Kalaha

Installation

This game requires additional Python libraries. Install them using pip command.

pip install -r requirements.txt

Using C++ extensions

To speed up best move search CMiniMax algorithm has been written in C++. Build them by using following command.

python setup.py build

Read more about using C++ in Python code here.

Running server

From source

Use server argument to start the server. By running following command server will listen for connections on 127.0.0.1:20202 with turn timeout of 60 seconds.

python -m kalaha server -H 127.0.0.1 -p 20202 -tt 60

Using Docker

Update environment variables in docker-compose.yml file and spin up containers.

docker-compose up

Running client

Human player

Start client and connect with server IP address and port autofilled.

python -m kalaha client -H 127.0.0.1 -p 20202

Auto player

Auto player can be configured via in-game GUI or command arguments. Note: You will not be able to make moves manually.

Start client, connect to server and enable autoplay. This command configures minimax algorithm to search for best move in tree of depth 4. Delay of auto move will be 5 seconds.

python -m kalaha client -ap -apd 5 -md 4

You can enable alpha-beta pruning by passing -ab argument.

python -m kalaha client -ap -apd 5 -md 4 -ab

You can enable iterative deepening by passing -id argument.

python -m kalaha client -ap -apd 5 -md 4 -id

To enable move highlighting, pass -hm argument.

python -m kalaha client -hm -md 4

FAQ

What game modes are available?

Human vs. Human, Human vs. Auto player, Auto player vs. Auto player.

Is input validation done on the server side or client side?

Every move including pit selection is validated on the server side.

How auto game works?

Autoplayer uses minimax algorithm to find the best move. Optimalizations were made to improve the performance and win rate.

• iterative deepening
• alpha-beta pruning
• tree is generated level by level when minimax traverses the tree
• auto player is written in C++

Important: Algorithm is run on client side. It is running in separate thread, but setting high tree depth can cause the client to freeze and exceed the player's time limit. Recommended tree depth is 4.

What Python libraries are used?

• qtpy - abstraction layer for PyQt5/PySide2
• PyQt5 - Qt5 bindings for Python
• qrainbowstyle - custom windows and widgets style
• QtPyNetwork - high level network library
• coloredlogs - colored logging