MapReduce自定义类的编写

在有些场景中，我们可以自己定义一个类用于传输或者处理key或者value，这个类必须实现序列化和反序列化，实现Writable接口。

假设我们需要处理一批记录，该记录中有三个字段：用户手机号(phone_num)，上行流量(upflow)和下行流量(downflow)，我们需要最终统计每一个手机号的所有upflow、downflow以及(upflow+downflow)，这种场景下用自定义类来处理就比较方便：

package com.wenhuan.defineclass;

import java.io.DataInput;
import java.io.DataOutput;
import java.io.IOException;

import org.apache.hadoop.io.Writable;

public class FlowBean implements Writable{
    private int upflow;
    private int downflow;
    private int sumflow;
    
    public int getUpflow() {
        return upflow;
    }
    public void setUpflow(int upflow) {
        this.upflow = upflow;
    }
    public int getDownflow() {
        return downflow;
    }
    public void setDownflow(int downflow) {
        this.downflow = downflow;
    }
    public int getSumflow() {
        return sumflow;
    }
    public void setSumflow(int sumflow) {
        this.sumflow = sumflow;
    }
    
    @Override
    public String toString() {
        return upflow + "\t" + downflow + "\t" + sumflow;
    }
    public FlowBean() {
        super();
    }
    public FlowBean(int upflow, int downflow) {
        super();
        this.upflow = upflow;
        this.downflow = downflow;
        this.sumflow = this.upflow + this.downflow;
    }
    @Override
    public void write(DataOutput out) throws IOException {
        //序列化的方法
        out.writeInt(upflow);
        out.writeInt(downflow);
        out.writeInt(sumflow);
    }
    @Override
    public void readFields(DataInput in) throws IOException {
        //反序列化的方法
        this.upflow = in.readInt();
        this.downflow = in.readInt();
        this.sumflow = in.readInt();
    }
}

重写Mapper类

package com.wenhuan.defineclass;

import java.io.IOException;

import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Mapper;

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

    @Override
    protected void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException {
        String[] datas = value.toString().split("\t");
        String phoneNum = datas[0];
        FlowBean fb = new FlowBean(Integer.parseInt(datas[datas.length - 1]),Integer.parseInt(datas[datas.length - 2]));
        context.write(new Text(phoneNum), fb);
    }
}

重写Reducer类

package com.wenhuan.defineclass;

import java.io.IOException;

import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Reducer;

public class MyReducer extends Reducer<Text,FlowBean,Text,FlowBean>{
    @Override
    protected void reduce(Text key, Iterable<FlowBean> values, Context context) throws IOException, InterruptedException {
        int sumupflow = 0;
        int sumdownflow = 0;
        for(FlowBean fb:values) {
            sumupflow += fb.getUpflow();
            sumdownflow += fb.getDownflow();
        }
        FlowBean fb1 = new FlowBean(sumupflow,sumdownflow);
        context.write(key, fb1);
    }
}

MapReduce的排序与分组

maptask和reducetask之间，框架默认按map端输出key的字典顺序进行排序。

理论上，只要Map端的KEYOUT实现了WritableComparable接口，都可以满足排序和分组的需求；具体来说，WritableComparable接口中的compareTo方法为记录的排序提供依据。

自定义类方式定义排序逻辑

我们重写之前的FlowBean接口，使其具有排序功能：

package com.wenhuan.defineclass;

import java.io.DataInput;
import java.io.DataOutput;
import java.io.IOException;

import org.apache.hadoop.io.WritableComparable;

public class FlowBeanSort implements WritableComparable<FlowBeanSort>{
    private String phoneNum;
    private int upflow;
    private int downflow;
    private int sumflow;
    
    public String getPhoneNum() {
        return phoneNum;
    }
    public void setPhoneNum(String phoneNum) {
        this.phoneNum = phoneNum;
    }
    public int getUpflow() {
        return upflow;
    }
    public void setUpflow(int upflow) {
        this.upflow = upflow;
    }
    public int getDownflow() {
        return downflow;
    }
    public void setDownflow(int downflow) {
        this.downflow = downflow;
    }
    public int getSumflow() {
        return sumflow;
    }
    public void setSumflow(int sumflow) {
        this.sumflow = sumflow;
    }
    
    public FlowBeanSort() {
        super();
        // TODO Auto-generated constructor stub
    }
    public FlowBeanSort(String phoneNum, int upflow, int downflow, int sumflow) {
        super();
        this.phoneNum = phoneNum;
        this.upflow = upflow;
        this.downflow = downflow;
        this.sumflow = sumflow;
    }
    @Override
    public String toString() {
        return phoneNum + "\t" + upflow + "\t" + downflow + "\t" + sumflow;
    }
    @Override
    public void readFields(DataInput in) throws IOException {
        this.phoneNum = in.readUTF();
        this.upflow = in.readInt();
        this.downflow = in.readInt();
        this.sumflow = in.readInt();
        
