Spark RDD Transformation 简单用例（一）

map(func)

/**
 * Return a new RDD by applying a function to all elements of this RDD.
 */

def map[U: ClassTag](f: T => U): RDD[U]

map(func)

Return a new distributed dataset formed by passing each element of the source through a function func.

将原RDD中的每一个元素经过func函数映射为一个新的元素形成一个新的RDD。

示例：

其中sc.parallelize第二个参数标识RDD的分区数量

val rdd = sc.parallelize(1 to 9,2)
val rdd1=rdd.map(x=>x+1)

scala> val rdd = sc.parallelize(1 to 9,2)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[3] at parallelize at <console>:24

scala> rdd.take(20)
res3: Array[Int] = Array(1, 2, 3, 4, 5, 6, 7, 8, 9)

scala> val rdd1=rdd.map(x=>x+1)
rdd1: org.apache.spark.rdd.RDD[Int] = MapPartitionsRDD[4] at map at <console>:26

scala> rdd1.take(20)
res5: Array[Int] = Array(2, 3, 4, 5, 6, 7, 8, 9, 10)

filter(func)

/**
 * Return a new RDD containing only the elements that satisfy a predicate.
 */

def filter(f: T => Boolean): RDD[T]

filter(func)

Return a new dataset formed by selecting those elements of the source on which func returns true.

原RDD中通过func函数结果为true的元素转换成一个新的RDD。

val rdd = sc.parallelize(1 to 9,2)
val rdd1 = rdd.filter(_>=5)

scala> val rdd = sc.parallelize(1 to 9,2)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[7] at parallelize at <console>:24

scala> val rdd1 = rdd.filter(_>=5)
rdd1: org.apache.spark.rdd.RDD[Int] = MapPartitionsRDD[8] at filter at <console>:26

scala> rdd1.take(10)
res13: Array[Int] = Array(5, 6, 7, 8, 9)

flatMap(func)

/**
 *  Return a new RDD by first applying a function to all elements of this
 *  RDD, and then flattening the results.
 */

def flatMap[U: ClassTag](f: T => TraversableOnce[U]): RDD[U]

flatMap(func)

Similar to map, but each input item can be mapped to 0 or more output items (so func should return a Seq rather than a single item).

和map类似，但是每一个元素可能被映射为0个或多个元素（func函数应该返回一个Seq，而不是单个的元素）；实际上就是先进行map，然后再进行一次平滑(flat)处理。

val rdd = sc.parallelize(1 to 3,2)
val rdd1 = rdd.flatMap( _ to 5)

scala> val rdd = sc.parallelize(1 to 3,2)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[9] at parallelize at <console>:24

scala> val rdd1 = rdd.flatMap( _ to 5)
rdd1: org.apache.spark.rdd.RDD[Int] = MapPartitionsRDD[10] at flatMap at <console>:26

scala> rdd1.take(100)
res14: Array[Int] = Array(1, 2, 3, 4, 5, 2, 3, 4, 5, 3, 4, 5)

mapPartitions(func)

/**
 * Return a new RDD by applying a function to each partition of this RDD.
 *
 * `preservesPartitioning` indicates whether the input function preserves the partitioner, which
 * should be `false` unless this is a pair RDD and the input function doesn't modify the keys.
 */
def mapPartitions[U: ClassTag](
    f: Iterator[T] => Iterator[U],
    preservesPartitioning: Boolean = false): RDD[U]

mapPartitions(func)

Similar to map, but runs separately on each partition (block) of the RDD, so func must be of type Iterator<T> => Iterator<U> when running on an RDD of type T.

