How to remove #text elements (white spaces) from a org.w3c.dom.Element

[Max J. Power]

Hello all,

I have a problem with the amount of child elements of a certain element. The goal is to count all child elements, but I'm getting also the Text elements because of the white spaces after the nodes:

<some_node>
<node1>text1</node1>
<node2>text1</node2>
<node3>text1</node3>
</some_node>

So for "someNode.getChildNodes().getLength()" I will get 7 instead of three.

What is the best practice here? I tried a loop with a counter which will be increased if an nodeof type Element will be found, which works, but aint there better ways? I also dont want to use XSLT as I will have then to maintain two different technologies.
Would it be better to remove the white spaces at the beginning of the code and then work with a "correct" element? If yes, what is the best way here?

Thank you!

Max J. Power

[Max J. Power]

I think I found a solution. The following method will remove the white spaces from a Dom Node (a Node can be an Element or a Document):

	/**
	 * This method removes all white spaces between >  < elements for a node
	 * @param incomingNode
	 * @return
	 */
	public static Node removeWhiteSpacesFromNode(Node incomingNode) {
		try {
			//Transform to byte[]
			byte[] nodeBytes = Helper.xmlNodeToByteArray(incomingNode);
			//Create a new String
			String nodeString = new String(nodeBytes, "UTF-8");
			nodeString = nodeString.replaceAll(">\\s*<", "><");
			incomingNode = Helper.byteArrayToXmlNode(nodeString.getBytes());
		}catch(Exception e) {
			e.printStackTrace();
		}
		return incomingNode;
	}

After this transformation, the Node.getChildNodes().getLength() method returns the right value.

[Thomas Rochon]

Max, you are my hero ;)

[William Brogden]

If your only need it to count Elements, your original loop approach with a test for Node type will certainly be faster since it does not require any additional object creation.

Bill

[g tsuji]

If those text nodes are that much hated, rather than using some helper to convert the element and byto array back-and-forth (which I find "ridiculous", sorry!), you can do that loop serving both purposes of cleaning them out once for all thereafter and counting the element nodes as child of it.

int count_elem=0;
//given the "someNode"
NodeList nlist=someNode.getChildNodes();
for (int i=nlist.getLength()-1; i>-1; i--) {
    Node node=nlist.item(i);
    if (node.getNodeType()==Node.ELEMENT_NODE) {
        count_elem++;
    }
    if (node.getNodeType()==Node.TEXT_NODE) {
        //if you are not 100% it contains only ignorable whitespace, add some testing here
        node.getParentNode().removeChild(node);
    }
}

From there onward, the someNode will contain only element node as its child.

For this matter, I just want to point out in the DocumentBuilderFactory, there is a settable feature on IgnoringElementContentWhitespace. Unfortunately, there is some divided interpretation on the recommendations or some grossly overlook in the building of jdk 1.6, there is a regression of bug/feature in 1.6 to before 1.5. But, that's how it is for the moment.