Flink 窗口 - 技术栈

介绍：流式计算是一种被设计用于处理无限数据集的数据处理引擎，而无限数据集是指一种不断增长的本质上无限的数据集，而 window 是一种切割无限数据为有限块进行处理的手段，其分为两种类型：1、时间窗口，2：计数窗口

一、时间窗口

时间窗口根据窗口实现原理的不同分成三类：滚动窗口（Tumbling Window）、滑动窗口（Sliding Window）和会话窗口（Session Window）

1.1、滚动窗口（Tumbling Windows）

介绍：将数据依据固定的窗口长度（时间）对数据进行切片

特点：时间对齐，窗口长度固定，没有重叠

java 复制代码

package com.xx.window;

import lombok.AllArgsConstructor;
import lombok.Data;
import lombok.NoArgsConstructor;
import lombok.extern.slf4j.Slf4j;
import org.apache.flink.api.common.functions.MapFunction;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.datastream.WindowedStream;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.windowing.assigners.TumblingProcessingTimeWindows;
import org.apache.flink.streaming.api.windowing.time.Time;
import org.apache.flink.streaming.api.windowing.windows.TimeWindow;

/**
 * @author aqi
 * @since 2023/8/30 15:46
 */
@Slf4j
public class WindowReduceDemo {

    public static void main(String[] args) throws Exception {
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

        // 接收socket数据(windows使用:nc -lp 7777,linux使用:nc -lk 7777进行数据推送)
        SingleOutputStreamOperator<Demo> sensorDS = env
                .socketTextStream("127.0.0.1", 7777)
                .map(new DemoMapFunction());

        // 滚动窗口(固定窗口长度为:10秒,每隔10s统计一次)
        WindowedStream<Demo, String, TimeWindow> sensorWS = sensorDS
                .keyBy(Demo::getId)
                .window(TumblingProcessingTimeWindows.of(Time.seconds(10)));

        // 聚合(也可以使用别的算子进行聚合)
        SingleOutputStreamOperator<Demo> reduce = sensorWS.reduce(
                (value1, value2) -> new Demo(value1.getId(), value1.getValue() + value2.getValue())
        );
        // 打印计算结果
        reduce.print();
        // 触发计算
        env.execute();
    }
}

@Data
@AllArgsConstructor
@NoArgsConstructor
class Demo {

    private String id;

    private Long value;
}

class DemoMapFunction implements MapFunction<String, Demo> {

    @Override
    public Demo map(String value) {
        String[] datas = value.split(",");
        return new Demo(datas[0], Long.valueOf(datas[1]));
    }
}

1.2、滑动窗口（Sliding Windows）

介绍：滑动窗口是固定窗口的更广义的一种形式，滑动窗口由固定的窗口长度和滑动间隔组成

特点：时间对齐，窗口长度固定，有重叠

java 复制代码

package com.xx.window;

import lombok.AllArgsConstructor;
import lombok.Data;
import lombok.NoArgsConstructor;
import lombok.extern.slf4j.Slf4j;
import org.apache.flink.api.common.functions.MapFunction;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.datastream.WindowedStream;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.windowing.assigners.SlidingProcessingTimeWindows;
import org.apache.flink.streaming.api.windowing.assigners.TumblingProcessingTimeWindows;
import org.apache.flink.streaming.api.windowing.time.Time;
import org.apache.flink.streaming.api.windowing.windows.TimeWindow;

/**
 * @author aqi
 * @since 2023/8/30 15:46
 */
@Slf4j
public class WindowReduceDemo {

    public static void main(String[] args) throws Exception {
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

        // 接收socket数据(windows使用:nc -lp 7777,linux使用:nc -lk 7777进行数据推送)
        SingleOutputStreamOperator<Demo> sensorDS = env
                .socketTextStream("127.0.0.1", 7777)
                .map(new DemoMapFunction());

        // 滚动窗口(固定窗口长度为:10秒,每隔10s统计一次)
//        WindowedStream<Demo, String, TimeWindow> sensorWS = sensorDS
//                .keyBy(Demo::getId)
//                .window(TumblingProcessingTimeWindows.of(Time.seconds(10)));

        // 滑动窗口(每5秒钟统计一次,过去的10秒钟内的数据)
        WindowedStream<Demo, String, TimeWindow> sensorWS = sensorDS
                .keyBy(Demo::getId)
                .window(SlidingProcessingTimeWindows.of(Time.seconds(10), Time.seconds(2)));

        // 聚合(也可以使用别的算子进行聚合)
        SingleOutputStreamOperator<Demo> reduce = sensorWS.reduce(
                (value1, value2) -> new Demo(value1.getId(), value1.getValue() + value2.getValue())
        );
        // 打印计算结果
        reduce.print();
        // 触发计算
        env.execute();
    }
}

@Data
@AllArgsConstructor
@NoArgsConstructor
class Demo {

    private String id;

    private Long value;
}

class DemoMapFunction implements MapFunction<String, Demo> {

    @Override
    public Demo map(String value) {
        String[] datas = value.split(",");
        return new Demo(datas[0], Long.valueOf(datas[1]));
    }
}

