source: main/trunk/greenstone3/web/sites/localsite/collect/gberg/java/IndexXML.java

Last change on this file was 5956, checked in by kjdon, 21 years ago

new files for making a simple lucene collection from xml documents

  • Property svn:keywords set to Author Date Id Revision
File size: 866 bytes
Line 
1
2/**
3 *
4 * @author [email protected]
5 * @version
6 */
7
8import java.io.File;
9
10public class IndexXML extends Object {
11
12 File out_dir = null;
13 Indexer indexer = null;
14
15 public void setOutDir(File out_dir) {
16 this.out_dir = out_dir;
17 }
18
19 public void init() {
20 indexer = new Indexer(out_dir, true);
21 }
22 public void finish() {
23 indexer.finish();
24 }
25 public void indexFile(File file) {
26
27 if (file.isDirectory()) {
28 File[] files = file.listFiles();
29 for (int i=0; i<files.length; i++) {
30 if (files[i].isDirectory() || files[i].getName().endsWith(".xml")) {
31 indexFile(files[i]);
32 }
33 }
34
35 } else {
36 String name = file.getName();
37 name = name.substring(0, name.lastIndexOf('.'));
38 System.out.println("Indexing "+file.getPath()+" with id "+name);
39 this.indexer.index(name, file);
40 }
41 }
42
43}
Note: See TracBrowser for help on using the repository browser.