Java Code Examples for org.jsoup.nodes.Attribute

@Test public void parsesBooleanAttributes() {
      String html = "<a normal=\"123\" boolean empty=\"\"></a>";
      Element el = Jsoup.parse(html).select("a").first();
      
      assertEquals("123", el.attr("normal"));
      assertEquals("", el.attr("boolean"));
      assertEquals("", el.attr("empty"));
      
      List<Attribute> attributes = el.attributes().asList();
      assertEquals("There should be 3 attribute present", 3, attributes.size());
      
      // Assuming the list order always follows the parsed html
assertFalse("'normal' attribute should not be boolean", attributes.get(0) instanceof BooleanAttribute);        
assertTrue("'boolean' attribute should be boolean", attributes.get(1) instanceof BooleanAttribute);        
assertFalse("'empty' attribute should not be boolean", attributes.get(2) instanceof BooleanAttribute);        
      
      assertEquals(html, el.outerHtml());
  }
 

@Override
protected boolean isSafeAttribute(String tagName, Element el, Attribute attr) {

    //不允许 javascript 开头的 src 和 href
    if ("src".equalsIgnoreCase(attr.getKey()) || "href".equalsIgnoreCase(attr.getKey())) {
        String value = attr.getValue();
        if (StrUtil.isNotBlank(value) && value.toLowerCase().startsWith("javascript")) {
            return false;
        }
    }


    //允许 base64 的图片内容
    if ("img".equals(tagName) && "src".equals(attr.getKey()) && attr.getValue().startsWith("data:;base64")){
        return true;
    }

    return super.isSafeAttribute(tagName, el, attr);
}
 

@Override
public Node buildNodeFromHTML(Element element) {
    String tag = element.tagName().toLowerCase();
    Node node = getNodeInstance(EditorType.macro);
    node.content.add(tag);

    List<Attribute> attrs = element.attributes().asList();

    if (!attrs.isEmpty()) {
        node.macroSettings = new HashMap<>();
        for (Attribute attr : attrs) {
            node.macroSettings.put(attr.getKey(), attr.getValue());
        }
    }
    int index = editorCore.getChildCount();
    View view = editorCore.getEditorListener().onRenderMacro(tag, node.macroSettings, editorCore.getChildCount());

    if(view == null)
        view = getEmptyMacro(node.content.get(0), node.macroSettings);

    insertMacro(tag, view, node.macroSettings, index);
    return null;
}
 

/**
 * Remove the comments of the page 
 * 
 * @param node 
 */
private void removeMalformedAttributes(Node node) {
    // as we are removing child nodes while iterating, we cannot use a normal foreach over children,
    // or will get a concurrent list modification error.
    int i = 0;
    while (i < node.childNodes().size()) {
        Node child = node.childNode(i);
        for (Attribute attr : child.attributes()) {
            if (attr.getKey().startsWith("\"") && attr.getKey().endsWith("\"")) {
                child.removeAttr(attr.getKey());
            }
        }
        removeMalformedAttributes(child);
        i++;
    }
}
 

/**
 *
 * @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;
}
 

/**
 * 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(");");
    }
}
 

/**
 * Test if the supplied attribute is allowed by this whitelist for this tag
 * @param tagName tag to consider allowing the attribute in
 * @param el element under test, to confirm protocol
 * @param attr attribute under test
 * @return true if allowed
 */
protected boolean isSafeAttribute(String tagName, Element el, Attribute attr) {
    TagName tag = TagName.valueOf(tagName);
    AttributeKey key = AttributeKey.valueOf(attr.getKey());

    if (attributes.containsKey(tag)) {
        if (attributes.get(tag).contains(key)) {
            if (protocols.containsKey(tag)) {
                Map<AttributeKey, Set<Protocol>> attrProts = protocols.get(tag);
                // ok if not defined protocol; otherwise test
                return !attrProts.containsKey(key) || testValidProtocol(el, attr, attrProts.get(key));
            } else { // attribute found, no protocols defined, so OK
                return true;
            }
        }
    }
    // no attributes defined for tag, try :all tag
    return !tagName.equals(":all") && isSafeAttribute(":all", el, attr);
}
 

