@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 isIdSetReturnsTrue() {
Document model = new Document();
model.setId(ValueGenerator.getUniqueString(32));
assertNotNull(model.getId());
assertTrue(model.isIdSet());
}
@Test
public void testAddNewText() {
Document doc = Jsoup.parse("<div id=1><p>Hello</p></div>");
Element div = doc.getElementById("1");
div.appendText(" there & now >");
assertEquals("<p>Hello</p> there & now >", TextUtil.stripNewlines(div.html()));
}
@Test
public void testNormalisesText() {
String h = "<p>Hello<p>There.</p> \n <p>Here <b>is</b> \n s<b>om</b>e text.";
Document doc = Jsoup.parse(h);
String text = doc.text();
assertEquals("Hello There. Here is some text.", text);
}
@Test
public void insertChildrenArgumentValidation() {
Document doc = Jsoup.parse("<div id=1>Text <p>One</p> Text <p>Two</p></div><div id=2></div>");
Element div1 = doc.select("div").get(0);
Element div2 = doc.select("div").get(1);
List<Node> children = div1.childNodes();
try {
div2.insertChildren(6, children);
fail();
} catch (IllegalArgumentException e) {
}
try {
div2.insertChildren(-5, children);
fail();
} catch (IllegalArgumentException e) {
}
try {
div2.insertChildren(0, null);
fail();
} catch (IllegalArgumentException e) {
}
}
@Test
public void testGetChildText() {
Document doc = Jsoup.parse("<p>Hello <b>there</b> now");
Element p = doc.select("p").first();
assertEquals("Hello there now", p.text());
assertEquals("Hello now", p.ownText());
}
@Test
public void testClassNames() {
Document doc = Jsoup.parse("<div class=\"c1 c2\">C</div>");
Element div = doc.select("div").get(0);
assertEquals("c1 c2", div.className());
final Set<String> set1 = div.classNames();
final Object[] arr1 = set1.toArray();
assertTrue(arr1.length == 2);
assertEquals("c1", arr1[0]);
assertEquals("c2", arr1[1]);
// Changes to the set should not be reflected in the Elements getters
set1.add("c3");
assertTrue(2 == div.classNames().size());
assertEquals("c1 c2", div.className());
// Update the class names to a fresh set
final Set<String> newSet = new LinkedHashSet<String>(3);
newSet.addAll(set1);
newSet.add("c3");
div.classNames(newSet);
assertEquals("c1 c2 c3", div.className());
final Set<String> set2 = div.classNames();
final Object[] arr2 = set2.toArray();
assertTrue(arr2.length == 3);
assertEquals("c1", arr2[0]);
assertEquals("c2", arr2[1]);
assertEquals("c3", arr2[2]);
}
@Test
public void testKeepsPreTextInCode() {
String h = "<pre><code>code\n\ncode</code></pre>";
Document doc = Jsoup.parse(h);
assertEquals("code\n\ncode", doc.text());
assertEquals("<pre><code>code\n\ncode</code></pre>", doc.body().html());
}
@Test
public void testSetHtml() {
Document doc = Jsoup.parse("<div id=1><p>Hello</p></div>");
Element div = doc.getElementById("1");
div.html("<p>there</p><p>now</p>");
assertEquals("<p>there</p><p>now</p>", TextUtil.stripNewlines(div.html()));
}
@Test
public void testClonesClassnames() {
Document doc = Jsoup.parse("<div class='one two'></div>");
Element div = doc.select("div").first();
Set<String> classes = div.classNames();
assertEquals(2, classes.size());
assertTrue(classes.contains("one"));
assertTrue(classes.contains("two"));
Element copy = div.clone();
Set<String> copyClasses = copy.classNames();
assertEquals(2, copyClasses.size());
assertTrue(copyClasses.contains("one"));
assertTrue(copyClasses.contains("two"));
copyClasses.add("three");
copyClasses.remove("one");
assertTrue(classes.contains("one"));
assertFalse(classes.contains("three"));
assertFalse(copyClasses.contains("one"));
assertTrue(copyClasses.contains("three"));
assertEquals("", div.html());
assertEquals("", copy.html());
}
@Test
public void testHtmlContainsOuter() {
Document doc = Jsoup.parse("<title>Check</title> <div>Hello there</div>");
doc.outputSettings().indentAmount(0);
assertTrue(doc.html().contains(doc.select("title").outerHtml()));
assertTrue(doc.html().contains(doc.select("div").outerHtml()));
}
@Test
public void testClone() {