和map类似，该函数和map函数类似，只不过映射函数的参数由RDD中的每一个元素变成了RDD中每一个分区的迭代器。如果在映射的过程中需要频繁创建额外的对象，使用mapPartitions要比map高效的过。

计算每一个partition中元素个数

def countPartitionEle(it : Iterator[Int]) = {
    var result = List[Int]()
     var i = 0
     while(it.hasNext){
       i += 1
       it.next
     }
     result.::(i).iterator//::在列表开头增加元素i，元素i必须用小括号包含，然后创建一个迭代器
}

val rdd = sc.parallelize(1 to 10, 3)
val rdd1 = rdd.mapPartitions(countPartitionEle(_))
rdd1.take(10)

scala> def countPartitionEle(it : Iterator[Int]) = {
     | var result = List[Int]()
     | var i = 0
     | while(it.hasNext){
     | i += 1
     | it.next
     | }
     | result.::(i).iterator
     | }
countPartitionEle: (it: Iterator[Int])Iterator[Int]

scala> val rdd = sc.parallelize(1 to 10, 3)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[6] at parallelize at <console>:24

scala> val rdd1 = rdd.mapPartitions(countPartitionEle(_))
rdd1: org.apache.spark.rdd.RDD[Int] = MapPartitionsRDD[7] at mapPartitions at <console>:30

scala> rdd1.take(10)
res8: Array[Int] = Array(3, 3, 4)

mapPartitionsWithIndex(func)

/**
* Return a new RDD by applying a function to each partition of this RDD, while tracking the index
* of the original partition.
*
* `preservesPartitioning` indicates whether the input function preserves the partitioner, which
* should be `false` unless this is a pair RDD and the input function doesn't modify the keys.
*/
def mapPartitionsWithIndex[U: ClassTag](
f: (Int, Iterator[T]) => Iterator[U],
preservesPartitioning: Boolean = false): RDD[U]

mapPartitionsWithIndex(func)

Similar to mapPartitions, but also provides func with an integer value representing the index of the partition, so func must be of type (Int, Iterator<T>) => Iterator<U> when running on an RDD of type T.

和mapPartitions类似，也是针对每个分区处理，但是func函数需要两个入参，第一个表示partition分区索引，第二个入参表示每个分区的迭代器。

def func(index :Int, it : Iterator[Int]) = {
    var result = List[String]()
     var i = ""
     while(it.hasNext){
       i += it.next + ","
     }
     result.::(i.dropRight(1) + " at partition "+index+".").iterator
}

val rdd = sc.parallelize(1 to 10, 3)
val rdd1 = rdd.mapPartitionsWithIndex((x,it) => func(x,it))
rdd1.take(3)

scala> def func(index :Int, it : Iterator[Int]) = {
     |     var result = List[String]()
     |      var i = ""
     |      while(it.hasNext){
     |        i += it.next + ","
     |      }
     |      result.::(i.dropRight(1) + " at partition "+index+".").iterator
     | }
func: (index: Int, it: Iterator[Int])Iterator[String]



scala> val rdd = sc.parallelize(1 to 10, 3)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[1] at parallelize at <console>:24

scala> val rdd1 = rdd.mapPartitionsWithIndex((x,it) => func(x,it))
rdd1: org.apache.spark.rdd.RDD[String] = MapPartitionsRDD[2] at mapPartitionsWithIndex at <console>:28


scala> rdd1.take(3)
res2: Array[String] = Array(1,2,3 at partition 0. 4,5,6 at partition 1. 7,8,9,10 at partition 2.)

sample(withReplacement, fraction, seed)

/**
* Return a sampled subset of this RDD.
*
* @param withReplacement can elements be sampled multiple times (replaced when sampled out)
* @param fraction expected size of the sample as a fraction of this RDD's size
* without replacement: probability that each element is chosen; fraction must be [0, 1]
* with replacement: expected number of times each element is chosen; fraction must be >= 0
* @param seed seed for the random number generator
*/
def sample(
withReplacement: Boolean,
fraction: Double,
seed: Long = Utils.random.nextLong): RDD[T]

sample(withReplacement, fraction, seed)

Sample a fraction fraction of the data, with or without replacement, using a given random number generator seed.

