SPARK计算用户不同时段的消费金额

import com.alibaba.fastjson.JSON
import org.apache.spark.streaming.dstream.DStream
import org.apache.spark.streaming.{Seconds, StreamingContext}
import org.apache.spark.streaming.kafka010.{ConsumerStrategies, KafkaUtils, LocationStrategies}

object StreamingETL2 extends App {



  val ssc=Spark_Utils.apply

  //{"user_name": "47vm9sfSn14","user_region": "甘肃省","user_id": 44953796,"pay_way": "支付宝支付","pay_Time": "1524837923","user_active_id": 1,"user_prop_id": 8,"pay_money": 536,"pay_times": 113}

  case class pay_time(user_name:String,user_region:String,user_id:String,pay_way:String,pay_Time:String,user_active_id:Int,user_prop_id:Int,pay_money:Int,pay_times:Int)

  val topics = Array("recharge_topic_li")
  val stream = KafkaUtils.createDirectStream[String, String](
    ssc,
    LocationStrategies.PreferConsistent,//存储策略,如果broker与Kafka代理位于相同的主机上，那么使用PreferBrokers，它更愿意为该分区在Kafka leader上调度分区。最后，如果分区之间的负载有明显的倾斜，请使用PreferFixed
    ConsumerStrategies.Subscribe[String, String](topics, Spark_Utils.kafka_util)
  )

  //计算每个用户的消费总和
  val paymoney= stream.map(f=>{

    val sf=f.value()
    val js=JSON.parseObject(sf,classOf[pay_time])
    Tuple2(js.user_id,js.pay_money)
  }).reduceByKey(_+_)
  

  //打标签 计算每个用户的消费次数
  val paycount=stream.map(f=>{

    val sf=f.value()
    val js=JSON.parseObject(sf,classOf[pay_time])
    (js.user_id,1)

  }).reduceByKey(_+_)



  //计算每5秒每个用户的平均消费金额

  def payavg(usersum:DStream[(String,Int)],usercount:DStream[(String,Int)])={
//    implicit def Int2Double(x:DStream[String],y:DStream[Int])=(x:DStream[String],y.map(f=>f.toDouble))

    //这里与另一个join 可以拿取相同的字段 
    val payment=usersum.join(usercount).map(f=>{
      // 消费/次数
      val user=f._1
      val money =f._2._1
      val count=f._2._2

      (user,0.1*money/count)
    })
    payment

  }

  ssc.checkpoint("window")

  //每5秒求用户前一分钟的消费
  val paymoneys=paymoney.reduceByKeyAndWindow(_+_,_-_,Seconds(5),Seconds(60))

  //每5秒求用户前一分钟的消费次数
  val paycounts= paycount.reduceByKeyAndWindow(_+_,_-_,Seconds(5),Seconds(60))

  val payavgs=payavg(paymoneys,paycounts).map(f=>println(f._1+"平均消费"+f._2)).print()

  ssc.start()
  ssc.awaitTermination()
}