Algorithm for Determining Tic Tac Toe Game Over

Question

I've written a game of tic-tac-toe in Java, and my current method of determining the end of the game accounts for the following possible scenarios for the game being over:

1. The board is full, and no winner has yet been declared: Game is a draw.
2. Cross has won.
3. Circle has won.

Unfortunately, to do so, it reads through a predefined set of these scenarios from a table. This isn't necessarily bad considering that there are only 9 spaces on a board, and thus the table is somewhat small, but is there a better algorithmic way of determining if the game is over? The determination of whether someone has won or not is the meat of the problem, since checking if 9 spaces are full is trivial.

The table method might be the solution, but if not, what is? Also, what if the board were not size n=9? What if it were a much larger board, say n=16, n=25, and so on, causing the number of consecutively placed items to win to be x=4, x=5, etc? A general algorithm to use for all n = { 9, 16, 25, 36 ... }?

Answer 1

If you have boarder field 5*5 for examle, I used next method of checking:

public static boolean checkWin(char symb) {
  int SIZE = 5;

        for (int i = 0; i < SIZE-1; i++) {
            for (int j = 0; j <SIZE-1 ; j++) {
                //vertical checking
            if (map[0][j] == symb && map[1][j] == symb && map[2][j] == symb && map[3][j] == symb && map[4][j] == symb) return true;      // j=0
            }
            //horisontal checking
            if(map[i][0] == symb && map[i][1] == symb && map[i][2] == symb && map[i][3] == symb && map[i][4] == symb) return true;  // i=0
        }
        //diagonal checking (5*5)
        if (map[0][0] == symb && map[1][1] == symb && map[2][2] == symb && map[3][3] == symb && map[4][4] == symb) return true;
        if (map[4][0] == symb && map[3][1] == symb && map[2][2] == symb && map[1][3] == symb && map[0][4] == symb) return true;

        return false; 
        }

I think, it's more clear, but probably is not the most optimal way.

Answer 2

Here is a solution I came up with, this stores the symbols as chars and uses the char's int value to figure out if X or O has won (look at the Referee's code)

public class TicTacToe {
    public static final char BLANK = '\u0000';
    private final char[][] board;
    private int moveCount;
    private Referee referee;

    public TicTacToe(int gridSize) {
        if (gridSize < 3)
            throw new IllegalArgumentException("TicTacToe board size has to be minimum 3x3 grid");
        board = new char[gridSize][gridSize];
        referee = new Referee(gridSize);
    }

    public char[][] displayBoard() {
        return board.clone();
    }

    public String move(int x, int y) {
        if (board[x][y] != BLANK)
            return "(" + x + "," + y + ") is already occupied";
        board[x][y] = whoseTurn();
        return referee.isGameOver(x, y, board[x][y], ++moveCount);
    }

    private char whoseTurn() {
        return moveCount % 2 == 0 ? 'X' : 'O';
    }

    private class Referee {
        private static final int NO_OF_DIAGONALS = 2;
        private static final int MINOR = 1;
        private static final int PRINCIPAL = 0;
        private final int gridSize;
        private final int[] rowTotal;
        private final int[] colTotal;
        private final int[] diagonalTotal;

        private Referee(int size) {
            gridSize = size;
            rowTotal = new int[size];
            colTotal = new int[size];
            diagonalTotal = new int[NO_OF_DIAGONALS];
        }

        private String isGameOver(int x, int y, char symbol, int moveCount) {
            if (isWinningMove(x, y, symbol))
                return symbol + " won the game!";
            if (isBoardCompletelyFilled(moveCount))
                return "Its a Draw!";
            return "continue";
        }

        private boolean isBoardCompletelyFilled(int moveCount) {
            return moveCount == gridSize * gridSize;
        }

        private boolean isWinningMove(int x, int y, char symbol) {
            if (isPrincipalDiagonal(x, y) && allSymbolsMatch(symbol, diagonalTotal, PRINCIPAL))
                return true;
            if (isMinorDiagonal(x, y) && allSymbolsMatch(symbol, diagonalTotal, MINOR))
                return true;
            return allSymbolsMatch(symbol, rowTotal, x) || allSymbolsMatch(symbol, colTotal, y);
        }

        private boolean allSymbolsMatch(char symbol, int[] total, int index) {
            total[index] += symbol;
            return total[index] / gridSize == symbol;
        }

        private boolean isPrincipalDiagonal(int x, int y) {
            return x == y;
        }

        private boolean isMinorDiagonal(int x, int y) {
            return x + y == gridSize - 1;
        }
    }
}

