public void nextRawValue(DataInputBuffer value) throws IOException {
final DataInputBuffer vb = kvIter.getValue();
final int vp = vb.getPosition();
final int vlen = vb.getLength() - vp;
value.reset(vb.getData(), vp, vlen);
bytesRead += vlen;
}
private static void rawValueToTextBytes(
DataOutputBuffer dataBuffer, DataInputBuffer inputBuffer, TextBytes textOut)
throws IOException {
inputBuffer.reset(dataBuffer.getData(), dataBuffer.getLength());
int newLength = WritableUtils.readVInt(inputBuffer);
textOut.set(inputBuffer.getData(), inputBuffer.getPosition(), newLength);
}
/**
* input/output simple records.
*
* @throws Exception if test was failed
*/
@SuppressWarnings("unchecked")
@Test
public void simple_record() throws Exception {
ModelLoader loader = generate();
Class<?> type = loader.modelType("Simple");
assertThat(type.isAnnotationPresent(ModelInputLocation.class), is(true));
assertThat(type.isAnnotationPresent(ModelOutputLocation.class), is(true));
ModelWrapper object = loader.newModel("Simple");
DataOutputBuffer output = new DataOutputBuffer();
ModelOutput<Object> modelOut =
(ModelOutput<Object>)
type.getAnnotation(ModelOutputLocation.class)
.value()
.getDeclaredConstructor(RecordEmitter.class)
.newInstance(new TsvEmitter(new OutputStreamWriter(output, "UTF-8")));
object.set("sid", 1L);
object.set("value", new Text("hello"));
modelOut.write(object.unwrap());
object.set("sid", 2L);
object.set("value", new Text("world"));
modelOut.write(object.unwrap());
object.set("sid", 3L);
object.set("value", null);
modelOut.write(object.unwrap());
modelOut.close();
DataInputBuffer input = new DataInputBuffer();
input.reset(output.getData(), output.getLength());
ModelInput<Object> modelIn =
(ModelInput<Object>)
type.getAnnotation(ModelInputLocation.class)
.value()
.getDeclaredConstructor(RecordParser.class)
.newInstance(new TsvParser(new InputStreamReader(input, "UTF-8")));
ModelWrapper copy = loader.newModel("Simple");
modelIn.readTo(copy.unwrap());
assertThat(copy.get("sid"), is((Object) 1L));
assertThat(copy.get("value"), is((Object) new Text("hello")));
modelIn.readTo(copy.unwrap());
assertThat(copy.get("sid"), is((Object) 2L));
assertThat(copy.get("value"), is((Object) new Text("world")));
modelIn.readTo(copy.unwrap());
assertThat(copy.get("sid"), is((Object) 3L));
assertThat(copy.getOption("value").isNull(), is(true));
assertThat(input.read(), is(-1));
modelIn.close();
}
public boolean nextRawKey(DataInputBuffer key) throws IOException {
if (kvIter.next()) {
final DataInputBuffer kb = kvIter.getKey();
final int kp = kb.getPosition();
final int klen = kb.getLength() - kp;
key.reset(kb.getData(), kp, klen);
bytesRead += klen;
return true;
}
return false;
}
static <T extends Writable> T read(T writable, byte[] bytes) {
DataInputBuffer buffer = new DataInputBuffer();
buffer.reset(bytes, bytes.length);
try {
writable.readFields(buffer);
assertThat("Enf of Stream", buffer.read(), is(-1));
} catch (IOException e) {
throw new AssertionError(e);
}
return writable;
}
@Override
@SuppressWarnings("unchecked")
public void run(final JobConf job, final TaskUmbilicalProtocol umbilical) throws IOException {
final Reporter reporter = getReporter(umbilical);
// start thread that will handle communication with parent
startCommunicationThread(umbilical);
int numReduceTasks = conf.getNumReduceTasks();
LOG.info("numReduceTasks: " + numReduceTasks);
MapOutputCollector collector = null;
if (numReduceTasks > 0) {
