How do I prevent a pair of XMLEvent[Reader|Writer] from splitting empty tags into start+end tags?

I have an understanding problem of how the following code works:

XMLInputFactory xif = XMLInputFactory.newFactory();
XMLOutputFactory xof = XMLOutputFactory.newFactory();

XMLEventReader reader = xif.createXMLEventReader(/* ... */);
XMLEventWriter writer = xof.createXMLEventWriter(/* ... */);

writer.add(reader);

The method writer.add([some reader]) reads all events from reader and writes them consequently. Sadly, following happens:

The input

<root><c/></root>

gets transformed to

<root><c><c/></root>

I know, from XML point of view, these are equal trees, but not for a human ;)

What can I do to get the same output?

FYI: I need a XMLEvent[Reader|Writer] pair later to filter "XML events".

Solution

According to the list of XMLEvents it seems like there is no way to make the distinction and it will indeed generate a StartElement and EndElement event. The consumer would need to optimize for the case when a StartElement is immediately followed by an EndElement.

This is apparently not the case of the XMLEventReader returned by the factory. If you want to optimize this behavior yourself, I see no other way than to do something like

check what's the concrete XMLEventReader implementation returned by createXMLEventReader
subclass the XMLEventReader implementation to optimize this case
subclass XMLInputFactory and override createXMLEventReader to return an instance of your XMLEventReader subclass

If this sounds too complicated (or doesn't work), I would suggest that you go with a solution that uses XMLStreamWriter. This one has a dedicated method writeEmptyElement.

(Or you can give a try to my home-made pretty printer, it's based on XMLStreamWriter)