Spark自定义排序

1，实现comparable接口和Serializable接口

class GoodsBean(  money: Double,  pid: String,  cid: String,  coverUrl: String,  title: String)
                extends Comparable[GoodsBean] with Serializable {

  override def compareTo(o: GoodsBean): Int = {
       this.money.compareTo(o.money)
  }

}

2，定义为case calss，并实现Orderd特质

case class GoodsBean(  money: Double,  pid: String,  cid: String,  coverUrl: String,  title: String)
                extends Ordered[GoodsBean]{
    override def compare(that: GoodsBean): Int = {
      this.money.compareTo(that.money)
    }
}

3，使用隐式转换，new Ordering

    implicit val ordering: Ordering[(String, Double, String)] = new Ordering[(String, Double, String)]{
      override def compare(x: (String, Double, String), y: (String, Double, String)): Int = {
        -x._2.compareTo(y._2)
      }
    }

简写：

implicit val ordering: Ordering[(String, Double, String)] = Ordering[Double].on[(String, Double, String)](-_._2)

4，使用元组的排序特点规则进行排序

元组排序先比较第一个字段，相等在比较第二个字段、如果还相等再比较下去