今天看Data-Intensive Text Processing with MapReduce 这本书的第三章的时候,里面有写到在map端优化wordcount。
对数据密集型数据进行分布式处理的时候,影响数据处理速度的非常重要的一个方面就是map的输出中间结果,在传送到reduce的过程中,很多的中间数据需要进行交换以及包括一些相应的处理,然后再交给相应的reduce。其中中间数据需要在网络中传输,另外中间数据在发送到网络上之前还要写到本地磁盘上,因为网络带宽和磁盘I/O是非常耗时的相比与其他的操作,所以减少中间数据的传输将会增加算法的执行效率,通过使用combiner函数或者其他的方式减少key-value对的个数。下面是一个改进的wordcount算法。
基本的思想是:
在map处理的时候定义一个关联数组,然后对文档进行处理,将<word,次数>加入到关联数组中,word存在,则将相应的次数加1,不存在则直接加入到关联数组中。所有的map任务结束后,然后再在run函数中输出处理结果。
伪代码:
class Mapper
method Map(docid a,doc d)
H =new AssociativeArray
for all term t 属于doc d do
H{t}=H{t}+1;
for all term t 属于 H do
EMIT(term t,count H{t})
class REDUCER
method REDUCE(term t,counts[c1,c2,...])
sum=0
for all count c 属于 counts[c1,c2,...] do
sum+=c
EMIT(term t,count sum)
代码如下:
import java.io.IOException;
import java.io.InputStream;
import java.net.URI;
import java.util.HashMap;
import java.util.Iterator;
import java.util.Map;
import java.util.StringTokenizer;
import java.util.Map.Entry;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FSDataInputStream;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Mapper.Context;
import org.apache.hadoop.mapreduce.lib.input.FileSplit;
import org.apache.hadoop.util.LineReader;
public class Mapper extends
org.apache.hadoop.mapreduce.Mapper<LongWritable, Text, Text, IntWritable> {
int c;
HashMap<String,IntWritable> map= new HashMap<String,IntWritable>();
@Override
protected void map(LongWritable key, Text value,
Context context)
throws IOException, InterruptedException {
String str=value.toString();
StringTokenizer token= new StringTokenizer(str);
while (token.hasMoreTokens()){
String value1=token.nextToken();
if (map.containsKey(value1)){
// System.out.println("ni");
int p=map.get(value1).get()+1;
map.remove(value1);
map.put(value1, new IntWritable(p));
}
else {
// System.out.println("ni");
map.put(value1, new IntWritable(1));
}
}
// TODO Auto-generated method stub
c++;
System.out.println(c);
}
@Override
protected void cleanup(org.apache.hadoop.mapreduce.Mapper.Context context)
throws IOException, InterruptedException {
// TODO Auto-generated method stub
System.out.println("cleanup");
super .cleanup(context);
}
@Override
public void run(Context context) throws IOException, InterruptedException {
// TODO Auto-generated method stub
super .run(context);
System.out.println("run");
Iterator it=map.entrySet().iterator();
while (it.hasNext()){
// System.out.println("nihe");
Map.Entry<String, IntWritable> entry=(Map.Entry<String, IntWritable>) it.next();
// System.out.println("nihe");
context.write( new Text(entry.getKey()), entry.getValue());
}
}
@Override
protected void setup(org.apache.hadoop.mapreduce.Mapper.Context context)
throws IOException, InterruptedException {
// TODO Auto-generated method stub
// System.out.println(context.getInputSplit().toString());
// System.out.println(context.getJobID());
// FileSplit input=(FileSplit)context.getInputSplit();
// String path=input.getPath().toString();
// Configuration conf= new Configuration();
// System.out.println(input.getPath().toString());
// FileSystem fs=FileSystem.get(URI.create(path), conf);
// FSDataInputStream filein=fs.open(input.getPath());
// LineReader in= new LineReader(filein,conf);
// Text line= new Text();
// int cd=in.readLine(line);
// System.out.println(line);
}
}
import java.io.IOException;
import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.Text;
public class Reducer extends
org.apache.hadoop.mapreduce.Reducer<Text, IntWritable, Text, IntWritable> {
@Override
protected void reduce(Text key, Iterable<IntWritable> values,
Context context)
throws IOException, InterruptedException {
// TODO Auto-generated method stub
int sum=0;
for (IntWritable it:values){
sum+=it.get();
}
context.write(key, new IntWritable(sum));
}
}
import java.io.IOException;
import java.net.URI;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
public class Word {
/**
* @param args
* @throws IOException
* @throws ClassNotFoundException
* @throws InterruptedException
*/
public static void main(String[] args) throws IOException, InterruptedException, ClassNotFoundException {
// TODO Auto-generated method stub
Job job= new Job();
Configuration conf= new Configuration();
Path in= new Path(args[0]);
Path out= new Path(args[1]);
FileSystem fs=FileSystem.get(URI.create(args[1]), conf);
fs.delete(out);
FileInputFormat.addInputPath(job, in);
FileOutputFormat.setOutputPath(job, out);
job.setMapperClass(Mapper. class );
job.setMapOutputKeyClass(Text. class );
job.setMapOutputValueClass(IntWritable. class );
job.waitForCompletion( false );
}
}