对原RDD进行采样，其中withReplacement表示是否有放回的抽样，fraction表示采样大小是原RDD的百分比，seed表示随机数生成器

scala> val rdd = sc.parallelize(1 to 10, 3)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[3] at parallelize at <console>:24

scala> val rdd1 = rdd.sample(true,0.5,0)
rdd1: org.apache.spark.rdd.RDD[Int] = PartitionwiseSampledRDD[4] at sample at <console>:26

scala> val rdd2 = rdd.sample(false,0.5,0)
rdd2: org.apache.spark.rdd.RDD[Int] = PartitionwiseSampledRDD[5] at sample at <console>:26

scala> rdd1.collect
res3: Array[Int] = Array(2)

scala> rdd2.collect
res4: Array[Int] = Array(1, 2, 4, 5, 6, 9)                                      


scala> val rdd1 = rdd.sample(true,0.5,1)
rdd1: org.apache.spark.rdd.RDD[Int] = PartitionwiseSampledRDD[7] at sample at <console>:26

scala> rdd1.collect
res6: Array[Int] = Array(1, 3, 7, 7, 8, 8, 9, 10)

union(otherDataset)

/**
* Return the union of this RDD and another one. Any identical elements will appear multiple
* times (use `.distinct()` to eliminate them).
*/
def union(other: RDD[T]): RDD[T]

union(otherDataset)

Return a new dataset that contains the union of the elements in the source dataset and the argument.

将两个RDD合并成一个RDD，相同的元素可能出现多次，可以使用distinct去重。

val rdd1 = sc.parallelize(1 to 5,2)
val rdd2 = sc.parallelize(1 to 5,3)
val rdd3 = sc.parallelize(2 to 8,3)
val rdd = rdd1.union(rdd2).union(rdd3)
rdd.collect
rdd.distinct.collect

scala> val rdd1 = sc.parallelize(1 to 5,2)
rdd1: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[8] at parallelize at <console>:24

scala> val rdd2 = sc.parallelize(1 to 5,3)
rdd2: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[9] at parallelize at <console>:24

scala> val rdd3 = sc.parallelize(2 to 8,3)
rdd3: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[10] at parallelize at <console>:24


scala> val rdd = rdd1.union(rdd2).union(rdd3)
rdd: org.apache.spark.rdd.RDD[Int] = UnionRDD[12] at union at <console>:30

scala> rdd.collect
res7: Array[Int] = Array(1, 2, 3, 4, 5, 1, 2, 3, 4, 5, 2, 3, 4, 5, 6, 7, 8)     

scala> rdd.distinct.collect
res8: Array[Int] = Array(8, 1, 2, 3, 4, 5, 6, 7)

intersection(otherDataset)

Return a new RDD that contains the intersection of elements in the source dataset and the argument.

两个RDD共同的元素组合一个新的RDD

val rdd1 = sc.parallelize(1 to 5,2)
val rdd2 = sc.parallelize(4 to 8,3)
val rdd = rdd1.intersection(rdd2)
rdd.collect

scala> val rdd1 = sc.parallelize(1 to 5,2)
rdd1: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[16] at parallelize at <console>:24

scala> val rdd2 = sc.parallelize(4 to 8,3)
rdd2: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[17] at parallelize at <console>:24

scala> val rdd = rdd1.intersection(rdd2)
rdd: org.apache.spark.rdd.RDD[Int] = MapPartitionsRDD[23] at intersection at <console>:28


scala> rdd.collect
res9: Array[Int] = Array(4, 5)

scala> rdd.partitions.length
res10: Int = 3

/**
* Return the intersection of this RDD and another one. The output will not contain any duplicate
* elements, even if the input RDDs did. Performs a hash partition across the cluster
*
* Note that this method performs a shuffle internally.
*
* @param numPartitions How many partitions to use in the resulting RDD
*/
def intersection(other: RDD[T], numPartitions: Int): RDD[T]

