public void process(Writable value) throws HiveException {
// A mapper can span multiple files/partitions.
// The serializers need to be reset if the input file changed
ExecMapperContext context = getExecContext();
if (context != null && context.inputFileChanged()) {
// The child operators cleanup if input file has changed
cleanUpInputFileChanged();
}
int childrenDone = 0;
for (MapOpCtx current : currentCtxs) {
Object row = null;
try {
row = current.readRow(value, context);
if (!current.forward(row)) {
childrenDone++;
}
} catch (Exception e) {
// TODO: policy on deserialization errors
String message = toErrorMessage(value, row, current.rowObjectInspector);
if (row == null) {
deserialize_error_count.set(deserialize_error_count.get() + 1);
throw new HiveException("Hive Runtime Error while processing writable " + message, e);
}
throw new HiveException("Hive Runtime Error while processing row " + message, e);
}
}
rowsForwarded(childrenDone, 1);
}
protected boolean isInputFileChangeSensitive(ExecMapperContext mapContext) {
return !(mapContext == null
|| mapContext.getLocalWork() == null
|| mapContext.getLocalWork().getInputFileChangeSensitive() == false);
}
@Override
@SuppressWarnings("unchecked")
public void init(JobConf job, OutputCollector output, Reporter reporter) throws Exception {
perfLogger.PerfLogBegin(CLASS_NAME, PerfLogger.SPARK_INIT_OPERATORS);
super.init(job, output, reporter);
rowObjectInspector = new ObjectInspector[Byte.MAX_VALUE];
ObjectInspector[] valueObjectInspector = new ObjectInspector[Byte.MAX_VALUE];
ObjectInspector keyObjectInspector;
ReduceWork gWork = Utilities.getReduceWork(job);
reducer = gWork.getReducer();
vectorized = gWork.getVectorMode();
reducer.setParentOperators(null); // clear out any parents as reducer is the
// root
isTagged = gWork.getNeedsTagging();
try {
keyTableDesc = gWork.getKeyDesc();
inputKeyDeserializer = ReflectionUtils.newInstance(keyTableDesc.getDeserializerClass(), null);
SerDeUtils.initializeSerDe(inputKeyDeserializer, null, keyTableDesc.getProperties(), null);
keyObjectInspector = inputKeyDeserializer.getObjectInspector();
valueTableDesc = new TableDesc[gWork.getTagToValueDesc().size()];
if (vectorized) {
final int maxTags = gWork.getTagToValueDesc().size();
keyStructInspector = (StructObjectInspector) keyObjectInspector;
batches = new VectorizedRowBatch[maxTags];
valueStructInspectors = new StructObjectInspector[maxTags];
valueStringWriters = new List[maxTags];
keysColumnOffset = keyStructInspector.getAllStructFieldRefs().size();
buffer = new DataOutputBuffer();
}
for (int tag = 0; tag < gWork.getTagToValueDesc().size(); tag++) {
// We should initialize the SerDe with the TypeInfo when available.
