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

private void process(Element parent, StreamReceiver receiver) {
    for (Element element : parent.children()) {
        receiver.startEntity(element.nodeName());
        Attributes attributes = element.attributes();
        for (Attribute attribute : attributes) {
            receiver.literal(attribute.getKey(), attribute.getValue());
        }
        if (element.children().isEmpty()) {
            String text = element.text().trim();
            String value = text.isEmpty() ? element.data() : text;
            if (!value.isEmpty()) {
                receiver.literal("value", value);
            }
        }
        process(element, receiver);
        receiver.endEntity();
    }
}
 

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);
}
 

/**
 * Creates a javascript which copies attributes from the {@code element} to
 * the created DOM element identified by {@code elementRef}. If {@code
 * element} contains a {@code src} attribute, its path is prefixed with
 * {@code basePath}.
 *
 * @param writer
 *            response writer
 * @param elementRef
 *            variable name of the element in javascript
 * @param element
 *            jsoup element from which to copy the attributes
 * @param basePath
 *            base path of {@code src} attributes (service url's path)
 * @throws IOException
 *             if {@code writer} is unable to write
 */
private void transferAttribute(Writer writer, String elementRef,
        Element element, String basePath) throws IOException {
    for (Attribute attribute : element.attributes()) {
        writer.append(elementRef).append(".setAttribute('")
                .append(attribute.getKey()).append("',");
        if (attribute.getValue() == null) {
            writer.append("''");
        } else {
            String path = attribute.getValue();
            if ("src".equals(attribute.getKey())) {
                path = modifyPath(basePath, path);
            }
            writer.append("'").append(path).append("'");
        }
        writer.append(");");
    }
}
 

/**
 *
 * @param element
 * @return wheter either one attribute of the current element, either its
 * text, either one attribute of one of its parent or the text of one of
 * its parents contains the "captcha" keyword
 */
private boolean parseAttributeToExtractCaptcha(Element element) {
    if (element.nodeName().equalsIgnoreCase(HTML_ELEMENT) || 
            element.nodeName().equalsIgnoreCase(BODY_ELEMENT)) {
        return false;
    }
    if (StringUtils.containsIgnoreCase(element.ownText(), CAPTCHA_KEY)) {
        return true;
    } else {
        for (Attribute attr : element.attributes()) {
            if (StringUtils.containsIgnoreCase(attr.getValue(), CAPTCHA_KEY)) {
                return true;
            }
        }
    }
    return false;
}
 

/**
 * Add the attributes to the given record definition
 *
 * <p>Uses Jsoup to parse the tag as if html
 *
 * @param recordDefinition the record definition
 * @param beginText the begin tag of the record definition
 */
private void addAttributes(TemplateRecordDefinition recordDefinition, String beginText) {

  Document doc = Jsoup.parseBodyFragment(beginText);
  Element fieldElement = doc.body().child(0);
  Attributes attributes = fieldElement.attributes();
  if (attributes.hasKey(REPEAT_ATTRIBUTE)) {
    String required = attributes.get(REPEAT_ATTRIBUTE);
    recordDefinition.setRepeat(
        Strings.isNullOrEmpty(required) ? true : Boolean.valueOf(required));
  }
}
 

@Override
public void readDesign(Element design, DesignContext designContext) {
    super.readDesign(design, designContext);

    Attributes attr = design.attributes();

    if (design.hasAttr("time-zone")) {
        setZoneId(ZoneId.of(DesignAttributeHandler.readAttribute("end-date", attr, String.class)));
    }

    if (design.hasAttr("time-format")) {
        setTimeFormat(TimeFormat.valueOf(
                "Format" + design.attr("time-format").toUpperCase()));
    }

    if (design.hasAttr("start-date")) {
        setStartDate(
                ZonedDateTime.ofInstant(DesignAttributeHandler.readAttribute("start-date", attr, Date.class)
                        .toInstant(), getZoneId()));
    }

    if (design.hasAttr("end-date")) {
        setEndDate(
                ZonedDateTime.ofInstant(DesignAttributeHandler.readAttribute("end-date", attr, Date.class)
                        .toInstant(), getZoneId()));
    }
}
 

@Test public void parsesRoughAttributeString() {
    String html = "<a id=\"123\" class=\"baz = 'bar'\" style = 'border: 2px'qux zim foo = 12 mux=18 />";
    // should be: <id=123>, <class=baz = 'bar'>, <qux=>, <zim=>, <foo=12>, <mux.=18>

    Element el = Jsoup.parse(html).getElementsByTag("a").get(0);
    Attributes attr = el.attributes();
    assertEquals(7, attr.size());
    assertEquals("123", attr.get("id"));
    assertEquals("baz = 'bar'", attr.get("class"));
    assertEquals("border: 2px", attr.get("style"));
    assertEquals("", attr.get("qux"));
    assertEquals("", attr.get("zim"));
    assertEquals("12", attr.get("foo"));
    assertEquals("18", attr.get("mux"));
}
 

@Test public void canStartWithEq() {
    String html = "<a =empty />";
    Element el = Jsoup.parse(html).getElementsByTag("a").get(0);
    Attributes attr = el.attributes();
    assertEquals(1, attr.size());
    assertTrue(attr.hasKey("=empty"));
    assertEquals("", attr.get("=empty"));
}
 

@Override
public Iterator<Attribute> getAttributeAxisIterator(Object contextNode, String localName, String namespacePrefix, String namespaceURI) {
    Element node = ((Element) contextNode);
    LinkedList<Attribute> attributes = new LinkedList<>();
    for (Attribute attribute : node.attributes()) {
        if (localName.equals(attribute.getKey())) {
            attributes.add(attribute);
        }
    }
    return attributes.iterator();
}
 

