Examples with ConnectorPageSink com.facebook.presto.spi.ConnectorPageSink used on opensource projects

Example 1 with ConnectorPageSink

use of com.facebook.presto.spi.ConnectorPageSink in project presto by prestodb.

the class AbstractTestHiveClient 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
        List<String> partitionNames = transaction.getMetastore(tableName.getSchemaName()).getPartitionNames(tableName.getSchemaName(), tableName.getTableName()).orElseThrow(() -> new PrestoException(HIVE_METASTORE_ERROR, "Partition metadata not available"));
        assertEqualsIgnoreOrder(partitionNames, CREATE_TABLE_PARTITIONED_DATA.getMaterializedRows().stream().map(row -> "ds=" + row.getField(CREATE_TABLE_PARTITIONED_DATA.getTypes().size() - 1)).collect(toList()));
        // verify the node versions in partitions
        Map<String, Optional<Partition>> partitions = getMetastoreClient(tableName.getSchemaName()).getPartitionsByNames(tableName.getSchemaName(), tableName.getTableName(), partitionNames);
        assertEquals(partitions.size(), partitionNames.size());
        for (String partitionName : partitionNames) {
            Partition partition = partitions.get(partitionName).get();
            assertEquals(partition.getParameters().get(HiveMetadata.PRESTO_VERSION_NAME), TEST_SERVER_VERSION);
            assertEquals(partition.getParameters().get(HiveMetadata.PRESTO_QUERY_ID_NAME), queryId);
        }
        // load the new table
        ConnectorSession session = newSession();
        ConnectorMetadata metadata = transaction.getMetadata();
        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());
    }
    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);
        stagingPathRoot = getStagingPathRoot(insertTableHandle);
        ConnectorPageSink sink = pageSinkProvider.createPageSink(transaction.getTransactionHandle(), session, insertTableHandle);
        sink.appendPage(CREATE_TABLE_PARTITIONED_DATA_2ND.toPage());
        getFutureValue(sink.finish());
        // 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
        Set<String> tempFiles = listAllDataFiles(getStagingPathRoot(insertTableHandle));
        assertTrue(!tempFiles.isEmpty());
        for (String filePath : tempFiles) {
            assertTrue(new Path(filePath).getName().startsWith(getFilePrefix(insertTableHandle)));
        }
        // rollback insert
        transaction.rollback();
    }
    // verify the data is unchanged
    try (Transaction transaction = newTransaction()) {
        ConnectorSession session = newSession();
        ConnectorMetadata metadata = transaction.getMetadata();
        ConnectorTableHandle tableHandle = getTableHandle(metadata, tableName);
        List<ColumnHandle> columnHandles = filterNonHiddenColumnHandles(metadata.getColumnHandles(session, tableHandle).values());
        MaterializedResult result = readTable(transaction, tableHandle, columnHandles, newSession(), 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
        assertTrue(listAllDataFiles(stagingPathRoot).isEmpty());
    }
}

Example 2 with ConnectorPageSink

use of com.facebook.presto.spi.ConnectorPageSink in project presto by prestodb.

the class AbstractTestHiveClient method testTableCreationRollback.

@Test
public void testTableCreationRollback() throws Exception {
    try {
        Path stagingPathRoot;
        try (Transaction transaction = newTransaction()) {
            ConnectorSession session = newSession();
            ConnectorMetadata metadata = transaction.getMetadata();
            // begin creating the table
            ConnectorTableMetadata tableMetadata = new ConnectorTableMetadata(temporaryCreateRollbackTable, CREATE_TABLE_COLUMNS, createTableProperties(RCBINARY));
            ConnectorOutputTableHandle outputHandle = metadata.beginCreateTable(session, tableMetadata, Optional.empty());
            // write the data
            ConnectorPageSink sink = pageSinkProvider.createPageSink(transaction.getTransactionHandle(), session, outputHandle);
            sink.appendPage(CREATE_TABLE_DATA.toPage());
            getFutureValue(sink.finish());
            // verify we have data files
            stagingPathRoot = getStagingPathRoot(outputHandle);
            assertFalse(listAllDataFiles(stagingPathRoot).isEmpty());
            // rollback the table
            transaction.rollback();
        }
        // verify all files have been deleted
        assertTrue(listAllDataFiles(stagingPathRoot).isEmpty());
        // verify table is not in the metastore
        try (Transaction transaction = newTransaction()) {
            ConnectorSession session = newSession();
            ConnectorMetadata metadata = transaction.getMetadata();
            assertNull(metadata.getTableHandle(session, temporaryCreateRollbackTable));
        }
    } finally {
        dropTable(temporaryCreateRollbackTable);
    }
}

Example 3 with ConnectorPageSink

use of com.facebook.presto.spi.ConnectorPageSink in project presto by prestodb.

