Package org.apache.storm.streams.processors

Class CoGroupByKeyProcessor<K,V1,V2>

java.lang.Object

org.apache.storm.streams.processors.CoGroupByKeyProcessor<K,V1,V2>

All Implemented Interfaces:

Serializable, BatchProcessor, Processor<Pair<K,?>>

public class CoGroupByKeyProcessor<K,V1,V2>
extends Object
implements BatchProcessor

co-group by key implementation.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field	Description
protected ProcessorContext	context

Constructor Summary

Constructors

Constructor	Description
CoGroupByKeyProcessor(String firstStream, String secondStream)

Method Summary

Modifier and Type	Method	Description
void	execute(Pair<K,?> input, String sourceStream)	Executes some operations on the input and possibly emits some results.
protected void	execute(T input)	Execute some operation on the input value.
void	finish()	This is triggered to signal the end of the current batch of values.
void	init(ProcessorContext context)	Initializes the processor.
protected <R> void	mayBeForwardAggUpdate(Supplier<R> result)	Forwards the result update to downstream processors.
void	punctuate(String stream)	Punctuation marks end of a batch which can be used to compute and pass the results of one stage in the pipeline to the next.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

context

protected ProcessorContext context

Constructor Detail

CoGroupByKeyProcessor

public CoGroupByKeyProcessor(String firstStream,
                             String secondStream)

Method Detail

execute

public void execute(Pair<K,?> input,
                    String sourceStream)

Executes some operations on the input and possibly emits some results. Processors that do not care about the source stream should override BaseProcessor.execute(Object) instead.

Specified by:

execute in interface Processor<K>

Parameters:

input - the input to be processed

sourceStream - the source stream id from where the input is received

finish

public void finish()

This is triggered to signal the end of the current batch of values. Sub classes can override this to emit the result of a batch of values, for e.g. to emit the result of an aggregate or join operation on a batch of values. If a processor does per-value operation like filter, map etc, they can choose to ignore this.

init

public void init(ProcessorContext context)

Initializes the processor. This is typically invoked from the underlying storm bolt's prepare method.

Specified by:

init in interface Processor<T>

Parameters:

context - the processor context

execute

protected void execute(T input)

Execute some operation on the input value. Sub classes can override this when then don't care about the source stream from where the input is received.

Parameters:

input - the input

punctuate

public void punctuate(String stream)

Punctuation marks end of a batch which can be used to compute and pass the results of one stage in the pipeline to the next. For e.g. emit the results of an aggregation.

Specified by:

punctuate in interface Processor<T>

Parameters:

stream - the stream id on which the punctuation arrived

mayBeForwardAggUpdate

protected final <R> void mayBeForwardAggUpdate(Supplier<R> result)

Forwards the result update to downstream processors. Processors that operate on a batch of tuples, like aggregation, join etc can use this to emit the partial results on each input if they are operating in non-windowed mode.

Type Parameters:

R - the result type

Parameters:

result - the result function