/**
 * Finds any namespaces defined in this element. Returns any tag prefix.
 */
private String updateNamespaces(org.jsoup.nodes.Element el) {
    // scan the element for namespace declarations
    // like: xmlns="blah" or xmlns:prefix="blah"
    Attributes attributes = el.attributes();
    for (Attribute attr : attributes) {
        String key = attr.getKey();
        String prefix;
        if (key.equals(xmlnsKey)) {
            prefix = "";
        } else if (key.startsWith(xmlnsPrefix)) {
            prefix = key.substring(xmlnsPrefix.length());
        } else {
            continue;
        }
        namespaces.put(prefix, attr.getValue());
    }

    // get the element prefix if any
    int pos = el.tagName().indexOf(":");
    return pos > 0 ? el.tagName().substring(0, pos) : "";
}
 

@Test public void parsesBooleanAttributes() {
      String html = "<a normal=\"123\" boolean empty=\"\"></a>";
      Element el = Jsoup.parse(html).select("a").first();
      
      assertEquals("123", el.attr("normal"));
      assertEquals("", el.attr("boolean"));
      assertEquals("", el.attr("empty"));
      
      List<Attribute> attributes = el.attributes().asList();
      assertEquals("There should be 3 attribute present", 3, attributes.size());
      
      // Assuming the list order always follows the parsed html
assertFalse("'normal' attribute should not be boolean", attributes.get(0) instanceof BooleanAttribute);        
assertTrue("'boolean' attribute should be boolean", attributes.get(1) instanceof BooleanAttribute);        
assertFalse("'empty' attribute should not be boolean", attributes.get(2) instanceof BooleanAttribute);        
      
      assertEquals(html, el.outerHtml());
  }
 

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);
}
 

/**
 * Finds any namespaces defined in this element. Returns any tag prefix.
 */
private String updateNamespaces(org.jsoup.nodes.Element el) {
    // scan the element for namespace declarations
    // like: xmlns="blah" or xmlns:prefix="blah"
    Attributes attributes = el.attributes();
    for (Attribute attr : attributes) {
        String key = attr.getKey();
        String prefix;
        if (key.equals(xmlnsKey)) {
            prefix = "";
        } else if (key.startsWith(xmlnsPrefix)) {
            prefix = key.substring(xmlnsPrefix.length());
        } else {
            continue;
        }
        namespaces.put(prefix, attr.getValue());
    }

    // get the element prefix if any
    int pos = el.tagName().indexOf(":");
    return pos > 0 ? el.tagName().substring(0, pos) : "";
}
 

@Test public void parsesBooleanAttributes() {
      String html = "<a normal=\"123\" boolean empty=\"\"></a>";
      Element el = Jsoup.parse(html).select("a").first();
      
      assertEquals("123", el.attr("normal"));
      assertEquals("", el.attr("boolean"));
      assertEquals("", el.attr("empty"));
      
      List<Attribute> attributes = el.attributes().asList();
      assertEquals("There should be 3 attribute present", 3, attributes.size());
      
      // Assuming the list order always follows the parsed html
assertFalse("'normal' attribute should not be boolean", attributes.get(0) instanceof BooleanAttribute);        
assertTrue("'boolean' attribute should be boolean", attributes.get(1) instanceof BooleanAttribute);        
assertFalse("'empty' attribute should not be boolean", attributes.get(2) instanceof BooleanAttribute);        
      
      assertEquals(html, el.outerHtml());
  }
 

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);
}
 

/**
 * Finds any namespaces defined in this element. Returns any tag prefix.
 */
private String updateNamespaces(org.jsoup.nodes.Element el) {
    // scan the element for namespace declarations
    // like: xmlns="blah" or xmlns:prefix="blah"
    Attributes attributes = el.attributes();
    for (Attribute attr : attributes) {
        String key = attr.getKey();
        String prefix;
        if (key.equals(xmlnsKey)) {
            prefix = "";
        } else if (key.startsWith(xmlnsPrefix)) {
            prefix = key.substring(xmlnsPrefix.length());
        } else {
            continue;
        }
        namespaces.put(prefix, attr.getValue());
    }

    // get the element prefix if any
    int pos = el.tagName().indexOf(":");
    return pos > 0 ? el.tagName().substring(0, pos) : "";
}
 