Document doc = Jsoup.parse("<div><p>One<p><span>Two</div>");
Element p = doc.select("p").get(1);
Element clone = p.clone();
assertNull(clone.parent()); // should be orphaned
assertEquals(0, clone.siblingIndex);
assertEquals(1, p.siblingIndex);
assertNotNull(p.parent());
clone.append("<span>Three");
assertEquals(
"<p><span>Two</span><span>Three</span></p>", TextUtil.stripNewlines(clone.outerHtml()));
assertEquals(
"<div><p>One</p><p><span>Two</span></p></div>",
TextUtil.stripNewlines(doc.body().html())); // not modified
doc.body().appendChild(clone); // adopt
assertNotNull(clone.parent());
assertEquals(
"<div><p>One</p><p><span>Two</span></p></div><p><span>Two</span><span>Three</span></p>",
TextUtil.stripNewlines(doc.body().html()));
}
@Test
public void dataset() {
Document doc =
Jsoup.parse(
"<div id=1 data-name=jsoup class=new data-package=jar>Hello</div><p id=2>Hello</p>");
Element div = doc.select("div").first();
Map<String, String> dataset = div.dataset();
Attributes attributes = div.attributes();
// size, get, set, add, remove
assertEquals(2, dataset.size());
assertEquals("jsoup", dataset.get("name"));
assertEquals("jar", dataset.get("package"));
dataset.put("name", "jsoup updated");
dataset.put("language", "java");
dataset.remove("package");
assertEquals(2, dataset.size());
assertEquals(4, attributes.size());
assertEquals("jsoup updated", attributes.get("data-name"));
assertEquals("jsoup updated", dataset.get("name"));
assertEquals("java", attributes.get("data-language"));
assertEquals("java", dataset.get("language"));
attributes.put("data-food", "bacon");
assertEquals(3, dataset.size());
assertEquals("bacon", dataset.get("food"));
attributes.put("data-", "empty");
assertEquals(null, dataset.get("")); // data- is not a data attribute
Element p = doc.select("p").first();
assertEquals(0, p.dataset().size());
}
@Test
public void insertChildrenAtPosition() {
Document doc =
Jsoup.parse(
"<div id=1>Text1 <p>One</p> Text2 <p>Two</p></div><div id=2>Text3 <p>Three</p></div>");
Element div1 = doc.select("div").get(0);
Elements p1s = div1.select("p");
Element div2 = doc.select("div").get(1);
assertEquals(2, div2.childNodeSize());
div2.insertChildren(-1, p1s);
assertEquals(2, div1.childNodeSize()); // moved two out
assertEquals(4, div2.childNodeSize());
assertEquals(3, p1s.get(1).siblingIndex()); // should be last
List<Node> els = new ArrayList<Node>();
Element el1 = new Element(Tag.valueOf("span"), "").text("Span1");
Element el2 = new Element(Tag.valueOf("span"), "").text("Span2");
TextNode tn1 = new TextNode("Text4", "");
els.add(el1);
els.add(el2);
els.add(tn1);
assertNull(el1.parent());
div2.insertChildren(-2, els);
assertEquals(div2, el1.parent());
assertEquals(7, div2.childNodeSize());
assertEquals(3, el1.siblingIndex());
assertEquals(4, el2.siblingIndex());
assertEquals(5, tn1.siblingIndex());
}
@Test
public void testOuterHtml() {
Document doc =
Jsoup.parse("<div title='Tags &c.'><img src=foo.png><p><!-- comment -->Hello<p>there");
assertEquals(
"<html><head></head><body><div title=\"Tags &c.\"><img src=\"foo.png\"><p><!-- comment -->Hello</p><p>there</p></div></body></html>",
TextUtil.stripNewlines(doc.outerHtml()));
}
@Test
public void testTagNameSet() {
Document doc = Jsoup.parse("<div><i>Hello</i>");
doc.select("i").first().tagName("em");
assertEquals(0, doc.select("i").size());
assertEquals(1, doc.select("em").size());
assertEquals("<em>Hello</em>", doc.select("div").first().html());
}
@Test
public void testElementSiblingIndexSameContent() {
Document doc = Jsoup.parse("<div><p>One</p>...<p>One</p>...<p>One</p>");
Elements ps = doc.select("p");
assertTrue(0 == ps.get(0).elementSiblingIndex());
assertTrue(1 == ps.get(1).elementSiblingIndex());
assertTrue(2 == ps.get(2).elementSiblingIndex());
}
@Test
public void testPrependElement() {
Document doc = Jsoup.parse("<div id=1><p>Hello</p></div>");
Element div = doc.getElementById("1");
div.prependElement("p").text("Before");
assertEquals("Before", div.child(0).text());
