use of com.facebook.presto.orc.StorageStripeMetadataSource in project presto by prestodb.
the class OrcTestingUtil method createReader.
public static OrcBatchRecordReader createReader(OrcDataSource dataSource, List<Long> columnIds, List<Type> types) throws IOException {
OrcReader orcReader = new OrcReader(dataSource, ORC, new StorageOrcFileTailSource(), new StorageStripeMetadataSource(), new RaptorOrcAggregatedMemoryContext(), createDefaultTestConfig(), false, NO_ENCRYPTION, DwrfKeyProvider.EMPTY, new RuntimeStats());
List<String> columnNames = orcReader.getColumnNames();
assertEquals(columnNames.size(), columnIds.size());
Map<Integer, Type> includedColumns = new HashMap<>();
int ordinal = 0;
for (long columnId : columnIds) {
assertEquals(columnNames.get(ordinal), String.valueOf(columnId));
includedColumns.put(ordinal, types.get(ordinal));
ordinal++;
}
return createRecordReader(orcReader, includedColumns);
}
use of com.facebook.presto.orc.StorageStripeMetadataSource in project presto by prestodb.
the class TestOrcFileRewriter method testRewriterDropThenAddDifferentColumns.
/**
* The following test add or drop different columns
*/
@Test
public void testRewriterDropThenAddDifferentColumns() throws Exception {
FunctionAndTypeManager functionAndTypeManager = createTestFunctionAndTypeManager();
DBI dbi = new DBI("jdbc:h2:mem:test" + System.nanoTime() + "_" + ThreadLocalRandom.current().nextInt());
dbi.registerMapper(new TableColumn.Mapper(functionAndTypeManager));
Handle dummyHandle = dbi.open();
File dataDir = Files.createTempDir();
StorageManager storageManager = createOrcStorageManager(dbi, dataDir);
List<Long> columnIds = ImmutableList.of(3L, 7L);
List<Type> columnTypes = ImmutableList.of(BIGINT, createVarcharType(20));
File file = new File(temporary, randomUUID().toString());
try (FileWriter writer = createFileWriter(columnIds, columnTypes, file, false)) {
List<Page> pages = rowPagesBuilder(columnTypes).row(1L, "1").row(2L, "2").row(3L, "3").row(4L, "4").build();
writer.appendPages(pages);
}
// Add a column
File newFile1 = new File(temporary, randomUUID().toString());
FileSystem fileSystem = new LocalOrcDataEnvironment().getFileSystem(DEFAULT_RAPTOR_CONTEXT);
OrcFileInfo info = createFileRewriter().rewrite(fileSystem, getColumnTypes(ImmutableList.of(3L, 7L, 10L), ImmutableList.of(BIGINT, createVarcharType(20), DOUBLE)), path(file), path(newFile1), new BitSet(5));
assertEquals(info.getRowCount(), 4);
assertEquals(readAllBytes(file.toPath()), readAllBytes(newFile1.toPath()));
// Drop a column
File newFile2 = new File(temporary, randomUUID().toString());
info = createFileRewriter().rewrite(fileSystem, getColumnTypes(ImmutableList.of(7L, 10L), ImmutableList.of(createVarcharType(20), DOUBLE)), path(newFile1), path(newFile2), new BitSet(5));
assertEquals(info.getRowCount(), 4);
// Optimized writer will keep the only column
OrcReader orcReader = new OrcReader(fileOrcDataSource(newFile2), ORC, new StorageOrcFileTailSource(), new StorageStripeMetadataSource(), new RaptorOrcAggregatedMemoryContext(), OrcTestingUtil.createDefaultTestConfig(), false, NO_ENCRYPTION, DwrfKeyProvider.EMPTY, new RuntimeStats());
orcReader.getColumnNames().equals(ImmutableList.of("7"));
// Add a column with the different ID with different type
File newFile3 = new File(temporary, randomUUID().toString());
info = createFileRewriter().rewrite(fileSystem, getColumnTypes(ImmutableList.of(7L, 10L, 13L), ImmutableList.of(createVarcharType(20), DOUBLE, createVarcharType(5))), path(newFile2), path(newFile3), new BitSet(5));
assertEquals(info.getRowCount(), 4);
