Refactoring Challenge: Amazing

Refactoring Challenge

I created this exercise for myself to push the limits of my refactoring abilities. I put it online because I enjoyed the exercise, and hope that others are similarly masochistic. It increased my confidence that refactoring can fix any code, no matter how bad.

First, I needed a prime example of an incomprehensible knot of spaghetti code, not too large but not too small. Shouldn't be hard to find. Then I would refactor mercilessly and completely untangle it. The idea is similar to the "Obfuscated C++" column from C++ Report, except that instead of trying to comprehend the code, I would try to make it comprehensible.

While looking for such an example, a childhood memory came back to me. About 20 years ago, when I was 13 or 14, I was mystified by a program in a book of BASIC programs. It was 138 lines of mostly GOTO statements, but it reliably drew mazes, with one exit and one way through. At the time, I was mightily impressed. Now, as a professional programmer, I was appalled, and decided to clear up the childhood mystery once and for all.

The program is listed on the right. If you're afraid I might have made a typo, you can check the original scan of the page. There is at least one actual bug in the code!

I wrote my solution in Java. (Don't cheat, but here it is for comparison). It took me about 6 or 7 hours. The first challenge is figuring out how to translate a zillion GOTO statements into a language that doesn't have a working GOTO statement. That is left as an exercise to the reader.

I don't feel I refactored it all the way. I only took it far enough to understand it. (And I fixed at least one bug). While I was refactoring, I kept feeling that I should be adding unit tests, because each refactoring carries some risk of screwing up. But I couldn't think of any good tests. It could be that I am not familiar enough with adding unit tests after the fact. It seems to me that testable units appeared only at the end, when the code took on a level of comprehensibility that I was satisfied with.

Notes on TRS-80 Model I BASIC:

CLS means "clear screen" — literally, clear the screen. I didn't bother to translate the details of entering maze dimensions.
RND is the random number generator. RND(2) is either 1 or 2, RND(3) is 1, 2, or 3, and RND(0) is a floating point number between 0 and 1.
An empty FOR loop was often used to pause the computer! Pinning the CPU was okay when there wasn't any such thing as multitasking.
DIM declares the dimensions of an array.
A semicolon at the end of a PRINT statement means don't automatically add a carriage return.

