Hadoop自定义outputformat输出文件格式

OutputFormat的使用场景：

为了实现控制最终文件的输出路径和输出格式,可以自定义OutputFormat。
　　　　例如:要在一个MapReducer程序中根据数据的不同输出结果到不同目录,这类灵活的输出要求可以通过自定义OutputFormat来实现。
　　自定义OutputFormat大致步骤:
　（1）自定义一个类继承FileOutputFormat;
　（2）改写RecordWriter,具体改写输出数据的write()方法。
　测试数据:
　

1). 实体类:

package com.root.outputformat;

import org.apache.hadoop.io.WritableComparable;

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

public class FilterBean implements WritableComparable<FilterBean> {

    private String Commodity;    //种类
    private Double price;    //商品价格

    //空参构造
    public FilterBean() {
        super();
    }

    public FilterBean(String commodity, Double price) {
        Commodity = commodity;
        this.price = price;
    }

    public int compareTo(FilterBean filterBean) {
        int result = Commodity.compareTo(filterBean.getCommodity());   //按照Ascill表进行比较
        //种类进行升序排序
        if (result > 0) {
            result = 1;
        } else if (result < 0) {
            result = -1;
        } else {
            //销售额进行降序排序

            if (price > filterBean.getPrice()) {
                result = -1;
            } else if (price < filterBean.getPrice()) {
                result = 1;
            } else {
                result = 0;
            }
        }
        return result;
    }

    public void write(DataOutput dataOutput) throws IOException {
        dataOutput.writeUTF(Commodity);
        dataOutput.writeDouble(price);
    }

    public void readFields(DataInput dataInput) throws IOException {
        Commodity = dataInput.readUTF();
        price = dataInput.readDouble();
    }

    public String getCommodity() {
        return Commodity;
    }

    public void setCommodity(String commodity) {
        Commodity = commodity;
    }

    public Double getPrice() {
        return price;
    }

    public void setPrice(Double price) {
        this.price = price;
    }

    @Override
    public String toString() {
        return Commodity + "\t"+ price + "\r\n";
    }
}

2)Map程序：

package com.root.outputformat;

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 java.io.IOException;

public class FilterMap extends Mapper<LongWritable, Text, FilterBean, NullWritable> {
    FilterBean filterbean = new FilterBean();

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

        //  0	Office Supplies	2	408.3

        //1.获取一行
        String line = value.toString();

        //2.切割
        String[] fields = line.split(",");

        //3.封装
        String Commodity = fields[1];

        Double price = Double.parseDouble(fields[3]);

        filterbean.setCommodity(Commodity);

        filterbean.setPrice(price);


        //4.写出
        context.write(filterbean, NullWritable.get());

    }
}

3）reduce程序

package com.root.outputformat;

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

import java.io.IOException;

public class FilterReduce extends Reducer<FilterBean, NullWritable, FilterBean, NullWritable> {


    @Override
    protected void reduce(FilterBean key, Iterable<NullWritable> values, Context context) throws IOException, InterruptedException {
        for (NullWritable value : values) {
            context.write(key,NullWritable.get());
        }
    }
}

4).自定义outputformat需要继承FileOutputFormat，重写getRecordWriter方法

package com.root.outputformat;

import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.mapreduce.RecordWriter;
import org.apache.hadoop.mapreduce.TaskAttemptContext;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;

import java.io.IOException;

public class FilterOutputFormat extends FileOutputFormat<FilterBean, NullWritable> {

    public RecordWriter<FilterBean, NullWritable> getRecordWriter(TaskAttemptContext taskAttemptContext) throws IOException, InterruptedException {


        return new FRecorWriter(taskAttemptContext);
    }
}

5).实现RecordWriter抽象类.

package com.root.outputformat;

import org.apache.hadoop.fs.FSDataOutputStream;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IOUtils;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.mapreduce.RecordWriter;
import org.apache.hadoop.mapreduce.TaskAttemptContext;

import java.io.IOException;

public class FRecorWriter extends RecordWriter<FilterBean, NullWritable> {
    FileSystem fs;
    FSDataOutputStream fosOffice, fosFurniture, fosTechnology, fosOther;

    public FRecorWriter(TaskAttemptContext taskAttemptContext) {


        try {

            //1. 获取文件系统
            fs = FileSystem.get(taskAttemptContext.getConfiguration());

