This is an interesting starting point for a python implementation, but it doesn't work correctly as far as I can tell. The list of tokens leaves out zero and includes a bunch of ascii characters that aren't valid hex. That second problem won't produce wrong results, but it'll spend a lot of time testing combinations that can't possibly appear in a SHA256 hex output. I also think the original tweet calculated without a newline.

More significantly, permutation in python is probably not the right tool for this. I believe it will not repeat token values at different positions, so for example you'll never end up testing "The SHA256 for this sentence begins with: f, f, and f", which is certainly a potentially valid result.