Query on Non-Alphameric Characters in Mapped URLs

I am unclear on how Google wants non-alphabetic/non-numeric characters in a URL treated in an XML sitemap. Reading what they say in their FAQs and such does not, for me, clarify anything.

It's hard to give an example, because I don't know how any odd character I might enter in this post would appear to someone else. Well, let's try; suppose the URL is something like--

http://mywonderfulsite.com/franCais/purE/good&bad.html

--where the upper-case C I show is really a c-cedilla character, and the upper-case E is really an e-acute (and don't overlook the ampersand in the filename).

Now I myself would never create such a nightmare, but if I am trying to make, say, a PHP script that handles mapping others' sites, how in heaven should it treat such stuff?

I at least gather that the ampersand wants to become the usual & substitute; and the same for quotation marks (single or double) and angle brackets.

Much seems to turn on their phrase "for hosting on a server that uses that encoding"--but how does one determine what encoding a server uses? The PHP environment variable "HTTP_ACCEPT_CHARSET" can--on my own server does--include both UTF-8 and ISO-8859-1 as acceptable. If that variable contains both, can one use either?

Or does one, then, simply URL-escape the filename as one reads it off the local file system? Suppose the filename contains a character, such as an ampersand, that is supposed to be escaped? Should--

some good & some bad.html

--be rendered as

some%20good%20&%20some%20bad.html

--or should the ampersand in the escaped entity also be URL-encoded?

Yechhh....