转载请标注原链接http://www.cnblogs.com/xczyd/p/8608906.html
在Hdfs学习笔记1 - 使用Java API访问远程hdfs集群中,我们已经可以完成了访问hdfs的配置。
接下来我们试图写一个最简单的map reduce程序。网上一般给的Demo都是统计词频(Word Count),
于是我们也简单先实现一下:
首先准备一个内容大致如下的test.txt文件:
aa
bbb
aaa
ab
ba
bb
bbb
bba
baa
aa
aaa
aa
aab
每行有且仅有一个单词,然后我们的程序需要完成的任务是统计每个词出现的次数。代码如下:
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.DoubleWritable;
import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.*;
import org.apache.hadoop.mapreduce.lib.input.TextInputFormat;
import org.apache.hadoop.mapreduce.lib.output.TextOutputFormat; import java.io.*;
import java.util.StringTokenizer; public class MRTest { public static class WordCountMap extends Mapper<Object, Text, Text, IntWritable> { private Text word = new Text();
private final static IntWritable one = new IntWritable(1); //key: 行号//value: 单词
@Override
protected void map(Object key, Text value, Context context) throws IOException,
InterruptedException {
StringTokenizer tokenizer = new StringTokenizer(value.toString());
while (tokenizer.hasMoreTokens()) {
String token = tokenizer.nextToken();
word.set(token);
context.write(word, one);
}
}
} public static class WordCountReduce extends Reducer<Text, IntWritable, Text,
IntWritable> { private IntWritable result = new IntWritable(); //key: 单词
//values: 在map阶段写入的one的数量
@Override
protected void reduce(Text key, Iterable<IntWritable> values, Context context)
throws IOException, InterruptedException {
int sum = 0;
for (IntWritable value: values) {
sum += value.get();
}
result.set(sum);
context.write(key, result);
}
} public static void main(String[] args) throws IOException, ClassNotFoundException,
InterruptedException { System.setProperty("HADOOP_USER_NAME", "root"); String hdfsUserName = "root";
String inputPath = "/jzhang4/test2.txt";
String outputPath = "/jzhang4/out"; Configuration conf = new Configuration(); conf.set("fs.default.name", "hdfs://hd-nn-test-01:9000");
conf.set("dfs.client.use.datanode.hostname","true"); FileSystem fs = FileSystem.get(conf);
fs.delete(new Path(outputPath), true); Job job = Job.getInstance(conf); job.setJarByClass(MRTest.class);
job.setJobName("jzhang4_avg_calc"); job.setMapperClass(WordCountMap.class);
job.setReducerClass(WordCountReduce.class);
TextInputFormat.setInputPaths(job, new Path(inputPath));
TextOutputFormat.setOutputPath(job, new Path(outputPath));
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(IntWritable.class); while (true) {
if (job.waitForCompletion(true)) {
break;
} else {
Thread.sleep(1000);
}
}
}
}
首先在map阶段,我们每读到一个word,就往context里面对应的word上添加一个1;
然后再reduce阶段,对于每一个word,再把这些1给累加起来;
main函数中有一些对于job的配置,主要是为了在IDE中运行此map reduce程序,
如果想要打成jar包放到服务器上去运行,或者是想在yarn上执行,需要调整配置,这些内容不在本篇讨论范围之内。
马不停蹄(事实上停了一天|||- -),立刻开始第二个Map Reduce程序。给定如下这么一个文件:
10
9
8
7
6
5
4
3
2
1
0
每行有且仅有一个数字。写一个map reduce程序来计算所有数字的平均值。
第一反应当然是照抄一下上面的程序,只是在map阶段往context里写入要统计的数字本身,
然后reduce阶段把这些数字加和再除以数字的数目即可。
但是老板说,这么操作过于naive。于是想到了map reduce中的计数器(Recorders),利用计数器可以
在reduce阶段什么也不做就得到均值。于是代码如下:
public static class CalculateAvgMap extends Mapper<Object, Text, Text,
IntWritable> { private final static IntWritable one = new IntWritable(1); public enum Recorders {
ItemCounter,
ValueCounter
} //key: 行号
//value: 数字
@Override
protected void map(Object key, Text value, Context context) throws IOException,
InterruptedException {
StringTokenizer tokenizer = new StringTokenizer(value.toString());
while (tokenizer.hasMoreTokens()) {
int number = Integer.parseInt(tokenizer.nextToken());
context.getCounter(Recorders.ItemCounter).increment(1);
context.getCounter(Recorders.ValueCounter).increment(number);
}
}
} public static class CalculateAvgReduce extends Reducer<Text, IntWritable, Text,
IntWritable> { private DoubleWritable result = new DoubleWritable(); @Override
protected void reduce(Text key, Iterable<IntWritable> values, Context context)
throws IOException, InterruptedException {
}
}
然后在适当的地方(比如main函数中)获得计数器的值即可:
public static void main(String[] args) throws IOException, ClassNotFoundException,
InterruptedException { System.setProperty("HADOOP_USER_NAME", "root"); String hdfsUserName = "root";
String inputPath = "/jzhang4/test2.txt";
String outputPath = "/jzhang4/out"; Configuration conf = new Configuration(); conf.set("fs.default.name", "hdfs://hd-nn-test-01:9000");
conf.set("dfs.client.use.datanode.hostname","true"); FileSystem fs = FileSystem.get(conf);
fs.delete(new Path(outputPath), true); Job job = Job.getInstance(conf); job.setJarByClass(MRTest.class);
job.setJobName("jzhang4_avg_calc"); job.setMapperClass(CalculateAvgMap.class);
job.setReducerClass(CalculateAvgReduce.class);
TextInputFormat.setInputPaths(job, new Path(inputPath));
TextOutputFormat.setOutputPath(job, new Path(outputPath));
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(IntWritable.class);
while (true) {
if (job.waitForCompletion(true)) {
break;
} else {
Thread.sleep(1000);
}
} Counters counters = job.getCounters(); Counter itemCounter = counters.findCounter(
CalculateAvgMap.Recorders.ItemCounter);
System.out.println(itemCounter.getName() + "\t" + itemCounter.getValue());
Counter valueCounter = counters.findCounter(
CalculateAvgMap.Recorders.ValueCounter);
System.out.println(valueCounter.getName() + "\t" + valueCounter.getValue());
}
这个实现有一定的问题,就是加合的结果会存在超界的风险。如果想要避免超界,比较合适的做法是利用bitmap的思想,
开32个或者64个Recorder,分别记录每一个bit的count数。当然在enum里面写茫茫多Recorder有点蠢,
在找到更优雅的办法之前,暂时先不写这么丑陋的代码了...