Removing elements from Swing HTMLDocument

Suppose you have an HTML document, like this:

<ul>
  <li id="LI1">Item 1</li>
  <li id="LI2">Item 2</li>
  <li id="LI3">Item 3</li>
</ul>
          

And you want to programmatically remove one of the list items (say, the second one).

For this, you load the document into Swing’s JEditorPane component:

JEditorPane p = new JEditorPane();
p.setContentType("text/html");
p.setText("HTML text from the above example");
    

As you know the HTML element’s ID, you can easily obtain a reference to the corresponding Element object:

HTMLDocument d = (HTMLDocument) p.getDocument();
Element e = d.getElement("LI2");
    

But how to remove this element from the document? A quick glance through the HTMLDocument API reveals the natural candidate for the task: the setOuterHTML method which allows to replace an arbitrary element with some new content. So you think, “if I specify an empty string as the new content, the element will be just removed from the document as there’s no replacement.” Pretty simple and elegant solution, really.

Except that it doesn’t work.

The setOuterHTML method is implemented so that if there’s no content in the replacement string, the target document won’t be altered at all. The method call is just a no-op in this case.

So how to actually remove an element from the document? Quite surprisingly, until recently there was no easy way of doing this.

Starting with JDK7 build 10, the Swing Text subsystem provides the new public method in the DefaultStyledDocument class: the removeElement method. It takes an element to remove as the sole parameter and removes the element from the document tree, as well as the corresponding text from the document content. So in order to remove the element, you just invoke this method:

d.removeElement(e);
    

Due to specifics of the default Element and Content implementations, there are some caveats:

• Empty branch elements are not allowed in the default implementation; so if you remove the last child of some element, the element itself will be removed as well, recursively (this means that if you need to replace the sole child, you should add the new child first and then remove the old one, not vice-versa).
• Element-less documents are not allowed in the default implementation; so you’ll get an exception if you try to remove the last leaf element in the document (because this will eventually require removal of the document root element).
• The default Content implementation requires the presence of the trailing newline character in the document; so if you remove the leaf element containing the trailing newline, it will be added to the preceding leaf element (unless that element already ends with newline).