Java Code Examples for org.apache.hadoop.mapred.RecordWriter#write()

@Test(enabled = true)
public void testWriteChunkData() throws Exception {
  NullWritable nada = NullWritable.get();
  MneDurableOutputSession<DurableChunk<?>> sess =
          new MneDurableOutputSession<DurableChunk<?>>(m_tacontext, null,
                  MneConfigHelper.DEFAULT_OUTPUT_CONFIG_PREFIX);
  MneDurableOutputValue<DurableChunk<?>> mdvalue =
          new MneDurableOutputValue<DurableChunk<?>>(sess);
  OutputFormat<NullWritable, MneDurableOutputValue<DurableChunk<?>>> outputFormat =
          new MneOutputFormat<MneDurableOutputValue<DurableChunk<?>>>();
  RecordWriter<NullWritable, MneDurableOutputValue<DurableChunk<?>>> writer =
          outputFormat.getRecordWriter(null, m_conf, null, null);
  DurableChunk<?> dchunk = null;
  Checksum cs = new CRC32();
  cs.reset();
  for (int i = 0; i < m_reccnt; ++i) {
    dchunk = genupdDurableChunk(sess, cs);
    Assert.assertNotNull(dchunk);
    writer.write(nada, mdvalue.of(dchunk));
  }
  m_checksum = cs.getValue();
  writer.close(null);
  sess.close();
}
 

@Test(enabled = true)
public void testWriteLongData() throws Exception {
  NullWritable nada = NullWritable.get();
  MneDurableOutputSession<Long> sess =
          new MneDurableOutputSession<Long>(m_tacontext, null,
                  MneConfigHelper.DEFAULT_OUTPUT_CONFIG_PREFIX);
  MneDurableOutputValue<Long> mdvalue =
          new MneDurableOutputValue<Long>(sess);
  OutputFormat<NullWritable, MneDurableOutputValue<Long>> outputFormat =
          new MneOutputFormat<MneDurableOutputValue<Long>>();
  RecordWriter<NullWritable, MneDurableOutputValue<Long>> writer =
          outputFormat.getRecordWriter(null, m_conf, null, null);
  Long val = null;
  for (int i = 0; i < m_reccnt; ++i) {
    val = m_rand.nextLong();
    m_sum += val;
    writer.write(nada, mdvalue.of(val));
  }
  writer.close(null);
  sess.close();
}
 

@Test(enabled = true)
public void testWritePersonData() throws Exception {
  NullWritable nada = NullWritable.get();
  MneDurableOutputSession<Person<Long>> sess =
          new MneDurableOutputSession<Person<Long>>(m_tacontext, null,
                  MneConfigHelper.DEFAULT_OUTPUT_CONFIG_PREFIX);
  MneDurableOutputValue<Person<Long>> mdvalue =
          new MneDurableOutputValue<Person<Long>>(sess);
  OutputFormat<NullWritable, MneDurableOutputValue<Person<Long>>> outputFormat =
          new MneOutputFormat<MneDurableOutputValue<Person<Long>>>();
  RecordWriter<NullWritable, MneDurableOutputValue<Person<Long>>> writer =
          outputFormat.getRecordWriter(null, m_conf, null, null);
  Person<Long> person = null;
  for (int i = 0; i < m_reccnt; ++i) {
    person = sess.newDurableObjectRecord();
    person.setAge((short) m_rand.nextInt(50));
    person.setName(String.format("Name: [%s]", Utils.genRandomString()), true);
    m_sumage += person.getAge();
    writer.write(nada, mdvalue.of(person));
  }
  writer.close(null);
  sess.close();
}
 

@Test(enabled = true)
public void testWriteBufferData() throws Exception {
  NullWritable nada = NullWritable.get();
  MneDurableOutputSession<DurableBuffer<?>> sess =
      new MneDurableOutputSession<DurableBuffer<?>>(null, m_conf,
          MneConfigHelper.DEFAULT_OUTPUT_CONFIG_PREFIX);
  MneDurableOutputValue<DurableBuffer<?>> mdvalue =
      new MneDurableOutputValue<DurableBuffer<?>>(sess);
  OutputFormat<NullWritable, MneDurableOutputValue<DurableBuffer<?>>> outputFormat =
      new MneOutputFormat<MneDurableOutputValue<DurableBuffer<?>>>();
  RecordWriter<NullWritable, MneDurableOutputValue<DurableBuffer<?>>> writer =
      outputFormat.getRecordWriter(m_fs, m_conf, null, null);
  DurableBuffer<?> dbuf = null;
  Checksum cs = new CRC32();
  cs.reset();
  for (int i = 0; i < m_reccnt; ++i) {
    dbuf = genupdDurableBuffer(sess, cs);
    Assert.assertNotNull(dbuf);
    writer.write(nada, mdvalue.of(dbuf));
  }
  m_checksum = cs.getValue();
  writer.close(null);
  sess.close();
}
 

@Test
public void testPartialReadingWithSeeks() throws IOException {
  // First we create a file of a small size and having two stripes.
  final String fileName = "file_with_long_col";
  final JobConf jobConf = new JobConf();
  jobConf.setLong(OrcConf.ConfVars.HIVE_ORC_STRIPE_SIZE.varname, 1024L);
  final Path filePath = new Path(this.stagingDir.toString(), fileName);
  final RecordWriter hiveRecordWriter = (RecordWriter) (new OrcOutputFormat().getHiveRecordWriter(
      jobConf, filePath, null, true, new Properties(), null));

  final OrcSerde orcSerde = new OrcSerde();
  ObjectInspector objectInspector = getObjectInspectorFor(
      ImmutableList.of("col1"),
      ImmutableList.of("bigint"));

  final Random rand = new Random();
  for (int i = 0; i < 1000; i++) {
    final List<Object> allColumns = new ArrayList<>();
    allColumns.add(rand.nextLong());
    Object obj = orcSerde.serialize(allColumns, objectInspector);
    hiveRecordWriter.write(NullWritable.get(), obj);
  }
  hiveRecordWriter.close(null);

  // Get all the stripes in the file written.
  final Configuration configuration = new Configuration();
  final FileSystem fileSystem = filePath.getFileSystem(configuration);
  final ReaderImpl readerImpl = new ReaderImpl(fileSystem, filePath, configuration);
  final ArrayList<StripeInformation> stripes = Lists.newArrayList(readerImpl.getStripes());
  assertTrue("Number of stripes produced should be >= 2", stripes.size() >= 2);

  // Read the file back and read with ReaderImpl over only the 2nd stripe.
  final boolean[] toRead = {true, true};
  final RecordReader recordReader = new ReaderImpl(fileSystem, filePath, configuration).rows(
      stripes.get(1).getOffset(), stripes.get(1).getDataLength(), toRead);
  OrcLazyRow row = null;
  final List<Long> longValuesForStripe = new ArrayList<>();
  while (recordReader.hasNext()) {
    row = (OrcLazyRow) recordReader.next(row);
    longValuesForStripe.add(row.getFieldValue(0).materializeLong());
  }

  // Seek to the beginning of the 2nd stripe and ensure that seeking works fine.
  recordReader.seekToRow(stripes.get(0).getNumberOfRows());
  assertEquals(recordReader.getRowNumber(), stripes.get(0).getNumberOfRows());
  row = (OrcLazyRow) recordReader.next(row);
  assertEquals((long) longValuesForStripe.get(0), row.getFieldValue(0).materializeLong());
}