Constructing a User-Friendly JTree from a DOM

The Java^TM Web Services Tutorial

Constructing a User-Friendly JTree from a DOM

Now that you know what a DOM looks like internally, you'll be better prepared to modify a DOM or construct one from scratch. Before going on to that, though, this section presents some modifications to the JTreeModel that let you produce a more user-friendly version of the JTree suitable for use in a GUI.

Compressing the Tree View

Displaying the DOM in tree form is all very well for experimenting and to learn how a DOM works. But it's not the kind of "friendly" display that most users want to see in a JTree. However, it turns out that very few modifications are needed to turn the TreeModel adapter into something that will present a user-friendly display. In this section, you'll make those modifications.

Note: The code discussed in this section is in DomEcho03.java. The file it operates on is slideSample01.xml. (The browsable version is slideSample01-xml.html.)

Make the Operation Selectable

When you modify the adapter, you're going to compress the view of the DOM, eliminating all but the nodes you really want to display. Start by defining a boolean variable that controls whether you want the compressed or uncompressed view of the DOM:
public class DomEcho extends JPanel
{
  static Document document; 
   boolean compress = true;
   static final int windowHeight = 460;
   ...
 
Identify Tree Nodes

The next step is to identify the nodes you want to show up in the tree. To do that, add the code highlighted below:
...
import org.w3c.dom.Document;
import org.w3c.dom.DOMException;
import org.w3c.dom.Node;

public class DomEcho extends JPanel
{
  ...

  public static void makeFrame() {
    ...
  }

  // An array of names for DOM node-type
  static final String[] typeName = {
    ...
  };

  static final int ELEMENT_TYPE = Node.ELEMENT_NODE;

  // The list of elements to display in the tree
  static String[] treeElementNames = {
    "slideshow",
    "slide",
    "title",         // For slideshow #1
    "slide-title",   // For slideshow #10
    "item",
  };

  boolean treeElement(String elementName) {
    for (int i=0; i<treeElementNames.length; i++) {
      if ( elementName.equals(treeElementNames[i]) )
        return true;
    }
    return false;
  } 
 
With this code, you set up a constant you can use to identify the ELEMENT node type, declared the names of the elements you want in the tree, and created a method tells whether or not a given element name is a "tree element". Since slideSample01.xml has title elements and slideSample10.xml has slide-title elements, you set up the contents of this arrays so it would work with either data file.

Note: The mechanism you are creating here depends on the fact that structure nodes like slideshow and slide never contain text, while text usually does appear in content nodes like item. Although those "content" nodes may contain subelements in slideShow10.xml, the DTD constrains those subelements to be XHTML nodes. Because they are XHTML nodes (an XML version of HTML that is constrained to be well-formed), the entire substructure under an item node can be combined into a single string and displayed in the htmlPane that makes up the other half of the application window. In the second part of this section, you'll do that concatenation, displaying the text and XHTML as content in the htmlPane.

Although you could simply reference the node types defined in the class, org.w3c.dom.Node, defining the ELEMENT_TYPE constant keeps the code a little more readable. Each node in the DOM has a name, a type, and (potentially) a list of subnodes. The functions that return these values are getNodeName(), getNodeType, and getChildNodes(). Defining our own constants will let us write code like this:
Node node = nodeList.item(i);
int type = node.getNodeType();
if (type == ELEMENT_TYPE) {
  ....
 
As a stylistic choice, the extra constants help us keep the reader (and ourselves!) clear about what we're doing. Here, it is fairly clear when we are dealing with a node object, and when we are dealing with a type constant. Otherwise, it would be fairly tempting to code something like, if (node == ELEMENT_NODE), which of course would not work at all.

Control Node Visibility

The next step is to modify the AdapterNode's childCount function so that it only counts "tree element" nodes--nodes which are designated as displayable in the JTree. Make the modifications highlighted below to do that:
public class DomEcho extends JPanel
{
  ...
  public class AdapterNode 
  { 
    ...
    public AdapterNode child(int searchIndex) {
      ... 
    }
    public int childCount() {
      if (!compress) {
        // Indent this
        return domNode.getChildNodes().getLength(); 
      } 
      int count = 0;
      for (int i=0;
        i<domNode.getChildNodes().getLength(); i++) 
      {
        org.w3c.dom.Node node =
          domNode.getChildNodes().item(i); 
        if (node.getNodeType() == ELEMENT_TYPE
        && treeElement( node.getNodeName() )) 
        {
          ++count;
        }
      }
      return count;
    }
  } // AdapterNode
 
The only tricky part about this code is checking to make sure the node is an element node before comparing the node. The DocType node makes that necessary, because it has the same name, "slideshow", as the slideshow element.

