Java Code Examples for org.jsoup.safety.Whitelist

@Override
public String[] getParameterValues(String name)
{
    String[] values = super.getParameterValues(name);
    if (values != null)
    {
        int length = values.length;
        String[] escapseValues = new String[length];
        for (int i = 0; i < length; i++)
        {
            // 防xss攻击和过滤前后空格
            escapseValues[i] = Jsoup.clean(values[i], Whitelist.relaxed()).trim();
        }
        return escapseValues;
    }
    return super.getParameterValues(name);
}
 

@Override
public String[] getParameterValues(String name)
{
    String[] values = super.getParameterValues(name);
    if (values != null)
    {
        int length = values.length;
        String[] escapseValues = new String[length];
        for (int i = 0; i < length; i++)
        {
            // 防xss攻击和过滤前后空格
            escapseValues[i] = Jsoup.clean(values[i], Whitelist.relaxed()).trim();
        }
        return escapseValues;
    }
    return super.getParameterValues(name);
}
 

public static void main(String[] args) {
	String baseUri = "http://www.baidu.com";
	String html = "<a href=\"http://www.baidu.com/gaoji/preferences.html\"name=\"tj_setting\">搜索设置</a>";
	String doc = Jsoup.clean(html, baseUri, Whitelist.none());
	System.out.println(doc);
	System.out.println("*******");
	doc = Jsoup.clean(html, baseUri, Whitelist.simpleText());
	System.out.println(doc);
	System.out.println("*******");
	doc = Jsoup.clean(html, baseUri, Whitelist.basic());
	System.out.println(doc);
	System.out.println("*******");
	doc = Jsoup.clean(html, baseUri, Whitelist.basicWithImages());
	System.out.println(doc);
	System.out.println("*******");
	doc = Jsoup.clean(html, baseUri, Whitelist.relaxed());
	System.out.println(doc);

}
 

public static void main(String[] args) {
    
    String d = "<span><div>test</div></span>";
    Document doc = Jsoup.parse(d);
    Element div = doc.select("div").first(); // <div></div>
    div.html("<p>lorem ipsum</p>"); // <div><p>lorem ipsum</p></div>
    div.prepend("<p>First</p>");
    div.append("<p>Last</p>");
    // now: <div><p>First</p><p>lorem ipsum</p><p>Last</p></div>
    div.appendElement(d);
    Element span = doc.select("span").first(); // <span>One</span>
    span.wrap("<li><a href='http://example.com/'></a></li>");
    // now: <li><a href="http://example.com"><span>One</span></a></li>
    System.out.println(doc.html());
    
    String s = Jsoup.clean(doc.html(), "", Whitelist.relaxed(), new OutputSettings().prettyPrint(false));
    
    System.out.println(s);
}
 

public TRECAquaintDocumentIndexer(String indexPath, String tokenFilterFile, boolean positional){
    super(indexPath, tokenFilterFile, positional);

    try {
        whiteList = Whitelist.relaxed();
        whiteList.addTags("docno");
        whiteList.addTags("doc");
        whiteList.addTags("headline");
        whiteList.addTags("text");
        whiteList.addTags("date_time");
        whiteList.addTags("slug");
    } catch (Exception e){
        System.out.println(" caught a " + e.getClass() +
                "\n with message: " + e.getMessage());
    }

    doc = new Document();
    initFields();
    initAQUAINTDoc();
}
 

@Override
public void loadURL(String url) {
	logger.debug("loading " + url);
	
	ThreadManager.getInstance().executeThread(()->{
			try {
				
				Whitelist w = Whitelist.basic();
				w.addTags("img");
				w.addAttributes("img", "src");
				
				String contf = Jsoup.clean(RequestBuilder.build().clean().url(url).method(METHOD.GET).setClient(client).toHtml().html(),w);
				browse.setText(contf);
			}
			catch(Exception e)
			{
				logger.error(e);
				browse.setText("Error " + e);
			}
			
	}, "loading " + url);
	
	
	
}
 

@Override
public int setErrorParameter(BeforeEnterEvent event,
        ErrorParameter<NotFoundException> parameter) {
    String path = event.getLocation().getPath();
    String additionalInfo = "";
    if (parameter.hasCustomMessage()) {
        additionalInfo = "Reason: " + parameter.getCustomMessage();
    }
    path = Jsoup.clean(path, Whitelist.none());
    additionalInfo = Jsoup.clean(additionalInfo, Whitelist.none());

    boolean productionMode = event.getUI().getSession().getConfiguration()
            .isProductionMode();

    String template = getErrorHtml(productionMode);
    template = template.replace("{{path}}", path);
    template = template.replace("{{additionalInfo}}", additionalInfo);
    if (template.contains("{{routes}}")) {
        template = template.replace("{{routes}}", getRoutes(event));
    }

    getElement().appendChild(new Html(template).getElement());
    return HttpServletResponse.SC_NOT_FOUND;
}
 

