Examples with Partition io.trino.plugin.hive.metastore.Partition used on opensource projects

Example 21 with Partition

use of io.trino.plugin.hive.metastore.Partition in project trino by trinodb.

the class TestProtoUtils method testPartition.

@Test
public void testPartition() {
    alluxio.grpc.table.layout.hive.PartitionInfo.Builder partitionInfo = TestingAlluxioMetastoreObjects.getTestingPartitionInfo();
    Partition partition = ProtoUtils.fromProto(partitionInfo.build());
    assertEquals(partitionInfo.getDataColsList().stream().map(ProtoUtils::fromProto).collect(Collectors.toList()), partition.getColumns());
    assertEquals(partitionInfo.getDbName(), partition.getDatabaseName());
    assertEquals(partitionInfo.getParametersMap(), partition.getParameters());
    assertEquals(partitionInfo.getValuesList(), partition.getValues());
    assertEquals(partitionInfo.getTableName(), partition.getTableName());
    Storage s = partition.getStorage();
    alluxio.grpc.table.layout.hive.Storage storage = TestingAlluxioMetastoreObjects.getTestingPartitionInfo().getStorage();
    assertEquals(storage.getSkewed(), s.isSkewed());
    assertEquals(ProtoUtils.fromProto(storage.getStorageFormat()), s.getStorageFormat());
    assertEquals(storage.getLocation(), s.getLocation());
    assertEquals(ProtoUtils.fromProto(partitionInfo.getParametersMap(), storage.getBucketProperty()), s.getBucketProperty());
    assertEquals(storage.getStorageFormat().getSerdelibParametersMap(), s.getSerdeParameters());
}

Example 22 with Partition

use of io.trino.plugin.hive.metastore.Partition in project trino by trinodb.

the class AbstractTestHive method listAllDataPaths.

public static List<String> listAllDataPaths(SemiTransactionalHiveMetastore metastore, String schemaName, String tableName) {
    ImmutableList.Builder<String> locations = ImmutableList.builder();
    Table table = metastore.getTable(schemaName, tableName).get();
    if (table.getStorage().getLocation() != null) {
        // For partitioned table, there should be nothing directly under this directory.
        // But including this location in the set makes the directory content assert more
        // extensive, which is desirable.
        locations.add(table.getStorage().getLocation());
    }
    Optional<List<String>> partitionNames = metastore.getPartitionNames(schemaName, tableName);
    if (partitionNames.isPresent()) {
        metastore.getPartitionsByNames(schemaName, tableName, partitionNames.get()).values().stream().map(Optional::get).map(partition -> partition.getStorage().getLocation()).filter(location -> !location.startsWith(table.getStorage().getLocation())).forEach(locations::add);
    }
    return locations.build();
}

Example 23 with Partition

use of io.trino.plugin.hive.metastore.Partition 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 24 with Partition

use of io.trino.plugin.hive.metastore.Partition in project trino by trinodb.

the class AbstractTestHive method doInsertIntoNewPartition.

