Just a thought. Is the output unique for every state in the game?
If not ( I suspect it isn't - ie you will be in many situations which return "You are in a room with a window, there is one door to the North and one door to the South" coresponding to completely different game states.) then a test suite mapping input to output doesn't prove your implementation in quite the same way. My approach might be to write a function in the C, Perl, and Lisp implementations that dumps the game state in a standard format. Use that in your tests. When you have proved that the same state sequence solves the game, then add your output - the same for every implementation. Interesting to read Knuths take on Adventure! BOL
in reply to Testing Zork-style game