同def intersection(other: RDD[T]): RDD[T]，numPartitions表示结果RDD的分区数量

scala> val rdd = rdd1.intersection(rdd2,1)
rdd: org.apache.spark.rdd.RDD[Int] = MapPartitionsRDD[41] at intersection at <console>:28

scala> rdd.partitions.length
res12: Int = 1

/**
* Return the intersection of this RDD and another one. The output will not contain any duplicate
* elements, even if the input RDDs did.
*
* Note that this method performs a shuffle internally.
*
* @param partitioner Partitioner to use for the resulting RDD
*/
def intersection(
other: RDD[T],
partitioner: Partitioner)(implicit ord: Ordering[T] = null): RDD[T]

自定义分区

自定义分区类必须继承Partitioner，方法numPartitions设置分区数量，getPartition获取分区索引。

class MyPartitioner(numParts:Int) extends org.apache.spark.Partitioner{
  override def numPartitions: Int = numParts
  override def getPartition(key: Any): Int = {
    key.toString.toInt%numPartitions
  }
}

val rdd1 = sc.parallelize(1 to 15,2)
val rdd2 = sc.parallelize(5 to 25,2)
val rdd = rdd1.intersection(rdd2,new MyPartitioner(5))
rdd.collect
rdd.partitions.length
def func(index :Int, it : Iterator[Int]) = {
     var result = List[String]()
     var i = ""
     while(it.hasNext){
       i += it.next + ","
     }
     result.::(i.dropRight(1) + " at partition "+index+".").iterator
}

val rdd3 = rdd.mapPartitionsWithIndex((x,it) => func(x,it))
rdd3.collect

scala> val rdd1 = sc.parallelize(1 to 15,2)
rdd1: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[54] at parallelize at <console>:27

scala> val rdd2 = sc.parallelize(5 to 25,2)
rdd2: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[55] at parallelize at <console>:27

scala> val rdd = rdd1.intersection(rdd2,newMyPartitioner(5))
<console>:31: error: not found: value newMyPartitioner
       val rdd = rdd1.intersection(rdd2,newMyPartitioner(5))
                                        ^

scala> val rdd = rdd1.intersection(rdd2,new MyPartitioner(5))
rdd: org.apache.spark.rdd.RDD[Int] = MapPartitionsRDD[61] at intersection at <console>:32

scala> rdd.collect
res25: Array[Int] = Array(15, 10, 5, 11, 6, 7, 12, 13, 8, 14, 9)


scala> rdd.partitions.length
res26: Int = 5

scala> def func(index :Int, it : Iterator[Int]) = {
     |     var result = List[String]()
     |      var i = ""
     |      while(it.hasNext){
     |        i += it.next + ","
     |      }
     |      result.::(i.dropRight(1) + " at partition "+index+".").iterator
     | }
func: (index: Int, it: Iterator[Int])Iterator[String]

scala> val rdd3 = rdd.mapPartitionsWithIndex((x,it) => func(x,it))
rdd3: org.apache.spark.rdd.RDD[String] = MapPartitionsRDD[62] at mapPartitionsWithIndex at <console>:36

scala> rdd3.collect
res27: Array[String] = Array(15,10,5 at partition 0., 11,6 at partition 1., 7,12 at partition 2., 13,8 at partition 3., 14,9 at partition 4.)

distinct([numTasks])

Return a new dataset that contains the distinct elements of the source dataset.

