@Test
public void testGetOutlinks() throws Exception {
if (testDOMs[0] == null) setup();
for (int i = 0; i < testPages.length; i++) {
ArrayList<Outlink> outlinks = new ArrayList<Outlink>();
if (i == SKIP) {
conf.setBoolean("parser.html.form.use_action", false);
utils.setConf(conf);
} else {
conf.setBoolean("parser.html.form.use_action", true);
utils.setConf(conf);
}
utils.getOutlinks(testBaseHrefURLs[i], outlinks, testDOMs[i]);
Outlink[] outlinkArr = new Outlink[outlinks.size()];
outlinkArr = outlinks.toArray(outlinkArr);
compareOutlinks(i, answerOutlinks[i], outlinkArr);
}
}
// won't work with Tika - the title is stored in the metadata but
// not put in the XHTML representation
@Test
public void testGetTitle() throws Exception {
if (testDOMs[0] == null) setup();
for (int i = 0; i < testPages.length; i++) {
StringBuffer sb = new StringBuffer();
utils.getTitle(sb, testDOMs[i]);
String title = sb.toString();
assertTrue(
"example "
+ i
+ " : expecting title: "
+ answerTitle[i]
+ System.getProperty("line.separator")
+ System.getProperty("line.separator")
+ "got title: "
+ title,
equalsIgnoreWhitespace(answerTitle[i], title));
}
}
