Examples with ConnectorPageSink io.trino.spi.connector.ConnectorPageSink used on opensource projects

Example 11 with ConnectorPageSink

use of io.trino.spi.connector.ConnectorPageSink in project trino by trinodb.

the class AbstractTestHive method doTestTransactionDeleteInsert.

private void doTestTransactionDeleteInsert(HiveStorageFormat storageFormat, SchemaTableName tableName, Domain domainToDrop, MaterializedResult insertData, MaterializedResult expectedData, TransactionDeleteInsertTestTag tag, boolean expectQuerySucceed, Optional<ConflictTrigger> conflictTrigger) throws Exception {
    Path writePath = null;
    Path targetPath = null;
    try (Transaction transaction = newTransaction()) {
        try {
            ConnectorMetadata metadata = transaction.getMetadata();
            ConnectorTableHandle tableHandle = getTableHandle(metadata, tableName);
            ConnectorSession session;
            rollbackIfEquals(tag, ROLLBACK_RIGHT_AWAY);
            // Query 1: delete
            session = newSession();
            HiveColumnHandle dsColumnHandle = (HiveColumnHandle) metadata.getColumnHandles(session, tableHandle).get("pk2");
            TupleDomain<ColumnHandle> tupleDomain = TupleDomain.withColumnDomains(ImmutableMap.of(dsColumnHandle, domainToDrop));
            Constraint constraint = new Constraint(tupleDomain, tupleDomain.asPredicate(), tupleDomain.getDomains().orElseThrow().keySet());
            tableHandle = applyFilter(metadata, tableHandle, constraint);
            tableHandle = metadata.applyDelete(session, tableHandle).get();
            metadata.executeDelete(session, tableHandle);
            rollbackIfEquals(tag, ROLLBACK_AFTER_DELETE);
            // Query 2: insert
            session = newSession();
            ConnectorInsertTableHandle insertTableHandle = metadata.beginInsert(session, tableHandle, ImmutableList.of(), NO_RETRIES);
            rollbackIfEquals(tag, ROLLBACK_AFTER_BEGIN_INSERT);
            writePath = getStagingPathRoot(insertTableHandle);
            targetPath = getTargetPathRoot(insertTableHandle);
            ConnectorPageSink sink = pageSinkProvider.createPageSink(transaction.getTransactionHandle(), session, insertTableHandle);
            sink.appendPage(insertData.toPage());
            rollbackIfEquals(tag, ROLLBACK_AFTER_APPEND_PAGE);
            Collection<Slice> fragments = getFutureValue(sink.finish());
            rollbackIfEquals(tag, ROLLBACK_AFTER_SINK_FINISH);
            metadata.finishInsert(session, insertTableHandle, fragments, ImmutableList.of());
            rollbackIfEquals(tag, ROLLBACK_AFTER_FINISH_INSERT);
            assertEquals(tag, COMMIT);
            if (conflictTrigger.isPresent()) {
                JsonCodec<PartitionUpdate> partitionUpdateCodec = JsonCodec.jsonCodec(PartitionUpdate.class);
                List<PartitionUpdate> partitionUpdates = fragments.stream().map(Slice::getBytes).map(partitionUpdateCodec::fromJson).collect(toList());
                conflictTrigger.get().triggerConflict(session, tableName, insertTableHandle, partitionUpdates);
            }
            transaction.commit();
            if (conflictTrigger.isPresent()) {
                assertTrue(expectQuerySucceed);
                conflictTrigger.get().verifyAndCleanup(session, tableName);
            }
        } catch (TestingRollbackException e) {
            transaction.rollback();
        } catch (TrinoException e) {
            assertFalse(expectQuerySucceed);
            if (conflictTrigger.isPresent()) {
                conflictTrigger.get().verifyAndCleanup(newSession(), tableName);
            }
        }
    }
    // check that temporary files are removed
    if (writePath != null && !writePath.equals(targetPath)) {
        HdfsContext context = new HdfsContext(newSession());
        FileSystem fileSystem = hdfsEnvironment.getFileSystem(context, writePath);
        assertFalse(fileSystem.exists(writePath));
    }
    try (Transaction transaction = newTransaction()) {
        // verify partitions
        List<String> partitionNames = transaction.getMetastore().getPartitionNames(tableName.getSchemaName(), tableName.getTableName()).orElseThrow(() -> new AssertionError("Table does not exist: " + tableName));
        assertEqualsIgnoreOrder(partitionNames, expectedData.getMaterializedRows().stream().map(row -> format("pk1=%s/pk2=%s", row.getField(1), row.getField(2))).distinct().collect(toImmutableList()));
        // 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(), expectedData.getMaterializedRows());
    }
}

