MapReduce之RecordReader组件源码解析及实例

简述

无论我们以怎样的方式从分片中读取一条记录，每读取一条记录都会调用RecordReader类；
系统默认的RecordReader是LineRecordReader，TextInputFormat；
LineRecordReader是用每行的偏移量作为map的key，每行的内容作为map的value；
而SequenceFileInputFormat的RecordReader是SequenceFileRecordReader；
应用场景：自定义读取每一条记录的方式；自定义读入key的类型，如希望读取的key是文件的路径或名字而不是该行在文件中的偏移量。

TextInputFormat源码如下：

package org.apache.hadoop.mapreduce.lib.input;
/** An {@link InputFormat} for plain text files.  Files are broken into lines.
 * Either linefeed or carriage-return are used to signal end of line.  Keys are
 * the position in the file, and values are the line of text.. */
public class TextInputFormat extends FileInputFormat<LongWritable, Text> {

  @Override
  public RecordReader<LongWritable, Text> 
    createRecordReader(InputSplit split,
                       TaskAttemptContext context) {
    String delimiter = context.getConfiguration().get(
        "textinputformat.record.delimiter");
    byte[] recordDelimiterBytes = null;
    if (null != delimiter)
      recordDelimiterBytes = delimiter.getBytes(Charsets.UTF_8);
    return new LineRecordReader(recordDelimiterBytes);
  }

  @Override
  protected boolean isSplitable(JobContext context, Path file) {
    final CompressionCodec codec =
      new CompressionCodecFactory(context.getConfiguration()).getCodec(file);
    if (null == codec) {
      return true;
    }
    return codec instanceof SplittableCompressionCodec;
  }
}

textinputformat.record.delimiter指的是读取一行的数据的终止符号，即遇到textinputformat.record.delimiter所包含的字符时，该一行的读取结束。
可以通过Configuration的set()方法来设置自定义的终止符，如果没有设置 textinputformat.record.delimiter，那么Hadoop就采用以CR，LF或者CRLF作为终止符，这一点可以查看LineReader的readDefaultLine方法 。

LineRecordReader源码如下：

package org.apache.hadoop.mapreduce.lib.input;

/**
 * Treats keys as offset in file and value as line. 
 */
public class LineRecordReader extends RecordReader<LongWritable, Text> {
     public void initialize(InputSplit genericSplit,
                         TaskAttemptContext context) throws IOException {
        ......
        start = split.getStart();
        end = start + split.getLength();
        final Path file = split.getPath();

        // open the file and seek to the start of the split
        final FileSystem fs = file.getFileSystem(job);
        fileIn = fs.open(file);
        ......
        // If this is not the first split, we always throw away first record
        // because we always (except the last split) read one extra line in
        // next() method.
        if (start != 0) {
            start += in.readLine(new Text(), 0, maxBytesToConsume(start));
        }
        this.pos = start;
        ......
    }
    ......
}

自定义RecordReader

1、继承抽象类RecordReader，实现RecordReader的一个实例；
2、实现自定义InputFormat类，重写InputFormat中createRecordReader()方法，返回值是自定义的RecordReader实例；
3、配置job.setInputFormatClass()设置自定义的InputFormat实例；

实例

数据：

10
20
30
40
50
60
70
……

要求：读取整个文件，分别计算奇数行与偶数行数据之和
奇数行之和：10+30+50+70=160
偶数行之和：20+40+60=120

package Recordreader;

import java.io.IOException;
import java.net.URI;
import java.net.URISyntaxException;
import java.util.List;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FSDataInputStream;
import org.apache.hadoop.fs.FileSystem;
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.InputSplit;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.JobContext;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Partitioner;
import org.apache.hadoop.mapreduce.Reducer;
import org.apache.hadoop.mapreduce.RecordReader;
import org.apache.hadoop.mapreduce.TaskAttemptContext;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.input.FileSplit;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import org.apache.hadoop.util.LineReader;

public class MyRecordReader {

    private final static String INPUT_PATH = "hdfs://liguodong:8020/inputsum";
    private final static String OUTPUT_PATH = "hdfs://liguodong:8020/outputsum";    

    public static class DefRecordReader extends RecordReader<LongWritable, Text>{

        private long start;//分片开始位置
        private long end;//分片结束位置
        private long pos;
        private FSDataInputStream fin = null;
        //自定义自己的key与value
        private LongWritable key = null;
        private Text value = null;
        //A class that provides a line reader from an input stream.
        private LineReader reader = null;

        @Override
        public void initialize(InputSplit split, TaskAttemptContext context)
                throws IOException, InterruptedException {
            FileSplit fileSplit = (FileSplit)split;
            start = fileSplit.getStart();
            end = start + fileSplit.getLength();
            Path path = fileSplit.getPath();//获取输入分片的路径
            Configuration conf = context.getConfiguration();
            //Return the FileSystem that owns this Path.
            FileSystem fs = path.getFileSystem(conf);
            fin = fs.open(path);
            reader = new LineReader(fin);
            pos = 1;
        }