/**
 * Test if the supplied attribute is allowed by this whitelist for this tag
 * @param tagName tag to consider allowing the attribute in
 * @param el element under test, to confirm protocol
 * @param attr attribute under test
 * @return true if allowed
 */
boolean isSafeAttribute(String tagName, Element el, Attribute attr) {
    TagName tag = TagName.valueOf(tagName);
    AttributeKey key = AttributeKey.valueOf(attr.getKey());

    if (attributes.containsKey(tag)) {
        if (attributes.get(tag).contains(key)) {
            if (protocols.containsKey(tag)) {
                Map<AttributeKey, Set<Protocol>> attrProts = protocols.get(tag);
                // ok if not defined protocol; otherwise test
                return !attrProts.containsKey(key) || testValidProtocol(el, attr, attrProts.get(key));
            } else { // attribute found, no protocols defined, so OK
                return true;
            }
        }
    }
    // no attributes defined for tag, try :all tag
    return !tagName.equals(":all") && isSafeAttribute(":all", el, attr);
}
 

private boolean testValidProtocol(Element el, Attribute attr, Set<Protocol> protocols) {
    // try to resolve relative urls to abs, and optionally update the attribute so output html has abs.
    // rels without a baseuri get removed
    String value = el.absUrl(attr.getKey());
    if (value.length() == 0)
        value = attr.getValue(); // if it could not be made abs, run as-is to allow custom unknown protocols
    if (!preserveRelativeLinks)
        attr.setValue(value);
    
    for (Protocol protocol : protocols) {
        String prot = protocol.toString() + ":";
        if (value.toLowerCase().startsWith(prot)) {
            return true;
        }
    }
    return false;
}
 

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();
    }
}
 

@Override
public boolean isAttribute(Object object) {
    if (object instanceof Attribute) {
        return true;
    } else {
        return false;
    }
}
 

@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();
}
 

@Override
public Attr getAttributeNode(String name) {
    if (element.attr(name) == null) {
        return null;
    }
    return NodeAdaptors.getAttr(new Attribute(name, element.attr(name)), element);
}
 

public AttributesAdaptor(Attributes attributes, Element element) {
    this.attributes = attributes;
    this.element = element;
    attrList = new ArrayList<Attr>();
    for (Attribute attribute : attributes) {
        attrList.add(new AttributeAdaptor(attribute,element));
    }
}
 

private void setAttribute(Attribute attribute) {
    String name = attribute.getKey();
    String value = attribute.getValue();
    if (value == null) {
        value = "";
    }
    getElement().setAttribute(name, value);
}
 

public void visit(DocumentType node) {
	Ast.Element.Builder b = Ast.Element.newBuilder();
	b.setKind(ElementKind.DOC_TYPE);
	b.setTag("!DOCTYPE");
	for (Attribute n : node.attributes()) {
		visit(n);
		b.addAttributes(attributes.pop());
	}
	elements.peek().add(b.build());
}
 

private boolean testValidProtocol(Element el, Attribute attr, Set<Protocol> protocols) {
    // try to resolve relative urls to abs, and optionally update the attribute so output html has abs.
    // rels without a baseuri get removed
    String value = el.absUrl(attr.getKey());
    if (value.length() == 0)
        value = attr.getValue(); // if it could not be made abs, run as-is to allow custom unknown protocols
    if (!preserveRelativeLinks)
        attr.setValue(value);
    
    for (Protocol protocol : protocols) {
        String prot = protocol.toString();

        if (prot.equals("#")) { // allows anchor links
            if (isValidAnchor(value)) {
                return true;
            } else {
                continue;
            }
        }

        prot += ":";

        if (value.toLowerCase().startsWith(prot)) {
            return true;
        }
    }
    return false;
}
 

/**
 * Test if the supplied attribute is allowed by this whitelist for this tag
 * @param tagName tag to consider allowing the attribute in
 * @param el element under test, to confirm protocol
 * @param attr attribute under test
 * @return true if allowed
 */
protected boolean isSafeAttribute(String tagName, Element el, Attribute attr) {
    TagName tag = TagName.valueOf(tagName);
    AttributeKey key = AttributeKey.valueOf(attr.getKey());

    Set<AttributeKey> okSet = attributes.get(tag);
    if (okSet != null && okSet.contains(key)) {
        if (protocols.containsKey(tag)) {
            Map<AttributeKey, Set<Protocol>> attrProts = protocols.get(tag);
            // ok if not defined protocol; otherwise test
            return !attrProts.containsKey(key) || testValidProtocol(el, attr, attrProts.get(key));
        } else { // attribute found, no protocols defined, so OK
            return true;
        }
    }
    // might be an enforced attribute?
    Map<AttributeKey, AttributeValue> enforcedSet = enforcedAttributes.get(tag);
    if (enforcedSet != null) {
        Attributes expect = getEnforcedAttributes(tagName);
        String attrKey = attr.getKey();
        if (expect.hasKeyIgnoreCase(attrKey)) {
            return expect.getIgnoreCase(attrKey).equals(attr.getValue());
        }
    }
    // no attributes defined for tag, try :all tag
    return !tagName.equals(":all") && isSafeAttribute(":all", el, attr);
}
 

