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

Example 1 with ConnectorPageSink

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

the class AbstractTestHive method testTableCreationRollback.

@Test
public void testTableCreationRollback() throws Exception {
    SchemaTableName temporaryCreateRollbackTable = temporaryTable("create_rollback");
    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(), NO_RETRIES);
            // 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);
            HdfsContext context = new HdfsContext(session);
            assertFalse(listAllDataFiles(context, stagingPathRoot).isEmpty());
            // rollback the table
            transaction.rollback();
        }
        // verify all files have been deleted
        HdfsContext context = new HdfsContext(newSession());
        assertTrue(listAllDataFiles(context, 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 2 with ConnectorPageSink

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

the class AbstractTestHive method doTestBucketSortedTables.

private void doTestBucketSortedTables(SchemaTableName table) throws IOException {
    int bucketCount = 3;
    int expectedRowCount = 0;
    try (Transaction transaction = newTransaction()) {
        ConnectorSession session = newSession();
        ConnectorMetadata metadata = transaction.getMetadata();
        // begin creating the table
        ConnectorTableMetadata tableMetadata = new ConnectorTableMetadata(table, ImmutableList.<ColumnMetadata>builder().add(new ColumnMetadata("id", VARCHAR)).add(new ColumnMetadata("value_asc", VARCHAR)).add(new ColumnMetadata("value_desc", BIGINT)).add(new ColumnMetadata("ds", VARCHAR)).build(), ImmutableMap.<String, Object>builder().put(STORAGE_FORMAT_PROPERTY, RCBINARY).put(PARTITIONED_BY_PROPERTY, ImmutableList.of("ds")).put(BUCKETED_BY_PROPERTY, ImmutableList.of("id")).put(BUCKET_COUNT_PROPERTY, bucketCount).put(SORTED_BY_PROPERTY, ImmutableList.builder().add(new SortingColumn("value_asc", ASCENDING)).add(new SortingColumn("value_desc", DESCENDING)).build()).buildOrThrow());
        ConnectorOutputTableHandle outputHandle = metadata.beginCreateTable(session, tableMetadata, Optional.empty(), NO_RETRIES);
        // write the data
        ConnectorPageSink sink = pageSinkProvider.createPageSink(transaction.getTransactionHandle(), session, outputHandle);
        List<Type> types = tableMetadata.getColumns().stream().map(ColumnMetadata::getType).collect(toList());
        ThreadLocalRandom random = ThreadLocalRandom.current();
        for (int i = 0; i < 50; i++) {
            MaterializedResult.Builder builder = MaterializedResult.resultBuilder(session, types);
            for (int j = 0; j < 1000; j++) {
                builder.row(sha256().hashLong(random.nextLong()).toString(), "test" + random.nextInt(100), random.nextLong(100_000), "2018-04-01");
                expectedRowCount++;
            }
            sink.appendPage(builder.build().toPage());
        }
        HdfsContext context = new HdfsContext(session);
        // verify we have enough temporary files per bucket to require multiple passes
        Path stagingPathRoot;
        if (isTemporaryStagingDirectoryEnabled(session)) {
            stagingPathRoot = new Path(getTemporaryStagingDirectoryPath(session).replace("${USER}", context.getIdentity().getUser()));
        } else {
            stagingPathRoot = getStagingPathRoot(outputHandle);
        }
        assertThat(listAllDataFiles(context, stagingPathRoot)).filteredOn(file -> file.contains(".tmp-sort.")).size().isGreaterThan(bucketCount * getHiveConfig().getMaxOpenSortFiles() * 2);
        // finish the write
        Collection<Slice> fragments = getFutureValue(sink.finish());
        // verify there are no temporary files
        for (String file : listAllDataFiles(context, stagingPathRoot)) {
            assertThat(file).doesNotContain(".tmp-sort.");
        }
        // finish creating table
        metadata.finishCreateTable(session, outputHandle, fragments, ImmutableList.of());
        transaction.commit();
    }
    // verify that bucket files are sorted
    try (Transaction transaction = newTransaction()) {
        ConnectorMetadata metadata = transaction.getMetadata();
        ConnectorSession session = newSession();
        metadata.beginQuery(session);
        ConnectorTableHandle tableHandle = getTableHandle(metadata, table);
        List<ColumnHandle> columnHandles = ImmutableList.copyOf(metadata.getColumnHandles(session, tableHandle).values());
        // verify local sorting property
        ConnectorTableProperties properties = metadata.getTableProperties(newSession(ImmutableMap.of("propagate_table_scan_sorting_properties", true, "bucket_execution_enabled", false)), tableHandle);
        Map<String, Integer> columnIndex = indexColumns(columnHandles);
        assertEquals(properties.getLocalProperties(), ImmutableList.of(new SortingProperty<>(columnHandles.get(columnIndex.get("value_asc")), ASC_NULLS_FIRST), new SortingProperty<>(columnHandles.get(columnIndex.get("value_desc")), DESC_NULLS_LAST)));
        assertThat(metadata.getTableProperties(newSession(), tableHandle).getLocalProperties()).isEmpty();
        List<ConnectorSplit> splits = getAllSplits(tableHandle, transaction, session);
        assertThat(splits).hasSize(bucketCount);
        int actualRowCount = 0;
        for (ConnectorSplit split : splits) {
            try (ConnectorPageSource pageSource = pageSourceProvider.createPageSource(transaction.getTransactionHandle(), session, split, tableHandle, columnHandles, DynamicFilter.EMPTY)) {
                String lastValueAsc = null;
                long lastValueDesc = -1;
                while (!pageSource.isFinished()) {
                    Page page = pageSource.getNextPage();
                    if (page == null) {
                        continue;
                    }
                    for (int i = 0; i < page.getPositionCount(); i++) {
                        Block blockAsc = page.getBlock(1);
                        Block blockDesc = page.getBlock(2);
                        assertFalse(blockAsc.isNull(i));
                        assertFalse(blockDesc.isNull(i));
                        String valueAsc = VARCHAR.getSlice(blockAsc, i).toStringUtf8();
                        if (lastValueAsc != null) {
                            assertGreaterThanOrEqual(valueAsc, lastValueAsc);
                            if (valueAsc.equals(lastValueAsc)) {
                                long valueDesc = BIGINT.getLong(blockDesc, i);
                                if (lastValueDesc != -1) {
                                    assertLessThanOrEqual(valueDesc, lastValueDesc);
                                }
                                lastValueDesc = valueDesc;
                            } else {
                                lastValueDesc = -1;
                            }
                        }
                        lastValueAsc = valueAsc;
                        actualRowCount++;
                    }
                }
            }
        }
        assertThat(actualRowCount).isEqualTo(expectedRowCount);
    }
}

