plugin.hive.HiveTableProperties.PARTITIONED_BY_PROPERTY) RcFilePageSource(io.trino.plugin.hive.rcfile.RcFilePageSource) Set(java.util.Set) MILLISECONDS(java.util.concurrent.TimeUnit.MILLISECONDS) SchemaTableName(io.trino.spi.connector.SchemaTableName) SortingProperty(io.trino.spi.connector.SortingProperty) ImmutableMap.toImmutableMap(com.google.common.collect.ImmutableMap.toImmutableMap) HiveColumnStatistics.createBooleanColumnStatistics(io.trino.plugin.hive.metastore.HiveColumnStatistics.createBooleanColumnStatistics) SchemaTablePrefix(io.trino.spi.connector.SchemaTablePrefix) Lists.reverse(com.google.common.collect.Lists.reverse) DATE(io.trino.spi.type.DateType.DATE) MoreObjects.toStringHelper(com.google.common.base.MoreObjects.toStringHelper) HiveColumnHandle.bucketColumnHandle(io.trino.plugin.hive.HiveColumnHandle.bucketColumnHandle) HivePrincipal(io.trino.plugin.hive.metastore.HivePrincipal) ConnectorTableLayout(io.trino.spi.connector.ConnectorTableLayout) ConnectorInsertTableHandle(io.trino.spi.connector.ConnectorInsertTableHandle) Slice(io.airlift.slice.Slice) NullableValue(io.trino.spi.predicate.NullableValue) Page(io.trino.spi.Page) BOOLEAN(io.trino.spi.type.BooleanType.BOOLEAN) MINUTES(java.util.concurrent.TimeUnit.MINUTES) JoinCompiler(io.trino.sql.gen.JoinCompiler) HiveUtil.columnExtraInfo(io.trino.plugin.hive.util.HiveUtil.columnExtraInfo) GroupByHashPageIndexerFactory(io.trino.operator.GroupByHashPageIndexerFactory) Float.floatToRawIntBits(java.lang.Float.floatToRawIntBits) ALLOW_INSECURE(com.google.common.io.RecursiveDeleteOption.ALLOW_INSECURE) UNGROUPED_SCHEDULING(io.trino.spi.connector.ConnectorSplitManager.SplitSchedulingStrategy.UNGROUPED_SCHEDULING) ThreadLocalRandom(java.util.concurrent.ThreadLocalRandom) ColumnHandle(io.trino.spi.connector.ColumnHandle) TEXTFILE(io.trino.plugin.hive.HiveStorageFormat.TEXTFILE) VARBINARY(io.trino.spi.type.VarbinaryType.VARBINARY) HIVE_INVALID_BUCKET_FILES(io.trino.plugin.hive.HiveErrorCode.HIVE_INVALID_BUCKET_FILES) HiveType.toHiveType(io.trino.plugin.hive.HiveType.toHiveType) TestingMetastoreLocator(io.trino.plugin.hive.metastore.thrift.TestingMetastoreLocator) STORAGE_FORMAT_PROPERTY(io.trino.plugin.hive.HiveTableProperties.STORAGE_FORMAT_PROPERTY) GoogleGcsConfigurationInitializer(io.trino.plugin.hive.gcs.GoogleGcsConfigurationInitializer) ConstraintApplicationResult(io.trino.spi.connector.ConstraintApplicationResult) RecordCursor(io.trino.spi.connector.RecordCursor) BlockTypeOperators(io.trino.type.BlockTypeOperators) LongStream(java.util.stream.LongStream) NO_REDIRECTIONS(io.trino.plugin.hive.HiveTableRedirectionsProvider.NO_REDIRECTIONS) DecimalType.createDecimalType(io.trino.spi.type.DecimalType.createDecimalType) HIVE_LONG(io.trino.plugin.hive.HiveType.HIVE_LONG) HiveTestUtils.getTypes(io.trino.plugin.hive.HiveTestUtils.getTypes) TRANSACTION_CONFLICT(io.trino.spi.StandardErrorCode.TRANSACTION_CONFLICT) TESTING_TYPE_MANAGER(io.trino.type.InternalTypeManager.TESTING_TYPE_MANAGER) ConnectorSession(io.trino.spi.connector.ConnectorSession) MoreFutures.getFutureValue(io.airlift.concurrent.MoreFutures.getFutureValue) UTC(org.joda.time.DateTimeZone.UTC) SESSION(io.trino.plugin.hive.HiveTestUtils.SESSION) SqlVarbinary(io.trino.spi.type.SqlVarbinary) DiscretePredicates(io.trino.spi.connector.DiscretePredicates) CharType(io.trino.spi.type.CharType) TableType(org.apache.hadoop.hive.metastore.TableType) CachingHiveMetastore.cachingHiveMetastore(io.trino.plugin.hive.metastore.cache.CachingHiveMetastore.cachingHiveMetastore) TINYINT(io.trino.spi.type.TinyintType.TINYINT) DateTimeTestingUtils.sqlTimestampOf(io.trino.testing.DateTimeTestingUtils.sqlTimestampOf) Assertions.assertThat(org.assertj.core.api.Assertions.assertThat) Assertions.assertGreaterThan(io.airlift.testing.Assertions.assertGreaterThan) HiveColumnStatistics.createBinaryColumnStatistics(io.trino.plugin.hive.metastore.HiveColumnStatistics.createBinaryColumnStatistics) MoreCollectors.onlyElement(com.google.common.collect.MoreCollectors.onlyElement) NoHdfsAuthentication(io.trino.plugin.hive.authentication.NoHdfsAuthentication) QueryAssertions.assertEqualsIgnoreOrder(io.trino.testing.QueryAssertions.assertEqualsIgnoreOrder) HiveGcsConfig(io.trino.plugin.hive.gcs.HiveGcsConfig) Iterables.concat(com.google.common.collect.Iterables.concat) Path(org.apache.hadoop.fs.Path) KILOBYTE(io.airlift.units.DataSize.Unit.KILOBYTE) TIMESTAMP_WITH_TIME_ZONE(io.trino.spi.type.TimestampWithTimeZoneType.TIMESTAMP_WITH_TIME_ZONE) AVRO(io.trino.plugin.hive.HiveStorageFormat.AVRO) StorageFormat(io.trino.plugin.hive.metastore.StorageFormat) RowType(io.trino.spi.type.RowType) HiveWriteUtils.getTableDefaultLocation(io.trino.plugin.hive.util.HiveWriteUtils.getTableDefaultLocation) ImmutableMap(com.google.common.collect.ImmutableMap) Predicate(java.util.function.Predicate) ThriftHiveMetastore(io.trino.plugin.hive.metastore.thrift.ThriftHiveMetastore) HiveSessionProperties.getTemporaryStagingDirectoryPath(io.trino.plugin.hive.HiveSessionProperties.getTemporaryStagingDirectoryPath) TrinoException(io.trino.spi.TrinoException) ArrayType(io.trino.spi.type.ArrayType) MaterializedResult.materializeSourceDataStream(io.trino.testing.MaterializedResult.materializeSourceDataStream) ConnectorBucketNodeMap(io.trino.spi.connector.ConnectorBucketNodeMap) ASC_NULLS_FIRST(io.trino.spi.connector.SortOrder.ASC_NULLS_FIRST) String.format(java.lang.String.format) SqlDate(io.trino.spi.type.SqlDate) Preconditions.checkState(com.google.common.base.Preconditions.checkState) SqlTimestampWithTimeZone(io.trino.spi.type.SqlTimestampWithTimeZone) DataSize(io.airlift.units.DataSize) HdfsContext(io.trino.plugin.hive.HdfsEnvironment.HdfsContext) List(java.util.List) DynamicFilter(io.trino.spi.connector.DynamicFilter) Assignment(io.trino.spi.connector.Assignment) Optional(java.util.Optional) ConnectorMetadata(io.trino.spi.connector.ConnectorMetadata) HivePrivilegeInfo(io.trino.plugin.hive.metastore.HivePrivilegeInfo) SqlStandardAccessControlMetadata(io.trino.plugin.hive.security.SqlStandardAccessControlMetadata) Logger(io.airlift.log.Logger) MetastoreConfig(io.trino.plugin.hive.metastore.MetastoreConfig) Type(io.trino.spi.type.