private boolean testValidProtocol(Element el, Attribute attr, Set<Protocol> protocols) {
    // try to resolve relative urls to abs, and optionally update the attribute so output html has abs.
    // rels without a baseuri get removed
    String value = el.absUrl(attr.getKey());
    if (value.length() == 0)
        value = attr.getValue(); // if it could not be made abs, run as-is to allow custom unknown protocols
    if (!preserveRelativeLinks)
        attr.setValue(value);
    
    for (Protocol protocol : protocols) {
        String prot = protocol.toString();

        if (prot.equals("#")) { // allows anchor links
            if (isValidAnchor(value)) {
                return true;
            } else {
                continue;
            }
        }

        prot += ":";

        if (lowerCase(value).startsWith(prot)) {
            return true;
        }
    }
    return false;
}
 

private void copyAttributes(org.jsoup.nodes.Node source, Element el) {
    for (Attribute attribute : source.attributes()) {
        // valid xml attribute names are: ^[a-zA-Z_:][-a-zA-Z0-9_:.]
        String key = attribute.getKey().replaceAll("[^-a-zA-Z0-9_:.]", "");
        if (key.matches("[a-zA-Z_:][-a-zA-Z0-9_:.]*"))
            el.setAttribute(key, attribute.getValue());
    }
}
 

/**
 * Test if the supplied attribute is allowed by this whitelist for this tag
 * @param tagName tag to consider allowing the attribute in
 * @param el element under test, to confirm protocol
 * @param attr attribute under test
 * @return true if allowed
 */
protected boolean isSafeAttribute(String tagName, Element el, Attribute attr) {
    TagName tag = TagName.valueOf(tagName);
    AttributeKey key = AttributeKey.valueOf(attr.getKey());

    Set<AttributeKey> okSet = attributes.get(tag);
    if (okSet != null && okSet.contains(key)) {
        if (protocols.containsKey(tag)) {
            Map<AttributeKey, Set<Protocol>> attrProts = protocols.get(tag);
            // ok if not defined protocol; otherwise test
            return !attrProts.containsKey(key) || testValidProtocol(el, attr, attrProts.get(key));
        } else { // attribute found, no protocols defined, so OK
            return true;
        }
    }
    // might be an enforced attribute?
    Map<AttributeKey, AttributeValue> enforcedSet = enforcedAttributes.get(tag);
    if (enforcedSet != null) {
        Attributes expect = getEnforcedAttributes(tagName);
        String attrKey = attr.getKey();
        if (expect.hasKeyIgnoreCase(attrKey)) {
            return expect.getIgnoreCase(attrKey).equals(attr.getValue());
        }
    }
    // no attributes defined for tag, try :all tag
    return !tagName.equals(":all") && isSafeAttribute(":all", el, attr);
}
 

private boolean testValidProtocol(Element el, Attribute attr, Set<Protocol> protocols) {
    // try to resolve relative urls to abs, and optionally update the attribute so output html has abs.
    // rels without a baseuri get removed
    String value = el.absUrl(attr.getKey());
    if (value.length() == 0)
        value = attr.getValue(); // if it could not be made abs, run as-is to allow custom unknown protocols
    if (!preserveRelativeLinks)
        attr.setValue(value);
    
    for (Protocol protocol : protocols) {
        String prot = protocol.toString();

        if (prot.equals("#")) { // allows anchor links
            if (isValidAnchor(value)) {
                return true;
            } else {
                continue;
            }
        }

        prot += ":";

        if (lowerCase(value).startsWith(prot)) {
            return true;
        }
    }
    return false;
}
 

private void copyAttributes(org.jsoup.nodes.Node source, Element el) {
    for (Attribute attribute : source.attributes()) {
        // valid xml attribute names are: ^[a-zA-Z_:][-a-zA-Z0-9_:.]
        String key = attribute.getKey().replaceAll("[^-a-zA-Z0-9_:.]", "");
        if (key.matches("[a-zA-Z_:][-a-zA-Z0-9_:.]*"))
            el.setAttribute(key, attribute.getValue());
    }
}