org.jsoup.Jsoup#clean

Source File: CommonUtil.java From blog-hunter with MIT License

5 votes

/**
 * 获取真实的网站介绍，最多只保留100个字符
 *
 * @param description 原博客的description
 * @param content     原博客的正文内容
 */
public static String getRealDescription(String description, String content) {
    if (StringUtils.isNotEmpty(description)) {
        return description.replaceAll("\r\n| ", "");
    }
    if (StringUtils.isNotEmpty(content)) {
        content = Jsoup.clean(content.trim(), Whitelist.simpleText());
        return content.length() > 100 ? content.substring(0, 100) : content;
    }
    return null;
}

Source File: HtmlParserTest.java From astor with GNU General Public License v2.0

5 votes

@Test public void selfClosingVoidIsNotAnError() {
    String html = "<p>test<br/>test<br/></p>";
    Parser parser = Parser.htmlParser().setTrackErrors(5);
    parser.parseInput(html, "");
    assertEquals(0, parser.getErrors().size());

    assertTrue(Jsoup.isValid(html, Whitelist.basic()));
    String clean = Jsoup.clean(html, Whitelist.basic());
    assertEquals("<p>test<br>test<br></p>", clean);
}

Source File: XssKillerUtil.java From OneBlog with GNU General Public License v3.0

5 votes

/**
 * 根据白名单，剔除多余的属性、标签
 *
 * @param xssStr
 * @return
 */
public static String clean(String xssStr) {
    if (null == xssStr || xssStr.isEmpty()) {
        return "";
    }
    return Jsoup.clean(xssStr, custome());
}

Source File: XssFilter.java From Aooms with Apache License 2.0

5 votes

/** 
 * 对用户输入内容进行过滤 
 * @return
 */  
public String filter(String value) {
    if(StrUtil.isNotBlank(value)){
    	return Jsoup.clean(value, user_content_filter);  
    }
    return value;
}

Source File: CleanerTest.java From astor with GNU General Public License v2.0

5 votes

@Test public void handlesAllPseudoTag() {
    String html = "<p class='foo' src='bar'><a class='qux'>link</a></p>";
    Whitelist whitelist = new Whitelist()
            .addAttributes(":all", "class")
            .addAttributes("p", "style")
            .addTags("p", "a");

    String clean = Jsoup.clean(html, whitelist);
    assertEquals("<p class=\"foo\"><a class=\"qux\">link</a></p>", clean);
}

Source File: CleanerTest.java From astor with GNU General Public License v2.0

5 votes

@Test public void handlesFramesets() {
    String dirty = "<html><head><script></script><noscript></noscript></head><frameset><frame src=\"foo\" /><frame src=\"foo\" /></frameset></html>";
    String clean = Jsoup.clean(dirty, Whitelist.basic());
    assertEquals("", clean); // nothing good can come out of that

    Document dirtyDoc = Jsoup.parse(dirty);
    Document cleanDoc = new Cleaner(Whitelist.basic()).clean(dirtyDoc);
    assertFalse(cleanDoc == null);
    assertEquals(0, cleanDoc.body().childNodeSize());
}

Source File: CleanerTest.java From astor with GNU General Public License v2.0

5 votes

@Test public void testRemoveEnforcedAttributes() {
    String h = "<div><p><A HREF='HTTP://nice.com'>Nice</a></p><blockquote>Hello</blockquote>";
    String cleanHtml = Jsoup.clean(h, Whitelist.basic().removeEnforcedAttribute("a", "rel"));

    assertEquals("<p><a href=\"http://nice.com\">Nice</a></p><blockquote>Hello</blockquote>",
            TextUtil.stripNewlines(cleanHtml));
}

Source File: PreviewTextUtils.java From mblog with GNU General Public License v3.0

4 votes

public static String removeHideHtml(String html) {
    if (html == null)
        return null;
    return Jsoup.clean(html, (new Whitelist()).addTags("hide"));
}

Source File: CleanerTest.java From astor with GNU General Public License v2.0

4 votes

@Test public void basicWithImagesTest() {
    String h = "<div><p><img src='http://example.com/' alt=Image></p><p><img src='ftp://ftp.example.com'></p></div>";
    String cleanHtml = Jsoup.clean(h, Whitelist.basicWithImages());
    assertEquals("<p><img src=\"http://example.com/\" alt=\"Image\"></p><p><img></p>", TextUtil.stripNewlines(cleanHtml));
}

Source File: CleanerTest.java From astor with GNU General Public License v2.0

4 votes

@Test public void testDropComments() {
    String h = "<p>Hello<!-- no --></p>";
    String cleanHtml = Jsoup.clean(h, Whitelist.relaxed());
    assertEquals("<p>Hello</p>", cleanHtml);
}

