4.RDD操作之Transform

RDD 两种类型的操作： Transform转化操作 和Action行动操作。

Transform操作会由一个RDD 生成一个新的RDD，这个过程中不进行实质计算，只有当第一次Action操作时才会真正计算。  称作Lazy计算，惰性计算。

比如：

scala> val a = sc.parallelize(1 to 9, 3)

scala> val b = a.map(x => x*2)            // map() 是Transform函数

scala> b.first                          // first() 是Action函数，此时才进行计算。

行动操作会对RDD 计算出一个结果，可以把结果返回，或把结果存储到外部存储系统（如HDFS）中。

RDD是类Iterator的数据结构，也具有Iterator类的Map()、filter()、flatMap()等高阶函数，这些函数是Scala里常用的。

1.1 Transform操作

分成单元素RDD和k-v元素RDD两种。

简单的有Map()、filter()、flatMap()，见下图：

flatMap() 对RDD中每个元素进行函数处理，和map函数不同的是，flatMap返回的可以是不同数据类型。

对单个RDD的Transform函数：

两个RDD之间的Transform函数：

我们打开spark-shell

[root@master ~]# spark-shell
Setting default log level to "WARN".
To adjust logging level use sc.setLogLevel(newLevel).
19/12/11 13:13:18 WARN spark.SparkContext: Use an existing SparkContext, some configuration may not take effect.
Spark context Web UI available at http://192.168.199.130:4040
Spark context available as 'sc' (master = local[*], app id = local-1576041198214).
Spark session available as 'spark'.
Welcome to
      ____              __
     / __/__  ___ _____/ /__
    _ / _ / _ `/ __/  '_/
   /___/ .__/\_,_/_/ /_/\_   version 2.0.0
      /_/
         
Using Scala version 2.11.8 (Java HotSpot(TM) 64-Bit Server VM, Java 1.7.0_67)
Type in expressions to have them evaluated.
Type :help for more information.

scala> val arr=Array("abc abc d d a c f g","sdkj kji jji jkl ooj hhu jkl","jki ihb jl hjihi jiow jkw bvjg lkjsdf","iqweio kljlf kljdfj slkj tkgj")
arr: Array[String] = Array(abc abc d d a c f g, sdkj kji jji jkl ooj hhu jkl, jki ihb jl hjihi jiow jkw bvjg lkjsdf, iqweio kljlf kljdfj slkj tkgj)

scala> val rdd = sc.parallelize(arr)
rdd: org.apache.spark.rdd.RDD[String] = ParallelCollectionRDD[0] at parallelize at <console>:26

scala> val rdd1 = rdd.flatMap(x=>x.split(" "))
rdd1: org.apache.spark.rdd.RDD[String] = MapPartitionsRDD[1] at flatMap at <console>:28

scala> rdd.collect
res0: Array[String] = Array(abc abc d d a c f g, sdkj kji jji jkl ooj hhu jkl, jki ihb jl hjihi jiow jkw bvjg lkjsdf, iqweio kljlf kljdfj slkj tkgj)

scala> rdd1.collect
res1: Array[String] = Array(abc, abc, d, d, a, c, f, g, sdkj, kji, jji, jkl, ooj, hhu, jkl, jki, ihb, jl, hjihi, jiow, jkw, bvjg, lkjsdf, iqweio, kljlf, kljdfj, slkj, tkgj)

scala> val rdd2 =rdd1.map(i=>(i,1))
rdd2: org.apache.spark.rdd.RDD[(String, Int)] = MapPartitionsRDD[2] at map at <console>:30

scala> rdd2.collect
res2: Array[(String, Int)] = Array((abc,1), (abc,1), (d,1), (d,1), (a,1), (c,1), (f,1), (g,1), (sdkj,1), (kji,1), (jji,1), (jkl,1), (ooj,1), (hhu,1), (jkl,1), (jki,1), (ihb,1), (jl,1), (hjihi,1), (jiow,1), (jkw,1), (bvjg,1), (lkjsdf,1), (iqweio,1), (kljlf,1), (kljdfj,1), (slkj,1), (tkgj,1))

scala> val rdd3=rdd2.reduceByKey((x,y)=>(x+y))
rdd3: org.apache.spark.rdd.RDD[(String, Int)] = ShuffledRDD[3] at reduceByKey at <console>:32

scala> rdd3.collect
res3: Array[(String, Int)] = Array((d,2), (kljlf,1), (kji,1), (kljdfj,1), (a,1), (ooj,1), (jkl,2), (sdkj,1), (slkj,1), (jji,1), (bvjg,1), (jkw,1), (hjihi,1), (jl,1), (hhu,1), (f,1), (jki,1), (lkjsdf,1), (abc,2), (iqweio,1), (g,1), (tkgj,1), (jiow,1), (c,1), (ihb,1))

scala>