10 CLS: PRINT @ 412,"AMAZING" 20 PRINT @ 538,"COPYRIGHT BY" 30 PRINT @ 587,"CREATIVE COMPUTING MORRISTOWN, NEW JERSEY" 40 FOR II=1 TO 1500: NEXT 50 CLS: INPUT"WHAT ARE YOUR WIDTH AND LENGTH"; H,V 60 IF H<>1 AND V<>1 THEN 90 70 PRINT "MEANINGLESS DIMENSIONS. TRY AGAIN" 80 FOR A=1 TO 500: NEXT A: GOTO 50 90 PRINT @ 522,"PRINTOUT IS IN PROGRESS, PLEASE BE PATIENT" 100 FOR II=1 TO 1500: NEXT 110 DIM W(H,V), V(H,V) 120 Q=0: Z=0: X=RND(H) 130 FOR I=1 TO H 140 IF I=X THEN 160 150 LPRINT ":--";: GOTO 170 160 LPRINT ": "; 170 NEXT I 180 LPRINT ":" 190 C=1: W(X,1)=C: C=C+1 200 R=X: S=1: GOTO 270 210 IF R<>H THEN 250 220 IF S<>V THEN 240 230 R=1: S=1: GOTO 260 240 R=1: S=S+1: GOTO 260 250 R=R+1 260 IF W(R,S)=0 THEN 210 270 IF R-1=0 THEN 600 280 IF W(R-1,S) <>0 THEN 600 290 IF S-1=0 THEN 430 300 IF W(R,S-1)<>0 THEN 430 310 IF R=H THEN 350 320 IF W(R+1,S)<>0 THEN 350 330 X=RND(3) 340 ON X GOTO 940,980,1020 350 IF S<>V THEN 380 360 IF Z=1 THEN 410 370 Q=1: GOTO 390 380 IF W(R,S+1)<>0 THEN 410 390 X=RND(3) 400 ON X GOTO 940,980,1090 410 X=RND(2) 420 ON X GOTO 940,980 430 IF R=H THEN 530 440 IF W(R+1,S)<>0 THEN 530 450 IF S<>V THEN 480 460 IF Z=1 THEN 510 470 Q=1: GOTO 490 480 IF W(R,S+1)<>0 THEN 510 490 X=RND(3) 500 ON X GOTO 940,1020,1090 510 X=RND(2) 520 ON X GOTO 940,1020 530 IF S<>V THEN 560 540 IF Z=1 THEN 590 550 Q=1: GOTO 570 560 IF W(R,S+1)<>0 THEN 590 570 X=INT(RND(0)*2+1) 580 ON X GOTO 940,1090 590 GOTO 940 600 IF S-1=0 THEN 790 610 IF W(R,S-1)<>0 THEN 790 620 IF R=H THEN 720 630 IF W(R+1,S)<>0 THEN 720 640 IF S<>V THEN 670 650 IF Z=1 THEN 700 660 Q=1: GOTO 680 670 IF W(R,S+1)<>0 THEN 700 680 X=RND(3) 690 ON X GOTO 980,1020,1090 700 X=RND(2) 710 ON X GOTO 980,1020 720 IF S<>V THEN 750 730 IF Z=1 THEN 780 740 Q=1: GOTO 760 750 IF W(R,S+1)<>0 THEN 780 760 X=RND(2) 770 ON X GOTO 980,1090 780 GOTO 980 790 IF R=H THEN 880 800 IF W(R+1,S)<>0 THEN 880 810 IF S<>V THEN 840 820 IF Z=1 THEN 870 830 Q=1: GOTO 990 840 IF W(R,S+1)<>0 THEN 870 850 X=RND(2) 860 ON X GOTO 1020,1090 870 GOTO 1020 880 IF S<>V THEN 910 890 IF Z=1 THEN 930 900 Q=1: GOTO 920 910 IF W(R,S+1)<>0 THEN 930 920 GOTO 1090 930 GOTO 1190 940 W(R-1,S)=C 950 C=C+1: V(R-1,S)=2: R=R-1 960 IF C=H*V+1 THEN 1200 970 Q=0: GOTO 270 980 W(R,S-1)=C 990 C=C+1 1000 V(R,S-1)=1: S=S-1: IF C=H*V+1 THEN 1200 1010 Q=0: GOTO 270 1020 W(R+1,S)=C 1030 C=C+1: IF V(R,S)=0 THEN 1050 1040 V(R,S)=3: GOTO 1060 1050 V(R,S)=2 1060 R=R+1 1070 IF C=H*V+1 THEN 1200 1080 GOTO 600 1090 IF Q=1 THEN 1150 1100 W(R,S+1)=C: C=C+1:IF V(R,S)=0 THEN 1120 1110 V(R,S)=3: GOTO 1130 1120 V(R,S)=1 1130 S=S+1: IF C=V*H+1 THEN 1200 1140 GOTO 270 1150 Z=1 1160 IF V(R,S)=0 THEN 1180 1170 V(R,S)=3: Q=0: GOTO 1190 1180 V(R,S)=1: Q=0: R=1: S=1: GOTO 260 1190 GOTO 210 1200 FOR J=1 TO V 1210 LPRINT "I"; 1220 FOR I=1 TO H 1230 IF V(I,J)<2 THEN 1260 1240 LPRINT " "; 1250 GOTO 1270 1260 LPRINT " I"; 1270 NEXT I 1280 LPRINT " " 1290 FOR I=1 TO H 1300 IF V(I,J)=0 THEN 1340 1310 IF V(I,J)=2 THEN 1340 1320 LPRINT ": "; 1330 GOTO 1350 1340 LPRINT ":--"; 1350 NEXT I 1360 LPRINT ":" 1370 NEXT J 1380 END