raylib 5.0.0.4

Python CFFI bindings for Raylib

Python Bindings for Raylib 5.0

Chatroom: Discord or Matrix

New CFFI API static bindings.

• Automatically generated to be as close as possible to original Raylib.
• Faster, fewer bugs and easier to maintain than ctypes.
• Commercial-friendly license.
• Docstrings and auto-completion.
• Now includes extra libraries: raymath, raygui, rlgl, physac and GLFW

Full documentation

Quickstart

pip3 install raylib

from pyray import *
init_window(800, 450, "Hello")
while not window_should_close():
    begin_drawing()
    clear_background(WHITE)
    draw_text("Hello world", 190, 200, 20, VIOLET)
    end_drawing()
close_window()

Installation

First make sure you have the latest pip installed:

python3 -m pip install --upgrade pip

Then install

python3 -m pip install setuptools
python3 -m pip install raylib

On most platforms it should install a binary wheel (Windows 10 x64, MacOS 12 x64/arm64, Linux Ubuntu2004 x64/arm64).

If yours isn't available then pip will attempt to build from source, in which case you will need to have Raylib development libs installed, e.g. using homebrew, apt, etc.

Raspberry Pi

Using on Rasperry Pi

Dynamic binding version

There is now a separate dynamic version of this binding:

python3 -m pip install raylib_dynamic

It works on some systems where the static version doesn't, but be sure to read these caveats before using it

Beta testing

If you find a bug, it may be fixed in the latest dev release. You can install an alpha or beta version by specifying the exact version number like this:

python3 -m pip install raylib==4.2.0.0.dev4

Problems?

If it doesn't work, try to build manually.. If that works then submit an issue to let us know what you did.

If you need help you can try asking on Discord.

If it still doesn't work, submit an issue.

How to use

There are two APIs, you can use either or both:

If you are familiar with C coding and the Raylib C library and you want to use an exact copy of the C API

Use the C API.

If you prefer a slightly more Pythonistic API and don't mind it might be slightly slower

Use the Python API.

Running in a web browser

Pygbag >=0.8.7 supports running in a web browser.

Make a folder my_project with a file main.py:

# /// script
# dependencies = [
#     "cffi",
#     "raylib"
# ]
# ///
import asyncio
import platform
from pyray import *

async def main():   # You must have an async main function
    init_window(500, 500, "Hello")
    platform.window.window_resize()  # You must add this line
    while not window_should_close():
        begin_drawing()
        clear_background(WHITE)
        draw_text("Hello world", 190, 200, 20, VIOLET)
        end_drawing()
        await asyncio.sleep(0) # You must call this in your main loop
    close_window()

asyncio.run(main())

Then to create the web files and launch a web server:

python3.12 -m pip install --user --upgrade pygbag
python3.12 -m pygbag --PYBUILD 3.12 --ume_block 0 --template noctx.tmpl my_project

Point your browser to http://localhost:8000

This is all done by Pygbag rather than by me, so you should probably contact them with any issues. Carefully read all their documentation.

It does work for most of these examples

App showcase

Tanki

Alloy Bloxel Editor

Add your app here!

RLZero

A related library (that is a work in progress!):

A simplified API for Raylib for use in education and to enable beginners to create 3d games

Help wanted

• Converting more examples from C to Python
• Testing on more platforms

License (updated)

The bindings are now under the Eclipse Public License, so you are free to statically link and use in non-free / proprietary / commercial projects!

Performance

If you need more performance, do in this order:

1. Use Pypy rather than standard CPython. It is much, much faster and will make more difference than any other optimisations you might do.

2. Every call to C is costly, so it's slightly faster if you use Python data structures and functions when calculating in your update loop and then only convert them to C data structures when you have to call the C functions for drawing.

3. The raylib.* functions are potentially slightly faster than the pyray.* equivalents, so if you need a tiny bit more performance you can switch your inner loop functions to these.

4. There is a version of Python that is faster than Pypy: GraalPy. However it's not fully compatible with all Python packages. It doesn't work with CFFI and so doesn't work with this binding. But it is compatible with the Java binding, Jaylib! There is an example of this here: https://github.com/electronstudio/megabunny/tree/master/raylib-python-jaylib

Bunnymark

Library	Implementation	Bunnies (60 FPS)	Percentage
Raylib 5.0	C	180000	100%
Raylib Python CFFI 5.0.0.2	Python 3.12	10500	5.8%
Raylib Python CFFI 5.0.0.2	Pypy 3.10	95000	53%
Raylib 3.7	C	168100	100%
Raylib Python CFFI 3.7	Pypy 3.7	33800	20%
Raylib Python CFFI 3.7	Python 3.9	7700	4.5%
Raylib Python CFFI 3.7	Python 3.9 Nuitka	8600	5.1%
Raylib Python CFFI 3.7 Dynamic	Python 3.9	6300	3.7%

See also https://github.com/electronstudio/megabunny/

Packaging your app

You can create a standalone binary using the Nuitka compiler. For example, here is how to package Bunnymark:

pip3 install nuitka
cd examples/textures
python3 -m nuitka --onefile --linux-onefile-icon resources/wabbit_alpha.png textures_bunnymark.py

Advert

RetroWar: 8-bit Party Battle is out now. Defeat up to 15 of your friends in a tournament of 80s-inspired retro mini games.

Coding Games With Pygame Zero & Python is a book for Python beginners.