Smallest chess program

19

6

Inspired by this recent article telling of a French programmer who wrote a 487 byte 2-player (human vs. human) chess program in Assembly, I wondered how small chess programs could be in other languages.

Details

  • The program must only accept legal chess moves.
  • Declaration of check/checkmate is not necessary (just desirable), although the first rule applies in cases of check.
  • As per the article, castling is not a necessary implementation.
  • Also, you do not have to implement en passant.
  • You do however, have to permit pawn promotion upon reaching eighth rank (you may opt to simply force conversion to queen).
  • It is up to you how the board is displayed - terminal ASCII, GUI, etc.
  • You must display whose turn it is.

The rules of chess can be found here - ignore specified tournament rules (e.g. timing, touching), and remember castling and en passant are not necessary implementations.

Kieran Hunt

Posted 2015-01-30T12:07:53.537

Reputation: 478

When you say "2 player" I understand its just moving pieces for 2 human players, not a bot like this: http://codegolf.stackexchange.com/q/12397/15599 . Must the program accept only legal moves? If so http://codegolf.stackexchange.com/q/10843/15599 is relevant. Finally http://codegolf.stackexchange.com/q/8911/15599 is about detecting check/mate. You need to specify more tightly what the program must do, otherwise it's acceptable to have a program that just prints the starting position, then accepts a first move like a1 h8 moving a rook diagonally to the opposite corner, jumping over the pawns

– Level River St – 2015-01-30T12:27:25.770

Does the input have to be in [a-h][1-8] [a-h][1-8] form or are there other acceptable ones? – randomra – 2015-01-30T12:54:38.737

It can be in any unambiguous form. – Kieran Hunt – 2015-01-30T12:58:04.337

11I think you should require castling and en passant! Why miss out some of the rules of chess? – None – 2015-01-30T13:01:34.147

Kieran, welcome to programming puzzles and code golf. The most important thing is to be clear, so that your challenge will be fair. @Lembik, regardless of whether castling and en passant is desirable or not, I wouldn't change the rules now. Changing rules may inconvenience someone who is already working on a solution, – Level River St – 2015-01-30T13:17:54.857

3Kieran, you may also want to check out the sandbox (see "featured on meta") for your next challenge. – Level River St – 2015-01-30T13:19:27.293

1I would suggest that you require an indication of whose turn it is to be displayed (as that hasn't been clarified yet.) Also, I'm a bit concerned about how people are going to interpret "any unambiguous form" for the input, but I am also quite curious about it. – Level River St – 2015-01-30T13:34:21.663

1Could you perhaps make or link to a complete list of all the rules required in this challenge, such that people with no prior chess experience can participate too? – flawr – 2015-01-30T14:44:30.803

BootChess is human vs. computer, afaik. – Gabriele D'Antona – 2015-01-30T20:11:46.880

@steveverrill Ooh, "any"... while 1:print(input()) -- The input has to be Python code that evaluates to the state of the board after the move. – KSFT – 2015-01-30T21:05:54.000

@friol in the article linked by the OP there is the following link olivier.poudade.free.fr/src/BootChess.asm and looking at the comments on the code, there does seem to be some kind of move search going on, so I guess you're right. Nevertheless, testing for check per this question is reasonably challenging. I'm predicting an answer below 487 bytes for this question, if we get one in a golfing language. But as you say, it wouldn't compare with the functionality of BootChess. Oliver should be congratulated for breaking the 33 year old record held by ZXchess - That's a real achievement! – Level River St – 2015-01-31T00:10:45.380

There is this reproduction of what I believe to be original ZX81 program. http://users.ox.ac.uk/~uzdm0006/scans/1kchess/ It seems it had an ability for computer to play moves - not just human-v-human.

– philcolbourn – 2015-02-04T03:31:06.483

Does the chess game have to display the board? Or can it play with only chess algebra? Output:1. Input:d4 Output: 1... (black's first move), etc? – mbomb007 – 2015-02-11T20:44:41.363

"It is up to you how the board is displayed - terminal ASCII, GUI, etc." - You do, but you can choose how that is implemented. – Kieran Hunt – 2015-02-11T20:59:59.310

Answers

21

C, 650 600

n=8,t=65,s,f,x,y,p,e,u=10,w=32,z=95;char a[95],b[95]="RNBKQBNR";v(){p=a[s]&z;y=f/u-s/u;x=f-s-y*u;e=x*x+y*y;n=s%u/8|f%u/8|a[s]/w-t/w|a[f]/w==t/w|!(p==75&e<3|p>80&x*y==0|p%5==1&x*x==y*y|p==78&e==5|p==80&x*(z-t)>0&(a[f]-w?e==2:e==1|e==4&s%5==1));if(!n&&p-78)for(e=(f-s)/abs(x*x>y*y?x:y),x=s;(x+=e)-f;)n|=a[x]-w;}main(){for(a[93]=40;n--;a[92]=47)sprintf(a,"%s%cP    p%c \n",a,b[n],b[n]+w);for(;1;){puts(a);for(n=1;n;){putchar(t);scanf("%d%d",&s,&f);v();memcpy(b,a,z);if(!n){a[f]=p-80|f%u%7?a[s]:t+16;a[s]=w;a[f]&z^75||(a[z-t/w]=f);f=a[z-t/w];t^=w;for(n=1,s=80;n&&s--;)v();if(n=!n)memcpy(a,b,z),t^=32;}}}}

To reduce the code for initializing the board, the display has White (uppercase) playing from left to right and Black (lowercase) playing from right to left. Input is in the form of two 2-digit decimal numbers (start position and finish position), giving file(0-7) and rank(0-7). For a bit of extra code (subtract 11 from each input), input could be made to comply with http://en.wikipedia.org/wiki/ICCF_numeric_notation (digits 1-8)

Here's a sample screenshot where Black has just advanced his Rook's Pawn. White tries various illegal moves with his queen, before finally capturing the pawn. Turn indicators are a for Black and A for White.

enter image description here

An interesting feature of my validation is the use of the square of the Euclidean distance. For the knight this is always 1^2+2^2=5, but I also use it for the King and the pawn.