Also here are my unit tests to validate it actually works

import static com.agilefaqs.tdd.demo.TicTacToe.BLANK;
import static org.junit.Assert.assertArrayEquals;
import static org.junit.Assert.assertEquals;

import org.junit.Test;

public class TicTacToeTest {
    private TicTacToe game = new TicTacToe(3);

    @Test
    public void allCellsAreEmptyInANewGame() {
        assertBoardIs(new char[][] { { BLANK, BLANK, BLANK },
                { BLANK, BLANK, BLANK },
                { BLANK, BLANK, BLANK } });
    }

    @Test(expected = IllegalArgumentException.class)
    public void boardHasToBeMinimum3x3Grid() {
        new TicTacToe(2);
    }

    @Test
    public void firstPlayersMoveMarks_X_OnTheBoard() {
        assertEquals("continue", game.move(1, 1));
        assertBoardIs(new char[][] { { BLANK, BLANK, BLANK },
                { BLANK, 'X', BLANK },
                { BLANK, BLANK, BLANK } });
    }

    @Test
    public void secondPlayersMoveMarks_O_OnTheBoard() {
        game.move(1, 1);
        assertEquals("continue", game.move(2, 2));
        assertBoardIs(new char[][] { { BLANK, BLANK, BLANK },
                { BLANK, 'X', BLANK },
                { BLANK, BLANK, 'O' } });
    }

    @Test
    public void playerCanOnlyMoveToAnEmptyCell() {
        game.move(1, 1);
        assertEquals("(1,1) is already occupied", game.move(1, 1));
    }

    @Test
    public void firstPlayerWithAllSymbolsInOneRowWins() {
        game.move(0, 0);
        game.move(1, 0);
        game.move(0, 1);
        game.move(2, 1);
        assertEquals("X won the game!", game.move(0, 2));
    }

    @Test
    public void firstPlayerWithAllSymbolsInOneColumnWins() {
        game.move(1, 1);
        game.move(0, 0);
        game.move(2, 1);
        game.move(1, 0);
        game.move(2, 2);
        assertEquals("O won the game!", game.move(2, 0));
    }

    @Test
    public void firstPlayerWithAllSymbolsInPrincipalDiagonalWins() {
        game.move(0, 0);
        game.move(1, 0);
        game.move(1, 1);
        game.move(2, 1);
        assertEquals("X won the game!", game.move(2, 2));
    }

    @Test
    public void firstPlayerWithAllSymbolsInMinorDiagonalWins() {
        game.move(0, 2);
        game.move(1, 0);
        game.move(1, 1);
        game.move(2, 1);
        assertEquals("X won the game!", game.move(2, 0));
    }

    @Test
    public void whenAllCellsAreFilledTheGameIsADraw() {
        game.move(0, 2);
        game.move(1, 1);
        game.move(1, 0);
        game.move(2, 1);
        game.move(2, 2);
        game.move(0, 0);
        game.move(0, 1);
        game.move(1, 2);
        assertEquals("Its a Draw!", game.move(2, 0));
    }

    private void assertBoardIs(char[][] expectedBoard) {
        assertArrayEquals(expectedBoard, game.displayBoard());
    }
}

Full solution: https://github.com/nashjain/tictactoe/tree/master/java

Answer 3

Heres my solution that I wrote for a project I'm working on in javascript. If you don't mind the memory cost of a few arrays it's probably the fastest and simplest solution you'll find. It assumes you know the position of the last move.