collector = new MapOutputBuffer(umbilical, job, reporter);
} else {
collector = new DirectMapOutputCollector(umbilical, job, reporter);
}
// reinstantiate the split
try {
instantiatedSplit =
(InputSplit) ReflectionUtils.newInstance(job.getClassByName(splitClass), job);
} catch (ClassNotFoundException exp) {
IOException wrap = new IOException("Split class " + splitClass + " not found");
wrap.initCause(exp);
throw wrap;
}
DataInputBuffer splitBuffer = new DataInputBuffer();
splitBuffer.reset(split.get(), 0, split.getSize());
instantiatedSplit.readFields(splitBuffer);
// if it is a file split, we can give more details
if (instantiatedSplit instanceof FileSplit) {
FileSplit fileSplit = (FileSplit) instantiatedSplit;
job.set("map.input.file", fileSplit.getPath().toString());
job.setLong("map.input.start", fileSplit.getStart());
job.setLong("map.input.length", fileSplit.getLength());
}
RecordReader rawIn = // open input
job.getInputFormat().getRecordReader(instantiatedSplit, job, reporter);
RecordReader in = new TrackedRecordReader(rawIn, getCounters());
MapRunnable runner = (MapRunnable) ReflectionUtils.newInstance(job.getMapRunnerClass(), job);
try {
runner.run(in, collector, reporter);
collector.flush();
} finally {
// close
in.close(); // close input
collector.close();
}
done(umbilical);
}
@Deprecated
private void readFields(byte[] bytes, int offset, int len) throws IOException {
if (bytes == null || len <= 0) {
throw new IllegalArgumentException("Can't build a writable with empty " + "bytes array");
}
DataInputBuffer in = new DataInputBuffer();
try {
in.reset(bytes, offset, len);
this.readFields(in);
} finally {
in.close();
}
}
/**
* all primitive types.
*
* @throws Exception if test was failed
*/
@SuppressWarnings("unchecked")
@Test
public void primitives() throws Exception {
ModelLoader loader = generate();
Class<?> type = loader.modelType("Primitives");
assertThat(type.isAnnotationPresent(ModelInputLocation.class), is(true));
assertThat(type.isAnnotationPresent(ModelOutputLocation.class), is(true));
ModelWrapper object = loader.newModel("Primitives");
object.set("type_boolean", true);
object.set("type_byte", (byte) 64);
object.set("type_short", (short) 256);
object.set("type_int", 100);
object.set("type_long", 200L);
object.set("type_float", 300.f);
object.set("type_double", 400.d);
object.set("type_decimal", new BigDecimal("1234.567"));
object.set("type_text", new Text("Hello, world!"));
object.set("type_date", new Date(2011, 3, 31));
object.set("type_datetime", new DateTime(2011, 3, 31, 23, 30, 1));
DataOutputBuffer output = new DataOutputBuffer();
ModelOutput<Object> modelOut =
(ModelOutput<Object>)
type.getAnnotation(ModelOutputLocation.class)
.value()
.getDeclaredConstructor(RecordEmitter.class)
.newInstance(new TsvEmitter(new OutputStreamWriter(output, "UTF-8")));
modelOut.write(object.unwrap());
modelOut.write(object.unwrap());
modelOut.write(object.unwrap());
modelOut.close();
DataInputBuffer input = new DataInputBuffer();
input.reset(output.getData(), output.getLength());
ModelInput<Object> modelIn =
(ModelInput<Object>)
type.getAnnotation(ModelInputLocation.class)
.value()
.getDeclaredConstructor(RecordParser.class)
.newInstance(new TsvParser(new InputStreamReader(input, "UTF-8")));
ModelWrapper copy = loader.newModel("Primitives");
modelIn.readTo(copy.unwrap());
assertThat(object.unwrap(), equalTo(copy.unwrap()));
assertThat(input.read(), is(-1));
modelIn.close();
}