Control Child Access

Finally, you need to modify the AdapterNode's child function to return the Nth item from the list of displayable nodes, rather than the Nth item from all nodes in the list. Add the code highlighted below to do that:
public class DomEcho extends JPanel
{
  ...
  public class AdapterNode 
  { 
    ...
    public int index(AdapterNode child) {
      ...
    }
    public AdapterNode child(int searchIndex) {
    //Note: JTree index is zero-based. 
    org.w3c.dom.Node node =
      domNode.getChildNodes()Item(searchIndex);
    if (compress) {
      // Return Nth displayable node
      int elementNodeIndex = 0;
      for (int i=0;
        i<domNode.getChildNodes().getLength(); i++) 
      {
        node = domNode.getChildNodes()Item(i);
        if (node.getNodeType() == ELEMENT_TYPE 
        && treeElement( node.getNodeName() )
        && elementNodeIndex++ == searchIndex) {
          break; 
        }
      }
    }
    return new AdapterNode(node); 
  } // child
}  // AdapterNode
 
There's nothing special going on here. It's a slightly modified version the same logic you used when returning the child count.

Check the Results

When you compile and run this version of the application on slideSample01.xml, and then expand the nodes in the tree, you see the results shown in Figure 8. The only nodes remaining in the tree are the high-level "structure" nodes.

Figure 8 Tree View with a Collapsed Hierarchy

Extra Credit

The way the application stands now, the information that tells the application how to compress the tree for display is "hard-coded". Here are some ways you could consider extending the app:

Use a Command-Line Argument

Whether you compress or don't compress the tree could be determined by a command line argument, rather than being a hard-coded boolean variable. On the other hand, the list the list of elements that goes into the tree is still hard coded, so maybe that option doesn't make much sense, unless...

Read the treeElement list from a file

If you read the list of elements to include in the tree from an external file, that would make the whole application command driven. That would be good. But wouldn't it be really nice to derive that information from the DTD or schema, instead? So you might want to consider...

Automatically Build the List

Watch out, though! As things stand right now, there are no standard DTD parsers! If you use a DTD, then, you'll need to write your parser to make sense out of its somewhat arcane syntax. You'll probably have better luck if you use a schema, instead of a DTD. The nice thing about schemas is that use XML syntax, so you can use an XML parser to read the schema the same way you use any other file.

As you analyze the schema, note that the JTree-displayable structure nodes are those that have no text, while the content nodes may contain text and, optionally, XHTML subnodes. That distinction works for this example, and will likely work for a large body of real-world applications. It's pretty easy to construct cases that will create a problem, though, so you'll have to be on the lookout for schema/DTD specifications that embed non-XHTML elements in text-capable nodes, and take the appropriate action.

Acting on Tree Selections

Now that the tree is being displayed properly, the next step is to concatenate the subtrees under selected nodes to display them in the htmlPane. While you're at it, you'll use the concatenated text to put node-identifying information back in the JTree.

Note: The code discussed in this section is in DomEcho04.java.

Identify Node Types

When you concatenate the subnodes under an element, the processing you do is going to depend on the type of node. So the first thing to is to define constants for the remaining node types. Add the code highlighted below to do that:
public class DomEcho extends JPanel
{
  ...
  // An array of names for DOM node-types
  static final String[] typeName = {
    ...
  };
  static final int ELEMENT_TYPE =   1;
  static final int ATTR_TYPE =            Node.ATTRIBUTE_NODE;
  static final int TEXT_TYPE =            Node.TEXT_NODE;
  static final int CDATA_TYPE =             Node.CDATA_SECTION_NODE;
  static final int ENTITYREF_TYPE =  
Node.ENTITY_REFERENCE_NODE;
  static final int ENTITY_TYPE =            Node.ENTITY_NODE;
  static final int PROCINSTR_TYPE =
             Node.PROCESSING_INSTRUCTION_NODE;
  static final int COMMENT_TYPE =   Node.COMMENT_NODE;
  static final int DOCUMENT_TYPE =  Node.DOCUMENT_NODE;
  static final int DOCTYPE_TYPE =            Node.DOCUMENT_TYPE_NODE;
  static final int DOCFRAG_TYPE =            
Node.DOCUMENT_FRAGMENT_NODE;
  static final int NOTATION_TYPE =            Node.NOTATION_NODE;
 
