Examples with OldApiWritableFileExtractor org.apache.gobblin.source.extractor.hadoop.OldApiWritableFileExtractor used on opensource projects

Example 1 with OldApiWritableFileExtractor

use of org.apache.gobblin.source.extractor.hadoop.OldApiWritableFileExtractor in project incubator-gobblin by apache.

the class HiveSerDeTest method testAvroOrcSerDes.

/**
 * This test uses Avro SerDe to deserialize data from Avro files, and use ORC SerDe
 * to serialize them into ORC files.
 */
@Test(groups = { "gobblin.serde" })
public void testAvroOrcSerDes() throws IOException, DataRecordException, DataConversionException {
    Properties properties = new Properties();
    properties.load(new FileReader("gobblin-core/src/test/resources/serde/serde.properties"));
    SourceState sourceState = new SourceState(new State(properties), ImmutableList.<WorkUnitState>of());
    OldApiWritableFileSource source = new OldApiWritableFileSource();
    List<WorkUnit> workUnits = source.getWorkunits(sourceState);
    Assert.assertEquals(workUnits.size(), 1);
    WorkUnitState wus = new WorkUnitState(workUnits.get(0));
    wus.addAll(sourceState);
    Closer closer = Closer.create();
    HiveWritableHdfsDataWriter writer = null;
    try {
        OldApiWritableFileExtractor extractor = closer.register((OldApiWritableFileExtractor) source.getExtractor(wus));
        HiveSerDeConverter converter = closer.register(new HiveSerDeConverter());
        writer = closer.register((HiveWritableHdfsDataWriter) new HiveWritableHdfsDataWriterBuilder<>().withBranches(1).withWriterId("0").writeTo(Destination.of(DestinationType.HDFS, sourceState)).writeInFormat(WriterOutputFormat.ORC).build());
        converter.init(wus);
        Writable record;
        while ((record = extractor.readRecord(null)) != null) {
            Iterable<Writable> convertedRecordIterable = converter.convertRecordImpl(null, record, wus);
            Assert.assertEquals(Iterators.size(convertedRecordIterable.iterator()), 1);
            writer.write(convertedRecordIterable.iterator().next());
        }
    } catch (Throwable t) {
        throw closer.rethrow(t);
    } finally {
        closer.close();
        if (writer != null) {
            writer.commit();
        }
        Assert.assertTrue(this.fs.exists(new Path(sourceState.getProp(ConfigurationKeys.WRITER_OUTPUT_DIR), sourceState.getProp(ConfigurationKeys.WRITER_FILE_NAME))));
        HadoopUtils.deletePath(this.fs, new Path(sourceState.getProp(ConfigurationKeys.WRITER_OUTPUT_DIR)), true);
    }
}