PenQuest Environment

This repository contains the code for the Reinforcement Learning environment for the game PenQuest.

Installation

To install this package please first install the package penquest-pkgs. Then install the this package, navigate into the folder and execute the following command:

    pip install -e .

The Penquest environment requires a 'default_config.ini' file. This might be already be present during installation and could be located in your environment at Lib/site-packages/penqest-pks. In case it is not, please create a 'default_config.ini' file in your working directory. In any other location, you need to provide the file path via the corresponding parameter.

The content of 'default_config.ini' should look like this:

[internal]
port = 50000

[external]
host = env.pen.quest
api_key = ""

[timeouts]
connection_start = 300
connection_restart = 2

Alternatively you can also provide the API key via the environment variable 'API_KEY'. For your convenience this is best done via an '.env' file.

Due to limited computation ressources, please contact the authors for an API key to access the PenQuest API.

Usage

The following is a minimal viable example usage of the environment for random actions:

import random as rand
import os
import gymnasium as gym
import penquest_env
from penquest_env import Scenario, SlotType, PlayerType, BotType

api_key = os.getenv("API_KEY")
OPTIONS = {
    "scenario": Scenario.INFRASTRUCUTRE_SCENARIO_MEDIUM_1_ALL_ACTIONS,
    'slot': SlotType.ATTACK,
    'seed': 1234,
    'players': [
        { 'type': PlayerType.BOT, 'bot_type': BotType.ADVANCED_BOT },
    ],
}

if __name__ == "__main__":
    penquest_env.start(api_key)
    env = gym.make('penquest_env/PenQuest-v0', options=OPTIONS)
    obs, info = env.reset(options=OPTIONS)
    done = False
    while not done:
        action = rand.choice(info["valid_actions"])
        obs, reward, terminated, truncated, info = env.step(action)
        done = terminated or truncated
    env.close()