Do something like this before passing your text to the XML parser?

s/&(\W|$)/&$1/g;
s/<([^/\w]|$)/&lt;$1/g;
s/(^|\W)>/$1&gt;/g;
[download]

But I think Re: Maximum parsing depth with XML::Parser? probably does a better job of this and implies the the greater-thans aren't a problem.

I recall a module like this for HTML. It would find common mistakes (like unquoted attributes) and fix them. Something like that would be even more useful as a module for XML since the spec says to reject invalid input.

        - tye (but my friends call me "Tye")