protected static void assertSerializable(HadoopSerialization ser, ITuple tuple, boolean debug)
throws IOException {
DataInputBuffer input = new DataInputBuffer();
DataOutputBuffer output = new DataOutputBuffer();
DatumWrapper<ITuple> wrapper = new DatumWrapper<ITuple>(tuple);
ser.ser(wrapper, output);
input.reset(output.getData(), 0, output.getLength());
DatumWrapper<ITuple> wrapper2 = new DatumWrapper<ITuple>();
wrapper2 = ser.deser(wrapper2, input);
if (debug) {
System.out.println("D:" + wrapper2.datum());
}
assertEquals(tuple, wrapper2.datum());
}
/** Used by child copy constructors. */
protected synchronized void copy(Writable other) {
if (other != null) {
try {
DataOutputBuffer out = new DataOutputBuffer();
other.write(out);
DataInputBuffer in = new DataInputBuffer();
in.reset(out.getData(), out.getLength());
readFields(in);
} catch (IOException e) {
throw new IllegalArgumentException("map cannot be copied: " + e.getMessage());
}
} else {
throw new IllegalArgumentException("source map cannot be null");
}
}
public DataInputBuffer getKey() throws IOException {
final int kvoff = kvoffsets[current % kvoffsets.length];
keybuf.reset(
kvbuffer,
kvindices[kvoff + KEYSTART],
kvindices[kvoff + VALSTART] - kvindices[kvoff + KEYSTART]);
return keybuf;
}
public DataInputBuffer getKey() throws IOException {
final int kvoff = offsetFor(current);
keybuf.reset(
kvbuffer,
kvmeta.get(kvoff + KEYSTART),
kvmeta.get(kvoff + VALSTART) - kvmeta.get(kvoff + KEYSTART));
return keybuf;
}
@Test
public void testReadWriteReplicaState() {
try {
DataOutputBuffer out = new DataOutputBuffer();
DataInputBuffer in = new DataInputBuffer();
for (HdfsServerConstants.ReplicaState repState : HdfsServerConstants.ReplicaState.values()) {
repState.write(out);
in.reset(out.getData(), out.getLength());
HdfsServerConstants.ReplicaState result = HdfsServerConstants.ReplicaState.read(in);
assertTrue("testReadWrite error !!!", repState == result);
out.reset();
in.reset();
}
} catch (Exception ex) {
fail("testReadWrite ex error ReplicaState");
}
}
/**
* Parses all the HRegionInfo instances from the passed in stream until EOF. Presumes the
* HRegionInfo's were serialized to the stream with {@link #toDelimitedByteArray()}
*
* @param bytes serialized bytes
* @param offset the start offset into the byte[] buffer
* @param length how far we should read into the byte[] buffer
* @return All the hregioninfos that are in the byte array. Keeps reading till we hit the end.
*/
public static List<HRegionInfo> parseDelimitedFrom(
final byte[] bytes, final int offset, final int length) throws IOException {
if (bytes == null) {
throw new IllegalArgumentException("Can't build an object with empty bytes array");
}
DataInputBuffer in = new DataInputBuffer();
List<HRegionInfo> hris = new ArrayList<HRegionInfo>();
try {
in.reset(bytes, offset, length);
while (in.available() > 0) {
HRegionInfo hri = parseFrom(in);
hris.add(hri);
}
} finally {
in.close();
}
return hris;
}
/** Write IFile by reading data from inputFile */
private void createIFile(WriterOptions options, KV_TRAIT trait) throws IOException {
Writer writer = IFile2.createWriter(options);
FSDataInputStream in = fs.open(inputCSVFile);
BufferedReader reader = new BufferedReader(new InputStreamReader(in));
while (reader.ready()) {
String line = reader.readLine();
String[] kv = line.split(",");
Iterator<String> it = Splitter.on(",").omitEmptyStrings().split(line).iterator();
String k = it.next();
String v = it.next();
key.reset(k.getBytes(), k.getBytes().length);