Concatenate Subnodes to Define Element Content

Next, you need to define add the method that concatenates the text and subnodes for an element and returns it as the element's "content". To define the content method, you'll need to add the big chunk of code highlighted below, but this is the last big chunk of code in the DOM tutorial!.
public class DomEcho extends JPanel
{
  ...
  public class AdapterNode 
  { 
    ...
    public String toString() {
    ...
    }
    public String content() {
      String s = "";
      org.w3c.dom.NodeList nodeList =
        domNode.getChildNodes();
      for (int i=0; i<nodeList.getLength(); i++) {
        org.w3c.dom.Node node = nodeList.item(i);
        int type = node.getNodeType();
        AdapterNode adpNode = new AdapterNode(node);
        if (type == ELEMENT_TYPE) { 
          if ( treeElement(node.getNodeName()) )
            continue;
          s += "<" + node.getNodeName() + ">";
          s += adpNode.content();
          s += "</" + node.getNodeName() + ">";
        } else if (type == TEXT_TYPE) {
          s += node.getNodeValue();
        } else if (type == ENTITYREF_TYPE) {
          // The content is in the TEXT node under it
          s += adpNode.content();
        } else if (type == CDATA_TYPE) { 
          StringBuffer sb = new StringBuffer(
            node.getNodeValue() );
          for (int j=0; j<sb.length(); j++) {
            if (sb.charAt(j) == '<') {
              sb.setCharAt(j, '&');
              sb.insert(j+1, "lt;");
              j += 3;
            } else if (sb.charAt(j) == '&') {
              sb.setCharAt(j, '&');
              sb.insert(j+1, "amp;");
              j += 4;
            }
          }
          s += "<pre>" + sb + "</pre>";
        }
      }
      return s;
    }
    ...
} // AdapterNode
 
Note: This code collapses EntityRef nodes, as inserted by the JAXP 1.1 parser that ins included in the 1.4 Java platform. With JAXP 1.2, that portion of the code is not necessary because entity references are converted to text nodes by the parser. Other parsers may well insert such nodes, however, so including this code "future proofs" your application, should you use a different parser in the future.

Although this code is not the most efficient that anyone ever wrote, it works and it will do fine for our purposes. In this code, you are recognizing and dealing with the following data types:

Element

For elements with names like the XHTML "em" node, you return the node's content sandwiched between the appropriate <em> and </em> tags. However, when processing the content for the slideshow element, for example, you don't include tags for the slide elements it contains so, when returning a node's content, you skip any subelements that are themselves displayed in the tree.

Text

No surprise here. For a text node, you simply return the node's value.

Entity Reference

Unlike CDATA nodes, Entity References can contain multiple subelements. So the strategy here is to return the concatenation of those subelements.

CDATA

Like a text node, you return the node's value. However, since the text in this case may contain angle brackets and ampersands, you need to convert them to a form that displays properly in an HTML pane. Unlike the XML CDATA tag, the HTML <pre> tag does not prevent the parsing of character-format tags, break tags and the like. So you have to convert left-angle brackets (<) and ampersands (&) to get them to display properly.

On the other hand, there are quite a few node types you are not processing with the code above. It's worth a moment to examine them and understand why:

Attribute

These nodes do not appear in the DOM, but are obtained by invoking getAttributes on element nodes.

Entity

These nodes also do not appear in the DOM. They are obtained by invoking getEntities on DocType nodes.

Processing Instruction

These nodes don't contain displayable data.

Comment

Ditto. Nothing you want to display here.

Document

This is the root node for the DOM. There's no data to display for that.

DocType

The DocType node contains the DTD specification, with or without external pointers. It only appears under the root node, and has no data to display in the tree.

Document Fragment

This node is equivalent to a document node. It's a root node that the DOM specification intends for holding intermediate results during cut/paste operations, for example. Like a document node, there's no data to display.

Notation

We're just flat out ignoring this one. These nodes are used to include binary data in the DOM. As discussed earlier in Referencing Binary Entities and Using the DTDHandler and EntityResolver, the MIME types (in conjunction with namespaces) make a better mechanism for that.

