@Zetmolm - of course, it was you that requested fb2 and Vaessa that requested txt... I just looked back at what I wrote about this previously
here. It seems I had ideas back then which I subsequently forgot about
I can't see a major downside to what I proposed (other than it being possibly slightly slower for mobi only users) as it looks similar to what I just stated above. Basically completely drop using the APNX algorithm for mobi books. Instead always convert to ePub format (if not already ePub) and use my existing algorithm for that. Which format it chooses to convert to ePub comes from your preferred input format order. So it should additionally cope with fb2, txt, lit or whatever - heck even pdf if you were silly enough to try it and that was your only or preferred format (we all know how slow and badly pdf -> epub conversions work out).
I shall take a look hopefully this weekend.