@Test public void testTwoHtml() throws IOException { File file = new File(testDataRoot, "two.html"); String htmlContent = FileUtil.readString(file); Document document = new LagartoDOMBuilder().parse(htmlContent); Node html = new NodeSelector(document).select("html").get(0); assertNotNull(html); Node body = new NodeSelector(html).selectFirst("body"); Element h1 = body.getFirstChildElement(); assertEquals("h1", h1.getNodeName()); Node comment1 = body.getFirstChild().getNextSibling(); assertEquals(Node.NodeType.COMMENT, comment1.getNodeType()); Element p = (Element) new NodeSelector(body).selectFirst("p"); assertEquals(h1, p.getPreviousSiblingElement()); assertEquals(h1, comment1.getNextSiblingElement()); assertNull(comment1.getNextSiblingName()); // check if filter works just for sub elements List<Node> p_ems = new NodeSelector(p).select("em"); assertEquals(1, p_ems.size()); Element script = (Element) new NodeSelector(html).selectFirst("script"); assertEquals("text/javascript", script.getAttribute("type")); assertTrue(document.check()); }
@Test public void testGroupOfSelectors() throws IOException { File file = new File(testDataRoot, "one.html"); String htmlContent = FileUtil.readString(file); Document document = new LagartoDOMBuilder().parse(htmlContent); List<Node> nodes = new NodeSelector(document).select("em, b, b"); assertEquals(9, nodes.size()); assertTrue(document.check()); }