assertEquals("Hello", div.child(1).text());
}
@Test
public void getNamespacedElementsByTag() {
Document doc = Jsoup.parse("<div><abc:def id=1>Hello</abc:def></div>");
Elements els = doc.getElementsByTag("abc:def");
assertEquals(1, els.size());
assertEquals("1", els.first().id());
assertEquals("abc:def", els.first().tagName());
}
@Test
public void testSetIndent() {
Document doc = Jsoup.parse("<div><p>Hello\nthere</p></div>");
doc.outputSettings().indentAmount(0);
assertEquals(
"<html>\n<head></head>\n<body>\n<div>\n<p>Hello there</p>\n</div>\n</body>\n</html>",
doc.html());
}
@Test
public void parentlessToString() {
Document doc = Jsoup.parse("<img src='foo'>");
Element img = doc.select("img").first();
assertEquals("<img src=\"foo\">", img.toString());
img.remove(); // lost its parent
assertEquals("<img src=\"foo\">", img.toString());
}
@Test
public void testBrHasSpace() {
Document doc = Jsoup.parse("<p>Hello<br>there</p>");
assertEquals("Hello there", doc.text());
assertEquals("Hello there", doc.select("p").first().ownText());
doc = Jsoup.parse("<p>Hello <br> there</p>");
assertEquals("Hello there", doc.text());
}
@Test
public void testFormatHtml() {
Document doc =
Jsoup.parse(
"<title>Format test</title><div><p>Hello <span>jsoup <span>users</span></span></p><p>Good.</p></div>");
assertEquals(
"<html>\n <head>\n <title>Format test</title>\n </head>\n <body>\n <div>\n <p>Hello <span>jsoup <span>users</span></span></p>\n <p>Good.</p>\n </div>\n </body>\n</html>",
doc.html());
}
@Test
public void testWrapWithRemainder() {
Document doc = Jsoup.parse("<div><p>Hello</p></div>");
Element p = doc.select("p").first();
p.wrap("<div class='head'></div><p>There!</p>");
assertEquals(
"<div><div class=\"head\"><p>Hello</p><p>There!</p></div></div>",
TextUtil.stripNewlines(doc.body().html()));
}
@Test
public void testGetElementsWithAttributeDash() {
Document doc =
Jsoup.parse(
"<meta http-equiv=content-type value=utf8 id=1> <meta name=foo content=bar id=2> <div http-equiv=content-type value=utf8 id=3>");
Elements meta = doc.select("meta[http-equiv=content-type], meta[charset]");
assertEquals(1, meta.size());
assertEquals("1", meta.first().id());
}
@Test
public void testNamespacedElements() {
// Namespaces with ns:tag in HTML must be translated to ns|tag in CSS.
String html = "<html><body><fb:comments /></body></html>";
Document doc = Jsoup.parse(html, "http://example.com/bar/");
Elements els = doc.select("fb|comments");
assertEquals(1, els.size());
assertEquals("html > body > fb|comments", els.get(0).cssSelector());
}
@Test
public void testHasText() {
Document doc = Jsoup.parse("<div><p>Hello</p><p></p></div>");
Element div = doc.select("div").first();
Elements ps = doc.select("p");
assertTrue(div.hasText());
assertTrue(ps.first().hasText());
assertFalse(ps.last().hasText());
}
@Test
public void testFormatOutline() {
Document doc =
Jsoup.parse(
"<title>Format test</title><div><p>Hello <span>jsoup <span>users</span></span></p><p>Good.</p></div>");
doc.outputSettings().outline(true);
assertEquals(
"<html>\n <head>\n <title>Format test</title>\n </head>\n <body>\n <div>\n <p>\n Hello \n <span>\n jsoup \n <span>users</span>\n </span>\n </p>\n <p>Good.</p>\n </div>\n </body>\n</html>",
doc.html());
}
@Test
public void testGetElementsWithAttributeValue() {
Document doc = Jsoup.parse("<div style='bold'><p><p><b style></b></p></div>");
List<Element> els = doc.getElementsByAttributeValue("style", "bold");
assertEquals(1, els.size());
assertEquals("div", els.get(0).tagName());
List<Element> none = doc.getElementsByAttributeValue("style", "none");
assertEquals(0, none.size());
}
@Test
public void testAppendRowToTable() {
Document doc = Jsoup.parse("<table><tr><td>1</td></tr></table>");
Element table = doc.select("tbody").first();
table.append("<tr><td>2</td></tr>");
assertEquals(
"<table><tbody><tr><td>1</td></tr><tr><td>2</td></tr></tbody></table>",
TextUtil.stripNewlines(doc.body().html()));
}