And then figure out how to convert an ascii art image that represents that knot into that internal format... and you do a "I'll do it later" which never comes around.
Yeah that seems pretty tricky. I'm somewhat convinced that the graph representation will work (it can represent the input correctly), but I'm not yet sure if it's actually useful or not for an algorithm to work through.
Consider the question of "what internal structure would you use to represent the trefoil knot?" - https://en.wikipedia.org/wiki/Trefoil_knot // https://mathworld.wolfram.com/TrefoilKnot.html
And then figure out how to convert an ascii art image that represents that knot into that internal format... and you do a "I'll do it later" which never comes around.