    }
    @Override
    public void write(DataOutput out) throws IOException {
        out.writeUTF(phoneNum);
        out.writeInt(upflow);
        out.writeInt(downflow);
        out.writeInt(sumflow);
    }
    @Override
    public int compareTo(FlowBeanSort o) {
        // 按照总流量倒序排序
        return o.getSumflow() - this.getSumflow();
    }   
}

对应的map和reduce:

package com.wenhuan.defineclass;

import java.io.IOException;

import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Reducer;

public class MyDefineSort {
    static class MyMapper extends Mapper<LongWritable,Text,FlowBeanSort,NullWritable>{
        @Override
        protected void map(LongWritable key, Text value,Context context) throws IOException, InterruptedException {
            String[] infos = value.toString().split("\t");
            FlowBeanSort fb = new FlowBeanSort(infos[0],Integer.parseInt(infos[1].trim()),Integer.parseInt(infos[2].trim()),Integer.parseInt(infos[3].trim()));
            context.write(fb, NullWritable.get());
        }
    }
    
    static class MyReducer extends Reducer<FlowBeanSort,NullWritable,FlowBeanSort,NullWritable>{
        @Override
        protected void reduce(FlowBeanSort key, Iterable<NullWritable> values,Context context) throws IOException, InterruptedException {
            
            for (NullWritable nl:values) {
                context.write(key, NullWritable.get());
            }
        }
    }
    ...
}

串行MapReduce解决重新排序问题

WritableComparator类中的compare方法则为记录的分组提供依据。

默认情况下，compare方法内部直接调用compareTo方法，所以默认情况下，分组和排序依赖的字段相同。compareTo返回值为0的数据将分为一组。

但是有些需求我们需要分组和排序依赖的字段分离，也即分组和排序依赖不同的字段。

以最简单的WordCount为例，最终结果默认是以word的字典顺序进行排序的，如果我们希望对count进行排序，那么在原本的wordcount程序后还需要再串行一个用于二次排序的MapReduce程序，该程序在Map阶段以count作为KEYOUT进行排序，在Reduce阶段再将word作为KEYOUT，从而实现重新排序：

package com.wenhuan.wordcount;

import java.io.IOException;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Reducer;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;

public class MySort {
    static class MyMapper extends Mapper<LongWritable,Text,IntWritable,Text>{

        @Override
        protected void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException {
            String[] datas = value.toString().split("\t");
            String word = datas[0];
            int count = Integer.parseInt(datas[1]);
            context.write(new IntWritable(count), new Text(word));
        }
    }
    
    static class MyReducer extends Reducer<IntWritable,Text,Text,IntWritable>{

        @Override
        protected void reduce(IntWritable key, Iterable<Text> values, Context context) throws IOException, InterruptedException {
            for(Text v:values) {
                context.write(v, key);
            }
        }
    }
    
    public static void main(String[] args) throws IOException, ClassNotFoundException, InterruptedException {
        //加载配置文件
        Configuration conf = new Configuration();
        //启动一个Job,封装maper和reducer
        Job job = Job.getInstance(conf);
        //设置计算程序的主驱动类，运行的时候打成jar包运行。
        job.setJarByClass(MySort.class);
        //设置Maper和Reduer类
        job.setMapperClass(MyMapper.class);
        job.setReducerClass(MyReducer.class);
        //设置mapper的输出类型
        job.setMapOutputKeyClass(IntWritable.class);
        job.setMapOutputValueClass(Text.class);
        //设置reducer的输出类型
        job.setOutputKeyClass(Text.class);
        job.setOutputValueClass(IntWritable.class);
        //设置输入路径和输出路径
        FileInputFormat.addInputPath(job, new Path(args[0]));
        FileOutputFormat.setOutputPath(job,new Path(args[1]));
        //提交,需要打印日志
        job.waitForCompletion(true);
    }
}

自定义类方式实现重新排序

当然，我们也可以自定义一个继承WritableComparator的类，然后重写compare和compareTo方法，也可以让排序和分组逻辑实现分离。

假设我们有如下数据：

math, huangxiaoming，85,87,86,75,88,94

english, huangdatou, 48,58,98,56,73,75

字段名依次是：课程、姓名、历次得分

我们需要求出每门课程参考学生成绩最高平均分的学生的信息，一个分组求最大值的需求，有两个步骤：分组和排序。

但是需要以课程字段进行分组，以课程+平均分进行排序，分组和排序的逻辑必须分离。

需要注意的是，shuffle过程中，排序逻辑在分组逻辑之前，所以如果需要先分组后排序，那么写排序逻辑时，分组字段也要参与排序。