Okay did a bit more searching and couldnt find anything about the local links and when they are parsed. But I did realise that parsing them myself wouldn't be too difficult.

So before I did the regex i needed to on the rich text i parsed the local links as follows:

    public class MyClass
    {
        public static string MakeGlossaryLink(string richText)
        {
            // take a rich text string, parse the local links then look for glossary links and replace them with anchors

            Regex ll = new Regex("(/{localLink:)(.*)}");
            MyClass = new MyClass();

            MatchEvaluator myEvaluator = new MatchEvaluator(MatchLocalLinks);

            var LocalLinksParsed = ll.Replace(richText, myEvaluator);
            return  HttpUtility.HtmlDecode(Regex.Replace(LocalLinksParsed, "(.*/glossary)/(.*)", "$1.aspx#$2"));
        }

        private static string MatchLocalLinks(Match m)
        {
            var noHome = Regex.Replace(umbraco.library.NiceUrl(int.Parse(m.Result("$2"))), ".*/home(.*)", "$2");
            return noHome;
        }

    }

One not is that after the local links had been parsed i had to remove the first /home.

Hope this helps someone else.