Java Code Examples for org.jsoup.select.Elements#add()

/**
 * 
 * @param sspHandler
 * @param elementHandler 
 * @param elementHandlerWithoutDataTableMarkup
 */
private void extractTableWithDataTableMarkup(
            ElementHandler<Element> elementHandler, 
            ElementHandler elementHandlerWithoutDataTableMarkup) {
    
    Elements elementsWithMarkup = new Elements();
    
    for (Element el : elementHandler.get()) {
        if (el.select(DATA_TABLE_MARKUP_CSS_LIKE_QUERY).size() > 0) {
            elementsWithMarkup.add(el);
        } else if (elementHandlerWithoutDataTableMarkup != null) {
            elementHandlerWithoutDataTableMarkup.add(el);
        }
    }
    elementHandler.clean().addAll(elementsWithMarkup);
}
 

/**
 * 
 * @param sspHandler
 * @param elementHandler 
 * @param elementHandlerWithoutDataTableMarkup
 */
private void extractTableWithDataTableMarkup(
            ElementHandler<Element> elementHandler, 
            ElementHandler<Element> elementHandlerWithoutDataTableMarkup) {
    
    Elements elementsWithMarkup = new Elements();
    
    for (Element el : elementHandler.get()) {
        if (el.select(DATA_TABLE_MARKUP_CSS_LIKE_QUERY).size() > 0) {
            elementsWithMarkup.add(el);
        } else if (elementHandlerWithoutDataTableMarkup != null) {
            elementHandlerWithoutDataTableMarkup.add(el);
        }
    }
    elementHandler.clean().addAll(elementsWithMarkup);
}
 

/**
 * 
 * @param sspHandler
 * @param elementHandler 
 * @param elementHandlerWithoutDataTableMarkup
 */
private void extractTableWithDataTableMarkup(
            ElementHandler<Element> elementHandler, 
            ElementHandler<Element> elementHandlerWithoutDataTableMarkup) {
    
    Elements elementsWithMarkup = new Elements();
    
    for (Element el : elementHandler.get()) {
        if (el.select(DATA_TABLE_MARKUP_CSS_LIKE_QUERY).size() > 0) {
            elementsWithMarkup.add(el);
        } else if (elementHandlerWithoutDataTableMarkup != null) {
            elementHandlerWithoutDataTableMarkup.add(el);
        }
    }
    elementHandler.clean().addAll(elementsWithMarkup);
}
 

private Elements filterElements(Elements elements, String[] rules) {
    if (rules == null || rules.length < 2) return elements;
    final Elements selectedEls = new Elements();
    for (Element ele : elements) {
        boolean isOk = false;
        switch (rules[0]) {
            case "class":
                isOk = ele.getElementsByClass(rules[1]).size() > 0;
                break;
            case "id":
                isOk = ele.getElementById(rules[1]) != null;
                break;
            case "tag":
                isOk = ele.getElementsByTag(rules[1]).size() > 0;
                break;
            case "text":
                isOk = ele.getElementsContainingOwnText(rules[1]).size() > 0;
                break;
        }
        if (isOk) {
            selectedEls.add(ele);
        }
    }
    return selectedEls;
}
 

private Elements filterElements(Elements elements, String[] rules) {
    if (rules == null || rules.length < 2) return elements;
    Elements selectedEls = new Elements();
    for (Element ele : elements) {
        boolean isOk = false;
        switch (rules[0]) {
            case "class":
                isOk = ele.getElementsByClass(rules[1]).size() > 0;
                break;
            case "id":
                isOk = ele.getElementById(rules[1]) != null;
                break;
            case "tag":
                isOk = ele.getElementsByTag(rules[1]).size() > 0;
                break;
            case "text":
                isOk = ele.getElementsContainingOwnText(rules[1]).size() > 0;
                break;
        }
        if (isOk) {
            selectedEls.add(ele);
        }
    }
    return selectedEls;
}
 

