|
1
|
|
package net.sf.flock.util;
|
|
2
|
|
|
|
3
|
|
import java.io.IOException;
|
|
4
|
|
import java.io.InputStream;
|
|
5
|
|
import java.io.StringReader;
|
|
6
|
|
import java.io.UnsupportedEncodingException;
|
|
7
|
|
import java.net.URL;
|
|
8
|
|
|
|
9
|
|
import net.sf.flock.FlockResourceException;
|
|
10
|
|
|
|
11
|
|
import org.apache.log4j.LogManager;
|
|
12
|
|
import org.apache.log4j.Logger;
|
|
13
|
|
import org.jdom.Document;
|
|
14
|
|
import org.jdom.JDOMException;
|
|
15
|
|
import org.jdom.input.SAXBuilder;
|
|
16
|
|
import org.xml.sax.EntityResolver;
|
|
17
|
|
import org.xml.sax.InputSource;
|
|
18
|
|
|
|
19
|
|
public class DocumentLoader {
|
|
20
|
|
|
|
21
|
|
private final static Logger LOGGER = LogManager.getLogger(DocumentLoader.class);
|
|
22
|
|
|
|
23
|
|
private final SAXBuilder saxBuilder;
|
|
24
|
|
|
|
25
|
0
|
public DocumentLoader() {
|
|
26
|
0
|
this.saxBuilder = new SAXBuilder(false); // no validation
|
|
27
|
0
|
this.saxBuilder.setEntityResolver(new NoOpEntityResolver()); // no DTD loading
|
|
28
|
|
}
|
|
29
|
|
|
|
30
|
0
|
public Document loadDocument(URL location) throws FlockResourceException {
|
|
31
|
0
|
URLLoader loader = null;
|
|
32
|
0
|
InputStream input = null;
|
|
33
|
0
|
try {
|
|
34
|
0
|
loader = new URLLoader(location);
|
|
35
|
0
|
loader.load();
|
|
36
|
0
|
LOGGER.info("URL "+location+" encoding : "+loader.getEncoding());
|
|
37
|
0
|
input = loader.getAsInputStream();
|
|
38
|
|
} catch (IOException io) {
|
|
39
|
0
|
throw new FlockResourceException("Error in getting document at "+location, io);
|
|
40
|
|
}
|
|
41
|
|
|
|
42
|
0
|
Document doc;
|
|
43
|
0
|
try {
|
|
44
|
0
|
doc = this.saxBuilder.build(input);
|
|
45
|
|
} catch (JDOMException e) {
|
|
46
|
0
|
try {
|
|
47
|
0
|
LOGGER.info("Possible missing encoding declaration, try ISO-8859-1 ["+location+"]");
|
|
48
|
0
|
doc = this.saxBuilder.build(loader.getAsReader("ISO-8859-1"));
|
|
49
|
0
|
LOGGER.info("... succeed");
|
|
50
|
|
} catch (UnsupportedEncodingException ue) {
|
|
51
|
0
|
throw new FlockResourceException("Unsupported encoding", ue);
|
|
52
|
|
} catch (JDOMException ex) {
|
|
53
|
0
|
throw new FlockResourceException("Unable to parse document at "+location, ex);
|
|
54
|
|
}
|
|
55
|
|
}
|
|
56
|
0
|
return doc;
|
|
57
|
|
}
|
|
58
|
|
|
|
59
|
|
private static class NoOpEntityResolver implements EntityResolver {
|
|
60
|
|
|
|
61
|
0
|
public InputSource resolveEntity(String publicId, String systemId) {
|
|
62
|
0
|
return new InputSource(new StringReader(""));
|
|
63
|
|
}
|
|
64
|
|
|
|
65
|
|
}
|
|
66
|
|
}
|
|
67
|
|
|