the class AbstractTestHiveClient 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<ColumnHandle> constraint = new Constraint<>(tupleDomain, convertToPredicate(tupleDomain));
            List<ConnectorTableLayoutResult> tableLayoutResults = metadata.getTableLayouts(session, tableHandle, constraint, Optional.empty());
            ConnectorTableLayoutHandle tableLayoutHandle = Iterables.getOnlyElement(tableLayoutResults).getTableLayout().getHandle();
            metadata.metadataDelete(session, tableHandle, tableLayoutHandle);
            rollbackIfEquals(tag, ROLLBACK_AFTER_DELETE);
            // Query 2: insert
            session = newSession();
            ConnectorInsertTableHandle insertTableHandle = metadata.beginInsert(session, tableHandle);
            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);
            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(tableName);
            }
        } catch (TestingRollbackException e) {
            transaction.rollback();
        } catch (PrestoException e) {
            assertFalse(expectQuerySucceed);
            if (conflictTrigger.isPresent()) {
                conflictTrigger.get().verifyAndCleanup(tableName);
            }
        }
    }
    // check that temporary files are removed
    if (writePath != null && !writePath.equals(targetPath)) {
        FileSystem fileSystem = hdfsEnvironment.getFileSystem("user", writePath);
        assertFalse(fileSystem.exists(writePath));
    }
    try (Transaction transaction = newTransaction()) {
        // verify partitions
        List<String> partitionNames = transaction.getMetastore(tableName.getSchemaName()).getPartitionNames(tableName.getSchemaName(), tableName.getTableName()).orElseThrow(() -> new PrestoException(HIVE_METASTORE_ERROR, "Partition metadata not available"));
        assertEqualsIgnoreOrder(partitionNames, expectedData.getMaterializedRows().stream().map(row -> format("pk1=%s/pk2=%s", row.getField(1), row.getField(2))).distinct().collect(toList()));
        // load the new table
        ConnectorSession session = newSession();
        ConnectorMetadata metadata = transaction.getMetadata();
        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 4 with ConnectorPageSink

use of com.facebook.presto.spi.ConnectorPageSink in project presto by prestodb.

the class AbstractTestHiveClient 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);
        ConnectorPageSink sink = pageSinkProvider.createPageSink(transaction.getTransactionHandle(), session, insertTableHandle);
        sink.appendPage(dataBefore.toPage());
        Collection<Slice> fragments = getFutureValue(sink.finish());
        metadata.finishInsert(session, insertTableHandle, fragments);
        transaction.commit();
    }
    // load the table and verify the data
    try (Transaction transaction = newTransaction()) {
        ConnectorSession session = newSession();
        ConnectorMetadata metadata = transaction.getMetadata();
        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 = new PrincipalPrivileges(ImmutableMultimap.<String, HivePrivilegeInfo>builder().put(session.getUser(), new HivePrivilegeInfo(HivePrivilege.SELECT, true)).put(session.getUser(), new HivePrivilegeInfo(HivePrivilege.INSERT, true)).put(session.getUser(), new HivePrivilegeInfo(HivePrivilege.UPDATE, true)).put(session.getUser(), new HivePrivilegeInfo(HivePrivilege.DELETE, true)).build(), ImmutableMultimap.of());
        Table oldTable = transaction.getMetastore(schemaName).getTable(schemaName, tableName).get();
        HiveTypeTranslator hiveTypeTranslator = new HiveTypeTranslator();
        List<Column> dataColumns = tableAfter.stream().filter(columnMetadata -> !columnMetadata.getName().equals("ds")).map(columnMetadata -> new Column(columnMetadata.getName(), toHiveType(hiveTypeTranslator, columnMetadata.getType()), Optional.empty())).collect(toList());
        Table.Builder newTable = Table.builder(oldTable).setDataColumns(dataColumns);
        transaction.getMetastore(schemaName).replaceView(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();
        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);
        ConnectorPageSink sink = pageSinkProvider.createPageSink(transaction.getTransactionHandle(), session, insertTableHandle);
        sink.appendPage(dataAfter.toPage());
        Collection<Slice> fragments = getFutureValue(sink.finish());
        metadata.finishInsert(session, insertTableHandle, fragments);
        transaction.commit();
        fail("expected exception");
    } catch (PrestoException e) {
        // expected
        assertEquals(e.getErrorCode(), HIVE_PARTITION_SCHEMA_MISMATCH.toErrorCode());
    }
}

Example 5 with ConnectorPageSink

use of com.facebook.presto.spi.ConnectorPageSink in project presto by prestodb.

the class TestHivePageSink method writeTestFile.

private static long writeTestFile(HiveClientConfig config, ExtendedHiveMetastore metastore, String outputPath) {
    HiveTransactionHandle transaction = new HiveTransactionHandle();
    ConnectorPageSink pageSink = createPageSink(transaction, config, metastore, new Path("file:///" + outputPath));
    List<LineItemColumn> columns = getTestColumns();
    List<Type> columnTypes = columns.stream().map(LineItemColumn::getType).map(TestHivePageSink::getHiveType).map(hiveType -> hiveType.getType(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) {
            nextPage.assureLoaded();
            pages.add(nextPage);
        }
    }
    MaterializedResult expectedResults = toMaterializedResult(getSession(config), columnTypes, ImmutableList.of(page));
    MaterializedResult results = toMaterializedResult(getSession(config), columnTypes, pages);
    assertEquals(results, expectedResults);
    return length;
}