Java Code Examples for org.jsoup.nodes.Element#removeAttr()

protected void processFontColor(org.jsoup.nodes.Document document) {
    Elements elements = document.getElementsByTag("font");
    for (Element element : elements) {
        String color = element.attr("color");
        if (color != null) {
            String style = StringUtils.trim(element.attr("style"));
            if (style != null) {
                if (StringUtils.endsWith(style, ";")) {
                    style = style + ";";
                }
                style = style + "color:" + color;
            } else {
                style = "color:" + color;
            }
            element.attr("style", style);
            element.removeAttr("color");
        }
    }
}
 

/**
 * 处理隐藏标签
 * @param html 富文本内容
 * @param visibleTagList 允许可见的隐藏标签
 * @return
 */
public String processHiddenTag(String html,List<Integer> visibleTagList){
	
	if(!StringUtils.isBlank(html)){
		Document doc = Jsoup.parseBodyFragment(html);
		Elements elements = doc.select("hide");  
		for (Element element : elements) {
			//隐藏标签类型
			String hide_type = element.attr("hide-type"); 
			//隐藏标签输入值
		//	String input_value = element.attr("input-value"); 
			
			if(visibleTagList.contains(Integer.parseInt(hide_type.trim()))){//如果允许可见
				/**
				//替换当前标签为<p>标签
				element.removeAttr("class"); 
				element.removeAttr("hide-type");
				element.removeAttr("input-value");
				element.tagName("p");**/
				//移除匹配的元素但保留他们的内容
				element.unwrap();  
				
			}else{
				if(hide_type.trim().equals(HideTagType.PASSWORD.getName().toString())){//输入密码可见
					element.removeAttr("input-value");
				}
				//清空元素的内容
				element.empty();	
			}
		}
		//prettyPrint(是否重新格式化)、outline(是否强制所有标签换行)、indentAmount(缩进长度)    doc.outputSettings().indentAmount(0).prettyPrint(false);
		doc.outputSettings().prettyPrint(false);
		html = doc.body().html();
	}
	return html;
}
 

/**
 * 处理视频播放器标签
 * @param html 富文本内容
 * @param tagId 话题标签  -1表示管理后台打开链接，不校验权限
 * @param secret 密钥
 * @return
 */
public String processVideoPlayer(String html,Long tagId,String secret){
	
	if(!StringUtils.isBlank(html)){
		Document doc = Jsoup.parseBodyFragment(html);
		Elements elements = doc.select("video");  
		for (Element element : elements) {
			//标签src属性
			String src = element.attr("src"); 

			element.removeAttr("src"); 
			//替换当前标签为<player>标签
			element.tagName("player");
			
			
			String url = "";
			if(secret != null && !"".equals(secret.trim())){
				url = SecureLink.createVideoRedirectLink(src,tagId,secret);
			}else{
				url = src;
			}
			element.attr("url",url); 
		
			
		}
		//prettyPrint(是否重新格式化)、outline(是否强制所有标签换行)、indentAmount(缩进长度)    doc.outputSettings().indentAmount(0).prettyPrint(false);
		doc.outputSettings().prettyPrint(false);
		html = doc.body().html();
	}
	return html;
}
 

private String getUpdatedFileContent(List<Vacancy> vacancies) {

        Document document = null;
        try {
            document = getDocument();

            Element templateOriginal = document.getElementsByClass("template").first();
            Element copyTemplate = templateOriginal.clone();
            copyTemplate.removeAttr("style");
            copyTemplate.removeClass("template");
            document.select("tr[class=vacancy]").remove().not("tr[class=vacancy template");

            for (Vacancy vacancy : vacancies) {
                Element localClone = copyTemplate.clone();
                localClone.getElementsByClass("city").first().text(vacancy.getCity());
                localClone.getElementsByClass("companyName").first().text(vacancy.getCompanyName());
                localClone.getElementsByClass("salary").first().text(vacancy.getSalary());
                Element link =localClone.getElementsByTag("a").first();
                link.text(vacancy.getTitle());
                link.attr("href", vacancy.getUrl());

                templateOriginal.before(localClone.outerHtml());
            }
        } catch (IOException e) {
            e.printStackTrace();
            return "Some exception occurred";
        }
        return document.html();
    }
 

