org.apache.crunch.impl.spark

Class SparkPipeline

java.lang.Object

org.apache.crunch.impl.dist.DistributedPipeline

org.apache.crunch.impl.spark.SparkPipeline

All Implemented Interfaces:

Pipeline

public class SparkPipeline
extends DistributedPipeline

Constructor Summary

Constructors

Constructor and Description
SparkPipeline(org.apache.spark.api.java.JavaSparkContext sparkContext, String appName)
SparkPipeline(org.apache.spark.api.java.JavaSparkContext sparkContext, String appName, Class<?> jarClass, org.apache.hadoop.conf.Configuration conf)
SparkPipeline(String sparkConnect, String appName)
SparkPipeline(String sparkConnect, String appName, Class<?> jarClass)
SparkPipeline(String sparkConnect, String appName, Class<?> jarClass, org.apache.hadoop.conf.Configuration conf)

Method Summary

Modifier and Type	Method and Description
<T> void	cache(PCollection<T> pcollection, CachingOptions options) Caches the given PCollection so that it will be processed at most once during pipeline execution.
<K,V> PTable<K,V>	create(Iterable<Pair<K,V>> contents, PTableType<K,V> ptype, CreateOptions options) Creates a PTable containing the values found in the given Iterable using an implementation-specific distribution mechanism.
<S> PCollection<S>	create(Iterable<S> contents, PType<S> ptype, CreateOptions options) Creates a PCollection containing the values found in the given Iterable using an implementation-specific distribution mechanism.
PipelineResult	done() Run any remaining jobs required to generate outputs and then clean up any intermediate data files that were created in this run or previous calls to run.
<S> PCollection<S>	emptyPCollection(PType<S> ptype) Creates an empty PCollection of the given PType.
<K,V> PTable<K,V>	emptyPTable(PTableType<K,V> ptype) Creates an empty PTable of the given PTable Type.
<T> Iterable<T>	materialize(PCollection<T> pcollection) Create the given PCollection and read the data it contains into the returned Collection instance for client use.
PipelineResult	run() Constructs and executes a series of MapReduce jobs in order to write data to the output targets.
PipelineExecution	runAsync() Constructs and starts a series of MapReduce jobs in order ot write data to the output targets, but returns a ListenableFuture to allow clients to control job execution.

Methods inherited from class org.apache.crunch.impl.dist.DistributedPipeline

cleanup, create, create, createIntermediateOutput, createTempPath, enableDebug, getConfiguration, getFactory, getMaterializeSourceTarget, getName, getNextAnonymousStageId, read, read, read, read, readTextFile, sequentialDo, setConfiguration, union, unionTables, write, write, writeTextFile

Methods inherited from class java.lang.Object

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

Constructor Detail

SparkPipeline

public SparkPipeline(String sparkConnect,
                     String appName)

SparkPipeline

public SparkPipeline(String sparkConnect,
                     String appName,
                     Class<?> jarClass)

SparkPipeline

public SparkPipeline(String sparkConnect,
                     String appName,
                     Class<?> jarClass,
                     org.apache.hadoop.conf.Configuration conf)

SparkPipeline

public SparkPipeline(org.apache.spark.api.java.JavaSparkContext sparkContext,
                     String appName)

SparkPipeline

public SparkPipeline(org.apache.spark.api.java.JavaSparkContext sparkContext,
                     String appName,
                     Class<?> jarClass,
                     org.apache.hadoop.conf.Configuration conf)

Method Detail

materialize

public <T> Iterable<T> materialize(PCollection<T> pcollection)

Description copied from interface: Pipeline

Create the given PCollection and read the data it contains into the returned Collection instance for client use.

Parameters:

pcollection - The PCollection to materialize

Returns:

the data from the PCollection as a read-only Collection

emptyPCollection

public <S> PCollection<S> emptyPCollection(PType<S> ptype)

Description copied from interface: Pipeline

Creates an empty PCollection of the given PType.

Specified by:

emptyPCollection in interface Pipeline

Overrides:

emptyPCollection in class DistributedPipeline

Parameters:

ptype - The PType of the empty PCollection

Returns:

A valid PCollection with no contents

emptyPTable

public <K,V> PTable<K,V> emptyPTable(PTableType<K,V> ptype)

Description copied from interface: Pipeline

Creates an empty PTable of the given PTable Type.

Specified by:

emptyPTable in interface Pipeline

Overrides:

emptyPTable in class DistributedPipeline

Parameters:

ptype - The PTableType of the empty PTable

Returns:

A valid PTable with no contents

create

public <S> PCollection<S> create(Iterable<S> contents,
                                 PType<S> ptype,
                                 CreateOptions options)

Description copied from interface: Pipeline

Creates a PCollection containing the values found in the given Iterable using an implementation-specific distribution mechanism.

Specified by:

create in interface Pipeline

Overrides:

create in class DistributedPipeline

Parameters:

contents - The values the new PCollection will contain

ptype - The PType of the PCollection

options - Additional options, such as the name or desired parallelism of the PCollection

Returns:

A PCollection that contains the given values

create

public <K,V> PTable<K,V> create(Iterable<Pair<K,V>> contents,
                                PTableType<K,V> ptype,
                                CreateOptions options)

Description copied from interface: Pipeline

Creates a PTable containing the values found in the given Iterable using an implementation-specific distribution mechanism.

Specified by:

create in interface Pipeline

Overrides:

create in class DistributedPipeline

Parameters:

contents - The values the new PTable will contain

ptype - The PTableType of the PTable

options - Additional options, such as the name or desired parallelism of the PTable

Returns:

A PTable that contains the given values

cache

public <T> void cache(PCollection<T> pcollection,
                      CachingOptions options)

Description copied from interface: Pipeline

Caches the given PCollection so that it will be processed at most once during pipeline execution.

Parameters:

pcollection - The PCollection to cache

options - The options for how the cached data is stored

run

public PipelineResult run()

Description copied from interface: Pipeline

Constructs and executes a series of MapReduce jobs in order to write data to the output targets.

runAsync

public PipelineExecution runAsync()

Description copied from interface: Pipeline

Constructs and starts a series of MapReduce jobs in order ot write data to the output targets, but returns a ListenableFuture to allow clients to control job execution.

Returns:

done

public PipelineResult done()

Description copied from interface: Pipeline

Run any remaining jobs required to generate outputs and then clean up any intermediate data files that were created in this run or previous calls to run.

Specified by:

done in interface Pipeline

Overrides:

done in class DistributedPipeline