An easy-to-use library for writing AI Bots for StarCraft II in Python 3. The ultimate goal is simplicity and ease of use, while still preserving all functionality. A really simple worker rush bot should be no more than twenty lines of code, not two hundred. However, this library intends to provide both high and low level abstractions.
This library (currently) covers only the raw scripted interface. At this time I don't intend to add support for graphics-based interfaces.
The documentation can be found here. For bot authors, looking directly at the files in the sc2 folder can also be of benefit: bot_ai.py, unit.py, units.py, client.py, game_info.py and game_state.py. Most functions in those files have docstrings, example usages and type hinting.
I am planning to change this fork more radically than the main repository, for bot performance benefits and to add functions to help new bot authors. This may break older bots in the future, however I try to add deprecation warnings to give a heads up notification. This means that the video tutorial made by sentdex is outdated and does no longer directly work with this fork.
For a list of ongoing changes and differences to the main repository of Dentosal, check here.
By installing this library you agree to be bound by the terms of the AI and Machine Learning License.
For this fork, you'll need Python 3.8 or newer.
Install the pypi package:
pip install --upgrade burnysc2
or directly from develop branch:
pip install poetry
pip install --upgrade --force-reinstall https://github.com/BurnySc2/python-sc2/archive/develop.zip
Both commands will use the sc2 library folder, so you will not be able to have Dentosal's and this fork installed at the same time, unless you use virtual environments or poetry.
You'll need a StarCraft II executable. If you are running Windows or macOS, just install SC2 from blizzard app.
You can install StarCraft II on Linux with Wine, Lutris or even the Linux binary, but the latter is headless so you cannot actually see the game. Starcraft II can be directly installed from Battlenet once it is downloaded with Lutris. By default, it will be installed here:
/home/burny/Games/battlenet/drive_c/Program Files (x86)/StarCraft II/
Next, set the following environment variables (either globally or within your development environment, e.g. Pycharm: Run -> Edit Configurations -> Environment Variables):
SC2PF=WineLinux
WINE=/usr/bin/wine
# Or a wine binary from lutris:
# WINE=/home/burny/.local/share/lutris/runners/wine/lutris-4.20-x86_64/bin/wine64
# Default Lutris StarCraftII Installation path:
SC2PATH='/home/burny/Games/battlenet/drive_c/Program Files (x86)/StarCraft II/'
When running WSL in Windows, python-sc2 detects WSL by default and starts Windows Starcraft 2 instead of Linux Starcraft 2.
If you wish to instead have the game played in Linux, you can disable this behavior by setting SC2_WSL_DETECT
environment variable to "0". You can do this inside python with the following code:
import os
os.environ["SC2_WSL_DETECT"] = "0"WSL version 1 should not require any configuration. You may be asked to allow Python through your firewall.
When running WSL version 2 you need to supply the following environment variables so that your bot can connect:
SC2CLIENTHOST=<your windows IP>
SC2SERVERHOST=0.0.0.0
If you are adding these to your .bashrc, you may need to export your environment variables by adding:
export SC2CLIENTHOST
export SC2SERVERHOSTYou can find your Windows IP using ipconfig /all from PowerShell.exe or CMD.exe.
You will need maps to run the library.
Official Blizzard map downloads are available from Blizzard/s2client-proto.
Extract these maps into their respective subdirectories in the SC2 maps directory.
e.g. install-dir/Maps/Ladder2017Season1/
Maps that are run on the SC2 AI Ladder and SC2 AI Arena can be downloaded from the sc2ai wiki and the aiarena wiki.
Extract these maps into the root of the SC2 maps directory (otherwise ladder replays won't work).
e.g. install-dir/Maps/AcropolisLE.SC2Map
After installing the library, a StarCraft II executable, and some maps, you're ready to get started. Simply run a bot file to fire up an instance of StarCraft II with the bot running. For example:
python examples/protoss/cannon_rush.pyAs promised, worker rush in less than twenty lines:
from sc2 import maps
from sc2.player import Bot, Computer
from sc2.main import run_game
from sc2.data import Race, Difficulty
from sc2.bot_ai import BotAI
class WorkerRushBot(BotAI):
async def on_step(self, iteration: int):
if iteration == 0:
for worker in self.workers:
worker.attack(self.enemy_start_locations[0])
run_game(maps.get("Abyssal Reef LE"), [
Bot(Race.Zerg, WorkerRushBot()),
Computer(Race.Protoss, Difficulty.Medium)
], realtime=True)This is probably the simplest bot that has any realistic chances of winning the game. I have ran it against the medium AI a few times, and once in a while, it wins.
You can find more examples in the examples/ folder.
The API supports a number of options for configuring how it operates.
Set this to 'True' if your bot is issueing commands using self.do(Unit(Ability, Target)) instead of Unit(Ability, Target).
class MyBot(BotAI):
def __init__(self):
self.unit_command_uses_self_do = TrueSetting this to true improves bot performance by a little bit.
class MyBot(BotAI):
def __init__(self):
self.raw_affects_selection = TrueThe distance calculation method:
- 0 for raw python
- 1 for scipy pdist
- 2 for scipy cdist
class MyBot(BotAI):
def __init__(self):
self.distance_calculation_method: int = 2On game start or in any frame actually, you can set the game step. This controls how often your bot's step method is called.
Do not set this in the __init__ function as the client will not have been initialized yet!
class MyBot(BotAI):
def __init__(self):
pass # don't set it here!
async def on_start(self):
self.client.game_step: int = 2You have questions but don't want to create an issue? Join the Starcraft 2 AI Discord server or aiarena.net Discord server. Questions about this repository can be asked in text channel #python. There are discussions and questions about SC2 bot programming and this repository every day.
If you have any issues, ideas or feedback, please create a new issue. Pull requests are also welcome!
Git commit messages use imperative-style messages, start with capital letter and do not have trailing commas.
To run pre-commit hooks (which run autoformatting and autosort imports) you can run
poetry run pre-commit install
poetry run pre-commit run --all-files