/*
 * Determines if the last move resulted in a win for either player
 * board: is an array representing the board
 * lastMove: is the boardIndex of the last (most recent) move
 *  these are the boardIndexes:
 *
 *   0 | 1 | 2
 *  ---+---+---
 *   3 | 4 | 5
 *  ---+---+---
 *   6 | 7 | 8
 * 
 * returns true if there was a win
 */
var winLines = [
    [[1, 2], [4, 8], [3, 6]],
    [[0, 2], [4, 7]],
    [[0, 1], [4, 6], [5, 8]],
    [[4, 5], [0, 6]],
    [[3, 5], [0, 8], [2, 6], [1, 7]],
    [[3, 4], [2, 8]],
    [[7, 8], [2, 4], [0, 3]],
    [[6, 8], [1, 4]],
    [[6, 7], [0, 4], [2, 5]]
];
function isWinningMove(board, lastMove) {
    var player = board[lastMove];
    for (var i = 0; i < winLines[lastMove].length; i++) {
        var line = winLines[lastMove][i];
        if(player === board[line[0]] && player === board[line[1]]) {
            return true;
        }
    }
    return false;
}

Answer 4

I am late the party, but I wanted to point out one benefit that I found to using a magic square, namely that it can be used to get a reference to the square that would cause the win or loss on the next turn, rather than just being used to calculate when a game is over.

Take this magic square:

4 9 2
3 5 7
8 1 6

First, set up an scores array that is incremented every time a move is made. See this answer for details. Now if we illegally play X twice in a row at [0,0] and [0,1], then the scores array looks like this:

[7, 0, 0, 4, 3, 0, 4, 0];

And the board looks like this:

X . .
X . .
. . .

Then, all we have to do in order to get a reference to which square to win/block on is:

get_winning_move = function() {
  for (var i = 0, i < scores.length; i++) {
    // keep track of the number of times pieces were added to the row
    // subtract when the opposite team adds a piece
    if (scores[i].inc === 2) {
      return 15 - state[i].val; // 8
    }
  }
}

In reality, the implementation requires a few additional tricks, like handling numbered keys (in JavaScript), but I found it pretty straightforward and enjoyed the recreational math.

Answer 5

I don't know Java that well, but I do know C, so I tried adk's magic square idea (along with Hardwareguy's search restriction).

// tic-tac-toe.c
// to compile:
//  % gcc -o tic-tac-toe tic-tac-toe.c
// to run:
//  % ./tic-tac-toe
#include <stdio.h>

// the two types of marks available
typedef enum { Empty=2, X=0, O=1, NumMarks=2 } Mark;
char const MarkToChar[] = "XO ";

// a structure to hold the sums of each kind of mark
typedef struct { unsigned char of[NumMarks]; } Sum;

// a cell in the board, which has a particular value
#define MAGIC_NUMBER 15
typedef struct {
  Mark mark;
  unsigned char const value;
  size_t const num_sums;
  Sum * const sums[4];
} Cell;

#define NUM_ROWS 3
#define NUM_COLS 3

// create a sum for each possible tic-tac-toe
Sum row[NUM_ROWS] = {0};
Sum col[NUM_COLS] = {0};
Sum nw_diag = {0};
Sum ne_diag = {0};

// initialize the board values so any row, column, or diagonal adds to
// MAGIC_NUMBER, and so they each record their sums in the proper rows, columns,
// and diagonals
Cell board[NUM_ROWS][NUM_COLS] = { 
  { 
    { Empty, 8, 3, { &row[0], &col[0], &nw_diag } },
    { Empty, 1, 2, { &row[0], &col[1] } },
    { Empty, 6, 3, { &row[0], &col[2], &ne_diag } },
  },
  { 
    { Empty, 3, 2, { &row[1], &col[0] } },
    { Empty, 5, 4, { &row[1], &col[1], &nw_diag, &ne_diag } },
    { Empty, 7, 2, { &row[1], &col[2] } },
  },
  { 
    { Empty, 4, 3, { &row[2], &col[0], &ne_diag } },
    { Empty, 9, 2, { &row[2], &col[1] } },
    { Empty, 2, 3, { &row[2], &col[2], &nw_diag } },
  }
};

// print the board
void show_board(void)
{
  size_t r, c;
  for (r = 0; r < NUM_ROWS; r++) 
  {
    if (r > 0) { printf("---+---+---\n"); }
    for (c = 0; c < NUM_COLS; c++) 
    {
      if (c > 0) { printf("|"); }
      printf(" %c ", MarkToChar[board[r][c].mark]);
    }
    printf("\n");
  }
}