@Test public void parsesRoughAttributeString() {
    String html = "<a id=\"123\" class=\"baz = 'bar'\" style = 'border: 2px'qux zim foo = 12 mux=18 />";
    // should be: <id=123>, <class=baz = 'bar'>, <qux=>, <zim=>, <foo=12>, <mux.=18>

    Element el = Jsoup.parse(html).getElementsByTag("a").get(0);
    Attributes attr = el.attributes();
    assertEquals(7, attr.size());
    assertEquals("123", attr.get("id"));
    assertEquals("baz = 'bar'", attr.get("class"));
    assertEquals("border: 2px", attr.get("style"));
    assertEquals("", attr.get("qux"));
    assertEquals("", attr.get("zim"));
    assertEquals("12", attr.get("foo"));
    assertEquals("18", attr.get("mux"));
}
 

@Test public void canStartWithEq() {
    String html = "<a =empty />";
    Element el = Jsoup.parse(html).getElementsByTag("a").get(0);
    Attributes attr = el.attributes();
    assertEquals(1, attr.size());
    assertTrue(attr.hasKey("=empty"));
    assertEquals("", attr.get("=empty"));
}
 

@Test public void parsesRoughAttributeString() {
    String html = "<a id=\"123\" class=\"baz = 'bar'\" style = 'border: 2px'qux zim foo = 12 mux=18 />";
    // should be: <id=123>, <class=baz = 'bar'>, <qux=>, <zim=>, <foo=12>, <mux.=18>

    Element el = Jsoup.parse(html).getElementsByTag("a").get(0);
    Attributes attr = el.attributes();
    assertEquals(7, attr.size());
    assertEquals("123", attr.get("id"));
    assertEquals("baz = 'bar'", attr.get("class"));
    assertEquals("border: 2px", attr.get("style"));
    assertEquals("", attr.get("qux"));
    assertEquals("", attr.get("zim"));
    assertEquals("12", attr.get("foo"));
    assertEquals("18", attr.get("mux"));
}
 

@Test public void canStartWithEq() {
    String html = "<a =empty />";
    Element el = Jsoup.parse(html).getElementsByTag("a").get(0);
    Attributes attr = el.attributes();
    assertEquals(1, attr.size());
    assertTrue(attr.hasKey("=empty"));
    assertEquals("", attr.get("=empty"));
}
 

/**
 * Handles the request filtering out unallowed tags. Note that an empty allowedTag we allow
 * all tags.
 */
private void handleRequest(
    Elements elements, HttpServletResponse response, String allowedTag, String allowedAttr)
        throws IOException {
  if (allowedTag.equalsIgnoreCase("script")) {
    elements.empty();
  }

  StringBuilder res = new StringBuilder();
  for (Element element : elements) {
    String tag = element.tagName();

    if (!allowedTag.isEmpty() && !allowedTag.equalsIgnoreCase(tag)) {
      continue;
    }

    if (!allowedAttr.isEmpty()) {
      Attributes attributes = element.attributes();
      for (Attribute attribute : attributes) {
        if (!attribute.getKey().equalsIgnoreCase(allowedAttr)) {
          Responses.sendError(response, "Invalid input attribute", 400);
          return;
        }
      }
    }
    res.append(element.toString());
  }
  Responses.sendXssed(response, res.toString());
}
 

@Test public void parsesEmptyString() {
    String html = "<a />";
    Element el = Jsoup.parse(html).getElementsByTag("a").get(0);
    Attributes attr = el.attributes();
    assertEquals(0, attr.size());
}
 

@Test public void parsesEmptyString() {
    String html = "<a />";
    Element el = Jsoup.parse(html).getElementsByTag("a").get(0);
    Attributes attr = el.attributes();
    assertEquals(0, attr.size());
}
 

@Test public void parsesEmptyString() {
    String html = "<a />";
    Element el = Jsoup.parse(html).getElementsByTag("a").get(0);
    Attributes attr = el.attributes();
    assertEquals(0, attr.size());
}
 

@Test public void parsesEmptyString() {
    String html = "<a />";
    Element el = Jsoup.parse(html).getElementsByTag("a").get(0);
    Attributes attr = el.attributes();
    assertEquals(0, attr.size());
}
 

@Override
public void doGet(HttpServletRequest request, HttpServletResponse response) throws IOException {
  if (request.getParameter("q") == null) {
    Responses.sendError(response, "Missing q parameter", 400);
    return;
  }

  String  q = request.getParameter("q");
  Document doc = Jsoup.parseBodyFragment(q);
  Element body = doc.body();
  Elements elements = body.getAllElements();
  elements.remove(body);
  if (elements.isEmpty()) {
    Responses.sendError(response, "Invalid input, no tags", 400);
    return;
  }

  StringBuilder res = new StringBuilder();
  for (Element element : elements) {
    boolean validElement = true;

    Attributes attributes = element.attributes();
    for (Attribute attribute : attributes) {
      if (attribute.getKey().toLowerCase().startsWith("on")
          || attribute.getKey().toLowerCase().equals("href")
          || attribute.getKey().toLowerCase().equals("src")) {
        validElement = false;
      }

      if (attribute.getKey().toLowerCase().equals("style")
          && attribute.getValue().toLowerCase().contains("expression")) {
        validElement = false;
      }
    }

    if (validElement) {
      res.append(element.toString());
    }
  }
  Responses.sendXssed(response, res.toString());
}