@PutMapping(value = "/{id}")
public Result edit(@PathVariable Integer id, @RequestBody Map<String, String> body) {
    User user = getApiUser();
    String title = body.get("title");
    String content = body.get("content");
    ApiAssert.notEmpty(title, "请输入标题");
    // 更新话题
    Topic topic = topicService.selectById(id);
    ApiAssert.isTrue(topic.getUserId().equals(user.getId()), "谁给你的权限修改别人的话题的？");
    topic.setTitle(Jsoup.clean(title, Whitelist.none().addTags("video")));
    topic.setContent(content);
    topic.setModifyTime(new Date());
    topicService.update(topic, null);
    topic.setContent(SensitiveWordUtil.replaceSensitiveWord(topic.getContent(), "*", SensitiveWordUtil.MinMatchType));
    return success(topic);
}
 

@Override
public void update(Topic topic, String tags) {
    topicMapper.updateById(topic);
    // 处理标签
    if (!StringUtils.isEmpty(tags)) {
        // 旧标签每个topicCount都-1
        tagService.reduceTopicCount(topic.getId());
        if (!StringUtils.isEmpty(tags)) {
            // 保存标签
            List<Tag> tagList = tagService.insertTag(Jsoup.clean(tags, Whitelist.none()));
            // 处理标签与话题的关联
            topicTagService.insertTopicTag(topic.getId(), tagList);
        }
    }
    // 索引话题
    indexedService.indexTopic(String.valueOf(topic.getId()), topic.getTitle(), topic.getContent());
}
 

public Map<String, Integer> calculateWordCount(final DocumentContentData documentContentData, final int maxResult) {

		final String html = documentContentData.getContent();
		
		final SimpleCorpus simpleCorpus = new SimpleCorpus(SimpleSentenceSplitter.getInstance(), new SimpleTokenizer(),
				new SwedishStopWords(), EnglishPunctuations.getInstance());

		simpleCorpus.add(new Text( Jsoup.clean(html, Whitelist.basic())));

		final Iterator<String> terms = simpleCorpus.getTerms();

		final Map<String, Integer> result = new HashMap<>();
		while (terms.hasNext()) {
			final String term = terms.next();
			result.put(term, simpleCorpus.getTermFrequency(term));
		}
		return result;
	}
 

@Override
protected void onCreate(Bundle savedInstanceState) {
    super.onCreate(savedInstanceState);
    setContentView(R.layout.activity_sanitize);

    final EditText inputText = (EditText) findViewById(R.id.input_text);
    inputText
            .setText("<p><a href='http://example.com/' onclick='doAttack()'>Link</a></p>");
    final EditText sanitizedText = (EditText) findViewById(R.id.sanitized_text);
    findViewById(R.id.sanitize_button).setOnClickListener(
            new OnClickListener() {
                @Override
                public void onClick(View v) {
                    String sanitized = Jsoup.clean(inputText.getText()
                            .toString(), Whitelist.basic());
                    sanitizedText.setText(sanitized);
                }
            });
}
 

/**
 * 提取纯文本
 * @param html 代码
 * @return string
 */
public static String getText(String html) {
    if (html == null) {
        return null;
    }
    return Jsoup.clean(html, Whitelist.none()).trim();
}
 

/**
 * 以下标签可以通过 (b, em, i, strong, u. 纯文本)
 * @param html 代码
 * @return string
 */
public static String getSimpleHtml(String html) {
    if (html == null) {
        return null;
    }
    return Jsoup.clean(html, Whitelist.simpleText());
}
 

/**
 * 文本过滤标签，只保留<br>标签
 * @param html
 * @return
 */
public String filterTag_br(String html) {  
	if(StringUtils.isBlank(html)) return ""; 
	
	Whitelist whitelist = Whitelist.none();//只保留文本，其他所有的html内容均被删除
	whitelist.addTags("br");
	
    return Jsoup.clean(html, whitelist); 
	//return Jsoup.clean(html,"", whitelist,new OutputSettings().prettyPrint(false)); //prettyPrint(是否重新格式化)
}
 

/**
 * 富文本过滤标签
 * @param request
 * @param html
 * @return
 */
public String filterTag(HttpServletRequest request,String html) {  
	if(StringUtils.isBlank(html)) return ""; 
	Whitelist whitelist = this.filterParameter(null);

    //return Jsoup.clean(html, Configuration.getUrl(request),whitelist); 
	return Jsoup.clean(html, Configuration.getUrl(request),whitelist,new OutputSettings().prettyPrint(false)); //prettyPrint(是否重新格式化)
}
 

