How about a base-N project? You configure it with your safe alphabet, it adapts to consume as many bits as it can for your 'encoding', outputs efficiency along

How about a base-N project? You configure it with your safe alphabet, it adapts to consume as many bits as it can for your 'encoding', outputs efficiency along with encoded data.