1.3、会话窗口（Session Windows）

介绍：由一系列事件组合一个指定时间长度的 timeout 间隙组成，也就是一段时间没有接收到新数据就会生成新的窗口

特点：时间无对齐

java 复制代码

package com.xx.window;

import lombok.AllArgsConstructor;
import lombok.Data;
import lombok.NoArgsConstructor;
import lombok.extern.slf4j.Slf4j;
import org.apache.flink.api.common.functions.MapFunction;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.datastream.WindowedStream;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.windowing.assigners.ProcessingTimeSessionWindows;
import org.apache.flink.streaming.api.windowing.time.Time;
import org.apache.flink.streaming.api.windowing.windows.TimeWindow;

/**
 * @author aqi
 * @since 2023/8/30 15:46
 */
@Slf4j
public class WindowReduceDemo {

    public static void main(String[] args) throws Exception {
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

        // 接收socket数据(windows使用:nc -lp 7777,linux使用:nc -lk 7777进行数据推送)
        SingleOutputStreamOperator<Demo> sensorDS = env
                .socketTextStream("127.0.0.1", 7777)
                .map(new DemoMapFunction());

        // 滚动窗口(固定窗口长度为:10秒,每隔10s统计一次)
//        WindowedStream<Demo, String, TimeWindow> sensorWS = sensorDS
//                .keyBy(Demo::getId)
//                .window(TumblingProcessingTimeWindows.of(Time.seconds(10)));


        // 滑动窗口(每5秒钟统计一次,过去的10秒钟内的数据)
//        WindowedStream<Demo, String, TimeWindow> sensorWS = sensorDS
//                .keyBy(Demo::getId)
//                .window(SlidingProcessingTimeWindows.of(Time.seconds(10), Time.seconds(2)));


        // 会话窗口(超时间隔5s)
        WindowedStream<Demo, String, TimeWindow> sensorWS = sensorDS
                .keyBy(Demo::getId)
                .window(ProcessingTimeSessionWindows.withGap(Time.seconds(5)));


        // 聚合(也可以使用别的算子进行聚合)
        SingleOutputStreamOperator<Demo> reduce = sensorWS.reduce(
                (value1, value2) -> new Demo(value1.getId(), value1.getValue() + value2.getValue())
        );
        // 打印计算结果
        reduce.print();
        // 触发计算
        env.execute();
    }
}

@Data
@AllArgsConstructor
@NoArgsConstructor
class Demo {

    private String id;

    private Long value;
}

class DemoMapFunction implements MapFunction<String, Demo> {

    @Override
    public Demo map(String value) {
        String[] datas = value.split(",");
        return new Demo(datas[0], Long.valueOf(datas[1]));
    }
}

1.4、总结

滚动窗口：TumblingProcessingTimeWindows.of(Time.seconds(10))

滑动窗口：SlidingProcessingTimeWindows.of(Time.seconds(10), Time.seconds(2))

会话窗口：ProcessingTimeSessionWindows.withGap(Time.seconds(5))

二、计数窗口

和时间窗口类似，同样也分为三种，使用方法也基本相同

1.1、滚动窗口（Tumbling Windows）

窗口长度=5个元素

java 复制代码

sensorKs.countWindow(5);

1.2、滑动窗口（Sliding Windows）

窗口长度=5个元素,滑动步长=2个元素

java 复制代码

sensorKs.countWindow(5, 2);

1.3、会话窗口（Session Windows）

三、窗口触发方式

3.1、增量聚合

来一条数据,计算一条数据,窗口触发的时候输出计算结果

函数：reduce、aggregate等，除了process都是增量函数

3.2、全窗口函数

数据来了不计算,存储起来,窗口触发的时候,计算并输出结果，并且可以获取到窗口信息、上下文信息等，灵活性非常的强

函数：process

java 复制代码

package com.xx.window;

import lombok.AllArgsConstructor;
import lombok.Data;
import lombok.NoArgsConstructor;
import lombok.extern.slf4j.Slf4j;
import org.apache.commons.lang3.time.DateFormatUtils;
import org.apache.flink.api.common.functions.MapFunction;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.datastream.WindowedStream;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.functions.windowing.ProcessWindowFunction;
import org.apache.flink.streaming.api.windowing.assigners.TumblingProcessingTimeWindows;
import org.apache.flink.streaming.api.windowing.time.Time;
import org.apache.flink.streaming.api.windowing.windows.TimeWindow;
import org.apache.flink.util.Collector;

/**
 * @author aqi
 * @since 2023/8/30 15:46
 */
@Slf4j
public class WindowReduceDemo {

    public static void main(String[] args) throws Exception {
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

        // 接收socket数据(windows使用:nc -lp 7777,linux使用:nc -lk 7777进行数据推送)
        SingleOutputStreamOperator<Demo> sensorDS = env
                .socketTextStream("127.0.0.1", 7777)
                .map(new DemoMapFunction());