/**
 * 获取同胞中同名元素的数量
 * @param e
 * @return
 */
public static int sameTagElNums(Element e,Scope scope){
    Elements context = new Elements();
    Elements els = e.parent().getElementsByTag(e.tagName());
    for (Element el:els){
        if (scope.context().contains(el)){
            context.add(el);
        }
    }
    return context.size();
}
 

private static Elements getImagesToInline(Document doc, ImageResource image) {
	Elements imgs = doc.select(MessageFormat.format(IMG_SELECTOR, image.getSrcUrl()));
	Elements found = new Elements();
	for (Element img : imgs) {
		// only apply inlining if mode matches
		if (isInlineModeAllowed(img, BASE64)) {
			found.add(img);
		}
	}
	return found;
}
 

private static Elements getImagesToAttach(Document doc, ImageResource image) {
	Elements imgs = doc.select(format(IMG_SELECTOR, image.getSrcUrl()));
	Elements found = new Elements();
	for (Element img : imgs) {
		// skip images that have skip-attach attribute
		if (isInlineModeAllowed(img, ATTACH)) {
			found.add(img);
		}
	}
	return found;
}
 

/**
 * Get sibling elements. If the element has no sibling elements, returns an empty list. An element is not a sibling
 * of itself, so will not be included in the returned list.
 * @return sibling elements
 */
public Elements siblingElements() {
    if (parentNode == null)
        return new Elements(0);

    List<Element> elements = parent().childElementsList();
    Elements siblings = new Elements(elements.size() - 1);
    for (Element el: elements)
        if (el != this)
            siblings.add(el);
    return siblings;
}
 

/**
 * Get sibling elements. If the element has no sibling elements, returns an empty list. An element is not a sibling
 * of itself, so will not be included in the returned list.
 * @return sibling elements
 */
public Elements siblingElements() {
    if (parentNode == null)
        return new Elements(0);

    List<Element> elements = parent().childElementsList();
    Elements siblings = new Elements(elements.size() - 1);
    for (Element el: elements)
        if (el != this)
            siblings.add(el);
    return siblings;
}
 

@Test
public void testShort() throws Exception {
    Elements context = new Elements();
    Element el = new Element("V");
    el.appendText("test .69");
    context.add(el);
    Num n = new Num();
    XValue v = n.call(Scope.create(context));
    logger.info("v = {}",v);
    Assert.assertEquals(0.69,v.asDouble(),0.00000000000001);
}
 

public String text(Element element) {
    // no patterns at all - return the text from the whole document
    if (inclusionPatterns.size() == 0 && excludedTags.size() == 0) {
        return _text(element);
    }

    Elements matches = new Elements();

    for (String pattern : inclusionPatterns) {
        matches = element.select(pattern);
        if (!matches.isEmpty())
            break;
    }

    // if nothing matches or no patterns were defined use the whole doc
    if (matches.isEmpty()) {
        matches.add(element);
    }

    final StringBuilder accum = new StringBuilder();

    for (Element node : matches) {
        accum.append(_text(node)).append("\n");
    }

    return accum.toString().trim();
}
 

/**
 * 返回下一个同胞节点(扩展) 语法 following-sibling-one
 *
 * @param e
 * @return
 */
public Elements followingSiblingOne(Element e) {
    Elements rs = new Elements();
    if (e.nextElementSibling() != null) {
        rs.add(e.nextElementSibling());
    }
    return rs;
}
 

public static Elements followingSibling(Element el){
    Elements rs = new Elements();
    Element tmp = el.nextElementSibling();
    while (tmp!=null){
        rs.add(tmp);
        tmp = tmp.nextElementSibling();
    }
    if (rs.size() > 0){
        return rs;
    }
    return null;
}
 

