Hey all,

I'm working on a web site using XSLT for the front end. Unfortunately, one non-Roman character slipped in (such as an e-acute) brings the whole thing down with "Undeclared Entity" errors. It's getting this error both in Sablotron (on the server) and loading the offending XML file directly into Firefox to display it.

Escaping the offending character (as é) doesn't do a whole lot of good since the XSLT parser decodes it internally and then chokes on the same error again. Double-escaping it (&eacute) works, but means I have to put 'disable-output-escaping="yes"' on every line of every XSLT file to display it properly, and code extra server-side software to keep "bad" things from getting unescaped too. But that just seems like way too much work!

So can XML not handle non-roman characters without going through all these steps? It seems very, very strange that XML would natively support UTF-8, yet claim any non-ASCII character to be malformed...

Any advice?

(I do start both the source XML files and the XSLT files with the usual '<?xml version="1.0" encoding="UTF-8"?>' declaration.)

Thanks!
Richard

If you want to use entity references such as & eacute; in XML, you have to declare them. You have a number of options: you can use numeric character references instead (for example & #xac;), or you can use the native character encoding of the character. Of course, you have to use the encoding you have declared in the XML declaration.

Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference

There lies my problem. The original character causing all these problems is the UTF-8 character itself, of an e with an acute accent mark over it (same character as &eacute;, but the original character.) And the <?xml declaration states UTF-8 as the character set.

In spite of the fact that both Sablotron and Firefox are reporting "undeclared entity" errors and refusing to render the page, there really is no XML entity. There's just a UTF-8 extended latin character that's nicely breaking things by being included.

Any ideas of what could be causing this?

Thanks!

Rich

You'll have to show some of the original code, if the files are utf-8 and you're using e-acute either typed directly or using a numeric charcter reference then it should work.
Can you post a sample of the original XML where the charcter appears, or post a link to prevent the forum's software messing it up?

--

Joe (Microsoft MVP - XML)