Example 3 with ConnectorPageSink

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

the class AbstractTestHiveFileSystem method createTable.

private void createTable(SchemaTableName tableName, HiveStorageFormat storageFormat) throws Exception {
    List<ColumnMetadata> columns = ImmutableList.<ColumnMetadata>builder().add(new ColumnMetadata("id", BIGINT)).build();
    MaterializedResult data = MaterializedResult.resultBuilder(newSession(), BIGINT).row(1L).row(3L).row(2L).build();
    try (Transaction transaction = newTransaction()) {
        ConnectorMetadata metadata = transaction.getMetadata();
        ConnectorSession session = newSession();
        // begin creating the table
        ConnectorTableMetadata tableMetadata = new ConnectorTableMetadata(tableName, columns, createTableProperties(storageFormat));
        ConnectorOutputTableHandle outputHandle = metadata.beginCreateTable(session, tableMetadata, Optional.empty(), NO_RETRIES);
        // write the records
        ConnectorPageSink sink = pageSinkProvider.createPageSink(transaction.getTransactionHandle(), session, outputHandle);
        sink.appendPage(data.toPage());
        Collection<Slice> fragments = getFutureValue(sink.finish());
        // commit the table
        metadata.finishCreateTable(session, outputHandle, fragments, ImmutableList.of());
        transaction.commit();
        // Hack to work around the metastore not being configured for S3 or other FS.
        // The metastore tries to validate the location when creating the
        // table, which fails without explicit configuration for file system.
        // We work around that by using a dummy location when creating the
        // table and update it here to the correct location.
        metastoreClient.updateTableLocation(database, tableName.getTableName(), locationService.getTableWriteInfo(((HiveOutputTableHandle) outputHandle).getLocationHandle(), false).getTargetPath().toString());
    }
    try (Transaction transaction = newTransaction()) {
        ConnectorMetadata metadata = transaction.getMetadata();
        ConnectorSession session = newSession();
        // 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()), columns);
        // verify the data
        metadata.beginQuery(session);
        ConnectorSplitSource splitSource = getSplits(splitManager, transaction, session, tableHandle);
        ConnectorSplit split = getOnlyElement(getAllSplits(splitSource));
        try (ConnectorPageSource pageSource = pageSourceProvider.createPageSource(transaction.getTransactionHandle(), session, split, tableHandle, columnHandles, DynamicFilter.EMPTY)) {
            MaterializedResult result = materializeSourceDataStream(session, pageSource, getTypes(columnHandles));
            assertEqualsIgnoreOrder(result.getMaterializedRows(), data.getMaterializedRows());
        }
        metadata.cleanupQuery(session);
    }
}