使用原RDD中的元素组成一个没有重复元素的RDD

/**
* Return a new RDD containing the distinct elements in this RDD.
*/
def distinct(numPartitions: Int)(implicit ord: Ordering[T] = null): RDD[T]

numPartitions表示结果RDD的分区数量

val a = Array(1,1,1,2,2,3,4,5)
val rdd = sc.parallelize(a,2)
rdd.collect
val rdd1 = rdd.distinct(1)
rdd1.collect
rdd1.partitions.length

scala> val a = Array(1,1,1,2,2,3,4,5)
a: Array[Int] = Array(1, 1, 1, 2, 2, 3, 4, 5)

scala> val rdd = sc.parallelize(a)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[42] at parallelize at <console>:26

scala> val rdd = sc.parallelize(a,2)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[43] at parallelize at <console>:26

scala> rdd.collect
res13: Array[Int] = Array(1, 1, 1, 2, 2, 3, 4, 5)                               

scala> val rdd1 = rdd.distinct(1)
rdd1: org.apache.spark.rdd.RDD[Int] = MapPartitionsRDD[46] at distinct at <console>:28

scala> rdd1.collect
res14: Array[Int] = Array(4, 1, 3, 5, 2)

scala> rdd1.partitions.length
res15: Int = 1

/**
* Return a new RDD containing the distinct elements in this RDD.
*/
def distinct(): RDD[T]

同distinct(numPartitions: Int)，不同的是结果RDD中partition数量依赖父RDD。

scala> val rdd1 = rdd.distinct()
rdd1: org.apache.spark.rdd.RDD[Int] = MapPartitionsRDD[49] at distinct at <console>:28

scala> rdd1.partitions.length
res16: Int = 2

scala> rdd1.collect
res17: Array[Int] = Array(4, 2, 1, 3, 5)

keyBy(func)

/**
 * Creates tuples of the elements in this RDD by applying `f`.
 */
def keyBy[K](f: T => K): RDD[(K, T)]

使用func为RDD每一个元素创建一个key-value对元素

val rdd = sc.parallelize(1 to 9 ,2)
val rdd1 = rdd.keyBy(_%3)
rdd1.collect

scala> val rdd = sc.parallelize(1 to 9 ,2)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[0] at parallelize at <console>:24

scala> val rdd1 = rdd.keyBy(_%3)
rdd1: org.apache.spark.rdd.RDD[(Int, Int)] = MapPartitionsRDD[1] at keyBy at <console>:26

scala> rdd1.collect
res0: Array[(Int, Int)] = Array((1,1), (2,2), (0,3), (1,4), (2,5), (0,6), (1,7), (2,8), (0,9))

/**
 * Group the values for each key in the RDD into a single sequence. Hash-partitions the
 * resulting RDD with the existing partitioner/parallelism level. The ordering of elements
 * within each group is not guaranteed, and may even differ each time the resulting RDD is
 * evaluated.
 *
 * Note: This operation may be very expensive. If you are grouping in order to perform an
 * aggregation (such as a sum or average) over each key, using [[PairRDDFunctions.aggregateByKey]]
 * or [[PairRDDFunctions.reduceByKey]] will provide much better performance.
 */
def groupByKey(): RDD[(K, Iterable[V])]

val rdd = sc.parallelize(1 to 9 ,2)
val rdd1 = rdd.keyBy(_%3)
val rdd2 = rdd1.groupByKey()
rdd2.collect

scala> val rdd = sc.parallelize(1 to 9 ,2)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[2] at parallelize at <console>:24

scala> val rdd1 = rdd.keyBy(_%3)
rdd1: org.apache.spark.rdd.RDD[(Int, Int)] = MapPartitionsRDD[3] at keyBy at <console>:26

scala> val rdd2 = rdd1.groupByKey()
rdd2: org.apache.spark.rdd.RDD[(Int, Iterable[Int])] = ShuffledRDD[4] at groupByKey at <console>:28

scala> rdd2.collect
res1: Array[(Int, Iterable[Int])] = Array((0,CompactBuffer(3, 6, 9)), (2,CompactBuffer(2, 5, 8)), (1,CompactBuffer(1, 4, 7)))

