pocketpy

pocketpy: portable python 3.x interpreter

pocketpy is a portable Python 3.x interpreter, written in C11. It aims to be an alternative to Lua for game scripting, with elegant syntax, powerful features and competitive performance. pocketpy has no dependencies other than the C standard library, which can be easily integrated into your C/C++ project. Developers are able to write Python bindings via C-API or pybind11 compatible interfaces.

Please see https://pocketpy.dev for details and try the following resources.

• Live Python Demo: Run Python code in your browser
• Live C Examples: Explore C-APIs in your browser
• Godot Extension: Use pocketpy in Godot Engine

Supported Platforms

pkpy should work on any platform with a C11 compiler. These platforms are officially tested.

• Windows 64-bit
• Linux 64-bit / 32-bit
• macOS 64-bit
• Android 64-bit / 32-bit
• iOS 64-bit
• Emscripten 32-bit
• Raspberry Pi OS 64-bit

Quick Start

You have two options to integrate pkpy into your project.

Use the single header file

Download the pocketpy.h and pocketpy.c on our GitHub Release page. And #include it in your project.

Use CMake

Clone the whole repository as a submodule into your project, In your CMakelists.txt, add the following lines:

add_subdirectory(pocketpy)
target_link_libraries(<your_target> pocketpy)

See CMakeLists.txt for details.

It is safe to use main branch in production if CI badge is green.

Compile Flags

To compile it with your project, these flags must be set:

• --std=c11 flag must be set
• For MSVC, /utf-8 and /experimental:c11atomics flag must be set
• NDEBUG macro should be defined for release build, or you will get poor performance

For amalgamated build, run python amalgamate.py to generate pocketpy.c and pocketpy.h in amalgamated/ directory.

Example

#include "pocketpy.h"
#include <stdio.h>

static bool int_add(int argc, py_Ref argv) {
    PY_CHECK_ARGC(2);
    PY_CHECK_ARG_TYPE(0, tp_int);
    PY_CHECK_ARG_TYPE(1, tp_int);
    py_i64 a = py_toint(py_arg(0));
    py_i64 b = py_toint(py_arg(1));
    py_newint(py_retval(), a + b);
    return true;
}

int main() {
    // Initialize pocketpy
    py_initialize();

    // Hello world!
    bool ok = py_exec("print('Hello world!')", "<string>", EXEC_MODE, NULL);
    if(!ok) goto __ERROR;

    // Create a list: [1, 2, 3]
    py_Ref r0 = py_getreg(0);
    py_newlistn(r0, 3);
    py_newint(py_list_getitem(r0, 0), 1);
    py_newint(py_list_getitem(r0, 1), 2);
    py_newint(py_list_getitem(r0, 2), 3);

    // Eval the sum of the list
    py_Ref f_sum = py_getbuiltin(py_name("sum"));
    py_push(f_sum);
    py_pushnil();
    py_push(r0);
    ok = py_vectorcall(1, 0);
    if(!ok) goto __ERROR;

    printf("Sum of the list: %d\n", (int)py_toint(py_retval()));  // 6

    // Bind native `int_add` as a global variable
    py_newnativefunc(r0, int_add);
    py_setglobal(py_name("add"), r0);

    // Call `add` in python
    ok = py_exec("add(3, 7)", "<string>", EVAL_MODE, NULL);
    if(!ok) goto __ERROR;

    py_i64 res = py_toint(py_retval());
    printf("Sum of 2 variables: %d\n", (int)res);  // 10

    py_finalize();
    return 0;

__ERROR:
    py_printexc();
    py_finalize();
    return 1;
}

Features

Name	Example	Supported
If Else	if..else..elif	✅
Loop	for/while/break/continue	✅
Function	def f(x,*args,y=1):	✅
Subclass	class A(B):	✅
List	[1, 2, 'a']	✅
ListComp	[i for i in range(5)]	✅
Slice	a[1:2], a[:2], a[1:]	✅
Tuple	(1, 2, 'a')	✅
Dict	{'a': 1, 'b': 2}	✅
F-String	f'value is {x}'	✅
Unpacking	a, b = 1, 2	✅
Star Unpacking	a, *b = [1, 2, 3]	✅
Exception	raise/try..catch..finally	✅
Dynamic Code	eval()/exec()	✅
Reflection	hasattr()/getattr()/setattr()	✅
Import	import/from..import	✅
Context Block	with <expr> as <id>:	✅
Type Annotation	def f(a:int, b:float=1)	✅
Generator	yield i	✅
Decorator	@cache	✅
Match Case	match code: case 200:	✅

Performance

Currently, pkpy is as fast as cpython 3.9. Performance results for cpython 3.9 are applicable to for pkpy.

See https://pocketpy.dev/performance/ for details.

And these are the results of the primes benchmark on Intel i5-12400F, WSL (Ubuntu 20.04 LTS), which roughly reflects the performance among c++, lua, pkpy and cpython.

name	version	time	file
c++	gnu++11	0.104s ■□□□□□□□□□□□□□□□	benchmarks/primes.cpp
lua	5.3.3	1.576s ■■■■■■■■■□□□□□□□	benchmarks/primes.lua
pkpy	1.2.7	2.385s ■■■■■■■■■■■■■□□□	benchmarks/primes.py
cpython	3.8.10	2.871s ■■■■■■■■■■■■■■■■	benchmarks/primes.py

Used By

	Description
godot-pocketpy	Godot extension for using pocketpy in Godot Engine.
TIC-80	TIC-80 is a fantasy computer for making, playing and sharing tiny games.
py-js	Python3 externals for Max / MSP.
crescent	Crescent is a cross-platform 2D fighting and beat-em-up game engine.
orxpy	Python extension for orx engine.
CANopenTerm	Open-source software tool for CANopen CC networks and devices.

Submit a pull request to add your project here.

Contribution

All kinds of contributions are welcome.

• Submit a Pull Request
  • fix a bug
  • add a new feature
• Open an Issue
  • any suggestions
  • any questions

If you find pkpy useful, consider star this repository (●'◡'●)

Sponsor this project

You can sponsor this project via these ways.

• Github Sponsors
• Buy me a coffee

Your sponsorship will help us develop pkpy continuously.

Reference

• cpython

The official implementation of Python programming language.

• byterun

An excellent learning material. It illustrates how Python's virtual machine works.

Star History

License

MIT License