KickJava   Java API By Example, From Geeks To Geeks.

Java > Open Source Codes > net > nutch > parse > rtf > TestRTFParser


1 package net.nutch.parse.rtf;
2
3 import junit.framework.TestCase;
4 import net.nutch.parse.Parse;
5 import net.nutch.parse.ParseException;
6 import net.nutch.parse.Parser;
7 import net.nutch.parse.ParserFactory;
8 import net.nutch.protocol.Content;
9 import net.nutch.protocol.Protocol;
10 import net.nutch.protocol.ProtocolException;
11 import net.nutch.protocol.ProtocolFactory;
12
13 import java.util.Properties JavaDoc;
14
15 /**
16  * Unit tests for TestRTFParser. (Adapted from John Xing msword unit tests).
17  *
18  * @author Andy Hedges
19  */
20 public class TestRTFParser extends TestCase {
21
22   private String JavaDoc fileSeparator = System.getProperty("file.separator");
23   // This system property is defined in ./src/plugin/build-plugin.xml
24 private String JavaDoc sampleDir = System.getProperty("test.data", ".");
25   // Make sure sample files are copied to "test.data" as specified in
26 // ./src/plugin/parse-rtf/build.xml during plugin compilation.
27 // Check ./src/plugin/parse-rtf/sample/README.txt for what they are.
28 private String JavaDoc rtfFile = "test.rtf";
29
30   public TestRTFParser(String JavaDoc name) {
31     super(name);
32   }
33
34   protected void setUp() {
35   }
36
37   protected void tearDown() {
38   }
39
40   public void testIt() throws ProtocolException, ParseException {
41
42     String JavaDoc urlString;
43     Protocol protocol;
44     Content content;
45     Parser parser;
46     Parse parse;
47
48     urlString = "file:" + sampleDir + fileSeparator + rtfFile;
49     protocol = ProtocolFactory.getProtocol(urlString);
50     content = protocol.getContent(urlString);
51
52     parser = ParserFactory.getParser(content.getContentType(), urlString);
53     parse = parser.getParse(content);
54     String JavaDoc text = parse.getText();
55     assertEquals("The quick brown fox jumps over the lazy dog", text.trim());
56
57     String JavaDoc title = parse.getData().getTitle();
58     Properties JavaDoc meta = parse.getData().getMetadata();
59     assertEquals("test rft document", title);
60     assertEquals("tests", meta.getProperty("subject"));
61
62
63
64   }
65
66
67 }
68
Popular Tags