value.reset(v.getBytes(), v.getBytes().length);
writer.append(key, value);
}
reader.close();
writer.close();
}
public int read() throws IOException {
int ret;
if (null == inbuf || -1 == (ret = inbuf.read())) {
if (!r.next(key, val)) {
return -1;
}
byte[] tmp = key.toString().getBytes();
outbuf.write(tmp, 0, tmp.length);
outbuf.write('\t');
tmp = val.toString().getBytes();
outbuf.write(tmp, 0, tmp.length);
outbuf.write('\n');
inbuf.reset(outbuf.getData(), outbuf.getLength());
outbuf.reset();
ret = inbuf.read();
}
return ret;
}
@SuppressWarnings("unchecked")
protected boolean lessThan(Object a, Object b) {
DataInputBuffer key1 = ((Segment<K, V>) a).getKey();
DataInputBuffer key2 = ((Segment<K, V>) b).getKey();
int s1 = key1.getPosition();
int l1 = key1.getLength() - s1;
int s2 = key2.getPosition();
int l2 = key2.getLength() - s2;
return comparator.compare(key1.getData(), s1, l1, key2.getData(), s2, l2) < 0;
}
private FileStatus getFileStatus(FileStatus fileStatus) throws IOException {
FileStatus status = new FileStatus();
buffer.reset();
DataOutputStream out = new DataOutputStream(buffer);
fileStatus.write(out);
in.reset(buffer.toByteArray(), 0, buffer.size());
status.readFields(in);
return status;
}
protected static void assertSerializable(
TupleSerializer ser, TupleDeserializer deser, DatumWrapper<ITuple> tuple, boolean debug)
throws IOException {
DataOutputBuffer output = new DataOutputBuffer();
ser.open(output);
ser.serialize(tuple);
ser.close();
DataInputBuffer input = new DataInputBuffer();
input.reset(output.getData(), 0, output.getLength());
DatumWrapper<ITuple> deserializedTuple = new DatumWrapper<ITuple>();
deser.open(input);
deserializedTuple = deser.deserialize(deserializedTuple);
deser.close();
if (debug) {
System.out.println("D:" + deserializedTuple.datum());
}
assertEquals(tuple.datum(), deserializedTuple.datum());
}
@Override
public int compare(byte[] b1, int s1, int l1, byte[] b2, int s2, int l2) {
try {
final int ret;
di.reset(b1, s1, l1);
final int x1 = WritableUtils.readVInt(di);
di.reset(b2, s2, l2);
final int x2 = WritableUtils.readVInt(di);
final int t1 = b1[s1 + x1];
final int t2 = b2[s2 + x2];
if (t1 == GridmixKey.REDUCE_SPEC || t2 == GridmixKey.REDUCE_SPEC) {
ret = t1 - t2;
} else {
assert t1 == GridmixKey.DATA;
assert t2 == GridmixKey.DATA;
ret = WritableComparator.compareBytes(b1, s1, x1, b2, s2, x2);
}
di.reset(reset, 0, 0);
return ret;
} catch (IOException e) {
throw new RuntimeException(e);
}
}
public void reset(byte[] buffer, int start, int length) {
this.buffer = buffer;
this.start = start;
this.length = length;
if (start + length > bufvoid) {
this.buffer = new byte[this.length];
final int taillen = bufvoid - start;
System.arraycopy(buffer, start, this.buffer, 0, taillen);
System.arraycopy(buffer, 0, this.buffer, taillen, length - taillen);
this.start = 0;
}
super.reset(this.buffer, this.start, this.length);
}
public static class SpecGroupingComparator implements RawComparator<GridmixKey> {
private final DataInputBuffer di = new DataInputBuffer();
private final byte[] reset = di.getData();
@Override
public int compare(GridmixKey g1, GridmixKey g2) {
final byte t1 = g1.getType();
final byte t2 = g2.getType();
if (t1 == GridmixKey.REDUCE_SPEC || t2 == GridmixKey.REDUCE_SPEC) {
return t1 - t2;
}
assert t1 == GridmixKey.DATA;
assert t2 == GridmixKey.DATA;
return g1.compareTo(g2);
}
@Override
public int compare(byte[] b1, int s1, int l1, byte[] b2, int s2, int l2) {
