Search in sources :

Example 1 with HIVE_STRING

use of io.trino.plugin.hive.HiveType.HIVE_STRING in project trino by trinodb.

the class AbstractTestHive method prepareInvalidBuckets.

private void prepareInvalidBuckets(HiveStorageFormat storageFormat, SchemaTableName tableName) throws Exception {
    createEmptyTable(tableName, storageFormat, ImmutableList.of(new Column("id", HIVE_LONG, Optional.empty()), new Column("name", HIVE_STRING, Optional.empty())), ImmutableList.of(), Optional.of(new HiveBucketProperty(ImmutableList.of("id"), BUCKETING_V1, 8, ImmutableList.of())));
    MaterializedResult.Builder dataBuilder = MaterializedResult.resultBuilder(SESSION, BIGINT, VARCHAR);
    for (long id = 0; id < 100; id++) {
        dataBuilder.row(id, String.valueOf(id));
    }
    insertData(tableName, dataBuilder.build());
    try (Transaction transaction = newTransaction()) {
        Set<String> files = listAllDataFiles(transaction, tableName.getSchemaName(), tableName.getTableName());
        Path bucket2 = files.stream().map(Path::new).filter(path -> path.getName().startsWith("000002_0_")).collect(onlyElement());
        Path bucket5 = files.stream().map(Path::new).filter(path -> path.getName().startsWith("000005_0_")).collect(onlyElement());
        HdfsContext context = new HdfsContext(newSession());
        FileSystem fileSystem = hdfsEnvironment.getFileSystem(context, bucket2);
        fileSystem.delete(bucket2, false);
        fileSystem.rename(bucket5, bucket2);
    }
}
Also used : Path(org.apache.hadoop.fs.Path) HiveSessionProperties.getTemporaryStagingDirectoryPath(io.trino.plugin.hive.HiveSessionProperties.getTemporaryStagingDirectoryPath) 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) HiveColumnHandle.createBaseColumn(io.trino.plugin.hive.HiveColumnHandle.createBaseColumn) Column(io.trino.plugin.hive.metastore.Column) ViewColumn(io.trino.spi.connector.ConnectorViewDefinition.ViewColumn) SortingColumn(io.trino.plugin.hive.metastore.SortingColumn) FileSystem(org.apache.hadoop.fs.FileSystem) HdfsContext(io.trino.plugin.hive.HdfsEnvironment.HdfsContext) MaterializedResult(io.trino.testing.MaterializedResult)

Example 2 with HIVE_STRING

use of io.trino.plugin.hive.HiveType.HIVE_STRING in project trino by trinodb.

the class TrinoHiveCatalog method createMaterializedView.

