MR--二次排序

1.原始数据
[hadoop@h101 mr2]$ cat mm
1 10
1 3
2 2
9 8
8 11
8 1

2.
import java.io.DataInput;
import java.io.DataOutput;
import java.io.IOException;
import java.util.StringTokenizer;

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.RawComparator;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.io.WritableComparable;
import org.apache.hadoop.io.WritableComparator;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Partitioner;
import org.apache.hadoop.mapreduce.Reducer;
import org.apache.hadoop.util.GenericOptionsParser;

public class SecondarySort {
 

  public static class IntPair
                      implements WritableComparable<IntPair> {
    private int first = 0;
    private int second = 0;
    
 
    public void set(int left, int right) {
      first = left;
      second = right;
    }
    public int getFirst() {
      return first;
    }
    public int getSecond() {
      return second;
    }
    /**
     * Read the two integers.
     * Encoded as: MIN_VALUE -> 0, 0 -> -MIN_VALUE, MAX_VALUE-> -1
     */
    @Override
    public void readFields(DataInput in) throws IOException {
      first = in.readInt() + Integer.MIN_VALUE;
      second = in.readInt() + Integer.MIN_VALUE;
    }
    @Override
    public void write(DataOutput out) throws IOException {
      out.writeInt(first - Integer.MIN_VALUE);
      out.writeInt(second - Integer.MIN_VALUE);
    }
    @Override
//注释1
    public int hashCode() {
      return first * 157 + second;
    }
    @Override
    public boolean equals(Object right) {
//注释2
      if (right instanceof IntPair) {
        IntPair r = (IntPair) right;
        return r.first == first && r.second == second;
      } else {
        return false;
      }
    }
    /** A Comparator that compares serialized IntPair. */
    public static class Comparator extends WritableComparator {
      public Comparator() {
        super(IntPair.class);
      }

      public int compare(byte[] b1, int s1, int l1,
                         byte[] b2, int s2, int l2) {
        return compareBytes(b1, s1, l1, b2, s2, l2);
      }
    }

    static {                                        // register this comparator
      WritableComparator.define(IntPair.class, new Comparator());
    }

    @Override
    public int compareTo(IntPair o) {
      if (first != o.first) {
        return first < o.first ? -1 : 1;
      } else if (second != o.second) {
        return second < o.second ? -1 : 1;
      } else {
        return 0;
      }
    }
  }
 
  /**
   * Partition based on the first part of the pair.
   */
  public static class FirstPartitioner extends Partitioner<IntPair,IntWritable>{
    @Override
    public int getPartition(IntPair key, IntWritable value,
                            int numPartitions) {
      return Math.abs(key.getFirst() * 127) % numPartitions;
    }
  }

  /**
   * Compare only the first part of the pair, so that reduce is called once
   * for each value of the first part.
   */
  public static class FirstGroupingComparator
                implements RawComparator<IntPair> {
    @Override
    public int compare(byte[] b1, int s1, int l1, byte[] b2, int s2, int l2) {
      return WritableComparator.compareBytes(b1, s1, Integer.SIZE/8,
                                             b2, s2, Integer.SIZE/8);
    }

    @Override
    public int compare(IntPair o1, IntPair o2) {
      int l = o1.getFirst();
      int r = o2.getFirst();
      return l == r ? 0 : (l < r ? -1 : 1);
    }
  }

  /**
   * Read two integers from each line and generate a key, value pair
   * as ((left, right), right).
   */
  public static class MapClass
         extends Mapper<LongWritable, Text, IntPair, IntWritable> {
    
    private final IntPair key = new IntPair();
    private final IntWritable value = new IntWritable();
    
    @Override
    public void map(LongWritable inKey, Text inValue,
                    Context context) throws IOException, InterruptedException {
      StringTokenizer itr = new StringTokenizer(inValue.toString());
      int left = 0;
      int right = 0;
      if (itr.hasMoreTokens()) {
        left = Integer.parseInt(itr.nextToken());
        if (itr.hasMoreTokens()) {
          right = Integer.parseInt(itr.nextToken());
        }
        key.set(left, right);
        value.set(right);
        context.write(key, value);
      }
    }
  }
   /**
   * A reducer class that just emits the sum of the input values.
   */
  public static class Reduce
         extends Reducer<IntPair, IntWritable, Text, IntWritable> {
    private static final Text SEPARATOR =
      new Text("------------------------------------------------");
    private final Text first = new Text();
    