Test for Check is done by backing up the board, carrying out the players move, and scanning all 64 possible opponent's moves.

The program has to be ended with Ctrl-C. I cant think of a more gracious way of doing it, other than putting in a special move to end the program. A game of chess actually ends when one player is unable to move on his turn (checkmate or stalemate) and that requires a lot of tests that arent required by the spec.

Commented code

n=8,t=65,s,f,x,y,p,e,u=10,w=32,z=95;                 // 8,10 height and width of board. w=ASCII space, also difference between ucase and lcase
                                                     // 95 is bitmask for conversion lowercase to uppercase, but also used as length of array, etc.
char a[95],b[95]="RNBKQBNR";                         // a is main board, b is backup board (but used at start to hold 1st row data.)

v(){                                                 // validate move in all aspects except check
  p=a[s]&z;                                          // p=uppercase(character on start square)
  y=f/u-s/u;                                         // signed distance in y direction
  x=f-s-y*u;                                         // and x direction
  e=x*x+y*y;                                         // square of Euclidean distance
  n=s%u/8|f%u/8|                                     // n=true if 2nd digit of input out of bounds OR
  a[s]/w-t/w|a[f]/w==t/w|                            // start sq not friendly piece OR finish sq is friendly piece (also eliminates case where start=finish)
  !(                                                 // OR NOT geometry valid
    p==75&e<3|                                       // 'K'(ASCII75) AND euclidean distance squared =1 or 2 OR
    p>80&x*y==0|                                     // 'Q'or'R' AND x or y = 0 OR
    p%5==1&x*x==y*y|                                 // 'Q'or'B' AND abs(x)=abs(y)
    p==78&e==5|                                      // 'N' AND euclidean distance squared = 5
    p==80&x*(z-t)>0&(a[f]-w?e==2:e==1|e==4&s%5==1)   // 'P'(ASCII80):x direction must correspond with case of player (z-t)
  );                                                 // if capturing e=2. Otherwise e=1 (except on start rows 1 and 6, e can be 4)
  if(!n&&p-78)                                       // if not yet invalid and piece not 'N'(ASCII78) 
    for(e=(f-s)/abs(x*x>y*y?x:y),x=s;(x+=e)-f;)      // Set e to the numeric difference to travel 1 square in right direction. Set x to start square
       n|=a[x]-w;                                    // and iterate x through all intervening squares, checking they are blank
}



main(){

  for(a[93]=40;n--;a[92]=47)                         // iterate n through 8 rows of board. vacant spaces in bracket are use to assign start positions of kings to a[92&93] 
    sprintf(a,"%s%cP    p%c \n",a,b[n],b[n]+w);      // build up start position, each row 10 squares wide, including filler space the end and newline

  for(;1;){                                          // loop forever   
    puts(a);                                         // display board
    for(n=1;n;){                                     // loop while move invalid
      putchar(t);                                    // display prompt 'A' for white 'a' for black
      scanf("%d%d",&s,&f);                           // get input
      v();                                           // validate move
      memcpy(b,a,z);                                 // backup board (and king position metadata)  
      if(!n){                                        // if move not yet invalid
        a[f]=p-80|f%u%7?a[s]:t+16;                   // if not a pawn on last row, content of finish square = start square, ELSE queen of correct case (t+16) 
        a[s]=w;                                      // start square becomes blank (ASCII32)
        a[f]&z^75||(a[z-t/w]=f);                     // if finish square king, update king position metadata
        f=a[z-t/w];                                  // to begin scanning to see if king in check, set f to current king position
        t^=w;                                        // and change colour
        for(n=1,s=80;n&&s--;)v();                    // for s=79..0 search for valid opponent move to capture king (stops with n=0)
        if(n=!n)memcpy(a,b,z),t^=32;                 // valid opponent threat on king means invalid player move. Invert n, recover board from backup and change colour back.
      }
    }    
  }
}

Level River St

Posted 2015-01-30T12:07:53.537

Reputation: 22 049

Great code! Two ideas: 32 into constant (you use it a lot), does s,f needs to be initialized? Spaces at empty squares are great imo. – randomra – 2015-02-10T15:50:32.860

@randomra thanks! I've added the test for Check now, so golfing is the next task. I use 32 for two different things: ASCII code for space, and to diffentiate between ASCII blocks of 32 (control/symbol/uppercase/lowercase.) I will make it a constant. Also, I use 95 as a bitmask for converting uppercase to lowercase (but I have contrived to use the same number for many other things.) Initialisation of s was a carryover from an earlier stage. I've removed it now. – Level River St – 2015-02-11T02:29:44.183

1You can concatenate assignments: e=y*y+x*x=f-s-u*y=f/u-s/u;. – randomra – 2015-02-12T19:08:10.617

@randomra Thanks, I thought that would give an "l-value required" compile error but it does compile and seems to run fine. Assuming the l-value is the last variable before the = does seem the only interpretation. Regarding e=y*y+x*(x=.. I'm a bit wary of execution order since I got the wrong answer from an argument modified within a printf statement, but in this case it seems to work fine on GCC/Cygwin. I'll wait for more changes before including it in the code. One thing still missing is a check for numbers outside the range 0..77 (which I dont see as required by the spec, but anyway..) – Level River St – 2015-02-12T21:17:19.700

1Two obvious changes: t^=32 to t^=w and a[95],b[95] to a[z],b[z]. The code is surprisingly readable. Probably because of the verbose comments! – randomra – 2015-02-12T22:32:24.210