@Override
public void createMaterializedView(ConnectorSession session, SchemaTableName schemaViewName, ConnectorMaterializedViewDefinition definition, boolean replace, boolean ignoreExisting) {
    Optional<io.trino.plugin.hive.metastore.Table> existing = metastore.getTable(schemaViewName.getSchemaName(), schemaViewName.getTableName());
    // It's a create command where the materialized view already exists and 'if not exists' clause is not specified
    if (!replace && existing.isPresent()) {
        if (ignoreExisting) {
            return;
        }
        throw new TrinoException(ALREADY_EXISTS, "Materialized view already exists: " + schemaViewName);
    }
    // Generate a storage table name and create a storage table. The properties in the definition are table properties for the
    // storage table as indicated in the materialized view definition.
    String storageTableName = "st_" + randomUUID().toString().replace("-", "");
    Map<String, Object> storageTableProperties = new HashMap<>(definition.getProperties());
    storageTableProperties.putIfAbsent(FILE_FORMAT_PROPERTY, DEFAULT_FILE_FORMAT_DEFAULT);
    SchemaTableName storageTable = new SchemaTableName(schemaViewName.getSchemaName(), storageTableName);
    List<ColumnMetadata> columns = definition.getColumns().stream().map(column -> new ColumnMetadata(column.getName(), typeManager.getType(column.getType()))).collect(toImmutableList());
    ConnectorTableMetadata tableMetadata = new ConnectorTableMetadata(storageTable, columns, storageTableProperties, Optional.empty());
    Transaction transaction = IcebergUtil.newCreateTableTransaction(this, tableMetadata, session);
    transaction.newAppend().commit();
    transaction.commitTransaction();
    // Create a view indicating the storage table
    Map<String, String> viewProperties = ImmutableMap.<String, String>builder().put(PRESTO_QUERY_ID_NAME, session.getQueryId()).put(STORAGE_TABLE, storageTableName).put(PRESTO_VIEW_FLAG, "true").put(TRINO_CREATED_BY, TRINO_CREATED_BY_VALUE).put(TABLE_COMMENT, ICEBERG_MATERIALIZED_VIEW_COMMENT).buildOrThrow();
    Column dummyColumn = new Column("dummy", HIVE_STRING, Optional.empty());
    io.trino.plugin.hive.metastore.Table.Builder tableBuilder = io.trino.plugin.hive.metastore.Table.builder().setDatabaseName(schemaViewName.getSchemaName()).setTableName(schemaViewName.getTableName()).setOwner(isUsingSystemSecurity ? Optional.empty() : Optional.of(session.getUser())).setTableType(VIRTUAL_VIEW.name()).setDataColumns(ImmutableList.of(dummyColumn)).setPartitionColumns(ImmutableList.of()).setParameters(viewProperties).withStorage(storage -> storage.setStorageFormat(VIEW_STORAGE_FORMAT)).withStorage(storage -> storage.setLocation("")).setViewOriginalText(Optional.of(encodeMaterializedViewData(fromConnectorMaterializedViewDefinition(definition)))).setViewExpandedText(Optional.of("/* Presto Materialized View */"));
    io.trino.plugin.hive.metastore.Table table = tableBuilder.build();
    PrincipalPrivileges principalPrivileges = isUsingSystemSecurity ? NO_PRIVILEGES : buildInitialPrivilegeSet(session.getUser());
    if (existing.isPresent() && replace) {
        // drop the current storage table
        String oldStorageTable = existing.get().getParameters().get(STORAGE_TABLE);
        if (oldStorageTable != null) {
            metastore.dropTable(schemaViewName.getSchemaName(), oldStorageTable, true);
        }
        // Replace the existing view definition
        metastore.replaceTable(schemaViewName.getSchemaName(), schemaViewName.getTableName(), table, principalPrivileges);
        return;
    }
    // create the view definition
    metastore.createTable(table, principalPrivileges);
}
Also used : FileSystem(org.apache.hadoop.fs.FileSystem) Throwables.throwIfUnchecked(com.google.common.base.Throwables.throwIfUnchecked) ICEBERG_TABLE_TYPE_VALUE(org.apache.iceberg.BaseMetastoreTableOperations.ICEBERG_TABLE_TYPE_VALUE) TABLE_TYPE_PROP(org.apache.iceberg.BaseMetastoreTableOperations.TABLE_TYPE_PROP) NOT_SUPPORTED(io.trino.spi.StandardErrorCode.NOT_SUPPORTED) TableNotFoundException(io.trino.spi.connector.TableNotFoundException) Column(io.trino.plugin.hive.metastore.Column) TABLE_NOT_FOUND(io.trino.spi.StandardErrorCode.TABLE_NOT_FOUND) Duration(java.time.Duration) Map(java.util.Map) ViewNotFoundException(io.trino.spi.connector.ViewNotFoundException) TABLE_COMMENT(io.trino.plugin.hive.HiveMetadata.TABLE_COMMENT) HdfsEnvironment(io.trino.plugin.hive.HdfsEnvironment) BaseTable(org.apache.iceberg.BaseTable) ImmutableList.toImmutableList(com.google.common.collect.ImmutableList.toImmutableList) Schema(org.apache.iceberg.Schema) IcebergUtil.validateTableCanBeDropped(io.trino.plugin.iceberg.IcebergUtil.validateTableCanBeDropped) SchemaTableName(io.trino.spi.connector.SchemaTableName) HiveMetadata(io.trino.plugin.hive.HiveMetadata) Stream(java.util.stream.Stream) TrinoPrincipal(io.trino.spi.security.TrinoPrincipal) STORAGE_TABLE(io.trino.plugin.hive.HiveMetadata.STORAGE_TABLE) CatalogSchemaTableName(io.trino.spi.connector.CatalogSchemaTableName) CatalogUtil.dropTableData(org.apache.iceberg.CatalogUtil.dropTableData) HivePrincipal(io.trino.plugin.hive.metastore.HivePrincipal) HiveUtil(io.trino.plugin.hive.util.HiveUtil) HiveViewNotSupportedException(io.trino.plugin.hive.HiveViewNotSupportedException) IcebergMaterializedViewDefinition(io.trino.plugin.iceberg.IcebergMaterializedViewDefinition) ColumnMetadata(io.trino.spi.connector.ColumnMetadata) ConnectorTableMetadata(io.trino.spi.connector.ConnectorTableMetadata) HIVE_INVALID_METADATA(io.trino.plugin.hive.HiveErrorCode.HIVE_INVALID_METADATA) HiveMetastore(io.trino.plugin.hive.metastore.HiveMetastore) SCHEMA_NOT_EMPTY(io.trino.spi.StandardErrorCode.SCHEMA_NOT_EMPTY) FILE_FORMAT_PROPERTY(io.trino.plugin.iceberg.IcebergTableProperties.FILE_FORMAT_PROPERTY) AbstractTrinoCatalog(io.trino.plugin.iceberg.catalog.AbstractTrinoCatalog) ViewReaderUtil.isPrestoView(io.trino.plugin.hive.ViewReaderUtil.isPrestoView) Table(org.apache.iceberg.Table) IOException(java.io.IOException) ConnectorSession(io.trino.spi.connector.ConnectorSession) CatalogName(io.trino.plugin.base.CatalogName) Failsafe(net.jodah.failsafe.Failsafe) HiveUtil.isHiveSystemSchema(io.trino.plugin.hive.util.HiveUtil.isHiveSystemSchema) ChronoUnit(java.time.temporal.ChronoUnit) ViewReaderUtil.isHiveOrPrestoView(io.trino.plugin.hive.ViewReaderUtil.isHiveOrPrestoView) MaterializedViewNotFoundException(io.trino.spi.connector.MaterializedViewNotFoundException) IcebergSchemaProperties.getSchemaLocation(io.trino.plugin.iceberg.IcebergSchemaProperties.getSchemaLocation) IcebergMaterializedViewDefinition.decodeMaterializedViewData(io.trino.plugin.iceberg.IcebergMaterializedViewDefinition.decodeMaterializedViewData) Database(io.trino.plugin.hive.metastore.Database) HiveSchemaProperties(io.trino.plugin.hive.HiveSchemaProperties) ConnectorMaterializedViewDefinition(io.trino.spi.connector.ConnectorMaterializedViewDefinition) SchemaNotFoundException(io.trino.spi.connector.SchemaNotFoundException) TableMetadata(org.apache.iceberg.TableMetadata) IcebergUtil(io.trino.plugin.iceberg.IcebergUtil) Locale(java.util.Locale) ALREADY_EXISTS(io.trino.spi.StandardErrorCode.ALREADY_EXISTS) Path(org.apache.hadoop.fs.Path) ViewReaderUtil.encodeViewData(io.trino.plugin.hive.ViewReaderUtil.encodeViewData) ConnectorViewDefinition(io.trino.spi.connector.ConnectorViewDefinition) ImmutableSet(com.google.common.collect.ImmutableSet) HiveWriteUtils.getTableDefaultLocation(io.trino.plugin.hive.util.HiveWriteUtils.getTableDefaultLocation) ImmutableMap(com.google.common.collect.ImmutableMap) IcebergTableOperationsProvider(io.trino.plugin.iceberg.catalog.IcebergTableOperationsProvider) ConcurrentHashMap(java.util.concurrent.ConcurrentHashMap) VIEW_STORAGE_FORMAT(io.trino.plugin.hive.metastore.StorageFormat.VIEW_STORAGE_FORMAT) TableAlreadyExistsException(io.trino.plugin.hive.TableAlreadyExistsException) TrinoException(io.trino.spi.TrinoException) RetryPolicy(net.jodah.failsafe.RetryPolicy) Collectors(java.util.stream.Collectors) Preconditions.checkState(com.google.common.base.Preconditions.checkState) PRESTO_VIEW_FLAG(io.trino.plugin.hive.ViewReaderUtil.PRESTO_VIEW_FLAG) INVALID_SCHEMA_PROPERTY(io.trino.spi.StandardErrorCode.INVALID_SCHEMA_PROPERTY) HdfsContext(io.trino.plugin.hive.HdfsEnvironment.HdfsContext) List(java.util.List) IcebergUtil.loadIcebergTable(io.trino.plugin.iceberg.IcebergUtil.loadIcebergTable) MetastoreUtil.buildInitialPrivilegeSet(io.trino.plugin.hive.metastore.MetastoreUtil.buildInitialPrivilegeSet) PartitionSpec(org.apache.iceberg.PartitionSpec) Optional(java.util.Optional) Logger(io.airlift.log.Logger) PartitionFields.toPartitionFields(io.trino.plugin.iceberg.PartitionFields.toPartitionFields) ColumnIdentity(io.trino.plugin.iceberg.ColumnIdentity) ViewReaderUtil(io.trino.plugin.hive.ViewReaderUtil) HashMap(java.util.HashMap) IcebergUtil.getIcebergTableWithMetadata(io.trino.plugin.iceberg.IcebergUtil.getIcebergTableWithMetadata) PARTITIONING_PROPERTY(io.trino.plugin.iceberg.IcebergTableProperties.PARTITIONING_PROPERTY) ImmutableList(com.google.common.collect.ImmutableList) Objects.requireNonNull(java.util.Objects.requireNonNull) CachingHiveMetastore(io.trino.plugin.hive.metastore.cache.CachingHiveMetastore) VIRTUAL_VIEW(org.apache.hadoop.hive.metastore.TableType.VIRTUAL_VIEW) IcebergMaterializedViewDefinition.encodeMaterializedViewData(io.trino.plugin.iceberg.IcebergMaterializedViewDefinition.encodeMaterializedViewData) NO_PRIVILEGES(io.trino.plugin.hive.metastore.PrincipalPrivileges.NO_PRIVILEGES) ViewAlreadyExistsException(io.trino.plugin.hive.ViewAlreadyExistsException) HIVE_STRING(io.trino.plugin.hive.HiveType.HIVE_STRING) IcebergMaterializedViewDefinition.fromConnectorMaterializedViewDefinition(io.trino.plugin.iceberg.IcebergMaterializedViewDefinition.fromConnectorMaterializedViewDefinition) UUID.randomUUID(java.util.UUID.randomUUID) Transaction(org.apache.iceberg.Transaction) DEFAULT_FILE_FORMAT_DEFAULT(org.apache.iceberg.TableProperties.DEFAULT_FILE_FORMAT_DEFAULT) PrincipalPrivileges(io.trino.plugin.hive.metastore.PrincipalPrivileges) TypeManager(io.trino.spi.type.TypeManager) ColumnMetadata(io.trino.spi.connector.ColumnMetadata) BaseTable(org.apache.iceberg.BaseTable) Table(org.apache.iceberg.Table) IcebergUtil.loadIcebergTable(io.trino.plugin.iceberg.IcebergUtil.loadIcebergTable) PrincipalPrivileges(io.trino.plugin.hive.metastore.PrincipalPrivileges) ConcurrentHashMap(java.util.concurrent.ConcurrentHashMap) HashMap(java.util.HashMap) SchemaTableName(io.trino.spi.connector.SchemaTableName) CatalogSchemaTableName(io.trino.spi.connector.CatalogSchemaTableName) Transaction(org.apache.iceberg.Transaction) Column(io.trino.plugin.hive.metastore.Column) TrinoException(io.trino.spi.TrinoException) ConnectorTableMetadata(io.trino.spi.connector.ConnectorTableMetadata)