    @Override
    public void reduce(IntPair key, Iterable<IntWritable> values,
                       Context context
                       ) throws IOException, InterruptedException {
      context.write(SEPARATOR, null);
      first.set(Integer.toString(key.getFirst()));
      for(IntWritable value: values) {
        context.write(first, value);
      }
    }
  }
  public static void main(String[] args) throws Exception {
    Configuration conf = new Configuration();
    String[] otherArgs = new GenericOptionsParser(conf, args).getRemainingArgs();
    if (otherArgs.length != 2) {
      System.err.println("Usage: secondarysort <in> <out>");
      System.exit(2);
    }
    Job job = new Job(conf, "secondary sort");
    job.setJarByClass(SecondarySort.class);
    job.setMapperClass(MapClass.class);
    job.setReducerClass(Reduce.class);

    // group and partition by the first int in the pair
    job.setPartitionerClass(FirstPartitioner.class);
    job.setGroupingComparatorClass(FirstGroupingComparator.class);

    // the map output is IntPair, IntWritable
    job.setMapOutputKeyClass(IntPair.class);
    job.setMapOutputValueClass(IntWritable.class);

    // the reduce output is Text, IntWritable
    job.setOutputKeyClass(Text.class);
    job.setOutputValueClass(IntWritable.class);
    
    FileInputFormat.addInputPath(job, new Path(otherArgs[0]));
    FileOutputFormat.setOutputPath(job, new Path(otherArgs[1]));
    System.exit(job.waitForCompletion(true) ? 0 : 1);
  }

}

 

 

 

*************************************
1.
要想保证元素不重复，可两个元素是否重复应该依据什么来判断呢？
这就是Object.equals方法了。但是，如果每增加一个元素就检查一次，那么当元素很多时，后添加到集合中的元素比较的次数就非常多了。
也就是说，如果集合中现在已经有1000个元素，那么第1001个元素加入集合时，它就要调用1000次equals方法。这显然会大大降低效率。    
于是，Java采用了哈希表的原理。哈希（Hash）实际上是个人名，由于他提出一哈希算法的概念，所以就以他的名字命名了。
哈希算法也称为散列算法，是将数据依特定算法直接指定到一个地址上。如果详细讲解哈希算法，那需要更多的文章篇幅，我在这里就不介绍了。
初学者可以这样理解，hashCode方法实际上返回的就是对象存储的物理地址。   
这样一来，当集合要添加新的元素时，先调用这个元素的hashCode方法，就一下子能定位到它应该放置的物理位置上。
如果这个位置上没有元素，它就可以直接存储在这个位置上，不用再进行任何比较了；如果这个位置上已经有元素了，
就调用它的equals方法与新元素进行比较，相同的话就不存了，不相同就散列其它的地址。
所以这里存在一个冲突解决的问题。这样一来实际调用equals方法的次数就大大降低了，几乎只需要一两次。

 

2.
instanceof是Java的一个二元操作符，和==，>，<是同一类东东。由于它是由字母组成的，所以也是Java的保留关键字。它的作用是测试它左边的对象是否是它右边的类的实例，返回boolean类型的数据。举个例子：

String s = "I AM an Object!";
boolean isObject = s instanceof Object;

我们声明了一个String对象引用，指向一个String对象，然后用instancof来测试它所指向的对象是否是Object类的一个实例，显然，这是真的，所以返回true，也就是isObject的值为True。

********************************************

[hadoop@h101 mr2]$ /usr/jdk1.7.0_25/bin/javac SecondarySort.java
[hadoop@h101 mr2]$ /usr/jdk1.7.0_25/bin/jar cvf ss.jar SecondarySort*class
[hadoop@h101 mr2]$ hadoop jar ss.jar SecondarySort hdfs://h101:9000/user/hadoop/mm hdfs://h101:9000/user/hadoop/aaa1

[hadoop@h101 mr2]$ hadoop fs -cat /user/hadoop/aaa1/part-r-00000
------------------------------------------------
1       3
1       10
------------------------------------------------
2       2
------------------------------------------------
8       1
8       11
------------------------------------------------
9       8