/**
 * 过滤所有的标签，只返回文本
 * @param html
 * @return
 */
public String filterText(String html) {  
	if(StringUtils.isBlank(html)) return ""; 
	return Jsoup.clean(html, Whitelist.none()); //只保留文本，其他所有的html内容均被删除
	
	//doc.text()或Jsoup.clean提取出文本，注意text会将p等标签转为空格而不是换行符，而clean默认会转为换行符。
	
	
	//只保留文本，其他所有的html内容均被删除
	//return Jsoup.clean(html, "",Whitelist.none(),new OutputSettings().prettyPrint(false)); //prettyPrint(是否重新格式化)
}
 

/**
 * 过滤标签并删除<hide>标签所有内容,只返回文本
 * @param html
 * @return
 */
public String filterHideText(String html) {  
	if(StringUtils.isBlank(html)) return ""; 
	String newHtml = this.deleteHiddenTag(html);
	if(StringUtils.isBlank(newHtml)) return ""; 
	//只保留文本，其他所有的html内容均被删除
	return Jsoup.clean(newHtml, Whitelist.none()); 
	//return Jsoup.clean(newHtml,"", Whitelist.none(),new OutputSettings().prettyPrint(false)); //prettyPrint(是否重新格式化)
}
 

public static String cleanHighlight(String aHighlight) {
    Whitelist wl = new Whitelist();
    wl.addTags("em");
    Document dirty = Jsoup.parseBodyFragment(aHighlight, "");
    Cleaner cleaner = new Cleaner(wl);
    Document clean = cleaner.clean(dirty);
    clean.select("em").tagName("mark");

    return clean.body().html();
}
 

@SuppressWarnings("rawtypes")
public void execute(Environment env, Map params, TemplateModel[] loopVars,
		TemplateDirectiveBody body) throws TemplateException, IOException {
	DefaultObjectWrapperBuilder builder = new DefaultObjectWrapperBuilder(Configuration.VERSION_2_3_28);
	// 获取页面的参数
	String content = params.get("content").toString();
	Integer num = Integer.parseInt(params.get("num").toString());
	content = Jsoup.clean(content, Whitelist.none());
	content = StringUtils.abbreviate(content, num);
	env.setVariable("info_content", builder.build().wrap(content));
	body.render(env.getOut());
}
 

public static String sanitize(String unsafeHtmlContent) {
	Whitelist whiteList = Whitelist.relaxed().preserveRelativeLinks(true);
	whiteList = whiteList.addTags("head", "html", "style", "body", "fieldsMap", "area");
	whiteList = whiteList.addAttributes(":all", "name", "class", "style", "id", "src", "type", "cellpadding",
			"cellspacing", "alt", "title", "shape", "coords", "width", "height", "dir");
	whiteList = whiteList.addProtocols("img", "src", "http", "https", "data", "cid");

	return Jsoup.clean(unsafeHtmlContent, whiteList);
}
 

public static String sanitizeSimpleText(String unsafeHtmlContent) {
	OutputSettings outputSettings = new OutputSettings().indentAmount(0).prettyPrint(false);
	Whitelist whiteList = Whitelist.simpleText().preserveRelativeLinks(false);
	String sanitized = Jsoup.clean(unsafeHtmlContent, "", whiteList, outputSettings);
	sanitized = StringEscapeUtils.unescapeHtml(sanitized);
	return sanitized;
}
 

@Override
public String[] getParameterValues(String name) {
    String[] values = super.getParameterValues(name);
    if (values != null) {
        int length = values.length;
        String[] escapseValues = new String[length];
        for (int i = 0; i < length; i++) {
            // 防xss攻击和过滤前后空格
            escapseValues[i] = Jsoup.clean(values[i], Whitelist.relaxed()).trim();
        }
        return escapseValues;
    }
    return super.getParameterValues(name);
}
 

public String getDescription(String page) {
    try {
        // Fetch the image page
        Response resp = Http.url(page)
                .referrer(this.url)
                .response();
        cookies.putAll(resp.cookies());

        // Try to find the description
        Elements els = resp.parse().select("td[class=alt1][width=\"70%\"]");
        if (els.isEmpty()) {
            LOGGER.debug("No description at " + page);
            throw new IOException("No description found");
        }
        LOGGER.debug("Description found!");
        Document documentz = resp.parse();
        Element ele = documentz.select("td[class=alt1][width=\"70%\"]").get(0); // This is where the description is.
        // Would break completely if FurAffinity changed site layout.
        documentz.outputSettings(new Document.OutputSettings().prettyPrint(false));
        ele.select("br").append("\\n");
        ele.select("p").prepend("\\n\\n");
        LOGGER.debug("Returning description at " + page);
        String tempPage = Jsoup.clean(ele.html().replaceAll("\\\\n", System.getProperty("line.separator")), "", Whitelist.none(), new Document.OutputSettings().prettyPrint(false));
        return documentz.select("meta[property=og:title]").attr("content") + "\n" + tempPage; // Overridden saveText takes first line and makes it the file name.
    } catch (IOException ioe) {
        LOGGER.info("Failed to get description " + page + " : '" + ioe.getMessage() + "'");
        return null;
    }
}
 

