org.jsoup.parser.Tag Java Examples

@Test public void handlesBaseUri() {
    Tag tag = Tag.valueOf("a");
    Attributes attribs = new Attributes();
    attribs.put("relHref", "/foo");
    attribs.put("absHref", "http://bar/qux");

    Element noBase = new Element(tag, "", attribs);
    assertEquals("", noBase.absUrl("relHref")); // with no base, should NOT fallback to href attrib, whatever it is
    assertEquals("http://bar/qux", noBase.absUrl("absHref")); // no base but valid attrib, return attrib

    Element withBase = new Element(tag, "http://foo/", attribs);
    assertEquals("http://foo/foo", withBase.absUrl("relHref")); // construct abs from base + rel
    assertEquals("http://bar/qux", withBase.absUrl("absHref")); // href is abs, so returns that
    assertEquals("", withBase.absUrl("noval"));

    Element dodgyBase = new Element(tag, "wtf://no-such-protocol/", attribs);
    assertEquals("http://bar/qux", dodgyBase.absUrl("absHref")); // base fails, but href good, so get that
    assertEquals("", dodgyBase.absUrl("relHref")); // base fails, only rel href, so return nothing 
}
 

@Override
public Element appendBody(Elements tempBody) {
    final Element articleBody = new Element(Tag.valueOf("div"), "");
    for (final Element pEl : tempBody) {
        if (pEl.select("div.open-message,div.jp-relatedposts,div.article-social").size() != 0) {
            continue;
        }
        if (pEl.tagName().equals("p")) {
            Element imgEl = pEl.select("img").first();
            if (imgEl != null) {
                String src = imgEl.attr("src");
                if (src.contains("data:image")) {
                    src = imgEl.attr("data-src");
                } else if (!src.contains("www.guxiaobei.com")) {
                    src = "http://www.guxiaobei.com" + src;
                }
                imgEl.attr("src", src);

                articleBody.appendChild(buildFigure(imgEl));
                continue;
            }
        }
        articleBody.appendChild(pEl);
    }
    return articleBody;
}
 

@Override
public Element appendBody(Elements tempBody) {
    final Element articleBody = new Element(Tag.valueOf("div"), "");
    for (Element body : tempBody) {
        if (body.tagName().equals("p")) {
            boolean skipRegister = body.select("p").text().contains("即刻注册SHOPIFY账户, 跟着我们精心准备的SHOPIFY教程开始外贸独立站之旅!");
            boolean skipCopyRight = body.classNames().contains("post-copyright");
            if (skipRegister || skipCopyRight) {
                continue;
            }
        }
        Element imgEl = body.select("img").first();
        if (imgEl != null) {
            articleBody.appendChild(buildFigure(imgEl));
            continue;
        }
        articleBody.appendChild(body);
    }
    return articleBody;
}
 

@Override
public Element appendBody(Elements tempBody) {
    final Element articleBody = new Element(Tag.valueOf("div"), "");
    String blockquote = tempBody.select("div.fetch-read>div.summary").text();
    buildBlockquote(blockquote, articleBody);
    Elements inner = tempBody.select("div.article-inner>*");
    for (Element pEl : inner) {
        if (pEl.select("div.fetch-present").size() != 0) {
            continue;
        }
        Element imgEl = pEl.select("p>img").first();
        if (imgEl != null) {
            Element figure = buildFigure(imgEl);
            if (imgEl.nextElementSibling() != null && imgEl.nextElementSibling().tagName().equals("p")) {
                Element figcaption = buildFigcaption(imgEl.nextElementSibling().text());
                figure.appendChild(figcaption);
                articleBody.appendChild(figure);
                continue;
            }
            articleBody.appendChild(figure);
            continue;
        }
        articleBody.appendChild(pEl);
    }
    return articleBody;
}
 

@Test public void handlesBaseUri() {
    Tag tag = Tag.valueOf("a");
    Attributes attribs = new Attributes();
    attribs.put("relHref", "/foo");
    attribs.put("absHref", "http://bar/qux");

    Element noBase = new Element(tag, "", attribs);
    assertEquals("", noBase.absUrl("relHref")); // with no base, should NOT fallback to href attrib, whatever it is
    assertEquals("http://bar/qux", noBase.absUrl("absHref")); // no base but valid attrib, return attrib

    Element withBase = new Element(tag, "http://foo/", attribs);
    assertEquals("http://foo/foo", withBase.absUrl("relHref")); // construct abs from base + rel
    assertEquals("http://bar/qux", withBase.absUrl("absHref")); // href is abs, so returns that
    assertEquals("", withBase.absUrl("noval"));

    Element dodgyBase = new Element(tag, "wtf://no-such-protocol/", attribs);
    assertEquals("http://bar/qux", dodgyBase.absUrl("absHref")); // base fails, but href good, so get that
    assertEquals("", dodgyBase.absUrl("relHref")); // base fails, only rel href, so return nothing 
}
 