@randomra 1. I can't believe I missed t^=32! a[z],b[z] won't compile outside a function. In a function, it will compile, but not if an initialization string is given (I had v() as a macro while I was investigating that, due to v() needing to be able to recognise a and b, but now I know the score I've put it as a function) 2. I like to comment my golfs well. Having put the effort in to writing and testing the code, I think it only makes sense to share it :-) – Level River St – 2015-02-12T23:44:33.347

This is quite impressive; very well done! – 6infinity8 – 2017-06-10T10:57:01.767

You can replace the t^=32 at the end with t^=w. Also, you can replace for(;1;) with for(;;). – Zacharý – 2017-06-16T15:57:57.347

568 – ceilingcat – 2019-07-04T00:52:47.730

13

Python 3, 1166 1071 993 bytes

I only realized I needed to stop kings from moving into check after I had otherwise finished, but here's my submission anyways

h=abs
m=range
l=print
o=" "
a=o*8
j=[list(s)for s in["RNBKQBNR","P"*8,a,a,a,a,"p"*8,"rnbkqbnr"]]
u=False
v=lambda c:(c.lower()==c)==u
while 1:
 l("b"if u else"w")
 for b in j:l(*b)
 q,r,s,t=[int(i)for i in input().split()];P=j[r][q];g=j[t][s];p=P.lower()
 def w():
  if g==o or not v(g):j[t][s]=P;j[r][q]=o;global u;u=not u
 if not v(P):break
 if p=="r"or p=="q":
  for a,b,c,d in[(q,r,s,t),(r,q,t,s)]:
   if a==c:
    x=h(d-b)//(d-b)
    for n in m(b+x,d,x):
     if j[n if b else r][q if b else n]!=o:break
    else:w()
 if p=="b"or p=="q":
  if h(q-s)==h(r-t):
   for n in m(1, h(q-s)):
    if j[r+(n if t>r else-n)][q+(n if s>q else-n)]!=o:break
   else:w()
 if p=="k"and h(q-s)<2 and h(r-t)<2 or(p=="n"and(h(q-s)==2 and h(r-t)==1 or h(r-t)==2 and h(q-s)==1)):w()
 if p=="p":
  f=t-r==(-1 if u else 1)
  if(g!=o and not v(g)and h(q-s)==1 and f)or(g==o and q==s and f)or(g==o and q==s and t-r==(-2 if u else 2)and r==(6 if u else 1)):
   w()
   if t==(0 if u else 7):j[t][s]="q"if u else"Q"

To play, enter four space-delimited numbers, the first 2 being the coordinates of the piece you want to move, and the second 2 being where you want it to move to.

Cameron McCluskie

Posted 2015-01-30T12:07:53.537

Reputation: 346

1for initializing your board, you can use this instead for a big savings: a=o*8;j=[list(s) for s in ["RNBKQBNR","P"*8,a,a,a,a,"p"*8,"rnbkqbnr"]] – pseudonym117 – 2015-02-09T21:16:27.880

also .spilt() defaults to splitting at whitespace, so you can use input().split() instead of input().split(" ") – pseudonym117 – 2015-02-09T21:18:18.267

You don't need a space between literals and keywords [in that order]. (Like 2and, 1or, etc.) – Zacharý – 2017-06-16T16:15:07.880

False can be 0!=0. – Zacharý – 2017-06-16T16:18:22.063

Haven't tested but I guess False can just be 0 - not 0 should work fine... – Artemis still doesn't trust SE – 2019-12-01T13:11:07.853

6

Sorry for the delay.

The program is called ChesSkelet. Currently it uses 352 bytes for both code and data. It's written in Z80 Assembly, and particularly for the ZX Xpectrum.

If you don't feel like compiling it and loading it in an emulator you can play online in ChesSkelet site (http://chesskelet.x10host.com).

The program is pretty simple, a big loop where: - alternatively (1) white inputs its move or (2) black runs it's micro AI to move. - board is updated in screen. again, there is a huge guide describing the program and techniques in the website.

; ----------------------------------------------------------------------------- ; CHESSKELET /tseske'let/ ; Alex Garcia (reeagbo), Boria Labs 2018-2019 ; Thanks, @MstrBlinky and @johan_koelman, for your contribution ; Developed with ZXSpin, Notepad++ ; ----------------------------------------------------------------------------- ; Compilation with ZXSpin (all versions) and SpectNetIde (not all versions) ; Run with RANDOMIZE USR 30000 ; ----------------------------------------------------------------------------- ; debug mode: 0 = no, 1 = yes debmod equ 0 ; gramod: 0 = minimal interface, 1 = basic interface, 2 = full interface gramod equ 0 ; feamod: 0 = no features (if fails at legadd 'ret'), 1 = all features feamod equ 0 ; ROM memory addresses clescr equ 3503 laskey equ 23560 ; memory micro-pages (256B, typically H register) used for simple memory access auxsth equ $7D piearh equ $7E movlih equ $7F boasth equ $80 boaath equ $81 boaoph equ $82 canlih equ $83 org 30000 ; code is not directly portable ;------------------------------------------------------------------------------ ; Common code before turns ;------------------------------------------------------------------------------ ; legal moves generation (3B) ----------------------------------------- befmov call genlis ; candidate move list, used for both sides ; switch sides on every loop (6B+1B) ---------------------------------- whomov ld l, h ; (H)L: $7F7F = movlih + gamsta, ++ ld a, (hl) ; load state xor h ; (@johan_koelman) switch turn: bla=0, whi=1 ld (hl), a ; save state back in memory if feamod>0 jp z, blamov else jr z, blamov ; if 0, jump to black moves, jp maybe endif ; clear screen (3B) whimov call clescr ; ROM routine set screen mode ; print board ----------------------------------------------------------------- priboa ; A, B = 0 at this point ; initialization (4B) ld h, boasth ; H(L)= $80, L always 0, load board ++ ld d, piearh ; D(E): piece array pointer, E o/w later priloo ; print colored squares (8B) if gramod>0 ; opt: print colored squares ld a, 19 ; set bright ASCII code rst 16 ; print value ld a, c ; (@MstrBlinky) C is always $21 inc c ; change C parity and %00000001 ; keep Ab0, alternatively 0/1 rst 16 ; print value endif ; print piece (10B) ld a, (hl) ; load piece and %00100000 ; keep color, pih ld b, a ; Bb5: isolate piece color ld e, (hl) ; load piece res 5, e ; uncolor, pih ld a, (de) ; load piece character sub b ; capitalize (-32) only for white pieces rst 16 ; print piece ; next square, end of rank/board detection (15B+1B) inc l ; next square jp m, pricoo ; (@johan_koelman) end of 16x8 board, A=128? ld a, l ; (@MstrBlinky) and $08 ; 8 if end of rank, 0 other cases jr z, priski ; skip if not end of the rank add a, l ; ld l, a ; return result to L ld a, 13 ; A=
    rst 16          ; print char