try {
final int ret;
di.reset(b1, s1, l1);
final int x1 = WritableUtils.readVInt(di);
di.reset(b2, s2, l2);
final int x2 = WritableUtils.readVInt(di);
final int t1 = b1[s1 + x1];
final int t2 = b2[s2 + x2];
if (t1 == GridmixKey.REDUCE_SPEC || t2 == GridmixKey.REDUCE_SPEC) {
ret = t1 - t2;
} else {
assert t1 == GridmixKey.DATA;
assert t2 == GridmixKey.DATA;
ret = WritableComparator.compareBytes(b1, s1, x1, b2, s2, x2);
}
di.reset(reset, 0, 0);
return ret;
} catch (IOException e) {
throw new RuntimeException(e);
}
}
}
private void sortAndSpill() throws IOException {
// approximate the length of the output file to be the length of the
// buffer + header lengths for the partitions
long size =
(bufend >= bufstart ? bufend - bufstart : (bufvoid - bufend) + bufstart)
+ partitions * APPROX_HEADER_LENGTH;
FSDataOutputStream out = null;
FSDataOutputStream indexOut = null;
try {
// create spill file
Path filename = mapOutputFile.getSpillFileForWrite(getTaskID(), numSpills, size);
out = localFs.create(filename);
// create spill index
Path indexFilename =
mapOutputFile.getSpillIndexFileForWrite(
getTaskID(), numSpills, partitions * MAP_OUTPUT_INDEX_RECORD_LENGTH);
indexOut = localFs.create(indexFilename);
final int endPosition = (kvend > kvstart) ? kvend : kvoffsets.length + kvend;
sorter.sort(MapOutputBuffer.this, kvstart, endPosition, reporter);
int spindex = kvstart;
InMemValBytes value = new InMemValBytes();
for (int i = 0; i < partitions; ++i) {
IFile.Writer<K, V> writer = null;
try {
long segmentStart = out.getPos();
writer = new Writer<K, V>(job, out, keyClass, valClass, codec);
if (null == combinerClass) {
// spill directly
DataInputBuffer key = new DataInputBuffer();
while (spindex < endPosition
&& kvindices[kvoffsets[spindex % kvoffsets.length] + PARTITION] == i) {
final int kvoff = kvoffsets[spindex % kvoffsets.length];
getVBytesForOffset(kvoff, value);
key.reset(
kvbuffer,
kvindices[kvoff + KEYSTART],
(kvindices[kvoff + VALSTART] - kvindices[kvoff + KEYSTART]));
writer.append(key, value);
++spindex;
}
} else {
int spstart = spindex;
while (spindex < endPosition
&& kvindices[kvoffsets[spindex % kvoffsets.length] + PARTITION] == i) {
++spindex;
}
// Note: we would like to avoid the combiner if we've fewer
// than some threshold of records for a partition
if (spstart != spindex) {
combineCollector.setWriter(writer);
RawKeyValueIterator kvIter = new MRResultIterator(spstart, spindex);
combineAndSpill(kvIter, combineInputCounter);
}
}
// close the writer
writer.close();
// write the index as <offset, raw-length, compressed-length>
writeIndexRecord(indexOut, out, segmentStart, writer);
writer = null;
} finally {
if (null != writer) writer.close();
}
}
LOG.info("Finished spill " + numSpills);
++numSpills;
} finally {
if (out != null) out.close();
if (indexOut != null) indexOut.close();
}
}
@Test
public void testBuiltInGzipDecompressorExceptions() {
BuiltInGzipDecompressor decompresser = new BuiltInGzipDecompressor();
try {
decompresser.setInput(null, 0, 1);
} catch (NullPointerException ex) {
// expected
} catch (Exception ex) {
fail("testBuiltInGzipDecompressorExceptions npe error " + ex);
}
try {
decompresser.setInput(new byte[] {0}, 0, -1);
} catch (ArrayIndexOutOfBoundsException ex) {
// expected
} catch (Exception ex) {
fail("testBuiltInGzipDecompressorExceptions aioob error" + ex);
}
assertTrue("decompresser.getBytesRead error", decompresser.getBytesRead() == 0);
assertTrue("decompresser.getRemaining error", decompresser.getRemaining() == 0);