        @Override
        public boolean nextKeyValue() throws IOException, InterruptedException {

            if(key == null){
                key = new LongWritable();
            }
            key.set(pos);//设置key
            if(value == null){
                value = new Text();
            }
            //并没有跨块，跨文件，而是一个文件作为不可分割的 
            if(reader.readLine(value)==0){//一次读取行的内容,并设置值
                return false;
            }
            pos++;
            return true;
        }

        @Override
        public LongWritable getCurrentKey() throws IOException,
                InterruptedException {
            return key;
        }

        @Override
        public Text getCurrentValue() throws IOException, InterruptedException {
            return value;
        }

        /**
          * Get the progress within the split
          */
        @Override
        public float getProgress() throws IOException, InterruptedException {
            return 0;
        }

        @Override
        public void close() throws IOException {
            fin.close();
        }

    }

    public static class MyFileInputFormat extends FileInputFormat<LongWritable, Text>{

        @Override
        public RecordReader<LongWritable, Text> createRecordReader(
                InputSplit split, TaskAttemptContext context)
                throws IOException, InterruptedException {

            return new DefRecordReader();
        }

        @Override
        protected boolean isSplitable(JobContext context, Path filename) {
            return false;
        }
    }


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

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

            context.write(key, value);
        }
    }


    public static class DefPartitioner extends Partitioner<LongWritable,Text>{

        @Override
        public int getPartition(LongWritable key, Text value, int numPartitions) {
            //判断奇数行还是偶数行，确定分区
            if(key.get()%2==0){
                key.set(1);//偶数行key通通改为1
                return 1;
            }else {
                key.set(0);//奇数行key通通改为0
                return 0;
            }
        }

    }

    //接收来自不同分区的数据
    public static class MyReducer extends Reducer<LongWritable, Text,Text, IntWritable>{
        Text write_key = new Text();
        IntWritable write_value = new IntWritable();

        @Override
        protected void reduce(LongWritable key, Iterable<Text> values,
                Context context)
                throws IOException, InterruptedException {
            int sum=0;
            for (Text val : values) {
                sum += Integer.parseInt(val.toString());
            }
            if(key.get()==0){
                write_key.set("奇数行之和");
            }else {
                write_key.set("偶数行之和");
            }
            write_value.set(sum);
            context.write(write_key, write_value);
        }   
    }




    public static void main(String[] args) throws IOException, URISyntaxException, ClassNotFoundException, InterruptedException {
        //1、配置  
        Configuration conf = new Configuration();
        final FileSystem fileSystem = FileSystem.get(new URI(INPUT_PATH),conf);

        if(fileSystem.exists(new Path(OUTPUT_PATH)))
        {
            fileSystem.delete(new Path(OUTPUT_PATH),true);
        }
        Job job = Job.getInstance(conf, "Define RecordReader"); 

        //2、打包运行必须执行的方法
        job.setJarByClass(MyRecordReader.class);

        //3、输入路径
        FileInputFormat.addInputPath(job, new Path(INPUT_PATH));  

        //设置输入格式
        job.setInputFormatClass(MyFileInputFormat.class);

        //4、Map
        job.setMapperClass(MyMapper.class);
        //指定map的输出的<k,v>类型,如果<k3,v3>的类型与<k2,v2>的类型一致，那么可以省略。
        job.setMapOutputKeyClass(LongWritable.class);
        job.setMapOutputValueClass(Text.class);

        //5、Combiner
        //job.setCombinerClass(MyReducer.class);
        job.setPartitionerClass(DefPartitioner.class);

        //6、Reducer
        job.setReducerClass(MyReducer.class);
        job.setNumReduceTasks(2);//reduce个数默认是1

        //如果<k3,v3>的类型与<k2,v2>的类型不一致，要么都省略，要么都要写。
        job.setOutputKeyClass(Text.class);
        job.setOutputValueClass(IntWritable.class);

        //7、 输出路径
        FileOutputFormat.setOutputPath(job, new Path(OUTPUT_PATH));

        //8、提交作业
        System.exit(job.waitForCompletion(true) ? 0 : 1);
    }
}

[root@liguodong file]# vi inputsum
[root@liguodong file]# hdfs dfs -put inputsum /inputsum
[root@liguodong file]# hdfs dfs -cat /inputsum
10
20
30
40
50
60
70
[root@liguodong file]# yarn jar MyRecordReader.jar
[root@liguodong file]# hdfs dfs -ls /outputsum
Found 3 items
-rw-r--r--   1 root supergroup          0 2015-06-14 21:19 /outputsum/_SUCCESS
-rw-r--r--   1 root supergroup         20 2015-06-14 21:19 /outputsum/part-r-00000
-rw-r--r--   1 root supergroup         20 2015-06-14 21:19 /outputsum/part-r-00001
[root@liguodong file]# hdfs dfs -cat /outputsum/part-r-00000
奇数行之和      160
[root@liguodong file]# hdfs dfs -cat /outputsum/part-r-00001
偶数行之和      120