/**
 * Resolves an {@link AudioItem} from a string.
 * This can be a url or search terms
 *
 * @param player The music player
 * @param input  The string to get the AudioItem from.
 * @return {@link AudioItem} from the string.
 * @throws IllegalArgumentException If the Item couldn't be found due to it not existing on Youtube.
 * @throws IllegalStateException    If the Video is unavailable for Flare, for example if it was published by VEVO.
 */
public static AudioItem resolveItem(Player player, String input) throws IllegalArgumentException, IllegalStateException {
    Optional<AudioItem> item = Optional.empty();
    boolean failed = false;
    int backoff = 2;
    Throwable cause = null;
    for (int i = 0; i <= 2; i++) {
        try {
            item = Optional.ofNullable(player.resolve(input));
            failed = false;
            break;
        } catch (FriendlyException | InterruptedException | ExecutionException e) {
            failed = true;
            cause = e;
            if (e.getMessage().contains("Vevo")) {
                throw new IllegalStateException(Jsoup.clean(cause.getMessage(), Whitelist.none()), cause);
            }
            FlareBot.LOGGER.error(Markers.NO_ANNOUNCE, "Cannot get video '" + input + "'");
            try {
                Thread.sleep(backoff);
            } catch (InterruptedException ignored) {
            }
            backoff ^= 2;
        }
    }
    if (failed) {
        throw new IllegalStateException(Jsoup.clean(cause.getMessage(), Whitelist.none()), cause);
    } else if (!item.isPresent()) {
        throw new IllegalArgumentException();
    }
    return item.get();
}
 

@Override
protected String formatHtml(String content) {
    Element body = Jsoup.parse(content).body();
    Element contentDiv = body.select("div[dir='ltr']").first();
    if (contentDiv == null) {
        log.warn("Found no valid content in e-mail from Gmail, returning empty");
        return "";
    }

    while (contentDiv.children().size() > 0 && contentDiv.children().last().is("br")) {
        contentDiv.children().last().remove();
    }

    return Jsoup.clean(contentDiv.html(), Whitelist.basic());
}
 

@Override
protected String formatHtml(String content) {
    Element body = Jsoup.parse(content).body();
    body.select("blockquote[cite]").remove();
    body.select("div.moz-cite-prefix").remove();

    while (body.children().size() > 0 && body.children().last().is("br")) {
        body.children().last().remove();
    }

    return Jsoup.clean(body.html(), Whitelist.basic());
}
 

private static Whitelist whitelist(String extraTagToRemoveWhitelist, String extraTagToAddWhiteList)
{	
	Whitelist newWhiteList = whitelist;
	if(extraTagToRemoveWhitelist!=null && !extraTagToRemoveWhitelist.isEmpty())
		newWhiteList.removeTags(extraTagToRemoveWhitelist);
	if(extraTagToAddWhiteList!=null && !extraTagToAddWhiteList.isEmpty())
		newWhiteList.addTags(extraTagToAddWhiteList);
	return newWhiteList;
}
 

private static Whitelist whitelist(String[] extraTagsToRemoveWhitelist, String[] extraTagsToAddWhiteList)
{		
	Whitelist newWhiteList = whitelist;
	if(extraTagsToRemoveWhitelist!=null && extraTagsToRemoveWhitelist.length>0)
		newWhiteList.removeTags(extraTagsToRemoveWhitelist);
	if(extraTagsToAddWhiteList!=null && extraTagsToAddWhiteList.length>0)
		newWhiteList.addTags(extraTagsToAddWhiteList);
	return whitelist.removeTags(extraTagsToRemoveWhitelist);
}
 

private static List<String> parse(String input, Whitelist wl)
{
	String cleanInput=Jsoup.clean(input, "", wl, outputSettings);
	//System.out.println(input);
	Document document = Jsoup.parse(cleanInput);
	
	document.outputSettings(outputSettings);
	
	List<String> textList = new ArrayList<String>();

	readNodes(document.body().childNodes(), textList);
	return textList;
}
 

private static List<Map.Entry<String,String>> parseWithTags(String input, Whitelist wl)
{
	String cleanInput=Jsoup.clean(input, "", wl, outputSettings);
	Document document = Jsoup.parse(cleanInput);
	
	document.outputSettings(outputSettings);
	
	List<Map.Entry<String,String>> textListMap = new ArrayList<Map.Entry<String,String>>();

	readNodesWithTags(document.body().childNodes(), textListMap,"body");
	return textListMap;
}