Changeset 9 for sample/WordCount.java

Timestamp:

Jun 13, 2008, 5:45:02 PM (17 years ago)

Author:

waue

Message:

comment

File:

• sample/WordCount.java (modified) (5 diffs)

sample/WordCount.java

@@ -1 @@
-/*
- * map reduce sample code
+/**
+ * Program: WordCount.java
+ * Editor: Waue Chen 
+ * From :  NCHC. Taiwn
+ * Last Update Date: 06/13/2008
+ */
+
+/**
+ * Purpose : 
+ *  Store the result of WordCount.java from Hbase to Hadoop file system 
+ * 
+ * HowToUse : 
+ *  Make sure Hadoop file system is running correctly.
+ *  Put text file on the directory "/local_src/input" 
+ *  You can use the instruction to upload "/local_src/input" to HDFS input dir
+ *    $ bin/hadoop dfs -put /local_src/input input
+ *  Then modify the $filepath parameter in construtor to be correct and run this code.
+ *  
+ * 
+ * Check Result:
+ *  inspect http://localhost:50070 by web explorer
  */
 package tw.org.nchc.code;
@@ -23 +42 @@
 
 public class WordCount {
-
+  private String filepath;
+  private String outputPath;
+  
+  public WordCount(){
+    filepath = "/user/waue/input/";
+    outputPath = "counts1";
+  }
+  public WordCount(String path,String output){
+    filepath = path;
+    outputPath = output;
+  }
   // mapper: emits (token, 1) for every word occurrence
   private static class MapClass extends MapReduceBase implements
@@ -69 +98 @@
    */
   public static void main(String[] args) throws IOException {
-    String filename = "/user/waue/input/";
-    String outputPath = "sample-counts";
-    int mapTasks = 20;
+    WordCount wc = new WordCount();
+    
+    int mapTasks = 1;
     int reduceTasks = 1;
-
     JobConf conf = new JobConf(WordCount.class);
     conf.setJobName("wordcount");
@@ -80 +108 @@
     conf.setNumReduceTasks(reduceTasks);
 
-    conf.setInputPath(new Path(filename));
+    conf.setInputPath(new Path(wc.filepath));
     conf.setOutputKeyClass(Text.class);
     conf.setOutputValueClass(IntWritable.class);
-    conf.setOutputPath(new Path(outputPath));
+    conf.setOutputPath(new Path(wc.outputPath));
 
     conf.setMapperClass(MapClass.class);
@@ -90 +118 @@
     
     // Delete the output directory if it exists already
-    Path outputDir = new Path(outputPath);
+    Path outputDir = new Path(wc.outputPath);
     FileSystem.get(conf).delete(outputDir);