            //2.创建输出到Office Supplies.txt的输出流
            fosOffice = fs.create(new Path("F:\\scala\\Workerhdfs\\output10\\ffice Supplies.txt"));

            //3.创建输出到Furniture.txt的输出流
            fosFurniture = fs.create(new Path("F:\\scala\\Workerhdfs\\output10\\Furniture.txt"));

            //4.创建输出到Technology.txt的输出流
            fosTechnology = fs.create(new Path("F:\\scala\\Workerhdfs\\output10\\Technology.txt"));

            //5.创建输出到Other.txt的输出流
            fosOther = fs.create(new Path("F:\\scala\\Workerhdfs\\output10\\Other.txt"));
        } catch (IOException e) {
            e.printStackTrace();
        }


    }



    public void write(FilterBean filterBean, NullWritable nullWritable) throws IOException, InterruptedException {
        String line = filterBean.toString();
        if (line.contains("Office Supplies")) {
            fosOffice.write(line.getBytes());
        } else if (line.contains("Furniture")) {
            fosFurniture.write(line.getBytes());
        } else if (line.contains("Technology")) {
            fosTechnology.write(line.getBytes());
        } else {
            fosOther.write(line.getBytes());
        }
    }

    public void close(TaskAttemptContext taskAttemptContext) throws IOException, InterruptedException {
        IOUtils.closeStream(fosOffice);
        IOUtils.closeStream(fosFurniture);
        IOUtils.closeStream(fosTechnology);
        IOUtils.closeStream(fosOther);
    }
}

根据操作系统选择合适的输出路径，我的测试环境是windows系统我都把文件写入这个目录下F:\scala\Workerhdfs\output10\,jar包运行到linux系统需要更换路径

6）二次排序(辅助排序):

package com.root.outputformat;

import org.apache.hadoop.io.WritableComparable;
import org.apache.hadoop.io.WritableComparator;

public class FilterGroupComparator extends WritableComparator {
      protected FilterGroupComparator(){
          super(FilterBean.class,true);
      }

    @Override
    public int compare(WritableComparable a, WritableComparable b) {

        //只要种类相同，就认为是相同的key

        FilterBean aBean = (FilterBean) a;
        FilterBean bBean = (FilterBean) b;

        int result = aBean.getCommodity().compareTo(bBean.getCommodity());
        if (result > 0) {
            result = 1;
        } else if (result < 0) {
            result = -1;
        } else {
            result = 0;
        }
        return result;

    }
}

这一块看不懂可以看我之前二次排序的案例(小编也是无聊，才将二次排序融入到自定义outputformat案例中)

7).主程序:

package com.root.outputformat;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;

import java.io.IOException;

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

  args = new String[]{"F:\\scala\\Workerhdfs\\input5","F:\\scala\\Workerhdfs\\output10"};

        //1.获取job对象
        Configuration conf = new Configuration();
        Job job = Job.getInstance(conf);

        //2.设置jar路径
        job.setJarByClass(FilterDriver.class);

        //3.关联mapper和reducer
        job.setMapperClass(FilterMap.class);
        job.setReducerClass(FilterReduce.class);

        //4 设置mapper输出的key和value类型
        job.setMapOutputKeyClass(FilterBean.class);
        job.setMapOutputValueClass(NullWritable.class);

        //5. 设置最终输出的key和value类型
        job.setOutputKeyClass(FilterBean.class);
        job.setOutputValueClass(NullWritable.class);

        //设置 reduce端的分组
        job.setGroupingComparatorClass(FilterGroupComparator.class);

        //自定义的输出格式组件设置到job中
        job.setOutputFormatClass(FilterOutputFormat.class);

        //6.设置输出路径
        FileInputFormat.setInputPaths(job, new Path(args[0]));


        FileOutputFormat.setOutputPath(job, new Path(args[1]));


        //7.提交job
        boolean result = job.waitForCompletion(true);
        System.exit(result ? 0 : 1);
    }
}

由于我的测试文件中第二列除了Furniture,Office Supplies,Technology字段没有其他的字段，Other.txt文件内容为空

效果如下:

数据有点多，linux系统运行此脚本,查看每个文件的前五行(Other查看所有,反正它也是空)
打包在yarn上运行

查看文件内容前五行：