private void doInsertIntoNewPartition(HiveStorageFormat storageFormat, SchemaTableName tableName) throws Exception {
    // creating the table
    doCreateEmptyTable(tableName, storageFormat, CREATE_TABLE_COLUMNS_PARTITIONED);
    // insert the data
    String queryId = insertData(tableName, CREATE_TABLE_PARTITIONED_DATA);
    Set<String> existingFiles;
    try (Transaction transaction = newTransaction()) {
        // verify partitions were created
        Table table = metastoreClient.getTable(tableName.getSchemaName(), tableName.getTableName()).orElseThrow(() -> new TableNotFoundException(tableName));
        List<String> partitionNames = transaction.getMetastore().getPartitionNames(tableName.getSchemaName(), tableName.getTableName()).orElseThrow(() -> new AssertionError("Table does not exist: " + tableName));
        assertEqualsIgnoreOrder(partitionNames, CREATE_TABLE_PARTITIONED_DATA.getMaterializedRows().stream().map(row -> "ds=" + row.getField(CREATE_TABLE_PARTITIONED_DATA.getTypes().size() - 1)).collect(toImmutableList()));
        // verify the node versions in partitions
        Map<String, Optional<Partition>> partitions = getMetastoreClient().getPartitionsByNames(table, partitionNames);
        assertEquals(partitions.size(), partitionNames.size());
        for (String partitionName : partitionNames) {
            Partition partition = partitions.get(partitionName).get();
            assertEquals(partition.getParameters().get(PRESTO_VERSION_NAME), TEST_SERVER_VERSION);
            assertEquals(partition.getParameters().get(PRESTO_QUERY_ID_NAME), queryId);
        }
        // load the new table
        ConnectorSession session = newSession();
        ConnectorMetadata metadata = transaction.getMetadata();
        metadata.beginQuery(session);
        ConnectorTableHandle tableHandle = getTableHandle(metadata, tableName);
        List<ColumnHandle> columnHandles = filterNonHiddenColumnHandles(metadata.getColumnHandles(session, tableHandle).values());
        // verify the data
        MaterializedResult result = readTable(transaction, tableHandle, columnHandles, session, TupleDomain.all(), OptionalInt.empty(), Optional.of(storageFormat));
        assertEqualsIgnoreOrder(result.getMaterializedRows(), CREATE_TABLE_PARTITIONED_DATA.getMaterializedRows());
        // test rollback
        existingFiles = listAllDataFiles(transaction, tableName.getSchemaName(), tableName.getTableName());
        assertFalse(existingFiles.isEmpty());
        // test statistics
        for (String partitionName : partitionNames) {
            HiveBasicStatistics partitionStatistics = getBasicStatisticsForPartition(transaction, tableName, partitionName);
            assertEquals(partitionStatistics.getRowCount().getAsLong(), 1L);
            assertEquals(partitionStatistics.getFileCount().getAsLong(), 1L);
            assertGreaterThan(partitionStatistics.getInMemoryDataSizeInBytes().getAsLong(), 0L);
            assertGreaterThan(partitionStatistics.getOnDiskDataSizeInBytes().getAsLong(), 0L);
        }
    }
    Path stagingPathRoot;
    try (Transaction transaction = newTransaction()) {
        ConnectorSession session = newSession();
        ConnectorMetadata metadata = transaction.getMetadata();
        ConnectorTableHandle tableHandle = getTableHandle(metadata, tableName);
        // "stage" insert data
        ConnectorInsertTableHandle insertTableHandle = metadata.beginInsert(session, tableHandle, ImmutableList.of(), NO_RETRIES);
        stagingPathRoot = getStagingPathRoot(insertTableHandle);
        ConnectorPageSink sink = pageSinkProvider.createPageSink(transaction.getTransactionHandle(), session, insertTableHandle);
        sink.appendPage(CREATE_TABLE_PARTITIONED_DATA_2ND.toPage());
        Collection<Slice> fragments = getFutureValue(sink.finish());
        metadata.finishInsert(session, insertTableHandle, fragments, ImmutableList.of());
        // verify all temp files start with the unique prefix
        HdfsContext context = new HdfsContext(session);
        Set<String> tempFiles = listAllDataFiles(context, getStagingPathRoot(insertTableHandle));
        assertTrue(!tempFiles.isEmpty());
        for (String filePath : tempFiles) {
            assertThat(new Path(filePath).getName()).startsWith(session.getQueryId());
        }
        // rollback insert
        transaction.rollback();
    }
    // verify the data is unchanged
    try (Transaction transaction = newTransaction()) {
        ConnectorSession session = newSession();
        ConnectorMetadata metadata = transaction.getMetadata();
        metadata.beginQuery(session);
        ConnectorTableHandle tableHandle = getTableHandle(metadata, tableName);
        List<ColumnHandle> columnHandles = filterNonHiddenColumnHandles(metadata.getColumnHandles(session, tableHandle).values());
        MaterializedResult result = readTable(transaction, tableHandle, columnHandles, session, TupleDomain.all(), OptionalInt.empty(), Optional.empty());
        assertEqualsIgnoreOrder(result.getMaterializedRows(), CREATE_TABLE_PARTITIONED_DATA.getMaterializedRows());
        // verify we did not modify the table directory
        assertEquals(listAllDataFiles(transaction, tableName.getSchemaName(), tableName.getTableName()), existingFiles);
        // verify temp directory is empty
        HdfsContext context = new HdfsContext(session);
        assertTrue(listAllDataFiles(context, stagingPathRoot).isEmpty());
    }
}

Example 25 with Partition

use of io.trino.plugin.hive.metastore.Partition in project trino by trinodb.

the class RegisterPartitionProcedure method doRegisterPartition.

private void doRegisterPartition(ConnectorSession session, ConnectorAccessControl accessControl, String schemaName, String tableName, List<String> partitionColumn, List<String> partitionValues, String location) {
    if (!allowRegisterPartition) {
        throw new TrinoException(PERMISSION_DENIED, "register_partition procedure is disabled");
    }
    SemiTransactionalHiveMetastore metastore = hiveMetadataFactory.create(session.getIdentity(), true).getMetastore();
    HdfsContext hdfsContext = new HdfsContext(session);
    SchemaTableName schemaTableName = new SchemaTableName(schemaName, tableName);
    Table table = metastore.getTable(schemaName, tableName).orElseThrow(() -> new TableNotFoundException(schemaTableName));
    accessControl.checkCanInsertIntoTable(null, schemaTableName);
    checkIsPartitionedTable(table);
    checkPartitionColumns(table, partitionColumn);
    Optional<Partition> partition = metastore.unsafeGetRawHiveMetastoreClosure().getPartition(schemaName, tableName, partitionValues);
    if (partition.isPresent()) {
        String partitionName = FileUtils.makePartName(partitionColumn, partitionValues);
        throw new TrinoException(ALREADY_EXISTS, format("Partition [%s] is already registered with location %s", partitionName, partition.get().getStorage().getLocation()));
    }
    Path partitionLocation;
    if (location == null) {
        partitionLocation = new Path(table.getStorage().getLocation(), FileUtils.makePartName(partitionColumn, partitionValues));
    } else {
        partitionLocation = new Path(location);
    }
    if (!HiveWriteUtils.pathExists(hdfsContext, hdfsEnvironment, partitionLocation)) {
        throw new TrinoException(INVALID_PROCEDURE_ARGUMENT, "Partition location does not exist: " + partitionLocation);
    }
    metastore.addPartition(session, table.getDatabaseName(), table.getTableName(), buildPartitionObject(session, table, partitionValues, partitionLocation), partitionLocation, // no need for failed attempts cleanup
    Optional.empty(), PartitionStatistics.empty(), false);
    metastore.commit();
}