Examples with RCBINARY io.trino.plugin.hive.HiveStorageFormat.RCBINARY used on opensource projects

Example 1 with RCBINARY

use of io.trino.plugin.hive.HiveStorageFormat.RCBINARY in project trino by trinodb.

the class TestHiveFileFormats method testRCBinaryProjectedColumns.

@Test(dataProvider = "rowCount")
public void testRCBinaryProjectedColumns(int rowCount) throws Exception {
    // RCBinary does not support complex type as key of a map and interprets empty VARCHAR as nulls
    List<TestColumn> supportedColumns = TEST_COLUMNS.stream().filter(testColumn -> {
        String name = testColumn.getName();
        return !name.equals("t_map_null_key_complex_key_value") && !name.equals("t_empty_varchar");
    }).collect(toList());
    List<TestColumn> regularColumns = getRegularColumns(supportedColumns);
    List<TestColumn> partitionColumns = getPartitionColumns(supportedColumns);
    // Created projected columns for all regular supported columns
    ImmutableList.Builder<TestColumn> writeColumnsBuilder = ImmutableList.builder();
    ImmutableList.Builder<TestColumn> readeColumnsBuilder = ImmutableList.builder();
    generateProjectedColumns(regularColumns, writeColumnsBuilder, readeColumnsBuilder);
    List<TestColumn> writeColumns = writeColumnsBuilder.addAll(partitionColumns).build();
    List<TestColumn> readColumns = readeColumnsBuilder.addAll(partitionColumns).build();
    assertThatFileFormat(RCBINARY).withWriteColumns(writeColumns).withReadColumns(readColumns).withRowsCount(rowCount).withFileWriterFactory(new RcFileFileWriterFactory(HDFS_ENVIRONMENT, TESTING_TYPE_MANAGER, new NodeVersion("test"), HIVE_STORAGE_TIME_ZONE, STATS)).isReadableByPageSource(new RcFilePageSourceFactory(TESTING_TYPE_MANAGER, HDFS_ENVIRONMENT, STATS, new HiveConfig()));
}

Example 2 with RCBINARY

use of io.trino.plugin.hive.HiveStorageFormat.RCBINARY in project trino by trinodb.

the class AbstractTestHive method testStorePartitionWithStatistics.

