Package org.pipecraft.pipes.sync.inter.sample

Class ExactSamplerPipe<T>

java.lang.Object

org.pipecraft.pipes.sync.inter.FilterBasePipe<T>

org.pipecraft.pipes.sync.inter.sample.ExactSamplerPipe<T>

All Implemented Interfaces:

Closeable, AutoCloseable, BasePipe, Pipe<T>

public class ExactSamplerPipe<T>
extends FilterBasePipe<T>

A pipe sampling exactly m items of an input pipe containing n items, where n is known in advance. The algorithm is an online algorithm choosing a given item with probability {#remaining items to sample}/{#remaining items to visit}.

Author:

Eyal Schneider

Constructor Summary

Constructors

Constructor	Description
ExactSamplerPipe(Pipe<T> input, int n, int m)	Constructor
ExactSamplerPipe(Pipe<T> input, int n, int m, Random rnd)	Constructor

Method Summary

Modifier and Type	Method	Description
protected boolean	shouldSelect(T item)

Methods inherited from class org.pipecraft.pipes.sync.inter.FilterBasePipe

close, getProgress, next, peek, start

Methods inherited from class java.lang.Object

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

Constructor Detail

ExactSamplerPipe

public ExactSamplerPipe(Pipe<T> input,
                        int n,
                        int m,
                        Random rnd)

Constructor

Parameters:

input - The input pipe

n - The exact number of items in the input pipe. If inaccurate, sampling won't be complete or won't be uniform.

m - The number of items to sample

rnd - The randomizer to use, if consistency is required

ExactSamplerPipe

public ExactSamplerPipe(Pipe<T> input,
                        int n,
                        int m)

Constructor

Parameters:

input - The input pipe

n - The exact number of items in the input pipe. If inaccurate, sampling won't be complete or won't be uniform.

m - The number of items to sample

Method Detail

shouldSelect

protected boolean shouldSelect(T item)

Specified by:

shouldSelect in class FilterBasePipe<T>

Parameters:

item - The next item from the input stream

Returns:

true if and only if the item should be selected