Source File: ArticleProcessor.java From symphonyx with Apache License 2.0

4 votes

/**
 * Gets article preview content.
 *
 * <p>
 * Renders the response with a json object, for example,
 * <pre>
 * {
 *     "html": ""
 * }
 * </pre>
 * </p>
 *
 * @param request the specified http servlet request
 * @param response the specified http servlet response
 * @param context the specified http request context
 * @param articleId the specified article id
 * @throws Exception exception
 */
@RequestProcessing(value = "/article/{articleId}/preview", method = HTTPRequestMethod.GET)
@Before(adviceClass = StopwatchStartAdvice.class)
@After(adviceClass = StopwatchEndAdvice.class)
public void getArticlePreviewContent(final HttpServletRequest request, final HttpServletResponse response,
        final HTTPRequestContext context, final String articleId) throws Exception {
    context.renderJSON(true).renderJSONValue("html", "");

    final JSONObject article = articleQueryService.getArticle(articleId);
    if (null == article) {
        context.renderFalseResult();

        return;
    }

    final int length = Integer.valueOf("150");
    String content = article.optString(Article.ARTICLE_CONTENT);
    final String authorId = article.optString(Article.ARTICLE_AUTHOR_ID);
    final JSONObject author = userQueryService.getUser(authorId);

    if (null != author && UserExt.USER_STATUS_C_INVALID == author.optInt(UserExt.USER_STATUS)
            || Article.ARTICLE_STATUS_C_INVALID == article.optInt(Article.ARTICLE_STATUS)) {
        context.renderJSONValue("html", langPropsService.get("articleContentBlockLabel"));

        return;
    }

    final Set<String> userNames = userQueryService.getUserNames(content);
    final JSONObject currentUser = userQueryService.getCurrentUser(request);
    final String currentUserName = null == currentUser ? "" : currentUser.optString(User.USER_NAME);
    final String authorName = author.optString(User.USER_NAME);
    if (Article.ARTICLE_TYPE_C_DISCUSSION == article.optInt(Article.ARTICLE_TYPE)
            && !authorName.equals(currentUserName)) {
        boolean invited = false;
        for (final String userName : userNames) {
            if (userName.equals(currentUserName)) {
                invited = true;

                break;
            }
        }

        if (!invited) {
            String blockContent = langPropsService.get("articleDiscussionLabel");
            blockContent = blockContent.replace("{user}", "<a href='" + Latkes.getServePath()
                    + "/member/" + authorName + "'>" + authorName + "</a>");

            context.renderJSONValue("html", blockContent);

            return;
        }
    }

    content = Emotions.convert(content);
    content = Markdowns.toHTML(content);

    content = Jsoup.clean(content, Whitelist.none());
    if (content.length() >= length) {
        content = StringUtils.substring(content, 0, length)
                + " ....";
    }

    context.renderJSONValue("html", content);
}

Source File: MyHtmlUtils.java From spring-boot with Apache License 2.0

4 votes

public static String getBasicHtml(String html) {
    if (html == null)
        return null;
    return Jsoup.clean(html, Whitelist.basic());
}

Source File: CleanerTest.java From astor with GNU General Public License v2.0

4 votes

@Test public void simpleBehaviourTest() {
    String h = "<div><p class=foo><a href='http://evil.com'>Hello <b id=bar>there</b>!</a></div>";
    String cleanHtml = Jsoup.clean(h, Whitelist.simpleText());

    assertEquals("Hello <b>there</b>!", TextUtil.stripNewlines(cleanHtml));
}

Source File: CleanerTest.java From astor with GNU General Public License v2.0

4 votes

@Test public void handlesControlCharactersAfterTagName() {
    String html = "<a/\06>";
    String clean = Jsoup.clean(html, Whitelist.basic());
    assertEquals("<a rel=\"nofollow\"></a>", clean);
}

Source File: CleanerTest.java From astor with GNU General Public License v2.0

4 votes

@Test public void resolvesRelativeLinks() {
    String html = "<a href='/foo'>Link</a><img src='/bar'>";
    String clean = Jsoup.clean(html, "http://example.com/", Whitelist.basicWithImages());
    assertEquals("<a href=\"http://example.com/foo\" rel=\"nofollow\">Link</a>\n<img src=\"http://example.com/bar\">", clean);
}

Source File: MyHtmlUtils.java From spring-boot with Apache License 2.0

4 votes

public static String getText(String html) {
    if (html == null)
        return null;
    return Jsoup.clean(html, Whitelist.none());
}

Source File: JsoupUtils.java From springboot-admin with Apache License 2.0

4 votes

public static String clear(String html) {
	if (StringUtils.isNotBlank(html))
		return Jsoup.clean(html, whitelist);

	return html;
}

Source File: CleanerTest.java From astor with GNU General Public License v2.0

4 votes

@Test public void testDropsUnknownTags() {
    String h = "<p><custom foo=true>Test</custom></p>";
    String cleanHtml = Jsoup.clean(h, Whitelist.relaxed());
    assertEquals("<p>Test</p>", cleanHtml);
}

Source File: CleanerTest.java From astor with GNU General Public License v2.0

4 votes

@Test public void testDropsUnknownTags() {
    String h = "<p><custom foo=true>Test</custom></p>";
    String cleanHtml = Jsoup.clean(h, Whitelist.relaxed());
    assertEquals("<p>Test</p>", cleanHtml);
}

Source File: CleanerTest.java From astor with GNU General Public License v2.0

4 votes

@Test public void simpleBehaviourTest2() {
    String h = "Hello <b>there</b>!";
    String cleanHtml = Jsoup.clean(h, Whitelist.simpleText());

    assertEquals("Hello <b>there</b>!", TextUtil.stripNewlines(cleanHtml));
}

Java Code Examples for org.jsoup.Jsoup#clean()