if gramod>0         ; opt: print colored squares, end of the rank
    inc c           ; change C parity
endif

priski  jr priloo       ; loop through all squares

; print coords (28B+6B)--------------------------------------------------------
pricoo              ; (@MstrBlinky simplified it)
if gramod>0         ; opt: print board coords
    ld bc, $0709        ; B: loop count, C: fixed rank/col

nextce  ld a, $16           ; ASCII control code for AT         
    rst 16              ; print it
        ld a, b         ; set rank
        rst 16              ; print it
        ld a, c         ; set column
        rst 16              ; print it
    ld a, '8'       ; base rank
    sub b           ; decrease rank character (8..1)
    rst 16          ; print rank value

    ld a, $16           ; ASCII control code for AT
        rst 16              ; print it
        ld a, c         ; set rank
        rst 16              ; print it
        ld a, b         ; sets column
        rst 16              ; print it
    ld a, 'a'       ; base column character
    add a, b        ; increase rank character (a..h)
    rst 16          ; print rank value

    dec b           ; loop 8 times
    jp p, nextce        ;
endif

if gramod>0         ; opt:  + "?" for input prompt
    ld a, 13        ; A: set  ASCII code
    rst 16          ; prints it to go to the next line for input
    ld a, '?'       ; set "?" ASCII code
    rst 16          ; print it
endif
; -----------------------------------------------------------------------------
; -----------------------------------------------------------------------------
;               WHITE MOVES 
; -----------------------------------------------------------------------------
; -----------------------------------------------------------------------------
; read chars from keyboard and stores them (16B(+4B+4B))-----------------------
    ; 4 times loop for coord input (3B)
    ld b, 4         ; loop count
    dec d           ; D(E)= $7D =auxsth, E always 0 ++

    ; read key from keyboard loop (8B)
realoo  ld hl, laskey           ; LASTKEY system variable ++
    xor a           ; A=0
    ld (hl), a      ; reset LASTKEY, two birds with 1 stone
wailoo  add a, (hl)             ; load latest value of LASTKEY.
    jr z, wailoo            ; loop until a key is pressed.

    ; skip move/switch sides (4B)
if feamod>0         ; opt: special move, switch sides to play black
    cp 's'          ; if "s" pressed at any time
    jp z, aftmov        ; skip white's move, ### jr maybe
endif
    ; save pressed key and print it (5B)
    inc de          ; (@MstrBlinky) next char, E = 1 to 5
    ld (de), a      ; save char in string
    rst 16          ; print it

    djnz realoo     ; loop for 4 input chars

    ; border reset (4B)
if gramod>1         ; opt: border reset after first white move
    ld a, 7         ; set back to white
    out (254), a        ; set border back to white
endif

; translate coords to square (17B) --------------------------------------------
movchk  ex de, hl       ; (@MstrBlinky routine) DE=end of input string

movloo  ld a, 56        ; rank calc = 8-(rank input-48) = 56-(HL)
    sub (hl)        ; A= 56 - (HL)
    rla         ; move it to high nibble (x16)
    rla         ;
    rla         ;
    rla         ;

    dec hl          ; (@MstrBlinky) run backwards through string
    add a, (hl)     ; rank + column (not 0-7 column)
    sub 'a'         ; make it a 0-7 column

    ld c, b         ; slide results through B and C
    ld b, a         ; at end of 2nd loop everything is in place

    dec l           ; (@MstrBlinky) beginning of input string?
    jr nz, movloo       ; if not, loop again

; search white move in legal move list (24B) ----------------------------------
if feamod>0         ; opt: validate white move
seamov  ld hl, canlis       ; canli pointer ++
    ld a, (hl)      ; number of candidates
    ;inc hl         ; skip to first candidate (+2 bytes)
    ;inc hl         ; removed v0.808, no move in those two bytes

sealoo  ld d, (hl)      ; origin candidate move
    inc hl          ; next byte
    ld e, (hl)      ; target candidate move
    inc hl          ; next byte, for next loop

    ex de, hl       ; candidate pair, DE: HL-canli pointer
    or a            ; reset carry
    sbc hl, bc      ; compare input move with cand. move (Z)
    ex de, hl       ; revert back, canli pointer
    jr z, aftsid        ; move match: jump out. ready to move
                ; B (origin sq), C (target sq) ready here
    dec a           ; count down
    jr nz, sealoo       ; loop until canli covered

    jp whimov       ; if not found, back to move input, jp maybe
else                ; opt: skip validate white move
    jr aftsid       ; Outputs: B: origin square, C: target square
endif
; -----------------------------------------------------------------------------
; -----------------------------------------------------------------------------
;               BLACK MOVES
; -----------------------------------------------------------------------------
; -----------------------------------------------------------------------------
blamov
chomov  ; preparations (7B)----------------------------------------------------
    ld hl, canlis       ; candidate list. No H reuse ++
    ld b, (hl)      ; number of candidates at position 0
    ld c, l         ; C=0, maximum valuation reset
    inc hl          ; skip 2 bytes to first candidate in list
    inc hl          ;
choloo  ; loop through candidates (6B) ----------------------------------------
    ld d, (hl)      ; D: origin candidate square
    inc hl          ; next candidate byte
    ld e, (hl)      ; E: target candidate square
    inc hl          ; next candidate byte
    push bc         ; BC released
    push hl         ; HL is different from here
    ; pieces valuation ----------------------------------------------------
    ; pieces collection (8B)
evatap  ld h, boasth        ; board base ++
    ld l, e         ; target square
    ld b, (hl)      ; black piece value
    ld l, d         ; origin square
    ld c, (hl)      ; white piece value
    res 5, c        ; uncolor white piece; pih