/**
 * 将页面转为{@link org.jsoup.nodes.Document}对象，xhtml 格式
 *
 * @param url
 * @return
 * @throws Exception
 */
protected Document url2xhtml(String url) throws Exception {
    Document doc = Jsoup.connect(url).get(); //获得

    if (logger.isDebugEnabled()) {
        logger.debug("baseUri: {}", doc.baseUri());
    }

    for (Element script : doc.getElementsByTag("script")) { //除去所有 script
        script.remove();
    }

    for (Element a : doc.getElementsByTag("a")) { //除去 a 的 onclick，href 属性
        a.removeAttr("onclick");
        a.removeAttr("href");
    }

    Elements links = doc.getElementsByTag("link"); //将link中的地址替换为绝对地址
    for (Element element : links) {
        String href = element.absUrl("href");

        if (logger.isDebugEnabled()) {
            logger.debug("href: {} -> {}", element.attr("href"), href);
        }

        element.attr("href", href);
    }

    doc.outputSettings()
            .syntax(Document.OutputSettings.Syntax.xml)
            .escapeMode(Entities.EscapeMode.xhtml);  //转为 xhtml 格式

    if (logger.isDebugEnabled()) {
        String[] split = doc.html().split("\n");
        for (int c = 0; c < split.length; c++) {
            logger.debug("line {}:\t{}", c + 1, split[c]);
        }
    }
    return doc;
}
 

private String processContentImages(String content, List<String> imageUrls) {

        Document doc = Jsoup.parse(content);
        Elements imgElements = doc.select("img");
        if (imgElements != null) {
            Iterator<Element> iterator = imgElements.iterator();
            while (iterator.hasNext()) {
                Element element = iterator.next();

                String imageUrl = element.hasAttr("src")
                        ? element.attr("src")
                        : element.attr("data-src");

//http://mmbiz.qpic.cn/mmbiz/4gZTdZfnQeDvQqCZFuVvYv8scGS7sEQTRETgISib1blz5iclAtnsccaJhaugmKc
// hhm8mFOtjnicibibumazy8wPS6Xg/640?tp=webp&wxfrom=5&wx_lazy=1&wx_co=1

                imageUrl = replaceLast(imageUrl, "/", "__");
                imageUrl = imageUrl.startsWith("http://")
                        ? imageUrl.replace("http://", "/attachment/")
                        : imageUrl.replace("https://", "/attachment/s");

                imageUrl = imageUrl.replace("?",".png?");

                element.removeAttr("data-src");
                element.attr("src",imageUrl);

                imageUrls.add(imageUrl);
            }
        }

        return doc.toString();
    }
 

private static void cleanAttributes(@NotNull Document d) {
	Elements allEles = d.select("*");
	for (Element code : allEles) {
		code.removeAttr("class");
		code.removeAttr("title");
		code.removeAttr("id");
	}
}
 

private static void cleanAttributes(@NotNull Document d) {
	Elements allEles = d.select("*");
	for (Element code : allEles) {
		code.removeAttr("class");
		code.removeAttr("title");
		code.removeAttr("id");
	}
}
 

/**
 * Transfers styles from the <code>data-cssstyle</code> attribute to the
 * <code>style</code> attribute.
 *
 * @param doc
 *            the html document
 */
private static void applyStyles(Document doc) {
	Elements allStyledElements = doc.getElementsByAttribute(TEMP_STYLE_ATTR);

	for (Element e : allStyledElements) {
		if (isInlineModeAllowed(e, InlineModes.STYLE_ATTR)) {
			String newStyle = e.attr(TEMP_STYLE_ATTR);
			String oldStyle = e.attr(STYLE_ATTR);
			e.attr(STYLE_ATTR, (newStyle.trim() + ";" + oldStyle.trim()).replaceAll(";+", ";").trim());
		}
		e.removeAttr(TEMP_STYLE_ATTR);
	}
}
 

/**
 * Remove an attribute from every matched element.
 * @param attributeKey The attribute to remove.
 * @return this (for chaining)
 */
public Elements removeAttr(String attributeKey) {
    for (Element element : this) {
        element.removeAttr(attributeKey);
    }
    return this;
}
 

