@Test
public void testPeterFull() {
String content =
"<DIV class=\"section\" id=\"forest-elephants\" >\n"
+ "<H1>Forest elephants</H1>\n"
+ "<P>In this section, we discuss the lesser known forest elephants.\n"
+ "...this section continues...\n"
+ "<DIV class=\"subsection\" id=\"forest-habitat\" >\n"
+ "<H2>Habitat</H2>\n"
+ "<P>Forest elephants do not live in trees but among them.\n"
+ "...this subsection continues...\n"
+ "</DIV>\n"
+ "</DIV>";
LagartoDOMBuilder lagartoDOMBuilder = new LagartoDOMBuilder();
Document doc = lagartoDOMBuilder.parse(content);
String expected =
"<div class=\"section\" id=\"forest-elephants\">\n"
+ "<h1>Forest elephants</h1>\n"
+ "<p>In this section, we discuss the lesser known forest elephants.\n"
+ "...this section continues...\n</p>"
+ "<div class=\"subsection\" id=\"forest-habitat\">\n"
+ "<h2>Habitat</h2>\n"
+ "<p>Forest elephants do not live in trees but among them.\n"
+ "...this subsection continues...\n</p>"
+ "</div>\n"
+ "</div>";
assertEquals(expected, doc.getHtml());
assertTrue(doc.check());
}
@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 testPeterSimple1() {
String content = "<div><h1>FORELE</h1><p>dicuss<div>xxx</div></div>";
LagartoDOMBuilder lagartoDOMBuilder = new LagartoDOMBuilder();
Document doc = lagartoDOMBuilder.parse(content);
assertEquals("<div><h1>FORELE</h1><p>dicuss</p><div>xxx</div></div>", doc.getHtml());
assertTrue(doc.check());
}
@Test
public void testTwoNodes2() {
String content = "<body><div>test<span><form>xxx</body>";
LagartoDOMBuilder lagartoDOMBuilder = new LagartoDOMBuilder();
Document doc = lagartoDOMBuilder.parse(content);
assertEquals("<body><div>test<span><form>xxx</form></span></div></body>", doc.getHtml());
assertTrue(doc.check());
}
@Test
public void testOneNodeWithBlanks() {
String content = "<body><div> <span>sss</span></body>";
LagartoDOMBuilder lagartoDOMBuilder = new LagartoDOMBuilder();
Document doc = lagartoDOMBuilder.parse(content);
assertEquals("<body><div> <span>sss</span></div></body>", doc.getHtml());
assertTrue(doc.check());
}
@Test
public void testEof2() {
String content = "<body><div>";
LagartoDOMBuilder lagartoDOMBuilder = new LagartoDOMBuilder();
Document doc = lagartoDOMBuilder.parse(content);
assertEquals("<body><div></div></body>", doc.getHtml());
assertTrue(doc.check());
}
@Test
public void testPeterSimple3WithSpaces() {
String content =
"<div> <h1>FORELE</h1> <p>dicuss <div> <h2>HAB</h2> <p>AMONG </div> </div>".toUpperCase();
LagartoDOMBuilder lagartoDOMBuilder = new LagartoDOMBuilder();
Document doc = lagartoDOMBuilder.parse(content);
assertEquals(
"<div> <h1>FORELE</h1> <p>DICUSS </p><div> <h2>HAB</h2> <p>AMONG </p></div> </div>",
doc.getHtml());
assertTrue(doc.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());
}