Normally CDATA sections are just an easy way to input text that might have special characters in it. For an XML processors there is absolutely no difference between a CDATA section and the same data encoded using < and its friends.

At least that's how things should be in a ideal world.

But yes, in this less-than-perfect world, you can use XML::Twig. It will preserve CDATA sections, it can also preserve attribute order (some software is known to care about that. use the keep_atts_order option) and numerical entities (use the keep_encoding option, with caution). It's not quite as simple to use as XML::Simple, but it should be easy enough to learn for the task you describe.