Examples with Page io.trino.spi.Page used on opensource projects

Example 41 with Page

use of io.trino.spi.Page in project trino by trinodb.

the class RcFilePageSource method getNextPage.

@Override
public Page getNextPage() {
    try {
        // advance in the current batch
        pageId++;
        // if the batch has been consumed, read the next batch
        int currentPageSize = rcFileReader.advance();
        if (currentPageSize < 0) {
            close();
            return null;
        }
        Block[] blocks = new Block[hiveColumnIndexes.length];
        for (int fieldId = 0; fieldId < blocks.length; fieldId++) {
            if (constantBlocks[fieldId] != null) {
                blocks[fieldId] = new RunLengthEncodedBlock(constantBlocks[fieldId], currentPageSize);
            } else {
                blocks[fieldId] = createBlock(currentPageSize, fieldId);
            }
        }
        return new Page(currentPageSize, blocks);
    } catch (TrinoException e) {
        closeAllSuppress(e, this);
        throw e;
    } catch (RcFileCorruptionException e) {
        closeAllSuppress(e, this);
        throw new TrinoException(HIVE_BAD_DATA, format("Corrupted RC file: %s", rcFileReader.getId()), e);
    } catch (IOException | RuntimeException e) {
        closeAllSuppress(e, this);
        throw new TrinoException(HIVE_CURSOR_ERROR, format("Failed to read RC file: %s", rcFileReader.getId()), e);
    }
}

Example 42 with Page

use of io.trino.spi.Page 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 43 with Page

use of io.trino.spi.Page in project trino by trinodb.

the class AbstractTestHiveFileFormats method createTestFileTrino.

public static FileSplit createTestFileTrino(String filePath, HiveStorageFormat storageFormat, HiveCompressionCodec compressionCodec, List<TestColumn> testColumns, ConnectorSession session, int numRows, HiveFileWriterFactory fileWriterFactory) {
    // filter out partition keys, which are not written to the file
    testColumns = testColumns.stream().filter(column -> !column.isPartitionKey()).collect(toImmutableList());
    List<Type> types = testColumns.stream().map(TestColumn::getType).map(HiveType::valueOf).map(type -> type.getType(TESTING_TYPE_MANAGER)).collect(toList());
    PageBuilder pageBuilder = new PageBuilder(types);
    for (int rowNumber = 0; rowNumber < numRows; rowNumber++) {
        pageBuilder.declarePosition();
        for (int columnNumber = 0; columnNumber < testColumns.size(); columnNumber++) {
            serializeObject(types.get(columnNumber), pageBuilder.getBlockBuilder(columnNumber), testColumns.get(columnNumber).getWriteValue(), testColumns.get(columnNumber).getObjectInspector(), false);
        }
    }
    Page page = pageBuilder.build();
    JobConf jobConf = new JobConf();
    configureCompression(jobConf, compressionCodec);
    Properties tableProperties = new Properties();
    tableProperties.setProperty("columns", testColumns.stream().map(TestColumn::getName).collect(Collectors.joining(",")));
    tableProperties.setProperty("columns.types", testColumns.stream().map(TestColumn::getType).collect(Collectors.joining(",")));
    Optional<FileWriter> fileWriter = fileWriterFactory.createFileWriter(new Path(filePath), testColumns.stream().map(TestColumn::getName).collect(toList()), StorageFormat.fromHiveStorageFormat(storageFormat), tableProperties, jobConf, session, OptionalInt.empty(), NO_ACID_TRANSACTION, false, WriterKind.INSERT);
    FileWriter hiveFileWriter = fileWriter.orElseThrow(() -> new IllegalArgumentException("fileWriterFactory"));
    hiveFileWriter.appendRows(page);
    hiveFileWriter.commit();
    return new FileSplit(new Path(filePath), 0, new File(filePath).length(), new String[0]);
}

Example 44 with Page

use of io.trino.spi.Page in project trino by trinodb.

the class MergingPageIterator method computeNext.

@Override
protected Page computeNext() {
    while (!pageBuilder.isFull() && pagePositions.hasNext()) {
        pagePositions.next().appendTo(pageBuilder);
    }
    if (pageBuilder.isEmpty()) {
        return endOfData();
    }
    Page page = pageBuilder.build();
    pageBuilder.reset();
    return page;
}

Example 45 with Page

use of io.trino.spi.Page in project trino by trinodb.

the class TestOrcPageSourceMemoryTracking method testScanFilterAndProjectOperator.

@Test
public void testScanFilterAndProjectOperator() {
    // Numbers used in assertions in this test may change when implementation is modified,
    // feel free to change them if they break in the future
    DriverContext driverContext = testPreparer.newDriverContext();
    SourceOperator operator = testPreparer.newScanFilterAndProjectOperator(driverContext);
    assertEquals(driverContext.getMemoryUsage(), 0);
    int totalRows = 0;
    while (totalRows < NUM_ROWS) {
        assertFalse(operator.isFinished());
        Page page = operator.getOutput();
        assertNotNull(page);
        // memory usage varies depending on stripe alignment
        long memoryUsage = driverContext.getMemoryUsage();
        assertTrue(memoryUsage < 1000 || (memoryUsage > 150_000 && memoryUsage < 630_000), format("Memory usage (%s) outside of bounds", memoryUsage));
        totalRows += page.getPositionCount();
    }
    // done... in the current implementation finish is not set until output returns a null page
    assertNull(operator.getOutput());
    assertTrue(operator.isFinished());
    assertBetweenInclusive(driverContext.getMemoryUsage(), 0L, 500L);
}