Spark中的二次排序(简化版)
1,代码解析:

运行结果:
(de,17,30)
(db,14,28)
(da,15,28)
(dc,17,28)
(dd,17,10)

对比此代码即可知ordering,on方法

|
object SortDemo6 {
def main(args: Array[String]): Unit = {
val conf = new SparkConf()
.setMaster("local")
.setAppName(SortDemo4.getClass.getSimpleName)
val sc = new SparkContext(conf)
val line: RDD[String] = sc.parallelize(List("dd 17 10", "de 17 30", "da 15 28", "db 14 28", "dc 17 28"))
val person = line.map(t => {
val str = t.split(" ")
val name = str(0)
val age = str(1).toInt
val fv = str(2).toInt
(name, age, fv)
})
/** * Ordering[(Int,Int)],里面的参数是要参加排序的 * on[(String,Int,Int)],里面的参数是原函数的参数 调用Ordering.on方法 */ implicit val obj=Ordering[(Int,Int)].on[(String,Int,Int)](t=>(-t._3,t._2))
val res = person.sortBy(t=>t)
res.foreach(println)
}
}
|