Sergey Groznyh's Blog

Removing elements from Swing HTML in JDK6

After I published the entry on removing elements from Swing HTMLDocument in JDK7, I got a question from a reader: but how to actually remove elements in JDK6? There's no easy way, but what is the hard way?

Well, there exist a bunch of methods, to various degrees of ugliness, let's look at one of them. It's neither pretty nor effective, but has the advantage of being simple enough.

The method is to serialize all the sibling elements using HTMLWriter class, concatenate the serialized strings, and re-parse the resulting text using setInnerHTML method with the parent element as the target.

The following code illustrates this approach.

void removeElement(HTMLDocument d, Element e) throws Exception {
    Element p = e.getParentElement();
    int n = p.getElementCount();
    String s = "";
    for (int i = 0; i < n; i++) {
        Element c = p.getElement(i);
        if (c != e) {
            s += getElementText(d, c);
        }
    }
    d.setInnerHTML(p, s);
}

The tricky thing is, how to serialize not the complete document but a single element only. There exist a constructor in the HTMLWriter class that allows to specify only a fragment of the document to serialize, and we could use element's start and end offsets as the fragment boundaries. But by default the writer will output all ancestors of the element up to the root to make serialized form the complete HTML document.

To overcome this, we need to override the getElementIterator method which returns the root of the element hierarchy to serialize. In the HTMLWriter implementation it returns the document root element. We'll implement it to return just the element to serialize.

String getElementText(HTMLDocument d, final Element e) throws Exception {
    StringWriter sw = new StringWriter();
    int p0 = e.getStartOffset();
    int p1 = e.getEndOffset();
    new HTMLWriter(sw, d, p0, p1 - p0) {
        protected ElementIterator getElementIterator() {
            return new ElementIterator(e);
        }
    }.write();
    return sw.toString();
}

Note that if we try to remove the sole child element, the behavior of this method differs from removeElement in JDK7. In JDK7 the parent element is removed as well, recursively. In the above example nothing is removed (to make the example simple). To get the JDK7 behavior, we need to traverse to the nearest parent element that is not a sole child and remove it instead of the original element.

Comments

This work is licensed under a Creative Commons License.