Spark学习摘记 —— Pair RDD行动操作API归纳

本文参考

参考《Spark快速大数据分析》动物书中的第四章"键值对操作",本篇是对RDD转化操作和行动操作API归纳的最后一篇

RDD转化操作API归纳:https://www.cnblogs.com/kuluo/p/12545374.html

RDD行动操作API归纳:https://www.cnblogs.com/kuluo/p/12550938.html

pair RDD转化操作API归纳:https://www.cnblogs.com/kuluo/p/12558563.html

环境

idea + spark 2.4.5 + scala 2.11.12

RDD均通过SparkContext的parallelize()函数创建

countByKey()函数

目的:

对每个键对应的元素分别计数

代码:

/*
* (a,3) (b,5) (c,4) (d,2)
*/
val
testList1 = List("a a a b b b", "b b c c c", "c d d")
/*
* (a,5) (b,4)
*/
val
testList2 = List("a a a a a b b", "b b")

val testRdd1 = sc.parallelize(testList1)
val testRdd2 = sc.parallelize(testList2)

val map = testRdd1.flatMap(_.split(" ")).map((_, 1)).reduceByKey(_ + _)
  .union(testRdd2.flatMap(_.split(" ")).map((_, 1)).reduceByKey(_ + _))
  .countByKey()

for ((x, y) <- map) {
  println(s"($x, $y)")
}

输出:

(d, 1)

(a, 2)

(b, 2)

(c, 1)

注意:

This method should only be used if the resulting map is expected to be small, as the whole thing is loaded into the driver's memory. To handle very large results, consider using rdd.mapValues(_ => 1L).reduceByKey(_ + _), which returns an RDD[T, Long] instead of a map.

countByKey()函数会将结果全部加载到驱动器进程中,不适合结果集较大时使用

我们在源码中可以看到它调用了collect()函数

def countByKey(): Map[K, Long] = self.withScope { self.mapValues(_ => 1L).reduceByKey(_ + _).collect().toMap }

因此在处理大数据量时,应当使用.mapValues(_ => 1L).reduceByKey(_ + _)两个函数返回一个RDD

 

collectAsMap()函数

目的:

collect()函数针对pair RDD的实现,将结果以映射表的形式返回

代码:

/*
* (a,3) (b,5) (c,4) (d,2)
*/
val
testList1 = List("a a a b b b", "b b c c c", "c d d")
/*
* (a,5) (b,4)
*/
val
testList2 = List("a a a a a b b", "b b")

val testRdd1 = sc.parallelize(testList1)
val testRdd2 = sc.parallelize(testList2)

val map = testRdd1.flatMap(_.split(" ")).map((_, 1)).reduceByKey(_ + _)
  .union(testRdd2.flatMap(_.split(" ")).map((_, 1)).reduceByKey(_ + _))
  .collectAsMap()

for ((x, y) <- map) {
  println(s"($x, $y)")
}

输出:

(b, 4)

(d, 2)

(a, 5)

(c, 4)

注意:

this doesn't return a multimap (so if you have multiple values to the same key, only one value per key is preserved in the map returned)

也正如本例所示,pair RDD中有重复的键时,collectByKey函数只会保留一个

因为内部调用了collect()函数,不适合结果集较大时使用

 

lookup()函数

目的:

返回给定键对应的所有值

代码:

/*
* (a,3) (b,5) (c,4) (d,2)
*/
val
testList1 = List("a a a b b b", "b b c c c", "c d d")
/*
* (a,5) (b,4)
*/
val
testList2 = List("a a a a a b b", "b b")

val testRdd1 = sc.parallelize(testList1)
val testRdd2 = sc.parallelize(testList2)

println(testRdd1.flatMap(_.split(" ")).map((_, 1)).reduceByKey(_ + _)
  .union(testRdd2.flatMap(_.split(" ")).map((_, 1)).reduceByKey(_ + _))
  .lookup("a"))

输出:

ArrayBuffer(3, 5)

原文地址:https://www.cnblogs.com/kuluo/p/12567221.html