0
|
1 package de.mpiwg.dwinter.fulltextIndexer.harvester;
|
|
2
|
|
3 import java.io.File;
|
|
4 import java.io.IOException;
|
|
5 import java.io.InputStream;
|
|
6
|
|
7 import org.xml.sax.EntityResolver;
|
|
8 import org.xml.sax.InputSource;
|
|
9 import org.xml.sax.SAXException;
|
|
10
|
|
11 public class MyResolver implements EntityResolver {
|
|
12
|
|
13 public InputSource resolveEntity(String publicId, String systemId)
|
|
14 throws SAXException, IOException {
|
|
15 // TODO Auto-generated method stub
|
|
16 // System.out.println("public:"+publicId);
|
|
17 // System.out.println("static"+systemId);
|
|
18 //File f = new File(
|
|
19 // "/Users/dwinter/Documents/Projekte/data-mining/developmentEclipse/fulltextIndexer/catalog/xhtml1-transitional.dtd");
|
|
20
|
|
21
|
|
22 //if (!f.exists()) {
|
|
23 // System.err.println("Cant't find xhtml-dtd: MyResolver");
|
|
24 // return null;
|
|
25 //}
|
|
26 //if (publicId.equals("-//W3C//DTD XHTML 1.0 Transitional//EN"))
|
|
27 // return new InputSource(f.getAbsolutePath());
|
|
28
|
|
29 if (publicId.equals("-//W3C//DTD XHTML 1.0 Transitional//EN")){
|
|
30 InputStream res = getClass().getResourceAsStream("/de/mpiwg/dwinter/fulltextIndexer/catalog/xhtml1-transitional.dtd");
|
|
31 return new InputSource(res);
|
|
32 } else if (publicId.equals("-//W3C//ENTITIES Latin 1 for XHTML//EN")){
|
|
33 InputStream res = getClass().getResourceAsStream("/de/mpiwg/dwinter/fulltextIndexer/catalog/xhtml-lat1.ent");
|
|
34 return new InputSource(res);
|
|
35 } else if (publicId.equals("-//W3C//ENTITIES Symbols for XHTML//EN")){
|
|
36 InputStream res = getClass().getResourceAsStream("/de/mpiwg/dwinter/fulltextIndexer/catalog/xhtml-symbol.ent");
|
|
37 return new InputSource(res);
|
|
38 } else if (publicId.equals("-//W3C//ENTITIES Special for XHTML//EN")){
|
|
39 InputStream res = getClass().getResourceAsStream("/de/mpiwg/dwinter/fulltextIndexer/catalog/xhtml-special.ent");
|
|
40 return new InputSource(res);
|
|
41 }
|
|
42 return new InputSource(systemId);
|
|
43 }
|
|
44
|
|
45 }
|