Type) VarcharType.createUnboundedVarcharType(io.trino.spi.type.VarcharType.createUnboundedVarcharType) CounterStat(io.airlift.stats.CounterStat) HashMap(java.util.HashMap) HiveBasicStatistics.createZeroStatistics(io.trino.plugin.hive.HiveBasicStatistics.createZeroStatistics) CSV(io.trino.plugin.hive.HiveStorageFormat.CSV) AtomicReference(java.util.concurrent.atomic.AtomicReference) VarcharType(io.trino.spi.type.VarcharType) HiveColumnStatistics(io.trino.plugin.hive.metastore.HiveColumnStatistics) ROLLBACK_AFTER_DELETE(io.trino.plugin.hive.AbstractTestHive.TransactionDeleteInsertTestTag.ROLLBACK_AFTER_DELETE) Assertions.assertThatThrownBy(org.assertj.core.api.Assertions.assertThatThrownBy) Verify.verify(com.google.common.base.Verify.verify) Assertions.assertLessThanOrEqual(io.airlift.testing.Assertions.assertLessThanOrEqual) Threads.daemonThreadsNamed(io.airlift.concurrent.Threads.daemonThreadsNamed) SemiTransactionalHiveMetastore(io.trino.plugin.hive.metastore.SemiTransactionalHiveMetastore) RecordPageSource(io.trino.spi.connector.RecordPageSource) Objects.requireNonNull(java.util.Objects.requireNonNull) RowFieldName(io.trino.spi.type.RowFieldName) JSON(io.trino.plugin.hive.HiveStorageFormat.JSON) RCBINARY(io.trino.plugin.hive.HiveStorageFormat.RCBINARY) NO_PRIVILEGES(io.trino.plugin.hive.metastore.PrincipalPrivileges.NO_PRIVILEGES) DELTA_LAKE_PROVIDER(io.trino.plugin.hive.util.HiveUtil.DELTA_LAKE_PROVIDER) ColumnStatistics(io.trino.spi.statistics.ColumnStatistics) FieldDereference(io.trino.spi.expression.FieldDereference) HiveTestUtils.getDefaultHiveFileWriterFactories(io.trino.plugin.hive.HiveTestUtils.getDefaultHiveFileWriterFactories) HiveTestUtils.getHiveSessionProperties(io.trino.plugin.hive.HiveTestUtils.getHiveSessionProperties) TupleDomain(io.trino.spi.predicate.TupleDomain) HiveWriteUtils.createDirectory(io.trino.plugin.hive.util.HiveWriteUtils.createDirectory) TestingConnectorSession(io.trino.testing.TestingConnectorSession) HiveS3Config(io.trino.plugin.hive.s3.HiveS3Config) Executors.newCachedThreadPool(java.util.concurrent.Executors.newCachedThreadPool) BUCKET_COUNT_PROPERTY(io.trino.plugin.hive.HiveTableProperties.BUCKET_COUNT_PROPERTY) SortingColumn(io.trino.plugin.hive.metastore.SortingColumn) SECONDS(java.util.concurrent.TimeUnit.SECONDS) REGULAR(io.trino.plugin.hive.HiveColumnHandle.ColumnType.REGULAR) TableScanRedirectApplicationResult(io.trino.spi.connector.TableScanRedirectApplicationResult) CounterStat(io.airlift.stats.CounterStat) ConnectorMaterializedViewDefinition(io.trino.spi.connector.ConnectorMaterializedViewDefinition) CatalogSchemaTableName(io.trino.spi.connector.CatalogSchemaTableName) TestingNodeManager(io.trino.testing.TestingNodeManager) ConnectorSession(io.trino.spi.connector.ConnectorSession) TestingConnectorSession(io.trino.testing.TestingConnectorSession) CatalogSchemaTableName(io.trino.spi.connector.CatalogSchemaTableName) SchemaTableName(io.trino.spi.connector.SchemaTableName) CatalogName(io.trino.plugin.base.CatalogName) GroupByHashPageIndexerFactory(io.trino.operator.GroupByHashPageIndexerFactory)