    ; origin attacked square (7B)
evaato  ld a, b         ; target piece always counts
    ld h, boaoph        ; H(L): attacked board base, L: unchanged ++
    bit 7, (hl)     ; target square attacked?
    jr z, evaatt        ; not attacked, skip counting origin piece

if feamod=1         ; opt: rows 2 do not move even if attacked  
    ld a, d         ; 0rrr0ccc, add origin square     
    and $70         ; filter ranks
    cp $60          ; is rank 6?
    ld a, b         ; target piece always counts
    jr z, evaexi        ; skip this move
endif

    ; count origin piece (1B) if attacked, general case
evaatc  add a, c        ; A: 00pppppp, count white

    ; target attacked square (6B)
evaatt  ld l, e         ; H(L): point at target square
    bit 7, (hl)     ; target square attacked?
    jr z, skiato        ; if target not attacked, skip
    sub c           ; if target attacked, count white out  

    ; compensate + prioritize piece valuation(6B)
skiato  ld h, $20       ; prepare H for later rotation and use for A
    add a, h        ; A: 00pppppp, compensate=K+1, pih                                             
    rlca            ; leave space for square weight
    rlca            ; A: pppppp00, piece addition is 5 bits
    ld b, a         ; B: piece addition value
evacol  ld a, e         ; A: 0rrr0ccc
    ; these two values below can be tuned for different opening schemes
    if feamod>0
        add a, 2        ; A: 0rrr0ccc
        and 5           ; A: 00000ccc
    else
        inc a           ; A: 0rrr0ccc
        and 4           ; A: 00000cc0 (weight: 0,0,0,4,4,4,4,0)
    endif

    ; ranks weight (ranks weight is 8..1, aiming for board's end)
evarnk  add hl, hl      ; HL: 00100000 0rrr0ccc (before)
    add hl, hl      ; 
    add hl, hl      ; HL: 000000rr r0ccc000 (after)
    sub h           ; A:  00000cww (w=r+c)
    add a, b        ; total value: pieces + weight

    ; maximum value comparison (12B)
evaexi  pop hl          ; recover canli
    pop bc          ; recover previous maximum value
    cp c            ; compare with current maximum
    jr c, chonoc        ; if current eval (A) <= max eval (C), skip

    ld c, a         ; update best evaluation
    pop af          ; remove old maximum to avoid cascades in stack
                ; ### initial push to compensate?
    push de         ; push best candidates so far

chonoc  dec b           ; decrease loop counter 2 by 2.
    djnz choloo     ; loop through all candidates (canto)

    pop bc          ; recover saved values (B: origin, C: target)

; -----------------------------------------------------------------------------
; -----------------------------------------------------------------------------
;               AFTER SIDES
; -----------------------------------------------------------------------------
; -----------------------------------------------------------------------------
; move piece (8B) -------------------------------------------------------------
    ; inputs here: B: origin square, C: target square
    ; write origin square and read piece in it (4B)
aftsid  ld h, boasth        ; point at board (canlih=$83 before) ++
    ld l, b         ; point at origin square

    ; castling, rook moves, v0.800 (26B)
    ; very innacurate as it may cause false moves
if feamod>0         ; opt: castling, rook move
casroo  ld a, (hl)      ; origin piece
    add a, b        ; + origin square
    sub c           ; - target square
caslon  cp $38          ; $36(king) + $74(ori) - $72(tar)= $38, pih
    jr nz, cassho       ; no long castling
    ld l, $70       ; long castling rook square (a1)
    ld (hl), d      ; erase rook (D=0 here)
    ld l, $73       ; rook destination (d1)
    ld (hl), $25        ; move rook, pih
cassho  cp $34          ; $36(king) + $74(ori) - $76(tar)= $34, pih
    jr nz, casend       ; no short castling
    ld l, $77       ; short castling rook square (h1)
    ld (hl), d      ; erase rook (D=0 here)
    ld l, $75       ; rook destination (f1)
    ld (hl), $25        ; move rook, pih
casend
endif

if feamod>0         ; opt: special move: prom, no under-prom (12B)
    ld a, c         ; A: 0rrr0ccc
    and %01110000       ; A: 0rrr0000
    add a, (hl)     ; A: 0rrxpppp
    cp $22          ; white pawn ($22) on rank 8 ($00), pih
    ld l, b         ; restore origin square
    ld d, (hl)      ; original piece
    ld (hl), 0      ; write origin piece
    jr nz, aftdes       ; if not a pawn, skip
    ld d, $27       ; make piece a queen, pih
else                ; opt: write origin piece, no promotion (3B)
    ld d, (hl)      ; D: get origin piece
    ld (hl), 0      ; write origin piece    
endif

    ; write target square with origin piece (5B)
aftdes  ld l, c         ; (H)L: target square