assertEquals(readAllBytes(newFile2.toPath()), readAllBytes(newFile3.toPath()));
// Get prepared for the final file; make sure it is accessible from storage manager
UUID uuid = randomUUID();
File newFile4 = getFileSystemPath(new File(dataDir, "data/storage"), uuid);
// Optimized ORC writer does not create the file itself
newFile4.getParentFile().mkdirs();
newFile4.createNewFile();
// Drop a column and add a column; also delete 3 rows
BitSet rowsToDelete = new BitSet(5);
rowsToDelete.set(0);
rowsToDelete.set(1);
rowsToDelete.set(3);
info = createFileRewriter().rewrite(fileSystem, getColumnTypes(ImmutableList.of(7L, 13L, 18L), ImmutableList.of(createVarcharType(20), createVarcharType(5), INTEGER)), path(newFile3), path(newFile4), rowsToDelete);
assertEquals(info.getRowCount(), 1);
ConnectorPageSource source = storageManager.getPageSource(DEFAULT_RAPTOR_CONTEXT, DEFAULT_HIVE_FILE_CONTEXT, uuid, Optional.empty(), false, OptionalInt.empty(), ImmutableList.of(13L, 7L, 18L), ImmutableList.of(createVarcharType(5), createVarcharType(20), INTEGER), TupleDomain.all(), READER_ATTRIBUTES);
Page page = null;
while (page == null) {
page = source.getNextPage();
}
assertEquals(page.getPositionCount(), 1);
// Column 13L
Block column0 = page.getBlock(0);
assertTrue(column0.isNull(0));
// Column 7L
Block column1 = page.getBlock(1);
assertEquals(createVarcharType(20).getSlice(column1, 0), utf8Slice("3"));
// Column 8L
Block column2 = page.getBlock(2);
assertTrue(column2.isNull(0));
// Remove all the columns
File newFile5 = new File(temporary, randomUUID().toString());
info = createFileRewriter().rewrite(fileSystem, getColumnTypes(ImmutableList.of(13L, 18L), ImmutableList.of(createVarcharType(5), INTEGER)), path(newFile4), path(newFile5), new BitSet(5));
// Optimized writer will drop the file
assertEquals(info.getRowCount(), 0);
assertFalse(newFile5.exists());
dummyHandle.close();
deleteRecursively(dataDir.toPath(), ALLOW_INSECURE);
}
use of com.facebook.presto.orc.StorageStripeMetadataSource in project presto by prestodb.
the class OrcStorageManager method getRowsFromUuid.
Optional<BitSet> getRowsFromUuid(FileSystem fileSystem, Optional<UUID> deltaShardUuid) {
if (!deltaShardUuid.isPresent()) {
return Optional.empty();
}
try (OrcDataSource dataSource = openShard(fileSystem, deltaShardUuid.get(), defaultReaderAttributes)) {
OrcAggregatedMemoryContext systemMemoryUsage = new RaptorOrcAggregatedMemoryContext();
OrcReader reader = new OrcReader(dataSource, ORC, orcFileTailSource, new StorageStripeMetadataSource(), new RaptorOrcAggregatedMemoryContext(), new OrcReaderOptions(defaultReaderAttributes.getMaxMergeDistance(), defaultReaderAttributes.getTinyStripeThreshold(), HUGE_MAX_READ_BLOCK_SIZE, defaultReaderAttributes.isZstdJniDecompressionEnabled()), false, NO_ENCRYPTION, DwrfKeyProvider.EMPTY, new RuntimeStats());
if (reader.getFooter().getNumberOfRows() >= Integer.MAX_VALUE) {
throw new IOException("File has too many rows");
}
try (OrcBatchRecordReader recordReader = reader.createBatchRecordReader(ImmutableMap.of(0, BIGINT), OrcPredicate.TRUE, DEFAULT_STORAGE_TIMEZONE, systemMemoryUsage, INITIAL_BATCH_SIZE)) {
BitSet bitSet = new BitSet();
while (recordReader.nextBatch() > 0) {
Block block = recordReader.readBlock(0);
for (int i = 0; i < block.getPositionCount(); i++) {
bitSet.set(toIntExact(block.getLong(i)));
}
}
return Optional.of(bitSet);
}
} catch (IOException | RuntimeException e) {
throw new PrestoException(RAPTOR_ERROR, "Failed to read file: " + deltaShardUuid, e);
}
}
use of com.facebook.presto.orc.StorageStripeMetadataSource in project presto by prestodb.