Example 12 with ConnectorPageSink

use of io.trino.spi.connector.ConnectorPageSink in project trino by trinodb.

the class AbstractTestHive method doTestMismatchSchemaTable.

protected void doTestMismatchSchemaTable(SchemaTableName schemaTableName, HiveStorageFormat storageFormat, List<ColumnMetadata> tableBefore, MaterializedResult dataBefore, List<ColumnMetadata> tableAfter, MaterializedResult dataAfter) throws Exception {
    String schemaName = schemaTableName.getSchemaName();
    String tableName = schemaTableName.getTableName();
    doCreateEmptyTable(schemaTableName, storageFormat, tableBefore);
    // insert the data
    try (Transaction transaction = newTransaction()) {
        ConnectorSession session = newSession();
        ConnectorMetadata metadata = transaction.getMetadata();
        ConnectorTableHandle tableHandle = getTableHandle(metadata, schemaTableName);
        ConnectorInsertTableHandle insertTableHandle = metadata.beginInsert(session, tableHandle, ImmutableList.of(), NO_RETRIES);
        ConnectorPageSink sink = pageSinkProvider.createPageSink(transaction.getTransactionHandle(), session, insertTableHandle);
        sink.appendPage(dataBefore.toPage());
        Collection<Slice> fragments = getFutureValue(sink.finish());
        metadata.finishInsert(session, insertTableHandle, fragments, ImmutableList.of());
        transaction.commit();
    }
    // load the table and verify the data
    try (Transaction transaction = newTransaction()) {
        ConnectorSession session = newSession();
        ConnectorMetadata metadata = transaction.getMetadata();
        metadata.beginQuery(session);
        ConnectorTableHandle tableHandle = getTableHandle(metadata, schemaTableName);
        List<ColumnHandle> columnHandles = metadata.getColumnHandles(session, tableHandle).values().stream().filter(columnHandle -> !((HiveColumnHandle) columnHandle).isHidden()).collect(toList());
        MaterializedResult result = readTable(transaction, tableHandle, columnHandles, session, TupleDomain.all(), OptionalInt.empty(), Optional.empty());
        assertEqualsIgnoreOrder(result.getMaterializedRows(), dataBefore.getMaterializedRows());
        transaction.commit();
    }
    // alter the table schema
    try (Transaction transaction = newTransaction()) {
        ConnectorSession session = newSession();
        PrincipalPrivileges principalPrivileges = testingPrincipalPrivilege(session);
        Table oldTable = transaction.getMetastore().getTable(schemaName, tableName).get();
        List<Column> dataColumns = tableAfter.stream().filter(columnMetadata -> !columnMetadata.getName().equals("ds")).map(columnMetadata -> new Column(columnMetadata.getName(), toHiveType(columnMetadata.getType()), Optional.empty())).collect(toList());
        Table.Builder newTable = Table.builder(oldTable).setDataColumns(dataColumns);
        transaction.getMetastore().replaceTable(schemaName, tableName, newTable.build(), principalPrivileges);
        transaction.commit();
    }
    // load the altered table and verify the data
    try (Transaction transaction = newTransaction()) {
        ConnectorSession session = newSession();
        ConnectorMetadata metadata = transaction.getMetadata();
        metadata.beginQuery(session);
        ConnectorTableHandle tableHandle = getTableHandle(metadata, schemaTableName);
        List<ColumnHandle> columnHandles = metadata.getColumnHandles(session, tableHandle).values().stream().filter(columnHandle -> !((HiveColumnHandle) columnHandle).isHidden()).collect(toList());
        MaterializedResult result = readTable(transaction, tableHandle, columnHandles, session, TupleDomain.all(), OptionalInt.empty(), Optional.empty());
        assertEqualsIgnoreOrder(result.getMaterializedRows(), dataAfter.getMaterializedRows());
        transaction.commit();
    }
    // insertions to the partitions with type mismatches should fail
    try (Transaction transaction = newTransaction()) {
        ConnectorSession session = newSession();
        ConnectorMetadata metadata = transaction.getMetadata();
        ConnectorTableHandle tableHandle = getTableHandle(metadata, schemaTableName);
        ConnectorInsertTableHandle insertTableHandle = metadata.beginInsert(session, tableHandle, ImmutableList.of(), NO_RETRIES);
        ConnectorPageSink sink = pageSinkProvider.createPageSink(transaction.getTransactionHandle(), session, insertTableHandle);
        sink.appendPage(dataAfter.toPage());
        Collection<Slice> fragments = getFutureValue(sink.finish());
        metadata.finishInsert(session, insertTableHandle, fragments, ImmutableList.of());
        transaction.commit();
        fail("expected exception");
    } catch (TrinoException e) {
        // expected
        assertEquals(e.getErrorCode(), HIVE_PARTITION_SCHEMA_MISMATCH.toErrorCode());
    }
}

