top5

package TeacherTopN2

import org.apache.spark.{SparkConf, SparkContext}
import org.apache.spark.rdd.RDD

import scala.collection.mutable

object TeacherN05 {
  def main(args: Array[String]): Unit = {


    val conf = new SparkConf().setAppName("Teacher").setMaster("local[*]")

    val sc = new SparkContext(conf)

    val lines = sc.textFile("F:\\spark\\计算每个学科最受欢迎Top2.log")

    val line: RDD[((String, String), Int)] = lines.map(it => {
      val strings = it.split("/")
      val teacher = strings(3)
      val subject = strings(2).split("[.]")(0)

      ((subject, teacher), 1)
    })

    val subDis: Array[String] = line.map(i => i._1._1).distinct().collect()

    val partition: MyTreeSetPartition = new MyTreeSetPartition(subDis)
    //使用reduceByKey进行分区,加聚合
    val rdd1: RDD[((String, String), Int)] = line.reduceByKey(partition, _ + _)

    val value1: RDD[((String, String), Int)] = rdd1.mapPartitions(it => {
      implicit val value = Ordering[Int].on[((String, String), Int)](i => -i._2)
      val tuples = new mutable.TreeSet[((String, String), Int)]()
      //因为mapPartition只有出发Action才执行,用map不行,只会返回空的迭代器,所有用foreach每条遍历出来
      it.foreach(i => {
        tuples += i
        if (tuples.size > 2) {
          tuples -= tuples.last
        }
      })
      tuples.iterator
    })
    println(value1.collect().toBuffer)



  }

}