the class HiveTestUtils method getDefaultHiveBatchPageSourceFactories.
public static Set<HiveBatchPageSourceFactory> getDefaultHiveBatchPageSourceFactories(HiveClientConfig hiveClientConfig, MetastoreClientConfig metastoreClientConfig) {
FileFormatDataSourceStats stats = new FileFormatDataSourceStats();
HdfsEnvironment testHdfsEnvironment = createTestHdfsEnvironment(hiveClientConfig, metastoreClientConfig);
return ImmutableSet.<HiveBatchPageSourceFactory>builder().add(new RcFilePageSourceFactory(FUNCTION_AND_TYPE_MANAGER, testHdfsEnvironment, stats)).add(new OrcBatchPageSourceFactory(FUNCTION_AND_TYPE_MANAGER, FUNCTION_RESOLUTION, hiveClientConfig, testHdfsEnvironment, stats, new StorageOrcFileTailSource(), StripeMetadataSourceFactory.of(new StorageStripeMetadataSource()))).add(new DwrfBatchPageSourceFactory(FUNCTION_AND_TYPE_MANAGER, FUNCTION_RESOLUTION, hiveClientConfig, testHdfsEnvironment, stats, new StorageOrcFileTailSource(), StripeMetadataSourceFactory.of(new StorageStripeMetadataSource()), NO_ENCRYPTION)).add(new ParquetPageSourceFactory(FUNCTION_AND_TYPE_MANAGER, FUNCTION_RESOLUTION, testHdfsEnvironment, stats, new MetadataReader())).add(new PageFilePageSourceFactory(testHdfsEnvironment, new BlockEncodingManager())).build();
}
use of com.facebook.presto.orc.StorageStripeMetadataSource in project presto by prestodb.
the class HiveTestUtils method getDefaultHiveSelectivePageSourceFactories.
public static Set<HiveSelectivePageSourceFactory> getDefaultHiveSelectivePageSourceFactories(HiveClientConfig hiveClientConfig, MetastoreClientConfig metastoreClientConfig) {
FileFormatDataSourceStats stats = new FileFormatDataSourceStats();
HdfsEnvironment testHdfsEnvironment = createTestHdfsEnvironment(hiveClientConfig, metastoreClientConfig);
return ImmutableSet.<HiveSelectivePageSourceFactory>builder().add(new OrcSelectivePageSourceFactory(FUNCTION_AND_TYPE_MANAGER, FUNCTION_RESOLUTION, ROW_EXPRESSION_SERVICE, hiveClientConfig, testHdfsEnvironment, stats, new StorageOrcFileTailSource(), StripeMetadataSourceFactory.of(new StorageStripeMetadataSource()), new TupleDomainFilterCache())).add(new DwrfSelectivePageSourceFactory(FUNCTION_AND_TYPE_MANAGER, FUNCTION_RESOLUTION, ROW_EXPRESSION_SERVICE, hiveClientConfig, testHdfsEnvironment, stats, new StorageOrcFileTailSource(), StripeMetadataSourceFactory.of(new StorageStripeMetadataSource()), new TupleDomainFilterCache(), NO_ENCRYPTION)).build();
}
Aggregations