Example 4 with ConnectorPageSink

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

the class AbstractTestHive method doInsert.

private void doInsert(HiveStorageFormat storageFormat, SchemaTableName tableName) throws Exception {
    // creating the table
    doCreateEmptyTable(tableName, storageFormat, CREATE_TABLE_COLUMNS);
    MaterializedResult.Builder resultBuilder = MaterializedResult.resultBuilder(SESSION, CREATE_TABLE_DATA.getTypes());
    for (int i = 0; i < 3; i++) {
        insertData(tableName, CREATE_TABLE_DATA);
        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
            resultBuilder.rows(CREATE_TABLE_DATA.getMaterializedRows());
            MaterializedResult result = readTable(transaction, tableHandle, columnHandles, session, TupleDomain.all(), OptionalInt.empty(), Optional.empty());
            assertEqualsIgnoreOrder(result.getMaterializedRows(), resultBuilder.build().getMaterializedRows());
            // statistics
            HiveBasicStatistics tableStatistics = getBasicStatisticsForTable(transaction, tableName);
            assertEquals(tableStatistics.getRowCount().getAsLong(), CREATE_TABLE_DATA.getRowCount() * (i + 1));
            assertEquals(tableStatistics.getFileCount().getAsLong(), i + 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();
        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);
        sink.appendPage(CREATE_TABLE_DATA.toPage());
        sink.appendPage(CREATE_TABLE_DATA.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(), CREATE_TABLE_DATA.getRowCount() * 5L);
        try (Transaction otherTransaction = newTransaction()) {
            // statistics, not visible from outside transaction
            HiveBasicStatistics otherTableStatistics = getBasicStatisticsForTable(otherTransaction, tableName);
            assertEquals(otherTableStatistics.getRowCount().getAsLong(), CREATE_TABLE_DATA.getRowCount() * 3L);
        }
        // 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(), resultBuilder.build().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(), CREATE_TABLE_DATA.getRowCount() * 3L);
        assertEquals(statistics.getFileCount().getAsLong(), 3L);
    }
}

Example 5 with ConnectorPageSink

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

the class AbstractTestHive method doCreateTable.

protected void doCreateTable(SchemaTableName tableName, HiveStorageFormat storageFormat) throws Exception {
    String queryId;
    try (Transaction transaction = newTransaction()) {
        ConnectorSession session = newSession();
        ConnectorMetadata metadata = transaction.getMetadata();
        queryId = session.getQueryId();
        // begin creating the table
        ConnectorTableMetadata tableMetadata = new ConnectorTableMetadata(tableName, CREATE_TABLE_COLUMNS, createTableProperties(storageFormat));
        ConnectorOutputTableHandle outputHandle = metadata.beginCreateTable(session, tableMetadata, Optional.empty(), NO_RETRIES);
        // write the data
        ConnectorPageSink sink = pageSinkProvider.createPageSink(transaction.getTransactionHandle(), session, outputHandle);
        sink.appendPage(CREATE_TABLE_DATA.toPage());
        Collection<Slice> fragments = getFutureValue(sink.finish());
        // verify all new files start with the unique prefix
        HdfsContext context = new HdfsContext(session);
        for (String filePath : listAllDataFiles(context, getStagingPathRoot(outputHandle))) {
            assertThat(new Path(filePath).getName()).startsWith(session.getQueryId());
        }
        // commit the table
        metadata.finishCreateTable(session, outputHandle, fragments, ImmutableList.of());
        transaction.commit();
    }
    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.of(storageFormat));
        assertEqualsIgnoreOrder(result.getMaterializedRows(), CREATE_TABLE_DATA.getMaterializedRows());
        // verify the node version and query ID in table
        Table table = getMetastoreClient().getTable(tableName.getSchemaName(), tableName.getTableName()).get();
        assertEquals(table.getParameters().get(PRESTO_VERSION_NAME), TEST_SERVER_VERSION);
        assertEquals(table.getParameters().get(PRESTO_QUERY_ID_NAME), queryId);
        // verify basic statistics
        HiveBasicStatistics statistics = getBasicStatisticsForTable(transaction, tableName);
        assertEquals(statistics.getRowCount().getAsLong(), CREATE_TABLE_DATA.getRowCount());
        assertEquals(statistics.getFileCount().getAsLong(), 1L);
        assertGreaterThan(statistics.getInMemoryDataSizeInBytes().getAsLong(), 0L);
        assertGreaterThan(statistics.getOnDiskDataSizeInBytes().getAsLong(), 0L);
    }
}