@Test public void handlesBaseUri() {
    Tag tag = Tag.valueOf("a");
    Attributes attribs = new Attributes();
    attribs.put("relHref", "/foo");
    attribs.put("absHref", "http://bar/qux");

    Element noBase = new Element(tag, "", attribs);
    assertEquals("", noBase.absUrl("relHref")); // with no base, should NOT fallback to href attrib, whatever it is
    assertEquals("http://bar/qux", noBase.absUrl("absHref")); // no base but valid attrib, return attrib

    Element withBase = new Element(tag, "http://foo/", attribs);
    assertEquals("http://foo/foo", withBase.absUrl("relHref")); // construct abs from base + rel
    assertEquals("http://bar/qux", withBase.absUrl("absHref")); // href is abs, so returns that
    assertEquals("", withBase.absUrl("noval"));

    Element dodgyBase = new Element(tag, "wtf://no-such-protocol/", attribs);
    assertEquals("http://bar/qux", dodgyBase.absUrl("absHref")); // base fails, but href good, so get that
    assertEquals("", dodgyBase.absUrl("relHref")); // base fails, only rel href, so return nothing 
}
 

/**
 * Test of buildTextFromElement method, of class SimpleTextElementBuilder.
 */
public void testBuildTextFromElementWithChildren() {
    LOGGER.debug("buildTextFromElementWithChildren");
    Element element = new Element(Tag.valueOf("div"), "");
    element.appendText("   text1   ");
    
    Element childElement = new Element(Tag.valueOf("div"), "");
    childElement.text("   child element text   ");

    Element childElement2 = new Element(Tag.valueOf("div"), "");
    childElement2.text("   child element text second level  ");
    childElement.appendChild(childElement2);
    
    element.appendChild(childElement);
    element.appendText("   text2   ");

    SimpleTextElementBuilder instance = new SimpleTextElementBuilder();
    String expResult = "text1 child element text child element text second level text2";
    String result = instance.buildTextFromElement(element);
    assertEquals(expResult, result);
}
 

private ElementMeta createSafeElement(Element sourceEl) {
    String sourceTag = sourceEl.tagName();
    Attributes destAttrs = new Attributes();
    Element dest = new Element(Tag.valueOf(sourceTag), sourceEl.baseUri(), destAttrs);
    int numDiscarded = 0;

    Attributes sourceAttrs = sourceEl.attributes();
    for (Attribute sourceAttr : sourceAttrs) {
        if (whitelist.isSafeAttribute(sourceTag, sourceEl, sourceAttr))
            destAttrs.put(sourceAttr);
        else
            numDiscarded++;
    }
    Attributes enforcedAttrs = whitelist.getEnforcedAttributes(sourceTag);
    destAttrs.addAll(enforcedAttrs);

    return new ElementMeta(dest, numDiscarded);
}
 

private ElementMeta createSafeElement(Element sourceEl) {
    String sourceTag = sourceEl.tagName();
    Attributes destAttrs = new Attributes();
    Element dest = new Element(Tag.valueOf(sourceTag), sourceEl.baseUri(), destAttrs);
    int numDiscarded = 0;

    Attributes sourceAttrs = sourceEl.attributes();
    for (Attribute sourceAttr : sourceAttrs) {
        if (whitelist.isSafeAttribute(sourceTag, sourceEl, sourceAttr))
            destAttrs.put(sourceAttr);
        else
            numDiscarded++;
    }
    Attributes enforcedAttrs = whitelist.getEnforcedAttributes(sourceTag);
    destAttrs.addAll(enforcedAttrs);

    return new ElementMeta(dest, numDiscarded);
}
 

@Test
public void testArticle() throws UIMAException {
  final JCas jCas = JCasSingleton.getJCasInstance();
  final StructuralAnnotations sa = new StructuralAnnotations();

  final Map<String, Class<?>> expectedArticle = new HashMap<>();
  expectedArticle.put("Sheet", Sheet.class);
  expectedArticle.put("Slide", Slide.class);
  expectedArticle.put("Page", Page.class);
  expectedArticle.put("Another", Page.class);

  for (final Map.Entry<String, Class<?>> e : expectedArticle.entrySet()) {
    final Element anchor = new Element(Tag.valueOf("article"), "");
    anchor.attr("class", e.getKey());

    final AnnotationCollector collector = new AnnotationCollector();
    sa.map(jCas, anchor, collector);

    if (e.getValue() != null) {
      assertTrue(e.getValue().isInstance(collector.getAnnotations().get(0)));
    } else {
      assertNull(collector.getAnnotations());
    }
  }
}
 

@Test
public void testMain() throws UIMAException {
  JCas jCas = JCasSingleton.getJCasInstance();
  SemanticHtml sa = new SemanticHtml();

