source: sample/hadoop-0.17/tw/org/nchc/demo/DemoWordCount.java

Last change on this file was 20, checked in by waue, 16 years ago

將改完的 hadoop 0.17版package 放來備份
目前繼續開發 hadoop 0.16 + hbase 1.3

File size: 3.6 KB
Line 
1/**
2 * Program: HBaseRecordPro.java
3 * Editor: Waue Chen
4 * From :  NCHC. Taiwn
5 * Last Update Date: 07/02/2008
6 * Upgrade to 0.17
7 */
8/*
9 * Cloud9: A MapReduce Library for Hadoop
10 */
11
12package tw.org.nchc.demo;
13
14import java.io.IOException;
15import java.util.Iterator;
16import java.util.StringTokenizer;
17
18import org.apache.hadoop.fs.FileSystem;
19import org.apache.hadoop.fs.Path;
20import org.apache.hadoop.io.IntWritable;
21import org.apache.hadoop.io.LongWritable;
22import org.apache.hadoop.io.Text;
23import org.apache.hadoop.mapred.JobClient;
24import org.apache.hadoop.mapred.JobConf;
25import org.apache.hadoop.mapred.MapReduceBase;
26import org.apache.hadoop.mapred.Mapper;
27import org.apache.hadoop.mapred.OutputCollector;
28import org.apache.hadoop.mapred.Reducer;
29import org.apache.hadoop.mapred.Reporter;
30
31import tw.org.nchc.code.Convert;
32
33/**
34 * <p>Simple word count demo. Counts words in the Bible+Shakespeare sample
35 * collection. Expected trace of MapReduce operation:</p>
36 *
37 * <pre>
38 * Map input records=156215
39 * Map output records=1734298
40 * Map input bytes=9068074
41 * Map output bytes=15919397
42 * Combine input records=1734298
43 * Combine output records=135372
44 * Reduce input groups=41788
45 * Reduce input records=135372
46 * Reduce output records=41788
47 * </pre>
48 *
49 */
50public class DemoWordCount {
51
52  // mapper: emits (token, 1) for every word occurrence
53  private static class MapClass extends MapReduceBase implements
54      Mapper<LongWritable, Text, Text, IntWritable> {
55
56    // reuse objects to save overhead of object creation
57    private final static IntWritable one = new IntWritable(1);
58    private Text word = new Text();
59
60    public void map(LongWritable key, Text value,
61        OutputCollector<Text, IntWritable> output, Reporter reporter)
62        throws IOException {
63      String line = ((Text) value).toString();
64      StringTokenizer itr = new StringTokenizer(line);
65      while (itr.hasMoreTokens()) {
66        word.set(itr.nextToken());
67        output.collect(word, one);
68      }
69    }
70  }
71
72  // reducer: sums up all the counts
73  private static class ReduceClass extends MapReduceBase implements
74      Reducer<Text, IntWritable, Text, IntWritable> {
75
76    // reuse objects
77    private final static IntWritable SumValue = new IntWritable();
78
79    public void reduce(Text key, Iterator<IntWritable> values,
80        OutputCollector<Text, IntWritable> output, Reporter reporter)
81        throws IOException {
82      // sum up values
83      int sum = 0;
84      while (values.hasNext()) {
85        sum += values.next().get();
86      }
87      SumValue.set(sum);
88      output.collect(key, SumValue);
89    }
90  }
91
92  private DemoWordCount() {
93  }
94 
95  /**
96   * Runs the demo.
97   */
98  public static void main(String[] args) throws IOException {
99    String filename = "/user/waue/test/132.txt";
100    String outputPath = "sample-counts";
101    int mapTasks = 20;
102    int reduceTasks = 1;
103
104    JobConf conf = new JobConf(DemoWordCount.class);
105    conf.setJobName("wordcount");
106
107    conf.setNumMapTasks(mapTasks);
108    conf.setNumReduceTasks(reduceTasks);
109    //0.16
110//    conf.setInputPath(new Path(filename));
111    Convert.setInputPath(conf, new Path(filename));
112    conf.setOutputKeyClass(Text.class);
113    conf.setOutputValueClass(IntWritable.class);
114    // 0.16
115//    conf.setOutputPath(new Path(outputPath));
116    Convert.setInputPath(conf, new Path(outputPath));
117    conf.setMapperClass(MapClass.class);
118    conf.setCombinerClass(ReduceClass.class);
119    conf.setReducerClass(ReduceClass.class);
120   
121    // Delete the output directory if it exists already
122    Path outputDir = new Path(outputPath);
123    // 0.16
124//    FileSystem.get(conf).delete(outputDir);
125    FileSystem.get(conf).delete(outputDir,true);
126    JobClient.runJob(conf);
127  }
128}
Note: See TracBrowser for help on using the repository browser.