    ; checkmate with exit (3B), board is not updated in screen
chkmat  bit 4, (hl)     ; captured piece is king ($16)?, pih 
    ret nz          ; (@johan_koelman) return prompt at check mate

aftnok  ld (hl), d      ; write target square
    call genlis     ; update attacked matrix after move
aftmov
; reverse board (22B)----------------------------------------------------------
revboa  ; push full board to stack (7B)
    inc h           ; H = $80 = boasth ++
    ld l, h         ; (H)L: end of board. trick: start from $8080
revlo1  dec l           ; countdown squares
    ld a, (hl)      ; read piece
    push af         ; copy piece to to stack
    jr nz, revlo1       ; loop down to beginning of the board

    ; collect board back ir reverse order + switch color (15B)
    ld l, $78       ; (H)L: end of board again
revlo2  pop af          ; collect piece from stack
    or a            ; is it an empty square?
    jr z, revski        ; if yes, skip
    xor %00100000       ; otherwise, reverse color (b5), pih
revski  dec l           ; countdown squares
    ld (hl), a      ; piece back into board
    jr nz, revlo2       ; loop until beginning of board

    jp befmov       ; back to white move, too far for jr

; -----------------------------------------------------------------------------
; -----------------------------------------------------------------------------
;               AUXILIARY ROUTINES
; -----------------------------------------------------------------------------
; -----------------------------------------------------------------------------

; genlis: generates list of legal moves (92B + 9B) ----------------------------
; it was not possible to use it in two different places, only procedure in code
genlis
bacata  ; backup attack board in reverse order, used in evaluation (13B)
    ld l, $FF       ; (H)L = $80FF (boaata-1), H always $80 
    ld de, boaopo + $78 ; DE: same thing, 1B passed end of board
bacloo  inc hl          ; HL: increase 16b counter to hop to next page
    dec e           ; E:  decrease 8b counter to hit Z flag
    ld a, (hl)      ; load attack status
    ld (hl), 0      ; clear attack status, no alternative!
    ld (de), a      ; backup attack status
    jr nz, bacloo       ; loop down to square $00
                ; exit values: DE=$8200, HL=$8177

    ; prepare environment (4B)
    inc d           ; D= $83= canlih
    xor a           ; reset
    ld (de), a      ; cantot= 0
    ld b, l         ; B= L = 77, SQUARE COUNT

    ; read piece from board (4B)
squloo  ld h, boasth        ; H: board base ++
    ld l, b         ; point at current loop square
    ld a, (hl)      ; read piece from board

    ; king castling, v0.800 (15B)
    ; only basic rule: no unmoved pieces or attacked squares check
    if feamod>0
    kincas  ld l, a     ; save A (can't use push AF, flags lost)
        add a, b    ; A: 0rrr0xxx + 000ppppp (uncolored white p.)
        cp $AA      ; king($36) at E1($74)= $AA, pih
        ld a, l     ; recover A
        jr nz, kinend   ; if no match, skip adding legal move

        ld c, $72   ; E1-C1 move, rook's move missing
        call legadd ; add king's move
        ld c, $76   ; E1-G1 move, rook's move missing
        call legadd ; add king's move and go on with king's moves
    kinend
    endif

    ; get move type and pointer to move list (6B)
squgon  dec h           ; H(L)= movlih, moves vector base ++
    add a, a        ; x4, each piece vector is 4B long
    add a, a        ;   
    ld l, a         ; (H)L points at the move vector now

    ld d, 2         ; 2 submoves per piece
subloo      ; byte 1 - move type (5B)
        ld a, (hl)      ; move type loaded
        or a            ; =cp 0, 2nd move type not used case
                    ; black/empty: move type=0 leads here
        jr z, squexi        ; ---v exit: square is done
        ld e, a         ; E: MOVE TYPE (B,C,D used here)

        ; pawn 2 squares forward - move type modified (8B)
        if feamod>0     ; opt: special move, pawn 2 sq. forward
        genpw2  add a, b    ; piece square + move type
            and %11111000   ; masked with relevant bits
            cp $88      ; $28(str.pawn)+$60(rnk 6) ### univocal
            jr nz, skppw2   ; if not, skip
            inc e       ; increase radius: 1 -> 2
        skppw2
        endif

        ; byte 2 - movlis delta (3B)
        inc hl          ; next piece sub-entry
        push hl         ; Save HL for 2nd loop  
        ld l, (hl)      ; pointer to move delta

vecloo          ; vector read (8B)
            ld c, b     ; TARGET SQUARE init
            ld a, (hl)  ; vector delta
            or a        ; =cp 0
            jr z, vecexi    ; ---v exit: vectors end with 0, next sq.
            push hl     ; save current delta
            push de     ; save move type + radius
                    ; E: variable radius within loop
            ld d, a     ; D: store delta within loop

celloo              ; prepare x88 check (7B)
                ld a, d     ; delta loaded
                add a, c    ; current target (sq. + delta)
                ld c, a     ; current target
                and $88     ; 0x88, famous OOB trick
                jr nz, vecnex   ; ---v exit: OOB, next vector

                ; read target square (3B)
                inc h       ; H(L)= $80 = boasth ++
                ld l, c     ; point at target square            
                ld a, (hl)  ; read target square content

                ; mark attacked ### str. pawn marked attacked
                    inc h       ; H(L)= $81 = boaath ++
                    ld (hl), h  ; mark attacked ($81)
                    dec h       ; H(L)= $80 = boasth ++

                dec h       ; H(L)= $79= movlih ++
                ; target is white (4B)
                bit 5, a    ; is it white?, pih
                jr nz, vecnex   ; ---v exit: WHITE b4=1, next vector