/**
 * Group the values for each key in the RDD into a single sequence. Hash-partitions the
 * resulting RDD with into `numPartitions` partitions. The ordering of elements within
 * each group is not guaranteed, and may even differ each time the resulting RDD is evaluated.
 *
 * Note: This operation may be very expensive. If you are grouping in order to perform an
 * aggregation (such as a sum or average) over each key, using [[PairRDDFunctions.aggregateByKey]]
 * or [[PairRDDFunctions.reduceByKey]] will provide much better performance.
 *
 * Note: As currently implemented, groupByKey must be able to hold all the key-value pairs for any
 * key in memory. If a key has too many values, it can result in an [[OutOfMemoryError]].
 */
def groupByKey(numPartitions: Int): RDD[(K, Iterable[V])]
同groupByKey(),只是指定了分区数量。

scala> val rdd = sc.parallelize(1 to 9 ,2)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[5] at parallelize at <console>:24

scala> val rdd1 = rdd.keyBy(_%3)
rdd1: org.apache.spark.rdd.RDD[(Int, Int)] = MapPartitionsRDD[6] at keyBy at <console>:26

scala> val rdd2 = rdd1.groupByKey(3)
rdd2: org.apache.spark.rdd.RDD[(Int, Iterable[Int])] = ShuffledRDD[7] at groupByKey at <console>:28

scala> rdd2.partitions.length
res2: Int = 3

/**
 * Group the values for each key in the RDD into a single sequence. Allows controlling the
 * partitioning of the resulting key-value pair RDD by passing a Partitioner.
 * The ordering of elements within each group is not guaranteed, and may even differ
 * each time the resulting RDD is evaluated.
 *
 * Note: This operation may be very expensive. If you are grouping in order to perform an
 * aggregation (such as a sum or average) over each key, using [[PairRDDFunctions.aggregateByKey]]
 * or [[PairRDDFunctions.reduceByKey]] will provide much better performance.
 *
 * Note: As currently implemented, groupByKey must be able to hold all the key-value pairs for any
 * key in memory. If a key has too many values, it can result in an [[OutOfMemoryError]].
 */
def groupByKey(partitioner: Partitioner): RDD[(K, Iterable[V])]

class MyPartitioner(numParts:Int) extends org.apache.spark.Partitioner{
  override def numPartitions: Int = numParts
  override def getPartition(key: Any): Int = {
    key.toString.toInt%numPartitions
  }
}

val rdd = sc.parallelize(1 to 9 ,2)
val rdd1 = rdd.keyBy(_%3)
rdd1.collect
val rdd2 = rdd1.groupByKey(new MyPartitioner(2))
rdd2.collect

scala> class MyPartitioner(numParts:Int) extends org.apache.spark.Partitioner{
     |   override def numPartitions: Int = numParts
     |   override def getPartition(key: Any): Int = {
     |     key.toString.toInt%numPartitions
     |   }
     | }
defined class MyPartitioner

scala> val rdd = sc.parallelize(1 to 9 ,2)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[8] at parallelize at <console>:24

scala> val rdd1 = rdd.keyBy(_%3)
rdd1: org.apache.spark.rdd.RDD[(Int, Int)] = MapPartitionsRDD[9] at keyBy at <console>:26

scala> val rdd2 = rdd1.groupByKey(new MyPartitioner(2))
rdd2: org.apache.spark.rdd.RDD[(Int, Iterable[Int])] = ShuffledRDD[10] at groupByKey at <console>:29

scala> rdd2.collect
res3: Array[(Int, Iterable[Int])] = Array((0,CompactBuffer(3, 6, 9)), (2,CompactBuffer(2, 5, 8)), (1,CompactBuffer(1, 4, 7)))

scala> rdd1.collect
res4: Array[(Int, Int)] = Array((1,1), (2,2), (0,3), (1,4), (2,5), (0,6), (1,7), (2,8), (0,9))

scala> rdd2.partitions.length
res5: Int = 2

groupBy(func)