  Map<String, Class<?>> expectedMain = new HashMap<>();
  expectedMain.put("time", Temporal.class);
  expectedMain.put("meter", Quantity.class);
  expectedMain.put("dfn", Buzzword.class);
  expectedMain.put("address", Location.class);
  expectedMain.put("abbr", Buzzword.class);
  expectedMain.put("cite", DocumentReference.class);

  for (Map.Entry<String, Class<?>> e : expectedMain.entrySet()) {
    Element element = new Element(Tag.valueOf(e.getKey()), "");

    AnnotationCollector collector = new AnnotationCollector();
    sa.map(jCas, element, collector);

    if (e.getValue() != null) {
      assertTrue(e.getValue().isInstance(collector.getAnnotations().get(0)));
    } else {
      assertNull(collector.getAnnotations());
    }
  }
}
 

@Test
public void testLink() throws UIMAException {
  final JCas jCas = JCasSingleton.getJCasInstance();
  final StructuralAnnotations sa = new StructuralAnnotations();

  final Element a1 = new Element(Tag.valueOf("a"), "");
  a1.attr("href", "http://example.com");
  final Element a2 = new Element(Tag.valueOf("a"), "");
  a2.attr("href", "/example.com");

  final AnnotationCollector collector = new AnnotationCollector();
  sa.map(jCas, a1, collector);
  sa.map(jCas, a2, collector);

  Annotation link = collector.getAnnotations().get(0);
  assertTrue(link instanceof Link);
  assertEquals("http://example.com", ((Link) link).getTarget());
  Annotation link2 = collector.getAnnotations().get(1);
  assertTrue(link2 instanceof Link);
  assertEquals("/example.com", ((Link) link2).getTarget());
}
 

private Element createStylesheetElement(String url) {
    final Element cssElement;
    if (url != null) {
        cssElement = new Element(Tag.valueOf("link"), "")
                .attr("rel", "stylesheet")
                .attr("type", CSS_TYPE_ATTRIBUTE_VALUE)
                .attr("href", url);
    } else {
        cssElement = new Element(Tag.valueOf("style"), "").attr("type",
                CSS_TYPE_ATTRIBUTE_VALUE);
    }
    return cssElement;
}
 

@Test 
   public void whiteSpaceClassElement(){
    Tag tag = Tag.valueOf("a");
    Attributes attribs = new Attributes();
    Element el = new Element(tag, "", attribs);
    
    attribs.put("class", "abc ");
    boolean hasClass = el.hasClass("ab");
    assertFalse(hasClass);
        
}
 

@Test
public void root() {
	Elements sel = html.select(":root");
	assertEquals(1, sel.size());
	assertNotNull(sel.get(0));
	assertEquals(Tag.valueOf("html"), sel.get(0).tag());

	Elements sel2 = html.select("body").select(":root");
	assertEquals(1, sel2.size());
	assertNotNull(sel2.get(0));
	assertEquals(Tag.valueOf("body"), sel2.get(0).tag());
}
 

@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 after() {
    Document doc = Jsoup.parse("<p>One <b>two</b> three</p>");
    Element newNode = new Element(Tag.valueOf("em"), "");
    newNode.appendText("four");

    doc.select("b").first().after(newNode);
    assertEquals("<p>One <b>two</b><em>four</em> three</p>", doc.body().html());

    doc.select("b").first().after("<i>five</i>");
    assertEquals("<p>One <b>two</b><i>five</i><em>four</em> three</p>", doc.body().html());
}
 

@Test public void after() {
    Document doc = Jsoup.parse("<p>One <b>two</b> three</p>");
    Element newNode = new Element(Tag.valueOf("em"), "");
    newNode.appendText("four");

    doc.select("b").first().after(newNode);
    assertEquals("<p>One <b>two</b><em>four</em> three</p>", doc.body().html());

    doc.select("b").first().after("<i>five</i>");
    assertEquals("<p>One <b>two</b><i>five</i><em>four</em> three</p>", doc.body().html());
}
 

/**
 * Replace link tags with style tags in order to keep the same inclusion
 * order
 *
 * @param doc
 *            the html document
 * @param cssContents
 *            the list of external css files with their content
 */
private static void internStyles(Document doc, List<ExternalCss> cssContents) {
	Elements els = doc.select(CSS_LINKS_SELECTOR);
	for (Element e : els) {
		if (isInlineModeAllowed(e, InlineModes.STYLE_ATTR)) {
			String path = e.attr(HREF_ATTR);
			ExternalCss css = getCss(cssContents, path);
			if (css != null) {
				Element style = new Element(Tag.valueOf(STYLE_TAG), "");
				style.appendChild(new DataNode(getCssContent(css)));
				e.replaceWith(style);
			}
		}
	}
}
 

