BoardGameEngines is a Python3 library that provides a variety of simple and fast board game engines.

The Goal of this project is to provide a simple and standard API that allows to easily train and test AI agents on multiple games.

Games are implemented in both native python for use on all platforms and Rust for better performances. By default, The engine is the Rust one, but the python one is always available under the PythonEngine of each game package.

• Ultimate Tic-Tac-Toe
• Checkers
• Avalam
• Quoridor [Python engine only]

• Chess
• Go
• Riversi
• Nine men's morris
• Backgammon

The Library provide the engines in the form of a BoardState class for each game and a generic Game class.

To create AI players for the games, an Abstract class AbsPlayer is given as a base for custom players. A random player RandomPlayer is also provided as an example and benchmark for custom players.

Here's a code sample to run a game between two random players:

from GameEngines import Game, RandomPlayer
from GameEngines.Avalam import BoardState as AvalamBoard
# To get the python engine instead of the default one
# from GameEngines.Avalam.PythonEngine import BoardState as AvalamBoard

game = Game(AvalamBoard, RandomPlayer(), RandomPlayer())
game.play(5) # plays the next 5 moves
game.play_full() # plays the rest of the game

print(game.winner) # shows the winner of the game

A to create a player, it must implement the play method. You can also give a name to your player by putting it in the _name parameter or by implementing the name property.

from GameEngines import AbsPlayer, AbsBoardState
from random import choice


class RandomPlayer(AbsPlayer):
    def __init__(self, p_name: str = None):
        self._name = p_name # giving a name to the player

    def play(self, board: AbsBoardState, moves: set, pid: int):
        return choice(list(moves)) # randomly choosing a move

    @property
    def name(self) -> str:
        return self._name

The play method gets the relevant BoardState, the list of legal move for it's next turn and it's player number.

The binding between Python and Rust is done using PyO3 and the Maturin toolchain.

once Maturin is installed, building the library can be done using

maturin develop

To build the python library, simply use

maturin build -r