Example 2 with GroupByHashPageIndexerFactory

use of io.trino.operator.GroupByHashPageIndexerFactory in project trino by trinodb.

the class TestHivePageSink method createPageSink.

private static ConnectorPageSink createPageSink(HiveTransactionHandle transaction, HiveConfig config, HiveMetastore metastore, Path outputPath, HiveWriterStats stats) {
    LocationHandle locationHandle = new LocationHandle(outputPath, outputPath, false, DIRECT_TO_TARGET_NEW_DIRECTORY);
    HiveOutputTableHandle handle = new HiveOutputTableHandle(SCHEMA_NAME, TABLE_NAME, getColumnHandles(), new HivePageSinkMetadata(new SchemaTableName(SCHEMA_NAME, TABLE_NAME), metastore.getTable(SCHEMA_NAME, TABLE_NAME), ImmutableMap.of()), locationHandle, config.getHiveStorageFormat(), config.getHiveStorageFormat(), ImmutableList.of(), Optional.empty(), "test", ImmutableMap.of(), NO_ACID_TRANSACTION, false, false);
    JsonCodec<PartitionUpdate> partitionUpdateCodec = JsonCodec.jsonCodec(PartitionUpdate.class);
    TypeOperators typeOperators = new TypeOperators();
    BlockTypeOperators blockTypeOperators = new BlockTypeOperators(typeOperators);
    HivePageSinkProvider provider = new HivePageSinkProvider(getDefaultHiveFileWriterFactories(config, HDFS_ENVIRONMENT), HDFS_ENVIRONMENT, PAGE_SORTER, HiveMetastoreFactory.ofInstance(metastore), new GroupByHashPageIndexerFactory(new JoinCompiler(typeOperators), blockTypeOperators), TESTING_TYPE_MANAGER, config, new HiveLocationService(HDFS_ENVIRONMENT), partitionUpdateCodec, new TestingNodeManager("fake-environment"), new HiveEventClient(), getHiveSessionProperties(config), stats);
    return provider.createPageSink(transaction, getHiveSession(config), handle);
}
Also used : JoinCompiler(io.trino.sql.gen.JoinCompiler) HivePageSinkMetadata(io.trino.plugin.hive.metastore.HivePageSinkMetadata) SchemaTableName(io.trino.spi.connector.SchemaTableName) BlockTypeOperators(io.trino.type.BlockTypeOperators) TestingNodeManager(io.trino.testing.TestingNodeManager) GroupByHashPageIndexerFactory(io.trino.operator.GroupByHashPageIndexerFactory) TypeOperators(io.trino.spi.type.TypeOperators) BlockTypeOperators(io.trino.type.BlockTypeOperators)

Example 3 with GroupByHashPageIndexerFactory

use of io.trino.operator.GroupByHashPageIndexerFactory in project trino by trinodb.

the class TestDeltaLakePageSink method createPageSink.

private static ConnectorPageSink createPageSink(Path outputPath, DeltaLakeWriterStats stats) {
    HiveTransactionHandle transaction = new HiveTransactionHandle(false);
    DeltaLakeConfig deltaLakeConfig = new DeltaLakeConfig();
    DeltaLakeOutputTableHandle tableHandle = new DeltaLakeOutputTableHandle(SCHEMA_NAME, TABLE_NAME, getColumnHandles(), outputPath.toString(), Optional.of(deltaLakeConfig.getDefaultCheckpointWritingInterval()), true);
    DeltaLakePageSinkProvider provider = new DeltaLakePageSinkProvider(new GroupByHashPageIndexerFactory(new JoinCompiler(new TypeOperators()), new BlockTypeOperators()), HDFS_ENVIRONMENT, JsonCodec.jsonCodec(DataFileInfo.class), stats, deltaLakeConfig, new TestingTypeManager(), new NodeVersion("test-version"));
    return provider.createPageSink(transaction, SESSION, tableHandle);
}
Also used : NodeVersion(io.trino.plugin.hive.NodeVersion) JoinCompiler(io.trino.sql.gen.JoinCompiler) BlockTypeOperators(io.trino.type.BlockTypeOperators) HiveTransactionHandle(io.trino.plugin.hive.HiveTransactionHandle) GroupByHashPageIndexerFactory(io.trino.operator.GroupByHashPageIndexerFactory) TestingTypeManager(io.trino.spi.type.TestingTypeManager) TypeOperators(io.trino.spi.type.TypeOperators) BlockTypeOperators(io.trino.type.BlockTypeOperators)

