1 package old;
  2 
  3 import java.io.IOException;
  4 import java.net.URI;
  5 import java.util.Iterator;
  6 
  7 import org.apache.hadoop.conf.Configuration;
  8 import org.apache.hadoop.fs.FileSystem;
  9 import org.apache.hadoop.fs.Path;
 10 import org.apache.hadoop.io.LongWritable;
 11 import org.apache.hadoop.io.Text;
 12 import org.apache.hadoop.mapred.FileInputFormat;
 13 import org.apache.hadoop.mapred.FileOutputFormat;
 14 import org.apache.hadoop.mapred.JobClient;
 15 import org.apache.hadoop.mapred.JobConf;
 16 import org.apache.hadoop.mapred.MapReduceBase;
 17 import org.apache.hadoop.mapred.Mapper;
 18 import org.apache.hadoop.mapred.OutputCollector;
 19 import org.apache.hadoop.mapred.Reducer;
 20 import org.apache.hadoop.mapred.Reporter;
 21 
 22 /**
 23  * 老API实现单词统计
 24  * 
 25  */
 26 /**
 27  * hadoop版本1.x的包一般是mapreduce
 28  * 
 29  * hadoop版本0.x的包一般是mapred
 30  * 
 31  */
 32 
 33 public class OldApp {
 34 
 35     static final String INPUT_PATH = "hdfs://chaoren:9000/hello";
 36     static final String OUT_PATH = "hdfs://chaoren:9000/out";
 37 
 38     public static void main(String[] args) throws Exception {
 39         Configuration conf = new Configuration();
 40         FileSystem fileSystem = FileSystem.get(new URI(INPUT_PATH), conf);
 41         Path outPath = new Path(OUT_PATH);
 42         if (fileSystem.exists(outPath)) {
 43             fileSystem.delete(outPath, true);
 44         }
 45         /**
 46          * 改动1：不再使用Job，而是使用JobConf
 47          * 
 48          * 改动2：类的包名不再使用mapreduce,而是使用mapred
 49          * 
 50          * 改动3：不再使用job.waitForCompletion(true)提交作业，而是使用JobClient.runJob(job);
 51          */
 52         JobConf job = new JobConf(conf, OldApp.class);
 53 
 54         // 1.1指定读取的文件位于哪里
 55         FileInputFormat.setInputPaths(job, INPUT_PATH);
 56         // 指定如何对输入的文件进行格式化，把输入文件每一行解析成键值对
 57         // job.setInputFormatClass(TextInputFormat.class);
 58 
 59         // 1.2指定自定义的map类
 60         job.setMapperClass(MyMapper.class);
 61         // map输出的<k,v>类型。如果<k3,v3>的类型与<k2,v2>类型一致，则可以省略
 62         // job.setOutputKeyClass(Text.class);
 63         // job.setOutputValueClass(LongWritable.class);
 64 
 65         // 1.3分区
 66         // job.setPartitionerClass(org.apache.hadoop.mapreduce.lib.partition.HashPartitioner.class);
 67         // 有一个reduce任务运行
 68         // job.setNumReduceTasks(1);
 69 
 70         // 1.4排序、分组
 71 
 72         // 1.5归约
 73 
 74         // 2.2指定自定义reduce类
 75         job.setReducerClass(MyReducer.class);
 76         // 指定reduce的输出类型
 77         job.setOutputKeyClass(Text.class);
 78         job.setOutputValueClass(LongWritable.class);
 79 
 80         // 2.3指定写出到哪里
 81         FileOutputFormat.setOutputPath(job, outPath);
 82         // 指定输出文件的格式化类
 83         // job.setOutputFormatClass(TextOutputFormat.class);
 84 
 85         // 把job提交给jobtracker运行
 86         JobClient.runJob(job);
 87     }
 88 
 89     /**
 90      * 新API：extends Mapper
 91      * 
 92      * 老API：extends MapReduceBase implements Mapper
 93      */
 94     static class MyMapper extends MapReduceBase implements
 95             Mapper<LongWritable, Text, Text, LongWritable> {
 96         public void map(LongWritable k1, Text v1,
 97                 OutputCollector<Text, LongWritable> collector, Reporter reporter)
 98                 throws IOException {
 99             String[] split = v1.toString().split("\t");
100             for (String word : split) {
101                 collector.collect(new Text(word), new LongWritable(1));
102             }
103         }
104     }
105 
106     static class MyReducer extends MapReduceBase implements
107             Reducer<Text, LongWritable, Text, LongWritable> {
108         public void reduce(Text k2, Iterator<LongWritable> v2s,
109                 OutputCollector<Text, LongWritable> collector, Reporter reporter)
110                 throws IOException {
111             long times = 0L;
112             while (v2s.hasNext()) {
113                 long temp = v2s.next().get();
114                 times += temp;
115             }
116             collector.collect(k2, new LongWritable(times));
117         }
118     }
119 
120 }