Example 13 with ConnectorPageSink

use of io.trino.spi.connector.ConnectorPageSink 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 14 with ConnectorPageSink

use of io.trino.spi.connector.ConnectorPageSink in project trino by trinodb.

the class AbstractTestHive method doInsertOverwriteUnpartitioned.

private void doInsertOverwriteUnpartitioned(SchemaTableName tableName) throws Exception {
    // create table with data
    doCreateEmptyTable(tableName, ORC, CREATE_TABLE_COLUMNS);
    insertData(tableName, CREATE_TABLE_DATA);
    // overwrite table with new data
    MaterializedResult.Builder overwriteDataBuilder = MaterializedResult.resultBuilder(SESSION, CREATE_TABLE_DATA.getTypes());
    MaterializedResult overwriteData = null;
    Map<String, Object> overwriteProperties = ImmutableMap.of("insert_existing_partitions_behavior", "OVERWRITE");
    for (int i = 0; i < 3; i++) {
        overwriteDataBuilder.rows(reverse(CREATE_TABLE_DATA.getMaterializedRows()));
        overwriteData = overwriteDataBuilder.build();
        insertData(tableName, overwriteData, overwriteProperties);
        // verify overwrite
        try (Transaction transaction = newTransaction()) {
            ConnectorSession session = newSession();
            ConnectorMetadata metadata = transaction.getMetadata();
            metadata.beginQuery(session);
            // load the new table
            ConnectorTableHandle tableHandle = getTableHandle(metadata, tableName);
            List<ColumnHandle> columnHandles = filterNonHiddenColumnHandles(metadata.getColumnHandles(session, tableHandle).values());
            // verify the metadata
            ConnectorTableMetadata tableMetadata = metadata.getTableMetadata(session, getTableHandle(metadata, tableName));
            assertEquals(filterNonHiddenColumnMetadata(tableMetadata.getColumns()), CREATE_TABLE_COLUMNS);
            // verify the data
            MaterializedResult result = readTable(transaction, tableHandle, columnHandles, session, TupleDomain.all(), OptionalInt.empty(), Optional.empty());
            assertEqualsIgnoreOrder(result.getMaterializedRows(), overwriteData.getMaterializedRows());
            // statistics
            HiveBasicStatistics tableStatistics = getBasicStatisticsForTable(transaction, tableName);
            assertEquals(tableStatistics.getRowCount().getAsLong(), overwriteData.getRowCount());
            assertEquals(tableStatistics.getFileCount().getAsLong(), 1L);
            assertGreaterThan(tableStatistics.getInMemoryDataSizeInBytes().getAsLong(), 0L);
            assertGreaterThan(tableStatistics.getOnDiskDataSizeInBytes().getAsLong(), 0L);
        }
    }
    // test rollback
    Set<String> existingFiles;
    try (Transaction transaction = newTransaction()) {
        existingFiles = listAllDataFiles(transaction, tableName.getSchemaName(), tableName.getTableName());
        assertFalse(existingFiles.isEmpty());
    }
    Path stagingPathRoot;
    try (Transaction transaction = newTransaction()) {
        ConnectorSession session = newSession(overwriteProperties);
        ConnectorMetadata metadata = transaction.getMetadata();
        ConnectorTableHandle tableHandle = getTableHandle(metadata, tableName);
        // "stage" insert data
        ConnectorInsertTableHandle insertTableHandle = metadata.beginInsert(session, tableHandle, ImmutableList.of(), NO_RETRIES);
        ConnectorPageSink sink = pageSinkProvider.createPageSink(transaction.getTransactionHandle(), session, insertTableHandle);
        for (int i = 0; i < 4; i++) {
            sink.appendPage(overwriteData.toPage());
        }
        Collection<Slice> fragments = getFutureValue(sink.finish());
        metadata.finishInsert(session, insertTableHandle, fragments, ImmutableList.of());
        // statistics, visible from within transaction
        HiveBasicStatistics tableStatistics = getBasicStatisticsForTable(transaction, tableName);
        assertEquals(tableStatistics.getRowCount().getAsLong(), overwriteData.getRowCount() * 4L);
        try (Transaction otherTransaction = newTransaction()) {
            // statistics, not visible from outside transaction
            HiveBasicStatistics otherTableStatistics = getBasicStatisticsForTable(otherTransaction, tableName);
            assertEquals(otherTableStatistics.getRowCount().getAsLong(), overwriteData.getRowCount());
        }
        // verify we did not modify the table directory
        assertEquals(listAllDataFiles(transaction, tableName.getSchemaName(), tableName.getTableName()), existingFiles);
        // verify all temp files start with the unique prefix
        stagingPathRoot = getStagingPathRoot(insertTableHandle);
        HdfsContext context = new HdfsContext(session);
        Set<String> tempFiles = listAllDataFiles(context, stagingPathRoot);
        assertTrue(!tempFiles.isEmpty());
        for (String filePath : tempFiles) {
            assertThat(new Path(filePath).getName()).startsWith(session.getQueryId());
        }
        // rollback insert
        transaction.rollback();
    }
    // verify temp directory is empty
    HdfsContext context = new HdfsContext(newSession());
    assertTrue(listAllDataFiles(context, stagingPathRoot).isEmpty());
    // 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(), overwriteData.getMaterializedRows());
        // verify we did not modify the table directory
        assertEquals(listAllDataFiles(transaction, tableName.getSchemaName(), tableName.getTableName()), existingFiles);
    }
    // verify statistics unchanged
    try (Transaction transaction = newTransaction()) {
        HiveBasicStatistics statistics = getBasicStatisticsForTable(transaction, tableName);
        assertEquals(statistics.getRowCount().getAsLong(), overwriteData.getRowCount());
        assertEquals(statistics.getFileCount().getAsLong(), 1L);
    }
}