Example 4 with GroupByHashPageIndexerFactory

use of io.trino.operator.GroupByHashPageIndexerFactory in project trino by trinodb.

the class AbstractTestHiveFileSystem method setup.

protected void setup(String host, int port, String databaseName, boolean s3SelectPushdownEnabled, HdfsConfiguration hdfsConfiguration) {
    database = databaseName;
    table = new SchemaTableName(database, "trino_test_external_fs");
    tableWithHeader = new SchemaTableName(database, "trino_test_external_fs_with_header");
    tableWithHeaderAndFooter = new SchemaTableName(database, "trino_test_external_fs_with_header_and_footer");
    String random = randomUUID().toString().toLowerCase(ENGLISH).replace("-", "");
    temporaryCreateTable = new SchemaTableName(database, "tmp_trino_test_create_" + random);
    config = new HiveConfig().setS3SelectPushdownEnabled(s3SelectPushdownEnabled);
    Optional<HostAndPort> proxy = Optional.ofNullable(System.getProperty("hive.metastore.thrift.client.socks-proxy")).map(HostAndPort::fromString);
    MetastoreLocator metastoreLocator = new TestingMetastoreLocator(proxy, HostAndPort.fromParts(host, port));
    HivePartitionManager hivePartitionManager = new HivePartitionManager(config);
    hdfsEnvironment = new HdfsEnvironment(hdfsConfiguration, new HdfsConfig(), new NoHdfsAuthentication());
    MetastoreConfig metastoreConfig = new MetastoreConfig();
    metastoreClient = new TestingHiveMetastore(new BridgingHiveMetastore(new ThriftHiveMetastore(metastoreLocator, new HiveConfig(), metastoreConfig, new ThriftMetastoreConfig(), hdfsEnvironment, false), new HiveIdentity(getHiveSession(config).getIdentity())), getBasePath(), hdfsEnvironment);
    locationService = new HiveLocationService(hdfsEnvironment);
    JsonCodec<PartitionUpdate> partitionUpdateCodec = JsonCodec.jsonCodec(PartitionUpdate.class);
    metadataFactory = new HiveMetadataFactory(new CatalogName("hive"), config, metastoreConfig, HiveMetastoreFactory.ofInstance(metastoreClient), hdfsEnvironment, hivePartitionManager, newDirectExecutorService(), heartbeatService, TESTING_TYPE_MANAGER, NOOP_METADATA_PROVIDER, locationService, partitionUpdateCodec, new NodeVersion("test_version"), new NoneHiveRedirectionsProvider(), ImmutableSet.of(new PartitionsSystemTableProvider(hivePartitionManager, TESTING_TYPE_MANAGER), new PropertiesSystemTableProvider()), new DefaultHiveMaterializedViewMetadataFactory(), SqlStandardAccessControlMetadata::new, NO_REDIRECTIONS, TableInvalidationCallback.NOOP);
    transactionManager = new HiveTransactionManager(metadataFactory);
    splitManager = new HiveSplitManager(transactionManager, hivePartitionManager, new NamenodeStats(), hdfsEnvironment, new CachingDirectoryLister(new HiveConfig()), new BoundedExecutor(executor, config.getMaxSplitIteratorThreads()), new CounterStat(), config.getMaxOutstandingSplits(), config.getMaxOutstandingSplitsSize(), config.getMinPartitionBatchSize(), config.getMaxPartitionBatchSize(), config.getMaxInitialSplits(), config.getSplitLoaderConcurrency(), config.getMaxSplitsPerSecond(), config.getRecursiveDirWalkerEnabled(), TESTING_TYPE_MANAGER);
    TypeOperators typeOperators = new TypeOperators();
    BlockTypeOperators blockTypeOperators = new BlockTypeOperators(typeOperators);
    pageSinkProvider = new HivePageSinkProvider(getDefaultHiveFileWriterFactories(config, hdfsEnvironment), hdfsEnvironment, PAGE_SORTER, HiveMetastoreFactory.ofInstance(metastoreClient), new GroupByHashPageIndexerFactory(new JoinCompiler(typeOperators), blockTypeOperators), TESTING_TYPE_MANAGER, config, locationService, partitionUpdateCodec, new TestingNodeManager("fake-environment"), new HiveEventClient(), getHiveSessionProperties(config), new HiveWriterStats());
    pageSourceProvider = new HivePageSourceProvider(TESTING_TYPE_MANAGER, hdfsEnvironment, config, getDefaultHivePageSourceFactories(hdfsEnvironment, config), getDefaultHiveRecordCursorProviders(config, hdfsEnvironment), new GenericHiveRecordCursorProvider(hdfsEnvironment, config), Optional.empty());
    onSetupComplete();
}
Also used : TestingMetastoreLocator(io.trino.plugin.hive.metastore.thrift.TestingMetastoreLocator) CounterStat(io.airlift.stats.CounterStat) MetastoreLocator(io.trino.plugin.hive.metastore.thrift.MetastoreLocator) TestingMetastoreLocator(io.trino.plugin.hive.metastore.thrift.TestingMetastoreLocator) ThriftMetastoreConfig(io.trino.plugin.hive.metastore.thrift.ThriftMetastoreConfig) NoHdfsAuthentication(io.trino.plugin.hive.authentication.NoHdfsAuthentication) HiveIdentity(io.trino.plugin.hive.authentication.HiveIdentity) HostAndPort(com.google.common.net.HostAndPort) BlockTypeOperators(io.trino.type.BlockTypeOperators) TestingNodeManager(io.trino.testing.TestingNodeManager) BridgingHiveMetastore(io.trino.plugin.hive.metastore.thrift.BridgingHiveMetastore) TypeOperators(io.trino.spi.type.TypeOperators) BlockTypeOperators(io.trino.type.BlockTypeOperators) JoinCompiler(io.trino.sql.gen.JoinCompiler) ThriftMetastoreConfig(io.trino.plugin.hive.metastore.thrift.ThriftMetastoreConfig) MetastoreConfig(io.trino.plugin.hive.metastore.MetastoreConfig) ThriftHiveMetastore(io.trino.plugin.hive.metastore.thrift.ThriftHiveMetastore) SchemaTableName(io.trino.spi.connector.SchemaTableName) BoundedExecutor(io.airlift.concurrent.BoundedExecutor) CatalogName(io.trino.plugin.base.CatalogName) GroupByHashPageIndexerFactory(io.trino.operator.GroupByHashPageIndexerFactory)