/**
 * 返回前一个同胞节点（扩展），语法 preceding-sibling-one
 *
 * @param e
 * @return
 */
public Elements precedingSiblingOne(Element e) {
    Elements rs = new Elements();
    if (e.previousElementSibling() != null) {
        rs.add(e.previousElementSibling());
    }
    return rs;
}
 

/**
 * Get sibling elements. If the element has no sibling elements, returns an empty list. An element is not a sibling
 * of itself, so will not be included in the returned list.
 * @return sibling elements
 */
public Elements siblingElements() {
    if (parentNode == null)
        return new Elements(0);

    List<Element> elements = parent().children();
    Elements siblings = new Elements(elements.size() - 1);
    for (Element el: elements)
        if (el != this)
            siblings.add(el);
    return siblings;
}
 

public static Elements precedingSibling(Element el){
    Elements rs = new Elements();
    Element tmp = el.previousElementSibling();
    while (tmp!=null){
        rs.add(tmp);
        tmp = tmp.previousElementSibling();
    }
    if (rs.size() > 0){
        return rs;
    }
    return null;
}
 

List<Node> parseFragment(String inputFragment, Element context, String baseUri, ParseErrorList errors) {
    // context may be null
    state = HtmlTreeBuilderState.Initial;
    initialiseParse(inputFragment, baseUri, errors);
    contextElement = context;
    fragmentParsing = true;
    Element root = null;

    if (context != null) {
        if (context.ownerDocument() != null) // quirks setup:
            doc.quirksMode(context.ownerDocument().quirksMode());

        // initialise the tokeniser state:
        String contextTag = context.tagName();
        if (StringUtil.in(contextTag, "title", "textarea"))
            tokeniser.transition(TokeniserState.Rcdata);
        else if (StringUtil.in(contextTag, "iframe", "noembed", "noframes", "style", "xmp"))
            tokeniser.transition(TokeniserState.Rawtext);
        else if (contextTag.equals("script"))
            tokeniser.transition(TokeniserState.ScriptData);
        else if (contextTag.equals(("noscript")))
            tokeniser.transition(TokeniserState.Data); // if scripting enabled, rawtext
        else if (contextTag.equals("plaintext"))
            tokeniser.transition(TokeniserState.Data);
        else
            tokeniser.transition(TokeniserState.Data); // default

        root = new Element(Tag.valueOf("html"), baseUri);
        doc.appendChild(root);
        stack.push(root);
        resetInsertionMode();

        // setup form element to nearest form on context (up ancestor chain). ensures form controls are associated
        // with form correctly
        Elements contextChain = context.parents();
        contextChain.add(0, context);
        for (Element parent: contextChain) {
            if (parent instanceof FormElement) {
                formElement = (FormElement) parent;
                break;
            }
        }
    }

    runParser();
    if (context != null)
        return root.childNodes();
    else
        return doc.childNodes();
}
 

/**
 * 去掉 指定的 tag 及内容
 * -
 * select 语法 ：tag1,tag2 ...
 * example :
 * select("script").remove().("style").remove().("head").remove()
 * or
 * select("script, style, .hidden").remove()
 * -
 * 此时 currentElements 会包含所有的标签，逐层递减，会有多个，而不是去掉标签之后的结果，只有第一个是
 *
 * @param tags
 * @return
 */
public ElementSelector removeTags(String... tags) {
    Elements temp = new Elements();

    for (String s : tags)   // 大小写不敏感，都能去掉
        this.currentElements.select(s).remove();   // currentElements 已经改变

    if (tags.length != 0) { // 避免无参数时，改变 currentElements
        temp.add(this.currentElements.first());     //此时 currentElements 会包含所有的标签，逐层递减，会有多个，而不是去掉标签之后的结果，只有第一个是
        this.currentElements = temp;
    }
    return this;
}
 

/**
 * 全部祖先节点和自身节点
 *
 * @param e
 * @return
 */
public Elements ancestorOrSelf(Element e) {
    Elements rs = e.parents();
    rs.add(e);
    return rs;
}