Examples with HoodieDataBlock org.apache.hudi.common.table.log.block.HoodieDataBlock used on opensource projects

Example 26 with HoodieDataBlock

use of org.apache.hudi.common.table.log.block.HoodieDataBlock in project hudi by apache.

the class TestHoodieLogFormat method testBasicAppendAndTraverseInReverse.

@ParameterizedTest
@ValueSource(booleans = { true, false })
public void testBasicAppendAndTraverseInReverse(boolean readBlocksLazily) throws IOException, URISyntaxException, InterruptedException {
    Writer writer = HoodieLogFormat.newWriterBuilder().onParentPath(partitionPath).withFileExtension(HoodieLogFile.DELTA_EXTENSION).withFileId("test-fileid1").overBaseCommit("100").withFs(fs).build();
    Schema schema = getSimpleSchema();
    List<IndexedRecord> records1 = SchemaTestUtil.generateTestRecords(0, 100);
    List<IndexedRecord> copyOfRecords1 = records1.stream().map(record -> HoodieAvroUtils.rewriteRecord((GenericRecord) record, schema)).collect(Collectors.toList());
    Map<HoodieLogBlock.HeaderMetadataType, String> header = new HashMap<>();
    header.put(HoodieLogBlock.HeaderMetadataType.INSTANT_TIME, "100");
    header.put(HoodieLogBlock.HeaderMetadataType.SCHEMA, schema.toString());
    HoodieDataBlock dataBlock = getDataBlock(DEFAULT_DATA_BLOCK_TYPE, records1, header);
    writer.appendBlock(dataBlock);
    writer.close();
    writer = HoodieLogFormat.newWriterBuilder().onParentPath(partitionPath).withFileExtension(HoodieLogFile.DELTA_EXTENSION).withFileId("test-fileid1").overBaseCommit("100").withFs(fs).build();
    List<IndexedRecord> records2 = SchemaTestUtil.generateTestRecords(0, 100);
    dataBlock = getDataBlock(DEFAULT_DATA_BLOCK_TYPE, records2, header);
    writer.appendBlock(dataBlock);
    writer.close();
    // Close and Open again and append 100 more records
    writer = HoodieLogFormat.newWriterBuilder().onParentPath(partitionPath).withFileExtension(HoodieLogFile.DELTA_EXTENSION).withFileId("test-fileid1").overBaseCommit("100").withFs(fs).build();
    List<IndexedRecord> records3 = SchemaTestUtil.generateTestRecords(0, 100);
    dataBlock = getDataBlock(DEFAULT_DATA_BLOCK_TYPE, records3, header);
    writer.appendBlock(dataBlock);
    writer.close();
    FileCreateUtils.createDeltaCommit(basePath, "100", fs);
    HoodieLogFileReader reader = new HoodieLogFileReader(fs, new HoodieLogFile(writer.getLogFile().getPath(), fs.getFileStatus(writer.getLogFile().getPath()).getLen()), SchemaTestUtil.getSimpleSchema(), bufferSize, readBlocksLazily, true);
    assertTrue(reader.hasPrev(), "Third block should be available");
    reader.moveToPrev();
    assertTrue(reader.hasPrev(), "Second block should be available");
    reader.moveToPrev();
    // After moving twice, this last reader.prev() should read the First block written
    assertTrue(reader.hasPrev(), "First block should be available");
    HoodieLogBlock prevBlock = reader.prev();
    HoodieDataBlock dataBlockRead = (HoodieDataBlock) prevBlock;
    List<IndexedRecord> recordsRead = getRecords(dataBlockRead);
    assertEquals(copyOfRecords1.size(), recordsRead.size(), "Read records size should be equal to the written records size");
    assertEquals(copyOfRecords1, recordsRead, "Both records lists should be the same. (ordering guaranteed)");
    assertFalse(reader.hasPrev());
    reader.close();
}

Example 27 with HoodieDataBlock

use of org.apache.hudi.common.table.log.block.HoodieDataBlock in project hudi by apache.

the class TestHoodieLogFormat method testDataBlockFormatAppendAndReadWithProjectedSchema.

@ParameterizedTest
@EnumSource(names = { "AVRO_DATA_BLOCK", "HFILE_DATA_BLOCK", "PARQUET_DATA_BLOCK" })
public void testDataBlockFormatAppendAndReadWithProjectedSchema(HoodieLogBlockType dataBlockType) throws IOException, URISyntaxException, InterruptedException {
    Writer writer = HoodieLogFormat.newWriterBuilder().onParentPath(partitionPath).withFileExtension(HoodieLogFile.DELTA_EXTENSION).withFileId("test-fileid1").overBaseCommit("100").withFs(fs).build();
    List<GenericRecord> records = SchemaTestUtil.generateTestGenericRecords(0, 1000);
    Schema schema = getSimpleSchema();
    Map<HoodieLogBlock.HeaderMetadataType, String> header = new HashMap<HoodieLogBlock.HeaderMetadataType, String>() {

