|
1
2
3 package net.nutch.analysis.lang;
4
5 import java.util.Properties  ;
6
7 import junit.framework.TestCase;
8 import net.nutch.parse.Parse;
9 import net.nutch.parse.Parser;
10 import net.nutch.parse.ParserFactory;
11 import net.nutch.protocol.Content;
12
13 public class TestHTMLLanguageParser extends TestCase {
14
15 private static String URL = "http://foo.bar/";
16
17 private static String BASE = "http://foo.bar/";
18
19 String docs[] = {
20 "<html lang=\"fi\"><head>document 1 title</head><body>jotain suomeksi</body></html>",
21 "<html><head><meta http-equiv=\"content-language\" content=\"en\"><title>document 2 title</head><body>this is english</body></html>",
22 "<html><head><meta name=\"dc.language\" content=\"en\"><title>document 3 title</head><body>this is english</body></html>" };
23
24 String metalanguages[] = { "fi", "en", "en" };
25
26
29 public void testMetaHTMLParsing() {
30
31 try {
32
33
34 for (int t = 0; t < docs.length; t++) {
35
36 Content content = getContent(docs[t]);
37 Parser parser = ParserFactory.getParser("text/html", URL);
38 Parse parse = parser.getParse(content);
39
40 assertEquals(metalanguages[t], (String ) parse.getData().get(
41 HTMLLanguageParser.META_LANG_NAME));
42
43 }
44 } catch (Exception e) {
45 e.printStackTrace(System.out);
46 fail(e.toString());
47 }
48
49 }
50
51 private Content getContent(String text) {
52 Properties p = new Properties ();
53 p.put("Content-Type", "text/html");
54
55 Content content = new Content(URL, BASE, text.getBytes(), "text/html", p);
56 return content;
57 }
58 } |
Popular Tags |
Java API By Example, From Geeks To Geeks.