/**
 * Remove an attribute from every matched element.
 * @param attributeKey The attribute to remove.
 * @return this (for chaining)
 */
public Elements removeAttr(String attributeKey) {
    for (Element element : this) {
        element.removeAttr(attributeKey);
    }
    return this;
}
 

/**
 * Remove an attribute from every matched element.
 * @param attributeKey The attribute to remove.
 * @return this (for chaining)
 */
public Elements removeAttr(String attributeKey) {
    for (Element element : this) {
        element.removeAttr(attributeKey);
    }
    return this;
}
 

/**
 * Remove an attribute from every matched element.
 * @param attributeKey The attribute to remove.
 * @return this (for chaining)
 */
public Elements removeAttr(String attributeKey) {
    for (Element element : contents) {
        element.removeAttr(attributeKey);
    }
    return this;
}
 

static void removeTrackingPixels(Context context, Document document) {
    Drawable d = ContextCompat.getDrawable(context, R.drawable.baseline_my_location_24);
    d.setTint(Helper.resolveColor(context, R.attr.colorWarning));

    Bitmap bm = Bitmap.createBitmap(d.getIntrinsicWidth(), d.getIntrinsicHeight(), Bitmap.Config.ARGB_8888);
    Canvas c = new Canvas(bm);
    d.setBounds(0, 0, c.getWidth(), c.getHeight());
    d.draw(c);

    ByteArrayOutputStream bos = new ByteArrayOutputStream();
    bm.compress(Bitmap.CompressFormat.PNG, 100, bos);

    StringBuilder sb = new StringBuilder();
    sb.append("data:image/png;base64,");
    sb.append(Base64.encodeToString(bos.toByteArray(), Base64.NO_WRAP));

    // Build list of allowed hosts
    List<String> hosts = new ArrayList<>();
    //for (Element img : document.select("img")) {
    //    String src = img.attr("src");
    //    if (!TextUtils.isEmpty(src) && !isTrackingPixel(img)) {
    //        Uri uri = Uri.parse(img.attr("src"));
    //        String host = uri.getHost();
    //        if (host != null && !hosts.contains(host))
    //            hosts.add(host);
    //    }
    //}

    // Images
    for (Element img : document.select("img")) {
        img.removeAttr("x-tracking");
        String src = img.attr("src");
        if (!TextUtils.isEmpty(src) && isTrackingPixel(img)) {
            Uri uri = Uri.parse(src);
            String host = uri.getHost();
            if (host != null && !hosts.contains(host)) {
                img.attr("src", sb.toString());
                img.attr("alt", context.getString(R.string.title_legend_tracking_pixel));
                img.attr("height", "24");
                img.attr("width", "24");
                img.attr("style", "display:block !important; width:24px !important; height:24px !important;");
                img.attr("x-tracking", src);
            }
        }
    }
}
 

/**
 * Remove attributes that are used only by Ogham:
 * <ul>
 * <li>{@link CssInlinerConstants#INLINE_MODE_ATTR}</li>
 * <li>{@link CssInlinerConstants#INLINED_ATTR}</li>
 * </ul>
 * 
 * @param html
 *            the html to clean
 * @return the cleaned html
 */
public static String removeOghamAttributes(String html) {
	Document doc = Jsoup.parse(html);
	Elements nodes = doc.select("["+INLINE_MODE_ATTR+"], ["+INLINED_ATTR+"]");
	for (Element node : nodes) {
		node.removeAttr(INLINE_MODE_ATTR);
		node.removeAttr(INLINED_ATTR);
	}
	return doc.outerHtml();
}
 

/**
 * Remove attributes that are used only by Ogham:
 * <ul>
 * <li>{@link ImageInlinerConstants#INLINE_MODE_ATTR}</li>
 * <li>{@link ImageInlinerConstants#INLINED_ATTR}</li>
 * </ul>
 * 
 * @param html
 *            the html to clean
 * @return the cleaned html
 */
public static String removeOghamAttributes(String html) {
	Document doc = Jsoup.parse(html);
	Elements imgs = doc.select("img");
	for (Element img : imgs) {
		img.removeAttr(INLINE_MODE_ATTR);
		img.removeAttr(INLINED_ATTR);
	}
	return doc.outerHtml();
}