decompresser.reset();
decompresser.end();
InputStream decompStream = null;
try {
// invalid 0 and 1 bytes , must be 31, -117
int buffSize = 1 * 1024;
byte buffer[] = new byte[buffSize];
Decompressor decompressor = new BuiltInGzipDecompressor();
DataInputBuffer gzbuf = new DataInputBuffer();
decompStream = new DecompressorStream(gzbuf, decompressor);
gzbuf.reset(new byte[] {0, 0, 1, 1, 1, 1, 11, 1, 1, 1, 1}, 11);
decompStream.read(buffer);
} catch (IOException ioex) {
// expected
} catch (Exception ex) {
fail("invalid 0 and 1 byte in gzip stream" + ex);
}
// invalid 2 byte, must be 8
try {
int buffSize = 1 * 1024;
byte buffer[] = new byte[buffSize];
Decompressor decompressor = new BuiltInGzipDecompressor();
DataInputBuffer gzbuf = new DataInputBuffer();
decompStream = new DecompressorStream(gzbuf, decompressor);
gzbuf.reset(new byte[] {31, -117, 7, 1, 1, 1, 1, 11, 1, 1, 1, 1}, 11);
decompStream.read(buffer);
} catch (IOException ioex) {
// expected
} catch (Exception ex) {
fail("invalid 2 byte in gzip stream" + ex);
}
try {
int buffSize = 1 * 1024;
byte buffer[] = new byte[buffSize];
Decompressor decompressor = new BuiltInGzipDecompressor();
DataInputBuffer gzbuf = new DataInputBuffer();
decompStream = new DecompressorStream(gzbuf, decompressor);
gzbuf.reset(new byte[] {31, -117, 8, -32, 1, 1, 1, 11, 1, 1, 1, 1}, 11);
decompStream.read(buffer);
} catch (IOException ioex) {
// expected
} catch (Exception ex) {
fail("invalid 3 byte in gzip stream" + ex);
}
try {
int buffSize = 1 * 1024;
byte buffer[] = new byte[buffSize];
Decompressor decompressor = new BuiltInGzipDecompressor();
DataInputBuffer gzbuf = new DataInputBuffer();
decompStream = new DecompressorStream(gzbuf, decompressor);
gzbuf.reset(new byte[] {31, -117, 8, 4, 1, 1, 1, 11, 1, 1, 1, 1}, 11);
decompStream.read(buffer);
} catch (IOException ioex) {
// expected
} catch (Exception ex) {
fail("invalid 3 byte make hasExtraField" + ex);
}
}
/**
* create inmemory segments
*
* @return
* @throws IOException
*/
public List<TezMerger.Segment> createInMemStreams() throws IOException {
int numberOfStreams = Math.max(2, rnd.nextInt(10));
LOG.info("No of streams : " + numberOfStreams);
SerializationFactory serializationFactory = new SerializationFactory(conf);
Serializer keySerializer = serializationFactory.getSerializer(keyClass);
Serializer valueSerializer = serializationFactory.getSerializer(valClass);
LocalDirAllocator localDirAllocator =
new LocalDirAllocator(TezRuntimeFrameworkConfigs.LOCAL_DIRS);
InputContext context = createTezInputContext();
MergeManager mergeManager =
new MergeManager(
conf,
fs,
localDirAllocator,
context,
null,
null,
null,
null,
null,
1024 * 1024 * 10,
null,
false,
-1);
DataOutputBuffer keyBuf = new DataOutputBuffer();
DataOutputBuffer valBuf = new DataOutputBuffer();
DataInputBuffer keyIn = new DataInputBuffer();
DataInputBuffer valIn = new DataInputBuffer();
keySerializer.open(keyBuf);
valueSerializer.open(valBuf);
List<TezMerger.Segment> segments = new LinkedList<TezMerger.Segment>();
for (int i = 0; i < numberOfStreams; i++) {
BoundedByteArrayOutputStream bout = new BoundedByteArrayOutputStream(1024 * 1024);
InMemoryWriter writer = new InMemoryWriter(bout);
Map<Writable, Writable> data = createData();
// write data
for (Map.Entry<Writable, Writable> entry : data.entrySet()) {
keySerializer.serialize(entry.getKey());
valueSerializer.serialize(entry.getValue());
keyIn.reset(keyBuf.getData(), 0, keyBuf.getLength());
