empty xml element is changed in the output while using dom4j-1.6.jar

[asfak mahamud]

The following method I have used to write a dom4j Document to the output.

 

/**
	 * @param file
	 * @param inputDocument
	 * @throws FileNotFoundException
	 * @throws UnsupportedEncodingException
	 * @throws IOException
	 */
	public void writeDocument(File file, Document inputDocument, String headerText)
       throws UnsupportedEncodingException, IOException {
		XMLWriter writer = new XMLWriter(new OutputStreamWriter(new FileOutputStream(file),"UTF-8"));
		writer.setEscapeText(false);
		writer.write(headerText);
		writer.write(inputDocument.getRootElement());
		writer.flush();
		writer.close();
	}

Now my problem is :

The element <abc></abc> in the in the Document is changed to <abc/> in the output file.

I want

1. <abc/> will be output as <abc/>
2. <abc></abc> will be output as <abc></abc>

i.e. I want the output must be as it is in the input.

I found the XMLWriter.java has a method

    
protected void writeEmptyElementClose(String qualifiedName)
            throws IOException {
        // Simply close up
        if (!format.isExpandEmptyElements()) {
            writer.write("/>");
        } else {
            writer.write("></");
            writer.write(qualifiedName);
            writer.write(">");
        }
    }

is responsible for this problem.

But I can not figure out what will be the fix. I need an urgent solution of this problem.

Regards.

[William Brogden]

As far as XML correctness is concerned those two forms for an empty element are identical.

All the XML parsers I know of will treat them the same, so what is your problem?

Bill

[asfak mahamud]

Thanks for your reply. I have a program that modifies the source xml files in some specific positions. My clients want the output xml file should be changed only in those specific portions.

So when the output file changes the elements out of those specific portions my clients get confused. They do not want this unwanted changes.


Is there any fix?

Regards.

[William Brogden]

The only "fix" I can think of would be to forget treating the input and output as XML and treat it as text.

A bit of a pain to program but it would ensure that <tag/> and <tag></tag> and all the various formatting crlfs and tabs that people don't think about when looking at an XML document don't get changed.

Bill

[Paul Clapham]

Well, it looks like the XMLWriter might have an option to output empty elements in the short form or in the long form, based on that code you posted. But it can't possibly have an option "output in the same form as the original document" because the XMLWriter can't know what the form in the original document was. That's because the parser doesn't keep track of that.

And that's because it doesn't matter. If your customer thinks it matters then they are making a mistake and you should educate them. It doesn't matter.

[asfak mahamud]

Thanks for your reply.

I have solved the problem almost like the way William Brogden has said.

1. At first I have taken the xml file and done a "find and replace operation" to all the "></" places with "a long special text" that must not occur in the xml file using apache common io.
2. Then I did my usual task and got the output file.
3. Then I did find and replace operation again upon both input and output file. This time I only replaced my "a long special text" with an empty string.

That's it.


Regards.
>

[William Brogden]

Ha! That is a rather elegant solution, thanks for reporting it.

Bill