// run the game, asking the player for inputs for each side
int main(int argc, char * argv[])
{
  size_t m;
  show_board();
  printf("Enter moves as \"<row> <col>\" (no quotes, zero indexed)\n");
  for( m = 0; m < NUM_ROWS * NUM_COLS; m++ )
  {
    Mark const mark = (Mark) (m % NumMarks);
    size_t c, r;

    // read the player's move
    do
    {
      printf("%c's move: ", MarkToChar[mark]);
      fflush(stdout);
      scanf("%d %d", &r, &c);
      if (r >= NUM_ROWS || c >= NUM_COLS)
      {
        printf("illegal move (off the board), try again\n");
      }
      else if (board[r][c].mark != Empty)
      {
        printf("illegal move (already taken), try again\n");
      }
      else
      {
        break;
      }
    }
    while (1);

    {
      Cell * const cell = &(board[r][c]);
      size_t s;

      // update the board state
      cell->mark = mark;
      show_board();

      // check for tic-tac-toe
      for (s = 0; s < cell->num_sums; s++)
      {
        cell->sums[s]->of[mark] += cell->value;
        if (cell->sums[s]->of[mark] == MAGIC_NUMBER)
        {
          printf("tic-tac-toe! %c wins!\n", MarkToChar[mark]);
          goto done;
        }
      }
    }
  }
  printf("stalemate... nobody wins :(\n");
done:
  return 0;
}

It compiles and tests well.

% gcc -o tic-tac-toe tic-tac-toe.c
% ./tic-tac-toe
     |   |
  ---+---+---
     |   |
  ---+---+---
     |   |
  Enter moves as " " (no quotes, zero indexed)
  X's move: 1 2
     |   |
  ---+---+---
     |   | X
  ---+---+---
     |   |
  O's move: 1 2
  illegal move (already taken), try again
  O's move: 3 3
  illegal move (off the board), try again
  O's move: 2 2
     |   |
  ---+---+---
     |   | X
  ---+---+---
     |   | O
  X's move: 1 0
     |   |
  ---+---+---
   X |   | X
  ---+---+---
     |   | O
  O's move: 1 1
     |   |
  ---+---+---
   X | O | X
  ---+---+---
     |   | O
  X's move: 0 0
   X |   |
  ---+---+---
   X | O | X
  ---+---+---
     |   | O
  O's move: 2 0
   X |   |
  ---+---+---
   X | O | X
  ---+---+---
   O |   | O
  X's move: 2 1
   X |   |
  ---+---+---
   X | O | X
  ---+---+---
   O | X | O
  O's move: 0 2
   X |   | O
  ---+---+---
   X | O | X
  ---+---+---
   O | X | O
  tic-tac-toe! O wins!
% ./tic-tac-toe
     |   |
  ---+---+---
     |   |
  ---+---+---
     |   |
  Enter moves as " " (no quotes, zero indexed)
  X's move: 0 0
   X |   |
  ---+---+---
     |   |
  ---+---+---
     |   |
  O's move: 0 1
   X | O |
  ---+---+---
     |   |
  ---+---+---
     |   |
  X's move: 0 2
   X | O | X
  ---+---+---
     |   |
  ---+---+---
     |   |
  O's move: 1 0
   X | O | X
  ---+---+---
   O |   |
  ---+---+---
     |   |
  X's move: 1 1
   X | O | X
  ---+---+---
   O | X |
  ---+---+---
     |   |
  O's move: 2 0
   X | O | X
  ---+---+---
   O | X |
  ---+---+---
   O |   |
  X's move: 2 1
   X | O | X
  ---+---+---
   O | X |
  ---+---+---
   O | X |
  O's move: 2 2
   X | O | X
  ---+---+---
   O | X |
  ---+---+---
   O | X | O
  X's move: 1 2
   X | O | X
  ---+---+---
   O | X | X
  ---+---+---
   O | X | O
  stalemate... nobody wins :(
%

That was fun, thanks!

Actually, thinking about it, you don't need a magic square, just a count for each row/column/diagonal. This is a little easier than generalizing a magic square to n × n matrices, since you just need to count to n.

Answer 6

a non-loop way to determine if the point was on the anti diag:

`if (x + y == n - 1)`