valIn.reset(valBuf.getData(), 0, valBuf.getLength());
writer.append(keyIn, valIn);
originalData.put(entry.getKey(), entry.getValue());
keyBuf.reset();
valBuf.reset();
keyIn.reset();
valIn.reset();
}
IFile.Reader reader =
new InMemoryReader(mergeManager, null, bout.getBuffer(), 0, bout.getBuffer().length);
segments.add(new TezMerger.Segment(reader, true));
data.clear();
writer.close();
}
return segments;
}
protected void spill(int mstart, int mend) throws IOException, InterruptedException {
// approximate the length of the output file to be the length of the
// buffer + header lengths for the partitions
final long size =
(bufend >= bufstart ? bufend - bufstart : (bufvoid - bufend) + bufstart)
+ partitions * APPROX_HEADER_LENGTH;
FSDataOutputStream out = null;
try {
// create spill file
final TezSpillRecord spillRec = new TezSpillRecord(partitions);
final Path filename = mapOutputFile.getSpillFileForWrite(numSpills, size);
out = rfs.create(filename);
int spindex = mstart;
final InMemValBytes value = createInMemValBytes();
for (int i = 0; i < partitions; ++i) {
IFile.Writer writer = null;
try {
long segmentStart = out.getPos();
writer = new Writer(conf, out, keyClass, valClass, codec, spilledRecordsCounter, null);
if (combiner == null) {
// spill directly
DataInputBuffer key = new DataInputBuffer();
while (spindex < mend && kvmeta.get(offsetFor(spindex) + PARTITION) == i) {
final int kvoff = offsetFor(spindex);
int keystart = kvmeta.get(kvoff + KEYSTART);
int valstart = kvmeta.get(kvoff + VALSTART);
key.reset(kvbuffer, keystart, valstart - keystart);
getVBytesForOffset(kvoff, value);
writer.append(key, value);
++spindex;
}
} else {
int spstart = spindex;
while (spindex < mend && kvmeta.get(offsetFor(spindex) + PARTITION) == i) {
++spindex;
}
// Note: we would like to avoid the combiner if we've fewer
// than some threshold of records for a partition
if (spstart != spindex) {
TezRawKeyValueIterator kvIter = new MRResultIterator(spstart, spindex);
if (LOG.isDebugEnabled()) {
LOG.debug("Running combine processor");
}
runCombineProcessor(kvIter, writer);
}
}
// close the writer
writer.close();
if (numSpills > 0) {
additionalSpillBytesWritten.increment(writer.getCompressedLength());
numAdditionalSpills.increment(1);
// Reset the value will be set during the final merge.
outputBytesWithOverheadCounter.setValue(0);
} else {
// Set this up for the first write only. Subsequent ones will be handled in the final
// merge.
outputBytesWithOverheadCounter.increment(writer.getRawLength());
}
// record offsets
final TezIndexRecord rec =
new TezIndexRecord(segmentStart, writer.getRawLength(), writer.getCompressedLength());
spillRec.putIndex(rec, i);
writer = null;
} finally {
if (null != writer) writer.close();
}
}
if (totalIndexCacheMemory >= indexCacheMemoryLimit) {
// create spill index file
Path indexFilename =
mapOutputFile.getSpillIndexFileForWrite(
numSpills, partitions * MAP_OUTPUT_INDEX_RECORD_LENGTH);
spillRec.writeToFile(indexFilename, conf);
} else {
indexCacheList.add(spillRec);
totalIndexCacheMemory += spillRec.size() * MAP_OUTPUT_INDEX_RECORD_LENGTH;
}
LOG.info("Finished spill " + numSpills);
++numSpills;
} finally {
if (out != null) out.close();
}
}
private static void rawValueToWritable(
RawRecordValue rawValue, DataInputBuffer inputBuffer, Writable typeOut) throws IOException {
inputBuffer.reset(rawValue.data.getData(), rawValue.data.getLength());
typeOut.readFields(inputBuffer);
}