• flink 自定义ProcessFunction方法


    import it.bigdata.flink.study.SensorReding
    import org.apache.flink.api.common.state.{ValueState, ValueStateDescriptor}
    import org.apache.flink.configuration.Configuration
    import org.apache.flink.streaming.api.functions.KeyedProcessFunction
    import org.apache.flink.streaming.api.scala._
    import org.apache.flink.util.Collector
    
    object ProcessFunctionTest {
      def main(args: Array[String]): Unit = {
        val env = StreamExecutionEnvironment.getExecutionEnvironment
        env.setParallelism(1)
    
        val inputStream = env.socketTextStream("10.18.35.155", 777)
    
        val dataStream = inputStream.map(data => {
          val agg = data.split(",")
          SensorReding(agg(0), agg(1).toLong, agg(2).toDouble)
        })
    //      .keyBy(_.id)
    //      .process(new MykeyedProcessFunction)
    
        val warningSteam = dataStream
          .keyBy(_.id)
          .process(new TempIncreWaring(10000L))
        warningSteam.print()
    
        env.execute("process function test")
      }
    }
    
    
    class TempIncreWaring(interval: Long) extends KeyedProcessFunction[String,SensorReding,String]{
      //定义状态:保存上一个温度值进行比较,保存注册定时器的时间戳用于删除
      lazy val lastTempState: ValueState[Double] = getRuntimeContext.getState(new ValueStateDescriptor[Double]("last-temp", classOf[Double]))
      lazy val timerTsState: ValueState[Long] = getRuntimeContext.getState(new ValueStateDescriptor[Long]("time-ts", classOf[Long]))
    
    
      override def processElement(value: SensorReding, ctx: KeyedProcessFunction[String, SensorReding, String]#Context, out: Collector[String]): Unit = {
        //先取出状态
        val lastTemp = lastTempState.value()
        val timerTs = timerTsState.value()
    
        //更新温度
        lastTempState.update(value.temperature)
    
        //当前温度值和上次温度进行比较
        if(value.temperature>lastTemp && timerTs == 0){
          //如果温度上升,且没有定时器,那么注册当前数据时间10s之后的定时器
          val ts = ctx.timerService().currentProcessingTime() + interval
          ctx.timerService().registerProcessingTimeTimer(ts) //设置10s之后的定时器
          timerTsState.update(ts)  //更新状态
        }else if (value.temperature<lastTemp){
          //如果温度下降,那么删除定时器
          ctx.timerService().deleteProcessingTimeTimer(timerTs)
          timerTsState.clear()   //清空状态
        }
      }
    
      override def onTimer(timestamp: Long, ctx: KeyedProcessFunction[String, SensorReding, String]#OnTimerContext, out: Collector[String]): Unit = {
        out.collect("传感器"+ctx.getCurrentKey+"的温度,连续"+interval/1000+"秒,连续上升")
        timerTsState.clear()
      }
    }
    
    
    //keyedProcessFunction功能测试
    class MykeyedProcessFunction extends KeyedProcessFunction[String,SensorReding,String]{
       var myState: ValueState[Int]= _
    
      //生命周期
      override def open(parameters: Configuration): Unit = {
         myState = getRuntimeContext.getState(new ValueStateDescriptor[Int]("mystate", classOf[Int]))
      }
    
      override def processElement(value: SensorReding, ctx: KeyedProcessFunction[String, SensorReding, String]#Context, out: Collector[String]): Unit ={
        ctx.getCurrentKey   //当前key
        ctx.timestamp()   //当前时间戳
    //    ctx.output()            //侧输出流
        ctx.timerService().currentWatermark()  //当前事件事件
        ctx.timerService().registerEventTimeTimer(ctx.timestamp() + 60000L)   //注册定时器(当前事件之后60秒的定时器)
        ctx.timerService().deleteEventTimeTimer(ctx.timestamp() + 60000L)  //删除定时器
    
      }
    
      //定时器之后需要做的操作
      override def onTimer(timestamp: Long, ctx: KeyedProcessFunction[String, SensorReding, String]#OnTimerContext, out: Collector[String]): Unit = super.onTimer(timestamp, ctx, out)
    }

    数据

    sensor_1,1547718101,35.8
    sensor_1,1547718102,22.2
    sensor_1,1547718101,55.3
    sensor_1,1547718102,24.1
    sensor_1,1547718103,57
    sensor_1,1547718103,58
    sensor_1,1547718103,59
    sensor_6,1547718101,15.4
    sensor_7,1547718102,6.7
    sensor_10,1547718205,38.1
    

      

    author@nohert
  • 相关阅读:
    计算机网络 实验之 面向连接和无连接的套接字到底有什么区别?
    计算机网络 实验之 Internet 套接字
    计算机网络 实验之 socket是什么?套接字是什么?
    PepperLa's Boast(单调队列优化二维dp)
    理想的正方形(单调队列在二维的应用)
    移相器以及相控阵雷达移相器位数的选择
    盲速和频闪——雷达
    多普勒效应----雷达
    线性调频(LFM)脉冲压缩-----------雷达
    雷达----脉冲压缩
  • 原文地址:https://www.cnblogs.com/gzgBlog/p/14928295.html
Copyright © 2020-2023  润新知