                ; target not white (3B)
                or a        ; =cp 0, is it empty?, pih
                jr z, taremp    ; if not 0, it's black: legal, no go on

tarbla              ; target is black (7B)
                bit 5, e    ; special move: pawn straight check
                jr nz, vecnex   ; ---v exit: no straight capture, next vector
                ld e, a     ; make radius=0 (=<8 in code, canonical: ld e, 0)
                jr legadj   ;

taremp              ; target is empty (14B)
                bit 4, e    ; special move: pawn on capture check
                jr nz, vecnex   ; ---v exit: no diagonal without capture, next vector     
                dec e       ; decrease radius
legadj
                if feamod=0 ; opt: legadd for basic model
                ; add candidate (B: current square, C: target square) (9B)
                    push hl
                    ld hl, canlis   ; HL: start of candidate list. No H reuse ++
                    inc (hl)    ; +2 to candidate counter to move to next
                    inc (hl)    ; first free position in list
                    ld l, (hl)  ; point at free position

                    ld (hl), b  ; 1) save origin square
                    inc hl      ; move to next byte
                    ld (hl), c  ; 2) save dest square                   
legend                  pop hl      ; recover HL=pointer to vector list
                else        ; opt: legadd call for full model
                    call legadd
                endif

                bit 3, e    ; if radius < 8 (Cb3=0), radius limit
                jr nz, celloo   ; ---^ cell loop

vecnex          ; next vector preparation (5B)
            pop de      ; DE: recover move type + radius
            pop hl      ; HL: recover current vector
            inc hl      ; HL: next vector
            jr vecloo   ; ---^ vector loop

vecexi  ; next square preparation (5B)
    pop hl          ; HL: recover pointer to sub-move list
    inc hl          ; HL: next byte, point at 2nd sub-move
    dec d           ; 2 sub-move iterations loop control
    jr nz, subloo       ; if not 2nd iteration, repeat loop
    ; end of loop (2B)      
squexi  djnz squloo     ; ---^ squares loop
    ret

; legadd: add legal move -------------------------------------------------------
if feamod>0         ; legadd for king castling
legadd              ; (B: current square, C: target square)
    push hl
    ld hl, canlis       ; HL: start of candidate list. No H reuse ++
    inc (hl)        ; +2 to candidate counter to move to next
    inc (hl)        ; first free position in list
    ld l, (hl)      ; point at free position

    ld (hl), b      ; 1) save origin square
    inc hl          ; move to next byte
    ld (hl), c      ; 2) save dest square                   
    pop hl          ; recover HL=pointer to vector list
    ;ret            ; <===== not removed with feamod=0
endif

; -----------------------------------------------------------------------------
; DATA ------------------------------------------------------------------------
; -----------------------------------------------------------------------------

; Memory page: 7700h ----------------------------------------------------------
org $7700
auxstr              ; input string stored here

; Memory page: 7E00h ----------------------------------------------------------
; used to convert values to pieces
org $7E00
if gramod=0         ; opt: space or dot depending on the size
piearr  defb '.'        ; $2B
else
piearr  defb ' '
endif
org $7E02
    defm "pnbr"     ; change this array to any language, pih
org $7E07
    defb 'q'        ; change this array to any language, pih
org $7E16
    defb 'k'        ; change this array to any language, pih

; Memory page: 7F00h ----------------------------------------------------------
; sub-moves and vectors 
org $7F00
; leave empty $00-$04-...-$24 for black pieces/empty square pointers
org $7F88           ; pawn: $22x4=$84
; piece, move type, vector list delta address (18B)
; move type / 0 / 0 / pawn straight / pawn diagonal / DDDD (real radius + 7)
movlis
pawgen  defb    $28, $E3    ; pawn straight
    defb    $18, $E7    ; pawn capture
org $7F8C
knigen  defb    $08, $EA    ;
org $7F90
bisgen  defb    $0E, $E5    ; bishop
org $7F94
roogen  defb    $0E, $E0    ; rook
org $7F9C
quegen  defb    $0E, $E0    ; queen
    defb    $0E, $E5    ;
org $7FD8
kingen  defb    $08, $E0    ; king: $($16+$20)x4=$D8
    defb    $08, $E5    ;

org $7FE0           ; vectors start at: $7FE0 (arbitrary)
; (y, x) move delta pairs (16B)
veclis
strvec  defb    $FF, $01    ; +0, straight vectors
    defb    $10, $F0    ; +3, straight pawn, last half line
org $7FE5
diavec  defb    $0F, $11    ; +5, diagonal vectors
    defb    $EF, $F1    ; +7, diagonal pawn
org $7FEA
knivec  defb    $E1, $F2    ; +10, knight vectors
    defb    $12, $21    ; knight moves listed clockwise
    defb    $1F, $0E    ;
    defb    $EE, $DF    ;
; board status: 0000000 / turn (B=0, W=1)
org $7F7F
gamsta

; Memory page: 8000h ----------------------------------------------------------
    ; board squares format: 00cppppp
    ; pppp (value) : pawn=2, knight=3, bishop=4, rook=5, queen=7, king=$16
    ; c (color): white=1, black=0
    ; initial board setup
if debmod=1         ;opt: fill board for debugging
    org $8000
    boasta  defb $00, $00, $00, $00, $00, $00, $00, $00 ; <--8
        defb $00, $00, $00, $00, $00, $00, $00, $00
        defb $00, $00, $00, $00, $02, $00, $00, $00 ; <--7
        defb $00, $00, $00, $00, $00, $00, $00, $00
        defb $00, $00, $00, $23, $00, $00, $00, $00 ; <--6
        defb $00, $00, $00, $00, $00, $00, $00, $00
        defb $00, $00, $00, $00, $00, $00, $00, $00 ; <--5
        defb $00, $00, $00, $00, $00, $00, $00, $00
        defb $00, $00, $00, $00, $00, $00, $00, $00 ; <--4
        defb $00, $00, $00, $00, $00, $00, $00, $00
        defb $00, $00, $00, $00, $00, $00, $00, $00 ; <--3
        defb $00, $00, $00, $00, $00, $00, $00, $00
        defb $00, $00, $00, $00, $00, $00, $00, $00 ; <--2
        defb $00, $00, $00, $00, $00, $00, $00, $00 
        defb $00, $00, $00, $00, $00, $00, $00, $00 ; <--1
else                ; opt: reduces board size for gameplay
    org $8000
    boasta  defb $05, $03, $04, $07, $16, $04, $03, $05 
    org $8010
        defb $02, $02, $02, $02, $02, $02, $02, $02
    org $8060           
        defb $22, $22, $22, $22, $22, $22, $22, $22
    org $8070
        defb $25, $23, $24, $27, $36, $24, $23, $25
endif

