Stop Xerces to resolve Entities

Question

Hi,
1) Question:
============
I've got the following problem. In one of my swing-projects i need the possiblity to edit XML-Code. Now when reading in the XML-File Xerces expands ENTITITES like & and in my Swing-App only & is shown. I have not found a property or something like this to prevent Xerces from expanding the entities. Does anybody know how to do this?
2) Question:
============
At the moment I have solved the problem from above by using the xml-Parser which comes with java-1.4. It does not expand Entities when setting
-----------------cut----------------------
factory.setExpandEntityReferences( false );
-----------------cut----------------------
Now there's the problem. For some of the XML-Files I need Schema-Validation which could only be done by Xerces (am I right?). Still at the moment when I add the xercesImpl.jar to my classpath the sun-parser does not resolve entities any more. Does anybody know what's going on there?
thx
tom

Answer 1

XML does not allow & by itself to appear in a text node. The rule is that it must be escaped, i.e. you must put & when you want that ampersand character. That rule, however, only specifies how the ampersand character must appear when it is in an XML file. When it is extracted from the XML file by software, Xerces or other, it will be treated as an ampersand.
So that's the rule. Presumably what you said about "expanding entities" was a confusion of that rule, but I couldn't really tell.
Anyway, what your Swing app should display for the ampersand character depends on what it is designed to display. If you are just putting raw XML into a text box and allowing the user to edit it, then you should put valid XML there, and that means the escaped version (&). But I wouldn't recommend that approach, because humans are likely to violate XML rules, like the escaping rule for example and many others. You are likely to end up with malformed XML if you do that.
However, if you have something like a tree structure with an element or an attribute at each node, then you should certainly display the unescaped version (&). This is what you will get when you extract a text node from a DOM anyway, and it's a correct representation of what is actually in the XML file from the user's point of view.
That's how you should treat the ampersand character, in my view. But the escaped ampersand character is normally not considered to be an entity. If you are declaring it as an entity in your schema, perhaps you should not do that as it seems to be leading to a lot of confusion.