rdd简单操作

1.原始数据 Key value Transformations(example: ((1, 2), (3, 4), (3, 6)))

  

 

 2. flatMap测试示例

object FlatMapTran {

  //与map相似，区别是源rdd中的元素经map处理后只能生成一个元素，而原有的rdd中的元素经过flatmap处理后可以生成多个元素
  def main(args: Array[String]) {
    
    val spark = SparkSession.builder().appName("FlatMapTran").master("local[1]").getOrCreate()
    val sc = spark.sparkContext;

    val lines = sc.parallelize(Array("hi shao", "scala test", "good", "every"))
    lines.foreach(println)

    val line2 = lines.map(line => line.split(" "))
    line2.foreach(println)

    val line3 = lines.map(line => (line,1))
    line3.foreach(println)

    val line4=lines.flatMap(line => line.split(" "))
    line4.foreach(println)
  }
}

　执行结果：　

hi shao
scala test
good
every
[Ljava.lang.String;@129af42
[Ljava.lang.String;@1c9136
[Ljava.lang.String;@1927273
[Ljava.lang.String;@3b9611
(hi shao,1)
(scala test,1)
(good,1)
(every,1)
hi
shao
scala
test
good
every

3.distinct、reducebykey、groupbykey

object RddDistinct {

  def main(args: Array[String]): Unit = {
    val spark = SparkSession.builder().appName("FlatMapTran").master("local[1]").getOrCreate()
    val sc = spark.sparkContext

    //val datas=sc.parallelize(List(("g","23"),(1,"shao"),("haha","23"),("g","23")))
    val datas=sc.parallelize(Array(("g","23"),(1,"shao"),("haha","23"),("g","23")))
    datas.distinct().foreach(println(_))
    /**结果：
      * (haha,23)
        (1,shao)
        (g,23)
      */

    datas.reduceByKey((x,y)=>x+y).foreach(println)
    /**结果：
      * (haha,23)
        (1,shao)
        (g,2323)
      */

    datas.groupByKey().foreach(println(_))
    /**结果：
      * (haha,CompactBuffer(23))
        (1,CompactBuffer(shao))
        (g,CompactBuffer(23, 23))
      *
      */
  }

}

4.combineByKey(create Combiner, merge Value, merge Combiners, partitioner)

    最常用的基于key的聚合函数,返回的类型可以与输入类型不一样许多基于key的聚合函数都用到了它,像 groupbykey0

    遍历 partition中的元素,元素的key,要么之前见过的,要么不是。如果是新元素,使用我们提供的 createcombiner()函数如果是这个partition中已经存在的key,

    就会使用 mergevalue()函数合计每个 partition的结果的时候,使用 merge Combiners()函数

object CombineByKeyTest {

  def main(args: Array[String]): Unit = {

    val spark = SparkSession.builder().appName("FlatMapTran").master("local[1]").getOrCreate()
    val sc = spark.sparkContext

    val scores=sc.parallelize(Array(("jack",99.0),("jack",80.0),("jack",85.0),("jack",89.0),("lily",95.0),("lily",87.0),("lily",87.0),("lily",77.0)))

    //combineByKey(create Combiner, mergevalue, merge Combiners, partitioner)
    //（创建合并器、合并值、合并合并合并器、分区器）
    val scores2=scores.combineByKey(score=>(1,score),
                                   (c1:(Int,Double),newScore)=>(c1._1+1,c1._2+newScore),
                                   (c1:(Int,Double),c2:(Int,Double))=>(c1._1+c2._1,c1._2+c2._2))
    /**
      * 结果：
      * (lily,(4,346.0))
        (jack,(4,353.0))
      */

    scores2.foreach(println(_))
    scores2.map(score=>{
      (score._1,score._2,score._2._2/score._2._1)
    }).foreach(println(_))
    /**
      * 结果：
      * (lily,(4,346.0),86.5)
        (jack,(4,353.0),88.25)
      */

    scores2.map{case (name,(num,totalScore))=>{
      (name,num,totalScore,totalScore/num)
    }}.foreach(println(_))
    /**
      * 结果：
      * (lily,4,346.0,86.5)
        (jack,4,353.0,88.25)
      */

  }

}