rust_reversi_core

A Rust library for the game of Reversi (Othello) including game engine, AI players, and arena for playing games.

This is also the core implementation for rust_reversi.

See also the documentation.

Overview

This project provides:

• Complete Reversi game rule engine
• Multiple AI player implementations
• Arena for playing games (both local and network)
• Alpha-beta search engine implementation

Features

Board Engine

• 8x8 Reversi board management
• Legal move validation
• Move execution and piece flipping
• Pass detection
• Win condition checking
• Fast bitboard-based implementation

AI Players

Multiple AI strategies are implemented:

• Random Player - Makes random legal moves
• Piece Evaluator - Evaluates based on piece count difference
• Matrix Evaluator - Uses position weights for evaluation

You can also use your own Evaluator that implements the Evaluator trait.

Arena Features

• Local game support
• Network play over TCP/IP
• Automatic execution of multiple games between players
• Statistics collection (win rates, piece counts)
• Progress bar visualization

Search Engine

• Alpha-beta pruning implementation
• Iterative deepening
• Timeout control
• Pluggable evaluation functions

Installation

cargo add rust_reversi_core

Usage

Basic usage:

use rust_reversi_core::board::Board;

// Create a new board
let mut board = Board::new();

// Get legal moves
let legal_moves = board.get_legal_moves_vec();

// Make a move
board.do_move(legal_moves[0]).unwrap();

Using AI players:

use rust_reversi_core::search::{AlphaBetaSearch, MatrixEvaluator};

// Setup evaluator and search
let evaluator = MatrixEvaluator::new(matrix);
let search = AlphaBetaSearch::new(depth, Box::new(evaluator));

// Get best move
let best_move = search.get_move(&board);

Arena Usage

Running local games:

use rust_reversi_core::arena::LocalArena;

let mut arena = LocalArena::new(command1, command2, true);
arena.play_n(100).unwrap();
let (wins1, wins2, draws) = arena.get_stats();

Network games:

use rust_reversi_core::arena::{NetworkArenaServer, NetworkArenaClient};

// Server
let mut server = NetworkArenaServer::new(100, true).unwrap();
server.start("127.0.0.1".to_string(), 12345).unwrap();

// Client
let mut client = NetworkArenaClient::new(command);
client.connect("127.0.0.1".to_string(), 12345).unwrap();

Project Structure

• src/board.rs - Core game logic and board representation
• src/search/ - Search algorithms and evaluation functions
• src/arena/ - Local and network game coordination
• tests/ - Test cases and example players

Testing

Run the test suite:

cargo test

The test suite includes both unit tests and integration tests with example AI players.

Benchmarking

The project includes benchmarks for core functionality using Criterion:

cargo bench

Available benchmarks:

• Board operations (board benchmark)

  • Full game playthrough with random moves

• Search algorithms (search benchmark)

  • Alpha-beta search with various evaluators (depth 4):
    • Piece count evaluator
    • Legal moves evaluator
    • Matrix-based evaluator
    • Custom evaluator example

Each evaluator is tested with a small probability (ε=0.01) of making random moves to add variety.

License

MIT License

Author

neodymium6

Contributing

Contributions are welcome! Feel free to:

• Report bugs
• Suggest features
• Submit pull requests