And yet he calls everybody a troll. I won't use regsubex because it's not much different from the while loop version you showed, consumes more bytes than my tokenize method, and not backward compatible to any version below 6.17. Besides regsubex and while loop, you got any novel idea in mind? I suppose not.