valueTableDesc[tag] = gWork.getTagToValueDesc().get(tag);
inputValueDeserializer[tag] =
ReflectionUtils.newInstance(valueTableDesc[tag].getDeserializerClass(), null);
SerDeUtils.initializeSerDe(
inputValueDeserializer[tag], null, valueTableDesc[tag].getProperties(), null);
valueObjectInspector[tag] = inputValueDeserializer[tag].getObjectInspector();
ArrayList<ObjectInspector> ois = new ArrayList<ObjectInspector>();
if (vectorized) {
/* vectorization only works with struct object inspectors */
valueStructInspectors[tag] = (StructObjectInspector) valueObjectInspector[tag];
ObjectPair<VectorizedRowBatch, StandardStructObjectInspector> pair =
VectorizedBatchUtil.constructVectorizedRowBatch(
keyStructInspector,
valueStructInspectors[tag],
gWork.getVectorScratchColumnTypeMap());
batches[tag] = pair.getFirst();
final int totalColumns =
keysColumnOffset + valueStructInspectors[tag].getAllStructFieldRefs().size();
valueStringWriters[tag] = new ArrayList<VectorExpressionWriter>(totalColumns);
valueStringWriters[tag].addAll(
Arrays.asList(
VectorExpressionWriterFactory.genVectorStructExpressionWritables(
keyStructInspector)));
valueStringWriters[tag].addAll(
Arrays.asList(
VectorExpressionWriterFactory.genVectorStructExpressionWritables(
valueStructInspectors[tag])));
rowObjectInspector[tag] = pair.getSecond();
} else {
ois.add(keyObjectInspector);
ois.add(valueObjectInspector[tag]);
// reducer.setGroupKeyObjectInspector(keyObjectInspector);
rowObjectInspector[tag] =
ObjectInspectorFactory.getStandardStructObjectInspector(
Utilities.reduceFieldNameList, ois);
}
}
} catch (Exception e) {
throw new RuntimeException(e);
}
ExecMapperContext execContext = new ExecMapperContext(job);
localWork = gWork.getMapRedLocalWork();
execContext.setJc(jc);
execContext.setLocalWork(localWork);
reducer.passExecContext(execContext);
reducer.setReporter(rp);
OperatorUtils.setChildrenCollector(
Arrays.<Operator<? extends OperatorDesc>>asList(reducer), output);
// initialize reduce operator tree
try {
LOG.info(reducer.dump(0));
reducer.initialize(jc, rowObjectInspector);
if (localWork != null) {
for (Operator<? extends OperatorDesc> dummyOp : localWork.getDummyParentOp()) {
dummyOp.setExecContext(execContext);
dummyOp.initialize(jc, null);
}
}
} catch (Throwable e) {
abort = true;
if (e instanceof OutOfMemoryError) {
// Don't create a new object if we are already out of memory
throw (OutOfMemoryError) e;
} else {
throw new RuntimeException("Reduce operator initialization failed", e);
}
}
perfLogger.PerfLogEnd(CLASS_NAME, PerfLogger.SPARK_INIT_OPERATORS);
}
public static Object[] populateVirtualColumnValues(
ExecMapperContext ctx,
List<VirtualColumn> vcs,
Object[] vcValues,
Deserializer deserializer) {
if (vcs == null) {
return vcValues;
}
if (vcValues == null) {
vcValues = new Object[vcs.size()];
}
for (int i = 0; i < vcs.size(); i++) {
VirtualColumn vc = vcs.get(i);
if (vc.equals(VirtualColumn.FILENAME)) {
if (ctx.inputFileChanged()) {
vcValues[i] = new Text(ctx.getCurrentInputPath().toString());
}
} else if (vc.equals(VirtualColumn.BLOCKOFFSET)) {
long current = ctx.getIoCxt().getCurrentBlockStart();
LongWritable old = (LongWritable) vcValues[i];
if (old == null) {
old = new LongWritable(current);
vcValues[i] = old;
continue;
}
if (current != old.get()) {
old.set(current);
}
} else if (vc.equals(VirtualColumn.ROWOFFSET)) {
long current = ctx.getIoCxt().getCurrentRow();
LongWritable old = (LongWritable) vcValues[i];
if (old == null) {
old = new LongWritable(current);
vcValues[i] = old;
continue;
}
if (current != old.get()) {
old.set(current);
}
} else if (vc.equals(VirtualColumn.RAWDATASIZE)) {
long current = 0L;
SerDeStats stats = deserializer.getSerDeStats();
if (stats != null) {
current = stats.getRawDataSize();
}
LongWritable old = (LongWritable) vcValues[i];
if (old == null) {
old = new LongWritable(current);
vcValues[i] = old;
continue;
}
if (current != old.get()) {
old.set(current);
}
} else if (vc.equals(VirtualColumn.ROWID)) {
if (ctx.getIoCxt().getRecordIdentifier() == null) {
vcValues[i] = null;
} else {
if (vcValues[i] == null) {
vcValues[i] = new Object[RecordIdentifier.Field.values().length];
}
RecordIdentifier.StructInfo.toArray(
ctx.getIoCxt().getRecordIdentifier(), (Object[]) vcValues[i]);
ctx.getIoCxt()
.setRecordIdentifier(null); // so we don't accidentally cache the value; shouldn't
// happen since IO layer either knows how to produce ROW__ID or not - but to be safe
}
}
}
return vcValues;
}