        // 滚动窗口(固定窗口长度为:10秒,每隔10s统计一次)
        WindowedStream<Demo, String, TimeWindow> sensorWS = sensorDS
                .keyBy(Demo::getId)
                .window(TumblingProcessingTimeWindows.of(Time.seconds(10)));

        SingleOutputStreamOperator<String> process = sensorWS.process(new ProcessWindowFunction<Demo, String, String, TimeWindow>() {

            /**
             * 全窗口函数的计算逻辑,窗口触发时才会调用一次,统一计算窗口的所有数据
             * @param s 分组的key
             * @param context 上下文
             * @param elements 存的数据
             * @param out 采集器
             */
            @Override
            public void process(String s, ProcessWindowFunction<Demo, String, String, TimeWindow>.Context context, Iterable<Demo> elements, Collector<String> out) {
                long start = context.window().getStart();
                long end = context.window().getEnd();

                String startWindow = DateFormatUtils.format(start, "yyyy-MM-dd HH:mm:ss");
                String endWindow = DateFormatUtils.format(end, "yyyy-MM-dd HH:mm:ss");

                long count = elements.spliterator().estimateSize();

                out.collect("key=" + s + "的窗口[" + startWindow + "," + endWindow + "]包含:" + count + "条数据===>" + elements);
            }
        });

        // 打印计算结果
        process.print();
        // 触发计算
        env.execute();
    }
}

@Data
@AllArgsConstructor
@NoArgsConstructor
class Demo {

    private String id;

    private Long value;
}

class DemoMapFunction implements MapFunction<String, Demo> {

    @Override
    public Demo map(String value) {
        String[] datas = value.split(",");
        return new Demo(datas[0], Long.valueOf(datas[1]));
    }
}

3.3、增量函数和全窗口函数组合使用

java 复制代码

package com.xx.window;

import com.xx.entity.WaterSensor;
import com.xx.functions.WaterSensorMapFunction;
import org.apache.commons.lang3.time.DateFormatUtils;
import org.apache.flink.api.common.functions.AggregateFunction;
import org.apache.flink.streaming.api.datastream.KeyedStream;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.datastream.WindowedStream;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.functions.windowing.ProcessWindowFunction;
import org.apache.flink.streaming.api.windowing.assigners.TumblingProcessingTimeWindows;
import org.apache.flink.streaming.api.windowing.time.Time;
import org.apache.flink.streaming.api.windowing.windows.TimeWindow;
import org.apache.flink.util.Collector;

/**
 * @author aqi
 * @since 2023/8/30 15:46
 */
public class WindowAggregateAndProcessDemo {

    public static void main(String[] args) throws Exception {
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

        SingleOutputStreamOperator<WaterSensor> sensorDS = env
                .socketTextStream("127.0.0.1", 7777)
                .map(new WaterSensorMapFunction());


        KeyedStream<WaterSensor, String> sensorKS = sensorDS.keyBy(WaterSensor::getId);

        WindowedStream<WaterSensor, String, TimeWindow> sensorWS = sensorKS.window(TumblingProcessingTimeWindows.of(Time.seconds(10)));

        SingleOutputStreamOperator<String> result = sensorWS.aggregate(
                // 第一个参数:输入数据的类型,第二个参数:累加器的类型,存储的中间计算结果的类型,第三个参数:输出的类型
                new AggregateFunction<WaterSensor, Integer, String>() {
                    @Override
                    public Integer createAccumulator() {
                        System.out.println("初始化累加器");
                        return null;
                    }

                    @Override
                    public Integer add(WaterSensor value, Integer accumulator) {
                        if (accumulator == null) {
                            accumulator = 0;
                        }
                        Integer add = value.getVc() + accumulator;
                        System.out.println("调用add方法,累加结果:" + add);
                        return add;
                    }

                    @Override
                    public String getResult(Integer accumulator) {
                        System.out.println("获取最终结果");
                        return accumulator.toString();
                    }

                    @Override
                    public Integer merge(Integer a, Integer b) {
                        System.out.println("调用merge方法");
                        return null;
                    }
                }, new ProcessWindowFunction<String, String, String, TimeWindow>() {
                    @Override
                    public void process(String s, ProcessWindowFunction<String, String, String, TimeWindow>.Context context, Iterable<String> elements, Collector<String> out) throws Exception {
                        long start = context.window().getStart();
                        long end = context.window().getEnd();

                        String startWindow = DateFormatUtils.format(start, "yyyy-MM-dd HH:mm:ss");
                        String endWindow = DateFormatUtils.format(end, "yyyy-MM-dd HH:mm:ss");

                        long count = elements.spliterator().estimateSize();

                        out.collect("key=" + s + "的窗口[" + startWindow + "," + endWindow + "]包含:" + count + "条数据===>" + elements);
                    }
                });

        result.print();
        env.execute();
    }
}