/**
* Return an RDD of grouped items. Each group consists of a key and a sequence of elements
* mapping to that key. The ordering of elements within each group is not guaranteed, and
* may even differ each time the resulting RDD is evaluated.
*
* Note: This operation may be very expensive. If you are grouping in order to perform an
* aggregation (such as a sum or average) over each key, using [[PairRDDFunctions.aggregateByKey]]
* or [[PairRDDFunctions.reduceByKey]] will provide much better performance.
*/
def groupBy[K](f: T => K)(implicit kt: ClassTag[K]): RDD[(K, Iterable[T])]

def func(x:Int) = {x%3}
val rdd = sc.parallelize(1 to 10,2)
val rdd1 = rdd.groupBy(func(_))
rdd1.collect

scala> def func(x:Int) = {x%3}
func: (x: Int)Int

scala> val rdd = sc.parallelize(1 to 10,2)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[64] at parallelize at <console>:27

scala> val rdd1 = rdd.groupBy(func(_))
rdd1: org.apache.spark.rdd.RDD[(Int, Iterable[Int])] = ShuffledRDD[66] at groupBy at <console>:31

scala> rdd1.collect
res29: Array[(Int, Iterable[Int])] = Array((0,CompactBuffer(3, 6, 9)), (2,CompactBuffer(2, 5, 8)), (1,CompactBuffer(1, 4, 7, 10)))

/**
 * Return an RDD of grouped elements. Each group consists of a key and a sequence of elements
 * mapping to that key. The ordering of elements within each group is not guaranteed, and
 * may even differ each time the resulting RDD is evaluated.
 *
 * Note: This operation may be very expensive. If you are grouping in order to perform an
 * aggregation (such as a sum or average) over each key, using [[PairRDDFunctions.aggregateByKey]]
 * or [[PairRDDFunctions.reduceByKey]] will provide much better performance.
 */
def groupBy[K](
    f: T => K,
    numPartitions: Int)(implicit kt: ClassTag[K]): RDD[(K, Iterable[T])]

同groupBy[K](f: T => K)，只是指定了分区数量。

def func(x:Int) = {x%3}
val rdd = sc.parallelize(1 to 10,2)
val rdd1 = rdd.groupBy(func(_),3)
rdd1.collect

scala> def func(x:Int) = {x%3}
func: (x: Int)Int

scala> val rdd = sc.parallelize(1 to 10,2)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[11] at parallelize at <console>:24

scala> val rdd1 = rdd.groupBy(func(_),3)
rdd1: org.apache.spark.rdd.RDD[(Int, Iterable[Int])] = ShuffledRDD[13] at groupBy at <console>:28

scala> rdd1.partitions.length
res6: Int = 3

scala> rdd1.collect
res7: Array[(Int, Iterable[Int])] = Array((0,CompactBuffer(3, 6, 9)), (1,CompactBuffer(1, 4, 7, 10)), (2,CompactBuffer(2, 5, 8)))

/**
 * Return an RDD of grouped items. Each group consists of a key and a sequence of elements
 * mapping to that key. The ordering of elements within each group is not guaranteed, and
 * may even differ each time the resulting RDD is evaluated.
 *
 * Note: This operation may be very expensive. If you are grouping in order to perform an
 * aggregation (such as a sum or average) over each key, using [[PairRDDFunctions.aggregateByKey]]
 * or [[PairRDDFunctions.reduceByKey]] will provide much better performance.
 */
def groupBy[K](f: T => K, p: Partitioner)(implicit kt: ClassTag[K], ord: Ordering[K] = null)
    : RDD[(K, Iterable[T])]

class MyPartitioner(numParts:Int) extends org.apache.spark.Partitioner{
  override def numPartitions: Int = numParts
  override def getPartition(key: Any): Int = {
    key.toString.toInt%numPartitions
  }
}
def func(x:Int) = {x%3}
val rdd = sc.parallelize(1 to 10,2)
val rdd1 = rdd.groupBy(func(_),new MyPartitioner(3))
rdd1.collect
rdd1.partitions.length

