Sapphire Chess

Sapphire Chess is a command line-based chess game with an algebraic notation input system, complete chess rules, a beautiful interface and a functional AI. It provides three game modes: Human vs. Human, Human vs. AI, and AI vs. AI.

Please, visit https://medium.com/@lucas.sorribes/nostromo-my-ruby-chess-journey-part-i-7ef544b547a5 for a very detailed account of how I wrote this game.

Supported Rubies: 3.1, 3.0, 2.7.

The game has not been tested on older versions, but it might still work.

Playing within the VS Code terminal is highly recommended because it renders the icons better.

• A beautiful board with easy-to-distinguish colors for white and black pieces.
• Fully functional AI
• Three levels of difficulty.
• Three game modes: human vs. computer, human vs. human.
• Full chess movement rules implementation, including castling and en passant, for both the human and the computer player.
• Accepts algebraic notation for movements, with human input validation.
• Material score display.
• Player's last move display.

Simply, build the image (it's very lightweight) with:

docker build -t sapphire-chess .

And run it with:

docker run -it sapphire-chess

Install with:

gem install 'saphhire-chess'

And execute from the command prompt with:

sapphire-chess

Currently, the icons are only loaded properly from within VS Code terminals, so, for the moment, I highly recommend executing this command from there.

You can also create a new .rb file with this content and execute it:

require 'sapphire-chess'

Engine.new.play

The difficulty mode is based on how many turns the computer can think ahead (the depth of the tree generated by AI#minimax, max. 3), the aggressiveness of the pieces, and the enemy's carelessness.

My recommendation is to play the game on the hard mode, as this was the original intention when I designed it. I introduced the other difficulties based on the suggestions from those who weren't able to beat the machine. However, the hard mode, due to the current Minimax implementation, is a little slow; I'm working on ways to improve the performance of this method.

The AI is based on a Minimax algorithm with Alpha-Beta pruning. This algorithm generates possible outcomes (children) for a provisional move: Each branch represents the enemy's possible move in the next turn as an answer to the provisional move; (i.e.: if current player is white [the maximizing player], it generates every possible movement for the next player, black [the minimizing player], who will choose the best possible move, and so on.

The best (relative to each player) possible outcome for each move will determine what move is chosen) The Alpha-Beta 'prunes' the tree: it makes the search more efficient, removing unnecessary branches, resulting in a faster process.

Lucas Sorribes, released under the MIT license.