07-flink-1.10.1- 用户自定义 flink source api

Posted 2021-09-25 逃跑的沙丁鱼

1 定义自己想要的流返回值数据类型

package com.study.liucf.unbounded.source

/**
 * @Author liucf
 * @Date 2021/9/8
 */
case class LiucfSensorReding(id:String,timestamp:Long,temperature:Double)

2 定义自己的读取数据源的类

package com.study.liucf.unbounded.source

import java.util.Random

import org.apache.flink.streaming.api.functions.source.SourceFunction


/**
 * @Author liucf
 * @Date 2021/9/8
 *
 *      我读取数据源类
 */
class LiucfSourceFunction() extends SourceFunction[LiucfSensorReding]{
  var running = true
  //模拟读取数据
  override def run(ctx: SourceFunction.SourceContext[LiucfSensorReding]): Unit = {
    //定义随机变量
    val random = new Random()
    //定义一组10个传感器初始温度
    var currentTemp = 1.to(10).map(r=>("sensor_"+r,random.nextDouble()*100))
    while(running){
      currentTemp = currentTemp.map(r=>(r._1,r._2+random.nextGaussian()))
      val currentTimestamp = System.currentTimeMillis()
      //逐条发出去
      currentTemp.foreach(r=>ctx.collect(LiucfSensorReding(r._1,currentTimestamp,r._2)))
    }
    Thread.sleep(1000)
  }

  //如果调用cancel方法，则running标准位会被设置成false,停止读取数据
  override def cancel(): Unit = {
     running = false
  }
}

3 flink使用自定义的数据源处理数据

package com.study.liucf.unbounded.source

import org.apache.flink.streaming.api.scala._

/**
 * @Author liucf
 * @Date 2021/9/8
 *
 *      自定义数据源读取数据
 */
object MySource {
  def main(args: Array[String]): Unit = {
    //创建flink执行环境
    val env = StreamExecutionEnvironment.getExecutionEnvironment
    //添加自定义数据源
    val ds: DataStream[LiucfSensorReding] = env.addSource(new LiucfSourceFunction())
    //输出到标准控制台
    ds.print()
    //启动flink运行
    env.execute("liucf defined source api")

  }

}

输出结果