Package org.apache.hadoop.io.erasurecode.rawcoder

Class DecodingValidator

java.lang.Object

org.apache.hadoop.io.erasurecode.rawcoder.DecodingValidator

@Private
public class DecodingValidator
extends java.lang.Object

A utility class to validate decoding.

Constructor Summary

Constructors

Constructor	Description
DecodingValidator(RawErasureDecoder decoder)

Method Summary

Modifier and Type	Method	Description
protected int	getNewErasedIndex()
protected int[]	getNewValidIndexes()
void	validate(java.nio.ByteBuffer[] inputs, int[] erasedIndexes, java.nio.ByteBuffer[] outputs)	Validate outputs decoded from inputs, by decoding an input back from the outputs and comparing it with the original one.
void	validate(ECChunk[] inputs, int[] erasedIndexes, ECChunk[] outputs)	Validate outputs decoded from inputs, by decoding an input back from those outputs and comparing it with the original one.

Methods inherited from class java.lang.Object

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

Constructor Detail

DecodingValidator

public DecodingValidator(RawErasureDecoder decoder)

Method Detail

validate

public void validate(java.nio.ByteBuffer[] inputs,
                     int[] erasedIndexes,
                     java.nio.ByteBuffer[] outputs)
              throws java.io.IOException

Validate outputs decoded from inputs, by decoding an input back from the outputs and comparing it with the original one. For instance, in RS (6, 3), let (d0, d1, d2, d3, d4, d5) be sources and (p0, p1, p2) be parities, and assume inputs = [d0, null (d1), d2, d3, d4, d5, null (p0), p1, null (p2)]; erasedIndexes = [1, 6]; outputs = [d1, p0]. Then 1. Create new inputs, erasedIndexes and outputs for validation so that the inputs could contain the decoded outputs, and decode them: newInputs = [d1, d2, d3, d4, d5, p0] newErasedIndexes = [0] newOutputs = [d0'] 2. Compare d0 and d0'. The comparison will fail with high probability when the initial outputs are wrong. Note that the input buffers' positions must be the ones where data are read: If the input buffers have been processed by a decoder, the buffers' positions must be reset before being passed into this method. This method does not change outputs and erasedIndexes.

Parameters:

inputs - input buffers used for decoding. The buffers' position are moved to the end after this method.

erasedIndexes - indexes of erased units used for decoding

outputs - decoded output buffers, which are ready to be read after the call

Throws:

java.io.IOException - raised on errors performing I/O.

validate

public void validate(ECChunk[] inputs,
                     int[] erasedIndexes,
                     ECChunk[] outputs)
              throws java.io.IOException

Validate outputs decoded from inputs, by decoding an input back from those outputs and comparing it with the original one.

Parameters:

inputs - input buffers used for decoding

erasedIndexes - indexes of erased units used for decoding

outputs - decoded output buffers

Throws:

java.io.IOException - raised on errors performing I/O.

getNewValidIndexes

@VisibleForTesting
protected int[] getNewValidIndexes()

getNewErasedIndex

@VisibleForTesting
protected int getNewErasedIndex()