protected void testStorePartitionWithStatistics(List<ColumnMetadata> columns, PartitionStatistics statsForAllColumns1, PartitionStatistics statsForAllColumns2, PartitionStatistics statsForSubsetOfColumns, PartitionStatistics emptyStatistics) throws Exception {
    SchemaTableName tableName = temporaryTable("store_partition_with_statistics");
    try {
        doCreateEmptyTable(tableName, ORC, columns);
        HiveMetastoreClosure metastoreClient = new HiveMetastoreClosure(getMetastoreClient());
        Table table = metastoreClient.getTable(tableName.getSchemaName(), tableName.getTableName()).orElseThrow(() -> new TableNotFoundException(tableName));
        List<String> partitionValues = ImmutableList.of("2016-01-01");
        String partitionName = makePartName(ImmutableList.of("ds"), partitionValues);
        Partition partition = createDummyPartition(table, partitionName);
        // create partition with stats for all columns
        metastoreClient.addPartitions(tableName.getSchemaName(), tableName.getTableName(), ImmutableList.of(new PartitionWithStatistics(partition, partitionName, statsForAllColumns1)));
        assertEquals(metastoreClient.getPartition(tableName.getSchemaName(), tableName.getTableName(), partitionValues).get().getStorage().getStorageFormat(), fromHiveStorageFormat(ORC));
        assertThat(metastoreClient.getPartitionStatistics(tableName.getSchemaName(), tableName.getTableName(), ImmutableSet.of(partitionName))).isEqualTo(ImmutableMap.of(partitionName, statsForAllColumns1));
        // alter the partition into one with other stats
        Partition modifiedPartition = Partition.builder(partition).withStorage(storage -> storage.setStorageFormat(fromHiveStorageFormat(RCBINARY)).setLocation(partitionTargetPath(tableName, partitionName))).build();
        metastoreClient.alterPartition(tableName.getSchemaName(), tableName.getTableName(), new PartitionWithStatistics(modifiedPartition, partitionName, statsForAllColumns2));
        assertEquals(metastoreClient.getPartition(tableName.getSchemaName(), tableName.getTableName(), partitionValues).get().getStorage().getStorageFormat(), fromHiveStorageFormat(RCBINARY));
        assertThat(metastoreClient.getPartitionStatistics(tableName.getSchemaName(), tableName.getTableName(), ImmutableSet.of(partitionName))).isEqualTo(ImmutableMap.of(partitionName, statsForAllColumns2));
        // alter the partition into one with stats for only subset of columns
        modifiedPartition = Partition.builder(partition).withStorage(storage -> storage.setStorageFormat(fromHiveStorageFormat(TEXTFILE)).setLocation(partitionTargetPath(tableName, partitionName))).build();
        metastoreClient.alterPartition(tableName.getSchemaName(), tableName.getTableName(), new PartitionWithStatistics(modifiedPartition, partitionName, statsForSubsetOfColumns));
        assertThat(metastoreClient.getPartitionStatistics(tableName.getSchemaName(), tableName.getTableName(), ImmutableSet.of(partitionName))).isEqualTo(ImmutableMap.of(partitionName, statsForSubsetOfColumns));
        // alter the partition into one without stats
        modifiedPartition = Partition.builder(partition).withStorage(storage -> storage.setStorageFormat(fromHiveStorageFormat(TEXTFILE)).setLocation(partitionTargetPath(tableName, partitionName))).build();
        metastoreClient.alterPartition(tableName.getSchemaName(), tableName.getTableName(), new PartitionWithStatistics(modifiedPartition, partitionName, emptyStatistics));
        assertThat(metastoreClient.getPartitionStatistics(tableName.getSchemaName(), tableName.getTableName(), ImmutableSet.of(partitionName))).isEqualTo(ImmutableMap.of(partitionName, emptyStatistics));
    } finally {
        dropTable(tableName);
    }
}

Example 3 with RCBINARY

use of io.trino.plugin.hive.HiveStorageFormat.RCBINARY in project trino by trinodb.

the class TestHiveFileFormats method testRCBinaryProjectedColumnsPageSource.

@Test(dataProvider = "rowCount")
public void testRCBinaryProjectedColumnsPageSource(int rowCount) throws Exception {
    // RCBinary does not support complex type as key of a map and interprets empty VARCHAR as nulls
    List<TestColumn> supportedColumns = TEST_COLUMNS.stream().filter(testColumn -> !testColumn.getName().equals("t_empty_varchar")).collect(toList());
    List<TestColumn> regularColumns = getRegularColumns(supportedColumns);
    List<TestColumn> partitionColumns = getPartitionColumns(supportedColumns);
    // Created projected columns for all regular supported columns
    ImmutableList.Builder<TestColumn> writeColumnsBuilder = ImmutableList.builder();
    ImmutableList.Builder<TestColumn> readeColumnsBuilder = ImmutableList.builder();
    generateProjectedColumns(regularColumns, writeColumnsBuilder, readeColumnsBuilder);
    List<TestColumn> writeColumns = writeColumnsBuilder.addAll(partitionColumns).build();
    List<TestColumn> readColumns = readeColumnsBuilder.addAll(partitionColumns).build();
    assertThatFileFormat(RCBINARY).withWriteColumns(writeColumns).withReadColumns(readColumns).withRowsCount(rowCount).withFileWriterFactory(new RcFileFileWriterFactory(HDFS_ENVIRONMENT, TESTING_TYPE_MANAGER, new NodeVersion("test"), HIVE_STORAGE_TIME_ZONE, STATS)).isReadableByPageSource(new RcFilePageSourceFactory(TESTING_TYPE_MANAGER, HDFS_ENVIRONMENT, STATS, new HiveConfig()));
}