; Memory page: 8100h ----------------------------------------------------------
org $8100
boaata              ; attacked squares board

; Memory page: 8200h ----------------------------------------------------------
org $8200
boaopo              ; reversed attacked squares board

; Memory page: 8300h ----------------------------------------------------------
; candidate move list at the very end of the program 
org $8300 
canlis  equ $
````

reeagbo

Posted 2015-01-30T12:07:53.537

Reputation: 61

3

><>, 1467 bytes

Well, it doesn't quite break the 487-byte record…

>   "  :W"av
"RNBQKBNR"a/
"PPPPPPPP"a/
"        "a/
"        "a/
"        "a/
"        "a/
"pppppppp"a/
"rnbqkbnr"a/
   l?!vo29.>
v?)b:i<~
\:8c*-:i:}4c*-:@g:o" "o{{oo" "o}i:o8c*-i:o4c*-
\{:"^")80g"^")-?X
\:"^")48**-
\&3[:}]$:@=3[:}]{:}=*?X&
\:"Q"=?v:"R"=?v
v      >      >&r:4[:}]$-r:4[:}]-{:?!v$:?!v~~&
 ["+-+10"]?$~137*.*731$~$?]"10+-+"[1)<    >)1
\:"Q"=?v:"B"=?v
v      >      >&$:@$:@-r$:@$:@$-{-r4[r$:@$:@+r$:@$:@+{-}]{:?!v$:?!v~~&
 )1["+-11"]?$137*.                     .*733]~~r?{"--11++"[1)<    >
 \>42b*p72b*p32b*p62b*p{{$:@$:@}}4[r
 >$zz$zz$:@{:}=$:@5[:}]=*?v$:@$:@g" "-?X
               .+9*a29$]~~<
\&$:@{:}-:*$:@5[:}]-:*+&
\:"N"=?v
  X?-5&<v
\:"K"=?v
  X?)3&<v
  )-?vX >$:@$:@g:" "=?v"^")80g"^"
     >049*.*942      ~<
\:"P"=?v
 )?X:1=\&~$:@{:}-:*$:@5[:}]80g"^")?$-:2
X      \?v2-?X?X$:@$:@80g"^"(2*1--g" "-?Xv
         >~:?v?X                         >$:@$:@g" "-?Xv
  @@?$~&049*.>1-?X$:@$:@g:" "=?X"^"(80g"^"(=?X         >:880g"^")7*-=&"Q"
\>&80g"^"):&48**+@p" "@p0
>:8%1+$:8,:1%-1+$}$:@$:@g"Kk"&:&?$~=?v~~1+:88*=?X
v                {{1030014103431434~{<
>l2=?v$:@$:@}}$@+2-a%@+2-a%$g"nN"&:&?$~=?X{{
\    >0010200121021222{{
>l2=?v$:@$:@}}$@+1-@+1-$g"kK"&:&?$~=?X{{
v    >$:@1+$:@&:&2*1--g"pP"&:&?$~=?X$:@1-$:@&:&2*1--g"pP"&:&?$~=?X
\"10++r10-+r01++r01+-r11++b11+-b11-+b11--b"
>l2=?v48*4a*6+p44*4a*6+p74a*6+\
   /} }@:$@:${{p+6*a46p+6*a4fp/
 -?\$ zz:9%?!v$zz:9%?!v$:@$:@g:"y"&:&48**-=?X:"q"&:&48**-=?X" "
^~~<         <        <
v    >~~"bW"&?$~80paaoo

Try it at the fish playground! (It makes use of a couple of bugs in that interpreter, so I can't promise it'll work anywhere else. You'll probably want to run it at maximum speed — it can take over a minute to do one move.)

When you run the code, it will print

rnbqkbnr
pppppppp




PPPPPPPP
RNBQKBNR

W:  

Capital letters represent white, lowercase represent black. You can then give your move as input in the form [a-h][1-8][a-h][1-8], e.g. e2e4, meaning "move the piece at e2 to e4". The program will then print, for example,

W:  P e2 e4

rnbqkbnr
pppppppp


    P   

PPPP PPP
RNBQKBNR

b:  

The main form of memory in ><> is the stack; however, that's not very practical for storing a chess board. Instead, I used ><>'s self-modifying capability to store the chess board as part of the source code itself, accessing it with g and p.

The bulk of the code is to check whether your move is legal. There are several things to check here:

  • Are you trying to move a piece that really exists, and is it yours (line 12, 0-indexed)?
  • Are the start and end squares different (line 14)?
  • Is the move in the right direction for the type of piece (orthogonal for rook and queen, lines 15–17; diagonal for bishop and queen, lines 18–20; L-shaped for knight, lines 24–26; 1 square for king, lines 27–28; or a pawn move, lines 31–35)?
  • Are the squares between the start and end all empty (for rook, bishop, queen — lines 21–23)?
  • Is the target square empty or occupied by an enemy (lines 29–30)?
  • Are you in check after the move (lines 36–47)?

If any of those questions have the wrong answer, the program throws an error and halts; otherwise, it edits the chessboard in its source code, prints it again and waits for the next move.

For the king and knight moves, I borrowed Level River St's trick of checking the squared Euclidean distance: for a king, it's less than 3, and for a knight, it's exactly 5.

Not a tree

Posted 2015-01-30T12:07:53.537

Reputation: 3 106