Package org.apache.beam.runners.spark.translation

Class GroupCombineFunctions

java.lang.Object

org.apache.beam.runners.spark.translation.GroupCombineFunctions

public class GroupCombineFunctions
extends Object

A set of group/combine functions to apply to Spark RDDs.

Constructor Summary

Constructors

Constructor	Description
GroupCombineFunctions()

Method Summary

Modifier and Type	Method	Description
static <InputT, OutputT, AccumT> SparkCombineFn.WindowedAccumulator<InputT,InputT,AccumT,?>	combineGlobally(org.apache.spark.api.java.JavaRDD<WindowedValue<InputT>> rdd, SparkCombineFn<InputT,InputT,AccumT,OutputT> sparkCombineFn, Coder<AccumT> aCoder, WindowingStrategy<?,?> windowingStrategy)	Apply a composite Combine.Globally transformation.
static <K, V, AccumT> org.apache.spark.api.java.JavaPairRDD<K,SparkCombineFn.WindowedAccumulator<KV<K,V>,V,AccumT,?>>	combinePerKey(org.apache.spark.api.java.JavaRDD<WindowedValue<KV<K,V>>> rdd, SparkCombineFn<KV<K,V>,V,AccumT,?> sparkCombineFn, Coder<K> keyCoder, Coder<V> valueCoder, Coder<AccumT> aCoder, WindowingStrategy<?,?> windowingStrategy)	Apply a composite Combine.PerKey transformation.
static <K, V> org.apache.spark.api.java.JavaRDD<KV<K,Iterable<WindowedValue<V>>>>	groupByKeyOnly(org.apache.spark.api.java.JavaRDD<WindowedValue<KV<K,V>>> rdd, Coder<K> keyCoder, WindowedValues.WindowedValueCoder<V> wvCoder, @Nullable org.apache.spark.Partitioner partitioner)	An implementation of GroupByKeyViaGroupByKeyOnly.GroupByKeyOnly for the Spark runner.
static <T> org.apache.spark.api.java.JavaRDD<WindowedValue<T>>	reshuffle(org.apache.spark.api.java.JavaRDD<WindowedValue<T>> rdd, WindowedValues.WindowedValueCoder<T> wvCoder)	An implementation of Reshuffle for the Spark runner.

Methods inherited from class java.lang.Object

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

Constructor Details

GroupCombineFunctions

public GroupCombineFunctions()

Method Details

groupByKeyOnly

public static <K,
V>
org.apache.spark.api.java.JavaRDD<KV<K,Iterable<WindowedValue<V>>>> groupByKeyOnly(org.apache.spark.api.java.JavaRDD<WindowedValue<KV<K,V>>> rdd,
 Coder<K> keyCoder,
 WindowedValues.WindowedValueCoder<V> wvCoder,
 @Nullable org.apache.spark.Partitioner partitioner)

An implementation of GroupByKeyViaGroupByKeyOnly.GroupByKeyOnly for the Spark runner.

combineGlobally

public static <InputT,
OutputT,
AccumT>
SparkCombineFn.WindowedAccumulator<InputT,InputT,AccumT,?> combineGlobally(org.apache.spark.api.java.JavaRDD<WindowedValue<InputT>> rdd,
 SparkCombineFn<InputT,InputT,AccumT,OutputT> sparkCombineFn,
 Coder<AccumT> aCoder,
 WindowingStrategy<?,?> windowingStrategy)

Apply a composite Combine.Globally transformation.

combinePerKey

public static <K,
V,
AccumT>
org.apache.spark.api.java.JavaPairRDD<K,SparkCombineFn.WindowedAccumulator<KV<K,V>,V,AccumT,?>> combinePerKey(org.apache.spark.api.java.JavaRDD<WindowedValue<KV<K,V>>> rdd,
 SparkCombineFn<KV<K,V>,V,AccumT,?> sparkCombineFn,
 Coder<K> keyCoder,
 Coder<V> valueCoder,
 Coder<AccumT> aCoder,
 WindowingStrategy<?,?> windowingStrategy)

Apply a composite Combine.PerKey transformation.

This aggregation will apply Beam's Combine.CombineFn via Spark's JavaPairRDD.combineByKey(Function, Function2, Function2) aggregation. For streaming, this will be called from within a serialized context (DStream's transform callback), so passed arguments need to be Serializable.

reshuffle

public static <T>
org.apache.spark.api.java.JavaRDD<WindowedValue<T>> reshuffle(org.apache.spark.api.java.JavaRDD<WindowedValue<T>> rdd,
 WindowedValues.WindowedValueCoder<T> wvCoder)

An implementation of Reshuffle for the Spark runner.