scala> class MyPartitioner(numParts:Int) extends org.apache.spark.Partitioner{
     |   override def numPartitions: Int = numParts
     |   override def getPartition(key: Any): Int = {
     |     key.toString.toInt%numPartitions
     |   }
     | }
defined class MyPartitioner

scala> def func(x:Int) = {x%3}
func: (x: Int)Int

scala> val rdd = sc.parallelize(1 to 10,2)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[14] at parallelize at <console>:24

scala> val rdd1 = rdd.groupBy(func(_),new MyPartitioner(3))
rdd1: org.apache.spark.rdd.RDD[(Int, Iterable[Int])] = ShuffledRDD[16] at groupBy at <console>:29

scala> rdd1.collect
res8: Array[(Int, Iterable[Int])] = Array((0,CompactBuffer(3, 6, 9)), (1,CompactBuffer(1, 4, 7, 10)), (2,CompactBuffer(2, 5, 8)))

scala> rdd1.partitions.length
res9: Int = 3

reduceByKey(func, [numTasks])

reduceByKey(func, [numTasks]) When called on a dataset of (K, V) pairs, returns a dataset of (K, V) pairs where the values for each key are aggregated using the given reduce function func, which must be of type (V,V) => V. Like in groupByKey, the number of reduce tasks is configurable through an optional second argument.

对每一个key的所有value使用func函数进行聚合

/**
 * Merge the values for each key using an associative and commutative reduce function. This will
 * also perform the merging locally on each mapper before sending results to a reducer, similarly
 * to a "combiner" in MapReduce. Output will be hash-partitioned with the existing partitioner/
 * parallelism level.
 */
def reduceByKey(func: (V, V) => V): RDD[(K, V)]

val words = Array("one", "two", "two", "three", "three", "three")
val rdd = sc.parallelize(words).map(word => (word, 1))
val rdd1 = rdd.reduceByKey(_ + _)
rdd1.collect

scala> val words = Array("one", "two", "two", "three", "three", "three")  
words: Array[String] = Array(one, two, two, three, three, three)

scala> val rdd = sc.parallelize(words).map(word => (word, 1))  
rdd: org.apache.spark.rdd.RDD[(String, Int)] = MapPartitionsRDD[18] at map at <console>:26

scala> val rdd1 = rdd.reduceByKey(_ + _) 
rdd1: org.apache.spark.rdd.RDD[(String, Int)] = ShuffledRDD[19] at reduceByKey at <console>:28

scala> rdd1.collect
res10: Array[(String, Int)] = Array((two,2), (one,1), (three,3))

/**
 * Merge the values for each key using an associative and commutative reduce function. This will
 * also perform the merging locally on each mapper before sending results to a reducer, similarly
 * to a "combiner" in MapReduce. Output will be hash-partitioned with numPartitions partitions.
 */
def reduceByKey(func: (V, V) => V, numPartitions: Int): RDD[(K, V)]
同上，只是指定了分区数量

scala> val rdd2 = rdd.reduceByKey(_ + _,3) 
rdd2: org.apache.spark.rdd.RDD[(String, Int)] = ShuffledRDD[20] at reduceByKey at <console>:28

scala> rdd2.collect
res11: Array[(String, Int)] = Array((two,2), (one,1), (three,3))                

scala> rdd2.partitions.length
res12: Int = 3

/**
 * Merge the values for each key using an associative and commutative reduce function. This will
 * also perform the merging locally on each mapper before sending results to a reducer, similarly
 * to a "combiner" in MapReduce.
 */
def reduceByKey(partitioner: Partitioner, func: (V, V) => V): RDD[(K, V)]

同上，使用partitioner自定义分区

posted @ 2016-09-01 20:54 alianblog 阅读(2059) 评论(0) 编辑收藏举报

会员力量，点亮园子希望

刷新页面返回顶部