Aggregations

GroupByHashPageIndexerFactory (io.trino.operator.GroupByHashPageIndexerFactory)4 TypeOperators (io.trino.spi.type.TypeOperators)3 JoinCompiler (io.trino.sql.gen.JoinCompiler)3 BlockTypeOperators (io.trino.type.BlockTypeOperators)3 HostAndPort (com.google.common.net.HostAndPort)2 CounterStat (io.airlift.stats.CounterStat)2 CatalogName (io.trino.plugin.base.CatalogName)2 SchemaTableName (io.trino.spi.connector.SchemaTableName)2 TestingNodeManager (io.trino.testing.TestingNodeManager)2 MoreObjects.toStringHelper (com.google.common.base.MoreObjects.toStringHelper)1 Preconditions.checkArgument (com.google.common.base.Preconditions.checkArgument)1 Preconditions.checkState (com.google.common.base.Preconditions.checkState)1 Verify.verify (com.google.common.base.Verify.verify)1 ImmutableList (com.google.common.collect.ImmutableList)1 ImmutableList.toImmutableList (com.google.common.collect.ImmutableList.toImmutableList)1 ImmutableMap (com.google.common.collect.ImmutableMap)1 ImmutableMap.toImmutableMap (com.google.common.collect.ImmutableMap.toImmutableMap)1 ImmutableMultimap (com.google.common.collect.ImmutableMultimap)1 ImmutableSet (com.google.common.collect.ImmutableSet)1 ImmutableSet.toImmutableSet (com.google.common.collect.ImmutableSet.toImmutableSet)1