Hmm, I have to admit that I don't understand what exactly you're talking about, because I have no idea what "OEM charset" is supposed to be.
Anyway, lets just say you have your sources in some strange, evil charset.
1. Do you think it will help if the editor alone understands your charset? If you don't translate them alltogether, the compiler will not be able to read your sources either, so it will be pretty useless.
2. Have you tried gnu recode? I have no clue about "OEM", but recode reads and writes just about everything that exists.
3. Have you considered writing a plugin that reads in all files from a project, runs the text through OEMToChar, and saves the text again? This would be a lot easier and more straightforward than modifying the editor and would solve the problem with the compiler, too.