Converting partial XML (Node List) to a String

Question

I have a need to extract a part of XML tree (everything under root) and convert it to a string. (The result string will be later pasted inside XSL-FO XSLT file.)

The above functionality is a part of Java 6 application. I use org.w3c.dom and javax.xml.transform. Still in the debugging process.

// ...
// Extract all child notes of root from foDoc
// and convert the result to String
Node root = foDoc.getElementsByTagNameNS("*","root").item(0);
NodeList body = root.getChildNodes();
String foResult;
try {
 foResult = nodeListToString(body);
} catch (TransformerException e) {
 logger.error(e);
 foResult = "";
}
// ...

private static String nodeListToString(NodeList nodes) throws TransformerException {
 StringBuilder result = new StringBuilder();
 int len = nodes.getLength();
 for(int i = 0; i < len; ++ i) {
 result.append(nodeToString(nodes.item(i)));
 }
 return result.toString();
}
private static String nodeToString(Node node) throws TransformerException {
 StringWriter buf = new StringWriter();
 Transformer xform = TransformerFactory.newInstance().newTransformer();
 xform.setOutputProperty(OutputKeys.OMIT_XML_DECLARATION, "yes");
 xform.transform(new DOMSource(node), new StreamResult(buf));
 return(buf.toString());
}

My questions:

1. Assuming that there will always be exactly one root node, how reliable is my code in extracting everything under it?

2. Am I missing any edge cases?

3. Anything just plain wrong?

Please disregard any exception handling inadequacies.

Edited on 10/06/2015

Since my originally posted code had a "just plain wrong" part, I've fixed it. There were no answers based on the old code.

• rolfl: How committed are you to using DOM? I seldom just jump ion and recommend alternative tools, but, I maintain the JDOM library, and this sort of activity is easy if you can use JDOM instead..... just saying.
• PM 77-1: @rolfl - Thank you for the reference. I will keep it in mind for future projects. This time I am making a small change to a plugin for a huge system and need to stay within its existing tooling.

Answer 1

Reading the documentation for Transformer, where it says:

A Transformer may be used multiple times. Parameters and output properties are preserved across transformations.

, I'd say that not only can the Transformer object be reused across multiple nodes, but you should also be able to use a single StringWriter instead of recreating it for every node; same goes for the DOMSource actually:

private static String nodeListToString(NodeList nodes) throws TransformerException {
 DOMSource source = new DOMSource();
 StringWriter writer = new StringWriter();
 StreamResult result = new StreamResult(writer);
 Transformer transformer = TransformerFactory.newInstance().newTransformer();
 transformer.setOutputProperty(OutputKeys.OMIT_XML_DECLARATION, "yes");
 for (int i = 0; i < nodes.getLength(); ++i) {
 source.setNode(nodes.item(i));
 transformer.transform(source, result);
 }
 return writer.toString();
}

That said, I'm only guessing that this should work.

Also, the len variable is probably going to be inlined automatically anyway, though you could check that with a disassembler just to be sure (it wouldn't make that much of a difference I suspect).

Otherwise looks good to me; if you've tested it with all kinds of node types and the output looks good, then I'd say that should work IMHO.

• PM 77-1: I use nodeToString() in other places of my code as well.
• ferada: Then keep it in addition to this nodeListToString? I'm just assuming that inlining and removing allocations helps with bigger documents.
• janos: I second that it looks strange to build a string in a StringWriter just so that it can be appended to a StringBuilder in the caller. It would be better to reuse the StringWriter. You could pass the StreamResult object as a 2nd parameter into nodeListToString