Example 15 with ConnectorPageSink

use of io.trino.spi.connector.ConnectorPageSink in project trino by trinodb.

the class TestHivePageSink method writeTestFile.

private static long writeTestFile(HiveConfig config, HiveMetastore metastore, String outputPath) {
    HiveTransactionHandle transaction = new HiveTransactionHandle(false);
    HiveWriterStats stats = new HiveWriterStats();
    ConnectorPageSink pageSink = createPageSink(transaction, config, metastore, new Path("file:///" + outputPath), stats);
    List<LineItemColumn> columns = getTestColumns();
    List<Type> columnTypes = columns.stream().map(LineItemColumn::getType).map(TestHivePageSink::getHiveType).map(hiveType -> hiveType.getType(TESTING_TYPE_MANAGER)).collect(toList());
    PageBuilder pageBuilder = new PageBuilder(columnTypes);
    int rows = 0;
    for (LineItem lineItem : new LineItemGenerator(0.01, 1, 1)) {
        rows++;
        if (rows >= NUM_ROWS) {
            break;
        }
        pageBuilder.declarePosition();
        for (int i = 0; i < columns.size(); i++) {
            LineItemColumn column = columns.get(i);
            BlockBuilder blockBuilder = pageBuilder.getBlockBuilder(i);
            switch(column.getType().getBase()) {
                case IDENTIFIER:
                    BIGINT.writeLong(blockBuilder, column.getIdentifier(lineItem));
                    break;
                case INTEGER:
                    INTEGER.writeLong(blockBuilder, column.getInteger(lineItem));
                    break;
                case DATE:
                    DATE.writeLong(blockBuilder, column.getDate(lineItem));
                    break;
                case DOUBLE:
                    DOUBLE.writeDouble(blockBuilder, column.getDouble(lineItem));
                    break;
                case VARCHAR:
                    createUnboundedVarcharType().writeSlice(blockBuilder, Slices.utf8Slice(column.getString(lineItem)));
                    break;
                default:
                    throw new IllegalArgumentException("Unsupported type " + column.getType());
            }
        }
    }
    Page page = pageBuilder.build();
    pageSink.appendPage(page);
    getFutureValue(pageSink.finish());
    File outputDir = new File(outputPath);
    List<File> files = ImmutableList.copyOf(outputDir.listFiles((dir, name) -> !name.endsWith(".crc")));
    File outputFile = getOnlyElement(files);
    long length = outputFile.length();
    ConnectorPageSource pageSource = createPageSource(transaction, config, outputFile);
    List<Page> pages = new ArrayList<>();
    while (!pageSource.isFinished()) {
        Page nextPage = pageSource.getNextPage();
        if (nextPage != null) {
            pages.add(nextPage.getLoadedPage());
        }
    }
    MaterializedResult expectedResults = toMaterializedResult(getHiveSession(config), columnTypes, ImmutableList.of(page));
    MaterializedResult results = toMaterializedResult(getHiveSession(config), columnTypes, pages);
    assertEquals(results, expectedResults);
    assertEquals(round(stats.getInputPageSizeInBytes().getAllTime().getMax()), page.getRetainedSizeInBytes());
    return length;
}