Display the Content in the JTree

With the content-concatenation out of the way, only a few small programming steps remain. The first is to modify toString so that it uses the node's content for identifying information. Add the code highlighted below to do that:
public class DomEcho extends JPanel
{
  ...
  public class AdapterNode 
  { 
    ...
    public String toString() {
      ...
      if (! nodeName.startsWith("#")) {
        s += ": " + nodeName;
      }
      if (compress) {
        String t = content().trim();
        int x = t.indexOf(");
        if (x >= 0) t = t.substring(0, x);
        s += " " + t;
        return s;
      }
      if (domNode.getNodeValue() != null) {
        ...
      }
      return s;
    } 
 
Wire the JTree to the JEditorPane

Returning now to the app's constructor, create a tree selection listener and use to wire the JTree to the JEditorPane:
public class DomEcho extends JPanel
{
  ...
  public DomEcho()
  {
    ...
    // Build right-side view
    JEditorPane htmlPane = new JEditorPane("text/html","");
    htmlPane.setEditable(false);
    JScrollPane htmlView = new JScrollPane(htmlPane);
    htmlView.setPreferredSize( 
      new Dimension( rightWidth, windowHeight ));

      tree.addTreeSelectionListener(
        new TreeSelectionListener() {
          public void valueChanged(TreeSelectionEvent e)
          {
            TreePath p = e.getNewLeadSelectionPath();
            if (p != null) {
              AdapterNode adpNode = 
                (AdapterNode)
                  p.getLastPathComponent();
              htmlPane.setText(adpNode.content());
            }
          }
        }
      ); 
 
Now, when a JTree node is selected, it's contents are delivered to the htmlPane.

Note: The TreeSelectionListener in this example is created using an anonymous inner-class adapter. If you are programming for the 1.1 version of the platform, you'll need to define an external class for this purpose.

If you compile this version of the app, you'll discover immediately that the htmlPane needs to be specified as final to be referenced in an inner class, so add the keyword highlighted below:
public DomEcho04()
{
  ...
  // Build right-side view
  final JEditorPane htmlPane = new
    JEditorPane("text/html","");
  htmlPane.setEditable(false);
  JScrollPane htmlView = new JScrollPane(htmlPane);
  htmlView.setPreferredSize( 
    new Dimension( rightWidth, windowHeight ));
 
Run the App

When you compile the application and run it on slideSample10.xml (the browsable version is slideSample10-xml.html), you get a display like that shown in Figure 9. Expanding the hierarchy shows that the JTree now includes identifying text for a node whenever possible.

Figure 9   Collapsed Hierarchy Showing Text in Nodes

Selecting an item that includes XHTML subelements produces a display like that shown in Figure 10:

Figure 10   Node with <em> Tag Selected

Selecting a node that contains an entity reference causes the entity text to be included, as shown in Figure 11:

Figure 11   Node with Entity Reference Selected

Finally, selecting a node that includes a CDATA section produces results like those shown in Figure 12:

Figure 12   Node with CDATA Component Selected

Extra Credit

Now that you have the application working, here are some ways you might think about extending it in the future:

Use Title Text to Identify Slides

Special case the slide element so that the contents of the title node is used as the identifying text. When selected, convert the title node's contents to a centered H1 tag, and ignore the title element when constructing the tree.

Convert Item Elements to Lists

Remove item elements from the JTree and convert them to HTML lists using <ul>, <li>, </ul> tags, including them in the slide's content when the slide is selected.

Handling Modifications

A full discussion of the mechanisms for modifying the JTree's underlying data model is beyond the scope of this tutorial. However, a few words on the subject are in order.

Most importantly, note that if you allow the user to modifying the structure by manipulating the JTree, you have take the compression into account when you figure out where to apply the change. For example, if you are displaying text in the tree and the user modifies that, the changes would have to be applied to text subelements, and perhaps require a rearrangement of the XHTML subtree.

When you make those changes, you'll need to understand more about the interactions between a JTree, it's TreeModel, and an underlying data model. That subject is covered in depth in the Swing Connection article, Understanding the TreeModel at http://java.sun.com/products/jfc/tsc/articles/jtree/index.html.

Finishing Up

You now understand pretty much what there is know about the structure of a DOM, and you know how to adapt a DOM to create a user-friendly display in a JTree. It has taken quite a bit of coding, but in return you have obtained valuable tools for exposing a DOM's structure and a template for GUI apps. In the next section, you'll make a couple of minor modifications to the code that turn the application into a vehicle for experimentation, and then experiment with building and manipulating a DOM.

Home
TOC
Index

This tutorial contains information on the 1.0 version of the Java Web Services Developer Pack.

All of the material in The Java Web Services Tutorial is copyright-protected and may not be published in other works without express written permission from Sun Microsystems.