Question about changing un/escape behavior

[phdavis1027]

Hi all,

Me again! I've circled back to iterating on serialization and deserialization again, and I have a combo question/proposition for a PR. My project talks to a server which speaks an idiosyncratic dialect of XML in which, for example, proper unescape behavior is to map " to ` (backtick). Furthermore, this dialect of XML does not encode ' (single quote/apostrophe) , but it does escape " (double quote). As it is, the unescape method evaluates standard named entities before it evaluates custom entities, so I can't implement my unescape behavior. Furthermore, escape only offers partial_escape and escape, without offering a resolver. I am willing to throw together a PR, but since there are security issues involved I first wanted to poll for thoughts on the changes I'm proposing, which are:

• In unescape_with, reverse the order of the checks for resolver and named entities, allowing users to override standard XML unescapes.

• Create an escape_with family of functions which allow this kind of customization going the other way.

EDIT: Just to make clear, since the previous issue I raised here I have abandoned the serde API and am hand-parsing my structs use Reader