        {
            put(HoodieLogBlock.HeaderMetadataType.INSTANT_TIME, "100");
            put(HoodieLogBlock.HeaderMetadataType.SCHEMA, schema.toString());
        }
    };
    // Init Benchmark to report number of bytes actually read from the Block
    BenchmarkCounter.initCounterFromReporter(HadoopMapRedUtils.createTestReporter(), fs.getConf());
    // NOTE: Have to use this ugly hack since List generic is not covariant in its type param
    HoodieDataBlock dataBlock = getDataBlock(dataBlockType, (List<IndexedRecord>) (List) records, header);
    writer.appendBlock(dataBlock);
    writer.close();
    Schema projectedSchema = HoodieAvroUtils.generateProjectionSchema(schema, Collections.singletonList("name"));
    List<GenericRecord> projectedRecords = HoodieAvroUtils.rewriteRecords(records, projectedSchema);
    try (Reader reader = HoodieLogFormat.newReader(fs, writer.getLogFile(), projectedSchema, true, false)) {
        assertTrue(reader.hasNext(), "First block should be available");
        HoodieLogBlock nextBlock = reader.next();
        HoodieDataBlock dataBlockRead = (HoodieDataBlock) nextBlock;
        Map<HoodieLogBlockType, Integer> expectedReadBytes = new HashMap<HoodieLogBlockType, Integer>() {

            {
                // not supported
                put(HoodieLogBlockType.AVRO_DATA_BLOCK, 0);
                // not supported
                put(HoodieLogBlockType.HFILE_DATA_BLOCK, 0);
                put(HoodieLogBlockType.PARQUET_DATA_BLOCK, 2605);
            }
        };
        List<IndexedRecord> recordsRead = getRecords(dataBlockRead);
        assertEquals(projectedRecords.size(), recordsRead.size(), "Read records size should be equal to the written records size");
        assertEquals(projectedRecords, recordsRead, "Both records lists should be the same. (ordering guaranteed)");
        assertEquals(dataBlockRead.getSchema(), projectedSchema);
        int bytesRead = (int) BenchmarkCounter.getBytesRead();
        assertEquals(expectedReadBytes.get(dataBlockType), bytesRead, "Read bytes have to match");
    }
}

Example 28 with HoodieDataBlock

use of org.apache.hudi.common.table.log.block.HoodieDataBlock in project hudi by apache.

the class TestHoodieLogFormat method testBasicWriteAndScan.

@Test
public void testBasicWriteAndScan() throws IOException, URISyntaxException, InterruptedException {
    Writer writer = HoodieLogFormat.newWriterBuilder().onParentPath(partitionPath).withFileExtension(HoodieLogFile.DELTA_EXTENSION).withFileId("test-fileid1").overBaseCommit("100").withFs(fs).build();
    Schema schema = getSimpleSchema();
    List<IndexedRecord> records = SchemaTestUtil.generateTestRecords(0, 100);
    List<IndexedRecord> copyOfRecords = records.stream().map(record -> HoodieAvroUtils.rewriteRecord((GenericRecord) record, schema)).collect(Collectors.toList());
    Map<HoodieLogBlock.HeaderMetadataType, String> header = new HashMap<>();
    header.put(HoodieLogBlock.HeaderMetadataType.INSTANT_TIME, "100");
    header.put(HoodieLogBlock.HeaderMetadataType.SCHEMA, getSimpleSchema().toString());
    HoodieDataBlock dataBlock = getDataBlock(DEFAULT_DATA_BLOCK_TYPE, records, header);
    writer.appendBlock(dataBlock);
    writer.close();
    Reader reader = HoodieLogFormat.newReader(fs, writer.getLogFile(), SchemaTestUtil.getSimpleSchema());
    assertTrue(reader.hasNext(), "We wrote a block, we should be able to read it");
    HoodieLogBlock nextBlock = reader.next();
    assertEquals(DEFAULT_DATA_BLOCK_TYPE, nextBlock.getBlockType(), "The next block should be a data block");
    HoodieDataBlock dataBlockRead = (HoodieDataBlock) nextBlock;
    List<IndexedRecord> recordsRead = getRecords(dataBlockRead);
    assertEquals(copyOfRecords.size(), recordsRead.size(), "Read records size should be equal to the written records size");
    assertEquals(copyOfRecords, recordsRead, "Both records lists should be the same. (ordering guaranteed)");
    reader.close();
}

Example 29 with HoodieDataBlock

use of org.apache.hudi.common.table.log.block.HoodieDataBlock in project hudi by apache.

the class TestHoodieLogFormat method testBasicAppend.