Aggregations

Preconditions.checkState (com.google.common.base.Preconditions.checkState)2 ImmutableList (com.google.common.collect.ImmutableList)2 ImmutableList.toImmutableList (com.google.common.collect.ImmutableList.toImmutableList)2 ImmutableMap (com.google.common.collect.ImmutableMap)2 ImmutableSet (com.google.common.collect.ImmutableSet)2 Logger (io.airlift.log.Logger)2 CatalogName (io.trino.plugin.base.CatalogName)2 HdfsContext (io.trino.plugin.hive.HdfsEnvironment.HdfsContext)2 HIVE_STRING (io.trino.plugin.hive.HiveType.HIVE_STRING)2 Column (io.trino.plugin.hive.metastore.Column)2 HiveMetastore (io.trino.plugin.hive.metastore.HiveMetastore)2 HivePrincipal (io.trino.plugin.hive.metastore.HivePrincipal)2 PrincipalPrivileges (io.trino.plugin.hive.metastore.PrincipalPrivileges)2 NO_PRIVILEGES (io.trino.plugin.hive.metastore.PrincipalPrivileges.NO_PRIVILEGES)2 HiveWriteUtils.getTableDefaultLocation (io.trino.plugin.hive.util.HiveWriteUtils.getTableDefaultLocation)2 NOT_SUPPORTED (io.trino.spi.StandardErrorCode.NOT_SUPPORTED)2 TrinoException (io.trino.spi.TrinoException)2 CatalogSchemaTableName (io.trino.spi.connector.CatalogSchemaTableName)2 ColumnMetadata (io.trino.spi.connector.ColumnMetadata)2 ConnectorMaterializedViewDefinition (io.trino.spi.connector.ConnectorMaterializedViewDefinition)2