Package org.springframework.batch.core.step.item

Class FaultTolerantChunkProcessor<I,O>

java.lang.Object
org.springframework.batch.core.step.item.SimpleChunkProcessor<I,O>
org.springframework.batch.core.step.item.FaultTolerantChunkProcessor<I,O>
All Implemented Interfaces:
ChunkProcessor<I>, org.springframework.beans.factory.InitializingBean
public class FaultTolerantChunkProcessor<I,O> extends SimpleChunkProcessor<I,O>
FaultTolerant implementation of the ChunkProcessor interface, that allows for skipping or retry of items that cause exceptions during writing.

  • Constructor Details

  • Method Details

    • setKeyGenerator

      public void setKeyGenerator(KeyGenerator keyGenerator)
      The KeyGenerator to use to identify failed items across rollback. Not used in the case of the buffering flag being true (the default).
      Parameters:
      keyGenerator - the KeyGenerator to set

    • setProcessSkipPolicy

      public void setProcessSkipPolicy(SkipPolicy SkipPolicy)
      Parameters:
      SkipPolicy - the SkipPolicy for item processing

    • setWriteSkipPolicy

      public void setWriteSkipPolicy(SkipPolicy SkipPolicy)
      Parameters:
      SkipPolicy - the SkipPolicy for item writing

    • setRollbackClassifier

      public void setRollbackClassifier(org.springframework.classify.Classifier<Throwable,Boolean> rollbackClassifier)
      A classifier that can distinguish between exceptions that cause rollback (return true) or not (return false).
      Parameters:
      rollbackClassifier - classifier

    • setChunkMonitor

      public void setChunkMonitor(ChunkMonitor chunkMonitor)
      Parameters:
      chunkMonitor - monitor

    • setBuffering

      public void setBuffering(boolean buffering)
      A flag to indicate that items have been buffered and therefore will always come back as a chunk after a rollback. Otherwise things are more complicated because after a rollback the new chunk might or might not contain items from the previous failed chunk.
      Parameters:
      buffering - true if items will be buffered

    • setProcessorTransactional

      public void setProcessorTransactional(boolean processorTransactional)
      Flag to say that the ItemProcessor is transactional (defaults to true). If false then the processor is only called once per item per chunk, even if there are rollbacks with retries and skips.
      Parameters:
      processorTransactional - the flag value to set

    • initializeUserData

      protected void initializeUserData(Chunk<I> inputs)
      Description copied from class: SimpleChunkProcessor
      Extension point for subclasses to allow them to memorise the contents of the inputs, in case they are needed for accounting purposes later. The default implementation sets up some user data to remember the original size of the inputs. If this method is overridden then some or all of SimpleChunkProcessor.isComplete(Chunk), SimpleChunkProcessor.getFilterCount(Chunk, Chunk) and SimpleChunkProcessor.getAdjustedOutputs(Chunk, Chunk) might also need to be, to ensure that the user data is handled consistently.
      Overrides:
      initializeUserData in class SimpleChunkProcessor<I,O>
      Parameters:
      inputs - the inputs for the process

    • getFilterCount

      protected int getFilterCount(Chunk<I> inputs, Chunk<O> outputs)
      Description copied from class: SimpleChunkProcessor
      Extension point for subclasses to calculate the filter count. Defaults to the difference between input size and output size.
      Overrides:
      getFilterCount in class SimpleChunkProcessor<I,O>
      Parameters:
      inputs - the inputs after transformation
      outputs - the outputs after transformation
      Returns:
      the difference in sizes
      See Also:

    • isComplete

      protected boolean isComplete(Chunk<I> inputs)
      Description copied from class: SimpleChunkProcessor
      Extension point for subclasses that want to store additional data in the inputs. Default just checks if inputs are empty.
      Overrides:
      isComplete in class SimpleChunkProcessor<I,O>
      Parameters:
      inputs - the input chunk
      Returns:
      true if it is empty
      See Also:

    • getAdjustedOutputs

      protected Chunk<O> getAdjustedOutputs(Chunk<I> inputs, Chunk<O> outputs)
      Description copied from class: SimpleChunkProcessor
      Extension point for subclasses that want to adjust the outputs based on additional saved data in the inputs. Default implementation just returns the outputs unchanged.
      Overrides:
      getAdjustedOutputs in class SimpleChunkProcessor<I,O>
      Parameters:
      inputs - the inputs for the transformation
      outputs - the result of the transformation
      Returns:
      the outputs unchanged
      See Also:

    • transform

      protected Chunk<O> transform(StepContribution contribution, Chunk<I> inputs) throws Exception
      Overrides:
      transform in class SimpleChunkProcessor<I,O>
      Throws:
      Exception

    • write

      protected void write(StepContribution contribution, Chunk<I> inputs, Chunk<O> outputs) throws Exception
      Description copied from class: SimpleChunkProcessor
      Simple implementation delegates to the SimpleChunkProcessor.doWrite(Chunk) method and increments the write count in the contribution. Subclasses can handle more complicated scenarios, e.g.with fault tolerance. If output items are skipped they should be removed from the inputs as well.
      Overrides:
      write in class SimpleChunkProcessor<I,O>
      Parameters:
      contribution - the current step contribution
      inputs - the inputs that gave rise to the outputs
      outputs - the outputs to write
      Throws:
      Exception - if there is a problem