@ParameterizedTest
@EnumSource(names = { "AVRO_DATA_BLOCK", "HFILE_DATA_BLOCK", "PARQUET_DATA_BLOCK" })
public void testBasicAppend(HoodieLogBlockType dataBlockType) throws IOException, InterruptedException, URISyntaxException {
    Writer writer = HoodieLogFormat.newWriterBuilder().onParentPath(partitionPath).withFileExtension(HoodieLogFile.DELTA_EXTENSION).withFileId("test-fileid1").overBaseCommit("100").withFs(fs).build();
    List<IndexedRecord> records = SchemaTestUtil.generateTestRecords(0, 100);
    Map<HeaderMetadataType, String> header = new HashMap<>();
    header.put(HoodieLogBlock.HeaderMetadataType.INSTANT_TIME, "100");
    header.put(HoodieLogBlock.HeaderMetadataType.SCHEMA, getSimpleSchema().toString());
    long pos = writer.getCurrentSize();
    HoodieDataBlock dataBlock = getDataBlock(dataBlockType, records, header);
    AppendResult result = writer.appendBlock(dataBlock);
    long size = writer.getCurrentSize();
    assertTrue(size > 0, "We just wrote a block - size should be > 0");
    assertEquals(size, fs.getFileStatus(writer.getLogFile().getPath()).getLen(), "Write should be auto-flushed. The size reported by FileStatus and the writer should match");
    assertEquals(size, result.size());
    assertEquals(writer.getLogFile(), result.logFile());
    assertEquals(0, result.offset());
    writer.close();
}

Example 30 with HoodieDataBlock

use of org.apache.hudi.common.table.log.block.HoodieDataBlock in project hudi by apache.

the class TestHoodieLogFormat method testAppendNotSupported.

/*
   * This is actually a test on concurrent append and not recovery lease. Commenting this out.
   * https://issues.apache.org/jira/browse/HUDI-117
   */
/**
 * @Test public void testLeaseRecovery() throws IOException, URISyntaxException, InterruptedException { Writer writer
 * = HoodieLogFormat.newWriterBuilder().onParentPath(partitionPath)
 * .withFileExtension(HoodieLogFile.DELTA_EXTENSION).withFileId("test-fileid1")
 * .overBaseCommit("100").withFs(fs).build(); List<IndexedRecord> records =
 * SchemaTestUtil.generateTestRecords(0, 100); Map<HoodieLogBlock.HeaderMetadataType, String> header =
 * Maps.newHashMap(); header.put(HoodieLogBlock.HeaderMetadataType.INSTANT_TIME, "100");
 * header.put(HoodieLogBlock.HeaderMetadataType.SCHEMA, getSimpleSchema().toString()); HoodieAvroDataBlock
 * dataBlock = new HoodieAvroDataBlock(records, header); writer = writer.appendBlock(dataBlock); long size1 =
 * writer.getCurrentSize(); // do not close this writer - this simulates a data note appending to a log dying
 * without closing the file // writer.close();
 * <p>
 * writer = HoodieLogFormat.newWriterBuilder().onParentPath(partitionPath)
 * .withFileExtension(HoodieLogFile.DELTA_EXTENSION).withFileId("test-fileid1").overBaseCommit("100")
 * .withFs(fs).build(); records = SchemaTestUtil.generateTestRecords(0, 100);
 * header.put(HoodieLogBlock.HeaderMetadataType.SCHEMA, getSimpleSchema().toString()); dataBlock = new
 * HoodieAvroDataBlock(records, header); writer = writer.appendBlock(dataBlock); long size2 =
 * writer.getCurrentSize(); assertTrue("We just wrote a new block - size2 should be > size1", size2 > size1);
 * assertEquals("Write should be auto-flushed. The size reported by FileStatus and the writer should match",
 * size2, fs.getFileStatus(writer.getLogFile().getPath()).getLen()); writer.close(); }
 */
@Test
public void testAppendNotSupported() throws IOException, URISyntaxException, InterruptedException {
    // Use some fs like LocalFileSystem, that does not support appends
    Path localPartitionPath = new Path("file://" + partitionPath);
    FileSystem localFs = FSUtils.getFs(localPartitionPath.toString(), HoodieTestUtils.getDefaultHadoopConf());
    Path testPath = new Path(localPartitionPath, "append_test");
    localFs.mkdirs(testPath);
    // Some data & append two times.
    List<IndexedRecord> records = SchemaTestUtil.generateTestRecords(0, 100);
    Map<HoodieLogBlock.HeaderMetadataType, String> header = new HashMap<>();
    header.put(HoodieLogBlock.HeaderMetadataType.INSTANT_TIME, "100");
    header.put(HoodieLogBlock.HeaderMetadataType.SCHEMA, getSimpleSchema().toString());
    HoodieDataBlock dataBlock = getDataBlock(DEFAULT_DATA_BLOCK_TYPE, records, header);
    for (int i = 0; i < 2; i++) {
        Writer writer = HoodieLogFormat.newWriterBuilder().onParentPath(testPath).withFileExtension(HoodieArchivedLogFile.ARCHIVE_EXTENSION).withFileId("commits.archive").overBaseCommit("").withFs(localFs).build();
        writer.appendBlock(dataBlock);
        writer.close();
    }
    // ensure there are two log file versions, with same data.
    FileStatus[] statuses = localFs.listStatus(testPath);
    assertEquals(2, statuses.length);
}