KickJava   Java API By Example, From Geeks To Geeks.

Java > Open Source Codes > org > contineo > core > text > analyze > test > AnalyzeFile


1 /*
2  * AnalyzeFile.java
3  *
4  * Created on 4. Dezember 2003, 21:51
5  */

6
7 package org.contineo.core.text.analyze.test;
8
9 import java.io.BufferedInputStream JavaDoc;
10 import java.io.File JavaDoc;
11 import java.io.FileInputStream JavaDoc;
12 import java.util.Collection JavaDoc;
13 import java.util.Iterator JavaDoc;
14
15 import org.contineo.core.text.analyze.Analyzer;
16 import org.contineo.core.text.analyze.Entry;
17 import org.contineo.core.text.analyze.de.GermanAnalyzer;
18 /**
19  *
20  * @author Michael Scholz
21  */

22 public class AnalyzeFile {
23     
24     /** Creates a new instance of AnalyzeFile */
25     public AnalyzeFile() {
26     }
27     
28     /**
29      * @param args the command line arguments
30      */

31     public static void main(String JavaDoc[] args) {
32         if (args.length != 2) {
33             System.out.println("You must specify a language ('de' or 'en') and the absolute path of the file.");
34         } else {
35             try {
36                 String JavaDoc filename = args[1];
37                 //Analyzer analyzer = new EnglishAnalyzer(4);
38
Analyzer analyzer = new GermanAnalyzer(4);
39                 File JavaDoc file = new File JavaDoc(filename);
40                 BufferedInputStream JavaDoc bis = new BufferedInputStream JavaDoc(new FileInputStream JavaDoc(file));
41                 StringBuffer JavaDoc content = new StringBuffer JavaDoc();
42                 int ichar = 0;
43                 while ((ichar = bis.read()) > 0) {
44                     content.append((char)ichar);
45                 }
46                 analyzer.analyze(content.toString());
47                 Collection JavaDoc coll = analyzer.getTopWords(10);
48                 long count = analyzer.getWordCount();
49                 int relevant = analyzer.relevantWords();
50                 System.out.println("words: " + count);
51                 System.out.println("relevant words: " + relevant);
52                 Iterator JavaDoc iter = coll.iterator();
53                 int i = 1;
54                 while (iter.hasNext()) {
55                     System.out.print("hit " + i + ": ");
56                     Entry entry = (Entry)iter.next();
57                     System.out.println(entry.getWord() + " - " + entry.getNumber());
58                     i++;
59                 }
60             } catch (Exception JavaDoc e) {
61             }
62         }
63     }
64     
65 }
66
Popular Tags