BvY Chess

Introduction

BvY Chess is a chess game built in Ruby. To play: download the repo, navigate to the main folder in the terminal and enter ./play.rb (you'll need to have Ruby installed, and you may need to enter chmod +x play.rb in the shell to make the file executable). A prompt will ask if this game is vs. the AI or vs. another person.

In the game: move the cursor around the board with the arrow keys. Hit the Spacebar to select which piece to move, and then again to select which space to move it to. If the space you select is not a valid move for your piece, you will be re-prompted to select a piece.

The program ends when a player is in checkmate (or, just hit Ctrl-C to terminate early).

Features

Move objects:

Taking some influence from my experiences with React.js/Redux - specifically, how the Redux architecture passes data down from an initial 'store' to various components - the potential moves (on a given turn) for the computer player are added to a hash, which is then passed to different methods which algorithmically filter the hash until it contains only the most desirable potential moves for that turn (i.e. a move that results in checkmate is prioritized over a move that results in check, a move that results in a captured bishop is prioritized over a move that results in a captured pawn, etc.).

This means that it would be easy to, for example, have the AI choose moves completely at random, or have the AI select from a set of less-desirable moves - it's just a matter of setting restrictions on which moves will be included in the final instance of the hash.

These are example methods which return AI move data to the 'Game' class. The AI is choosing from within a set of valuable moves, as defined in the move_hierarchy method.

def choose_piece
  all_pieces = playable_pieces(@color)
  @possible_pieces = move_hierarchy(all_pieces)
  @chosen_piece = choose_random_piece(@possible_pieces)
end

def play_move
  choose_random_move(@possible_pieces, @chosen_piece)
end

Single board:

Potential subsequent moves (necessary to determine moves for the computer player, and for determining whether or not a player is in checkmate) are calculated by modifying and then resetting a single instance of the board class. This ends up being less space-intensive than, for example, a solution which continually duplicates the board to calculate future moves.

For example, here's a method which checks to see if a given move (entered as coordinates) will result in a given color being put into check:

def puts_color_in_check?(coords, color)
  answer = :no
  piece = @grid[coords[1]][coords[0]]
  prev_piece = @grid[coords[3]][coords[2]]
  test_move(coords, piece)
  if in_check?(color)
    answer = :mate if checkmate?(color)
    answer = :check
  end
  reset_move(coords, piece, prev_piece)
  answer
end

Object-oriented piece construction:

The various pieces were constructed using object-oriented programming patterns to maintain consistency. For example, all pieces inherit from a Piece class with variables for color, position and type, while the Knight and King classes inherit methods from a Steppable module to determine valid moves.

Future

Going forward, I plan to build-out the computer player by using polytree data structures to store chains of potential moves, allowing the AI to choose its moves based on long-term outcome.