/**
 * Finds elements, including and recursively under this element, with the specified tag name.
 * @param tagName The tag name to search for (case insensitively).
 * @return a matching unmodifiable list of elements. Will be empty if this element and none of its children match.
 */
public Elements getElementsByTag(String tagName) {
    Validate.notEmpty(tagName);
    tagName = tagName.toLowerCase().trim();

    return Collector.collect(new Evaluator.Tag(tagName), this);
}
 

private static Element createElement(String tag, String content,
        String... attrs) {
    Element elm = new Element(Tag.valueOf(tag), "");
    if (content != null && !content.isEmpty()) {
        elm.appendChild(new DataNode(content));
    }
    for (int i = 0; i < attrs.length - 1; i += 2) {
        elm.attr(attrs[i], attrs[i + 1]);
    }
    return elm;
}
 

private static List<LinkContainer> work(Elements elements) {
    List<LinkContainer> result = new ArrayList<>(elements.size());
    String currentCategory = null;
    for(Element element: elements) {
        Tag tag = element.tag();
        if(isHeader(tag)) {
            currentCategory = element.text();
            System.out.println(currentCategory);
        }
        work(element.children());
    }
    return result;
}
 

@Test public void orphanNodeReturnsNullForSiblingElements() {
    Node node = new Element(Tag.valueOf("p"), "");
    Element el = new Element(Tag.valueOf("p"), "");

    assertEquals(0, node.siblingIndex());
    assertEquals(0, node.siblingNodes().size());

    assertNull(node.previousSibling());
    assertNull(node.nextSibling());

    assertEquals(0, el.siblingElements().size());
    assertNull(el.previousElementSibling());
    assertNull(el.nextElementSibling());
}
 

@Test public void orphanNodeReturnsNullForSiblingElements() {
    Node node = new Element(Tag.valueOf("p"), "");
    Element el = new Element(Tag.valueOf("p"), "");

    assertEquals(0, node.siblingIndex());
    assertEquals(0, node.siblingNodes().size());

    assertNull(node.previousSibling());
    assertNull(node.nextSibling());

    assertEquals(0, el.siblingElements().size());
    assertNull(el.previousElementSibling());
    assertNull(el.nextElementSibling());
}
 

@Test public void before() {
    Document doc = Jsoup.parse("<p>One <b>two</b> three</p>");
    Element newNode = new Element(Tag.valueOf("em"), "");
    newNode.appendText("four");

    doc.select("b").first().before(newNode);
    assertEquals("<p>One <em>four</em><b>two</b> three</p>", doc.body().html());

    doc.select("b").first().before("<i>five</i>");
    assertEquals("<p>One <em>four</em><i>five</i><b>two</b> three</p>", doc.body().html());
}
 

/**
 * Finds elements, including and recursively under this element, with the specified tag name.
 * @param tagName The tag name to search for (case insensitively).
 * @return a matching unmodifiable list of elements. Will be empty if this element and none of its children match.
 */
public Elements getElementsByTag(String tagName) {
    Validate.notEmpty(tagName);
    tagName = normalize(tagName);

    return Collector.collect(new Evaluator.Tag(tagName), this);
}
 

/**
 * Finds elements, including and recursively under this element, with the specified tag name.
 * @param tagName The tag name to search for (case insensitively).
 * @return a matching unmodifiable list of elements. Will be empty if this element and none of its children match.
 */
public Elements getElementsByTag(String tagName) {
    Validate.notEmpty(tagName);
    tagName = normalize(tagName);

    return Collector.collect(new Evaluator.Tag(tagName), this);
}
 

@Test
public void root() {
	Elements sel = html.select(":root");
	assertEquals(1, sel.size());
	assertNotNull(sel.get(0));
	assertEquals(Tag.valueOf("html"), sel.get(0).tag());

	Elements sel2 = html.select("body").select(":root");
	assertEquals(1, sel2.size());
	assertNotNull(sel2.get(0));
	assertEquals(Tag.valueOf("body"), sel2.get(0).tag());
}
 

@Test public void before() {
    Document doc = Jsoup.parse("<p>One <b>two</b> three</p>");
    Element newNode = new Element(Tag.valueOf("em"), "");
    newNode.appendText("four");

    doc.select("b").first().before(newNode);
    assertEquals("<p>One <em>four</em><b>two</b> three</p>", doc.body().html());

    doc.select("b").first().before("<i>five</i>");
    assertEquals("<p>One <em>four</em><i>five</i><b>two</b> three</p>", doc.body().html());
}
 

private static Element createJavaScriptElement(String sourceUrl,
        boolean defer, String type) {
    Element jsElement = new Element(Tag.valueOf(SCRIPT_TAG), "")
            .attr("type", type).attr(DEFER_ATTRIBUTE, defer);
    if (sourceUrl != null) {
        jsElement = jsElement.attr("src", sourceUrl);
    }
    return jsElement;
}