import org.apache.hadoop.io.Writable;

import org.apache.hadoop.io.WritableComparable;



import java.io.DataInput;

import java.io.DataOutput;

import java.io.IOException;

//如果MR中的JavaBean不是作为K2，则只需要实现Writable接口即可

public class FlowBean implements Writable{

    private Integer upFlow;    //上行包数

    private Integer  downFlow; //下行包数

    private Integer upCountFlow; //上行流量和

    private Integer downCountFlow; //下行流量和





    public FlowBean() {

    }



    public FlowBean(Integer upFlow, Integer downFlow, Integer upCountFlow, Integer downCountFlow) {

        this.upFlow = upFlow;

        this.downFlow = downFlow;

        this.upCountFlow = upCountFlow;

        this.downCountFlow = downCountFlow;

    }



    public Integer getUpFlow() {

        return upFlow;

    }



    @Override

    public String toString() {

        return upFlow +

                "\\t" + downFlow +

                "\\t" + upCountFlow +

                "\\t" + downCountFlow;

    }



    public void setUpFlow(Integer upFlow) {

        this.upFlow = upFlow;

    }



    public Integer getDownFlow() {

        return downFlow;

    }



    public void setDownFlow(Integer downFlow) {

        this.downFlow = downFlow;

    }



    public Integer getUpCountFlow() {

        return upCountFlow;

    }



    public void setUpCountFlow(Integer upCountFlow) {

        this.upCountFlow = upCountFlow;

    }



    public Integer getDownCountFlow() {

        return downCountFlow;

    }



    public void setDownCountFlow(Integer downCountFlow) {

        this.downCountFlow = downCountFlow;

    }



    //序列化

    @Override

    public void write(DataOutput out) throws IOException {

        out.writeInt(upFlow);

        out.writeInt(downFlow);

        out.writeInt(upCountFlow);

        out.writeInt(downCountFlow);

    }

    //反序列化

    @Override

    public void readFields(DataInput in) throws IOException {

        this.upFlow = in.readInt();

        this.downFlow = in.readInt();

        this.upCountFlow = in.readInt();

        this.downCountFlow = in.readInt();

    }

}

第二步：定义FlowMapper类

import org.apache.hadoop.io.LongWritable;

import org.apache.hadoop.io.Text;

import org.apache.hadoop.mapreduce.Mapper;



import java.io.IOException;



public class FlowCountMapper extends Mapper<LongWritable, Text,Text,FlowBean> {

    @Override

    protected void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException {

        //1:拆分行文本数据（拆分v1）

        String[] split = value.toString().split("\\t");



        //2:从拆分数组中得到手机号，得到K2

        String phoneNum = split[1];

        //3:从拆分数组中得到4个流量字段，并封装到FlowBean,得到V2

        FlowBean flowBean = new FlowBean();

        flowBean.setUpFlow(Integer.parseInt(split[6]));

        flowBean.setDownFlow(Integer.parseInt(split[7]));

        flowBean.setUpCountFlow(Integer.parseInt(split[8]));

        flowBean.setDownCountFlow(Integer.parseInt(split[9]));



        //4:将K2和V2写入上下文中

        context.write(new Text(phoneNum), flowBean);



    }

}

第三步：定义FlowCountReducer类

import org.apache.hadoop.io.Text;

import org.apache.hadoop.mapreduce.Reducer;



import java.io.IOException;



public class FlowCountReducer extends Reducer<Text,FlowBean,Text,FlowBean> {

    @Override

    protected void reduce(Text key, Iterable<FlowBean> values, Context context) throws IOException, InterruptedException {

        //1:定义四个变量，分别来存储上行包数，下行包数，上行流量总和，下行流量总和

        int upFlow = 0;

        int downFlow = 0;

        int upCountFlow = 0;

        int downCountFlow = 0;



        //2:遍历集合，将集合中每一个FlowBean的四个流量字段相加

        for (FlowBean flowBean : values) {

            upFlow += flowBean.getUpFlow();

            downFlow += flowBean.getDownFlow();

            upCountFlow += flowBean.getUpCountFlow();

            downCountFlow += flowBean.getDownCountFlow();

        }

        //3:K3就是原来的K2，V3就是新的FlowBean

        FlowBean flowBean = new FlowBean(upFlow, downFlow, upCountFlow, downCountFlow);



        //4:将K3和V3写入上下文中

        context.write(key, flowBean);



    }

}

第四步：程序main函数入口FlowCountRunner

import org.apache.hadoop.conf.Configuration;

import org.apache.hadoop.fs.Path;

import org.apache.hadoop.io.NullWritable;

import org.apache.hadoop.io.Text;

import org.apache.hadoop.mapreduce.Job;

import org.apache.hadoop.mapreduce.lib.input.TextInputFormat;

import org.apache.hadoop.mapreduce.lib.output.TextOutputFormat;



import java.io.IOException;



public class FlowCountRunner {

    public static void main(String[] args) throws IOException, ClassNotFoundException, InterruptedException {

        //1、创建建一个job任务对象

        Configuration configuration = new Configuration();





        Job job = Job.getInstance(configuration, "flowcount_demo");



        //2、指定job所在的jar包

        job.setJarByClass(FlowCountRunner.class);



        //3、指定源文件的读取方式类和源文件的读取路径

        job.setInputFormatClass(TextInputFormat.class); //按照行读取

        //TextInputFormat.addInputPath(job, new Path("hdfs://node1:8020/input/wordcount")); //只需要指定源文件所在的目录即可

        TextInputFormat.addInputPath(job, new Path("file:///E:\\\\input\\\\flowcount")); //只需要指定源文件所在的目录即可



        //4、指定自定义的Mapper类和K2、V2类型

        job.setMapperClass(FlowCountMapper.class); //指定Mapper类

        job.setMapOutputKeyClass(Text.class); //K2类型

        job.setMapOutputValueClass(FlowBean.class);//V2类型



        //5、指定自定义分区类（如果有的话）

        //6、指定自定义分组类（如果有的话）

        //7、指定自定义Combiner类(如果有的话)





        //设置ReduceTask个数



        //8、指定自定义的Reducer类和K3、V3的数据类型

        job.setReducerClass(FlowCountReducer.class); //指定Reducer类

        job.setOutputKeyClass(Text.class); //K3类型

        job.setOutputValueClass(FlowBean.class);  //V3类型



        //9、指定输出方式类和结果输出路径

        job.setOutputFormatClass(TextOutputFormat.class);

        //TextOutputFormat.setOutputPath(job, new  Path("hdfs://node1:8020/output/wordcount")); //目标目录不能存在，否则报错

        TextOutputFormat.setOutputPath(job, new  Path("file:///E:\\\\output\\\\flowcount")); //目标目录不能存在，否则报错



        //10、将job提交到yarn集群

        boolean bl = job.waitForCompletion(true); //true表示可以看到任务的执行进度



        //11.退出执行进程

        System.exit(bl?0:1);

    }

}

本博客大数据系列文章会一直每天更新，记得收藏加关注喔~

以上是关于2021年大数据Hadoop（二十四）：MapReduce高阶训练的主要内容，如果未能解决你的问题，请参考以下文章

2021年大数据ELK（二十四）：安装Kibana

2021年大数据Hadoop（二十七）：YARN运行流程

2021年大数据Hadoop（二十一）：MapReuce的Combineer

2021年大数据常用语言Scala（二十四）：函数式编程过滤 filter

2021年大数据Hadoop（十四）：HDFS的高可用机制

2021年大数据Hadoop（二十六）：Yarn三大组件介绍