RG3

RG3 - is a backend & frontend for processing & analyzing C++ code. It provides information about types into Python frontend to future codegen stage. We've using LLVM as our backend to prepare code.

See our GitHub for more details.

Install

Make sure that your system has clang (any version):

• macOS: you need to install XCode (tested on 15.x but should work everywhere)
• Window: you need to install clang 17.x or later and add it into PATH
• Linux: gcc & g++ at least 13 version
• Other platforms & archs: Contact us in our GitHub.

It's a better way to use RG3 inside virtualenv:

python3 -m venv venv
source ./venv/bin/activate
pip install rg3py

Usage:

Sample code analyze code from buffer:

from rg3py import CodeAnalyzer, CppStandard, CppCompilerIssueKind, CppTypeKind, CppClass, FunctionArgument
from typing import List


def prepare_args(args: List[FunctionArgument]) -> str:
    return ','.join(f'{arg.type_info.get_name()} {arg.name}' for arg in args)


analyzer: CodeAnalyzer = CodeAnalyzer.make()
analyzer.set_code("""
    namespace my::cool::name_space {
        /**
        * @runtime
        **/
        enum class ECoolEnum : int {
            CE_FIRST_ENTRY = 0,
            CE_ANOTHER_ENTRY = 0xFFEE,
            CE_DUMMY = 256
        };

        /// @runtime
        struct MyGeniusStruct
        {};

        struct SomeThirdPartyStruct
        {
            float fProperty = 42.f;
            MyGeniusStruct sGenius {};

            static bool IsGeniusDesc(bool bCanReplace) const;
        };

        template <typename T> struct ThirdPartyRegistrator {};
        template <> struct 
            __attribute__((annotate("RG3_RegisterRuntime")))
            __attribute__((annotate("RG3_RegisterField[fProperty:Property]")))
            __attribute__((annotate("RG3_RegisterField[sGenius:GeniusData]")))
            __attribute__((annotate("RG3_RegisterFunction[IsGeniusDesc]")))
        ThirdPartyRegistrator<SomeThirdPartyStruct>
        {
            using Type = SomeThirdPartyStruct;
        };
    }
    """)

analyzer.set_cpp_standard(CppStandard.CXX_17)
analyzer.analyze()

for t in analyzer.types:
    print(f"We have a type {t.pretty_name} ({t.kind})")

    if t.kind == CppTypeKind.TK_STRUCT_OR_CLASS:
        as_class: CppClass = t
        for prop in as_class.properties:
            print(f"\tProperty {prop.name} (aka {prop.alias}) of type {prop.type_info.get_name()}")

        for func in as_class.functions:
            args_as_str: str = prepare_args(func.arguments)
            print(f"\tFunction {'static' if func.is_static else ''} {func.return_type.get_name()} {func.name}({args_as_str}){' const' if func.is_const else ''}")

expected output is

We have a type my::cool::name_space::ECoolEnum (TK_ENUM)
We have a type my::cool::name_space::MyGeniusStruct (TK_STRUCT_OR_CLASS)
We have a type my::cool::name_space::SomeThirdPartyStruct (TK_STRUCT_OR_CLASS)
	Property fProperty (aka Property) of type float
	Property sGenius (aka GeniusData) of type MyGeniusStruct
	Function static bool IsGeniusDesc(bool bCanReplace)

Another good use case is constexpr evaluation feature:

import rg3py
from typing import Union, List, Dict

evaluator: rg3py.CodeEvaluator = rg3py.CodeEvaluator.make_from_system_env()

evaluator.set_compiler_config({
    "cpp_standard" : rg3py.CppStandard.CXX_20
})

result: Union[Dict[str, any], List[rg3py.CppCompilerIssue]] = evaluator.eval("""
        #include <type_traits>
        
        class Simple {};
        class Base {};
        class Inherited : public Base {};
        
        constexpr bool r0 = std::is_base_of_v<Base, Inherited>;
        constexpr bool r1 = std::is_base_of_v<Base, Simple>;
        """, ["r0", "r1"])

print(result)

output will be

{'r1': False, 'r0': True}

and that's great feature to make some checks like type should be inherited from A, have some methods and etc...

And this is independent of your current environment, only C++ STL library should be found!

Features

• Supported Windows (x86_64), Linux (x86_64) and macOS (x86_64 and ARM64)
• Supported C++03, 11, 14, 17, 20, 23, 26
• Supported threads in analysis on native side (see Tests/PyIntegration/test.py test: test_analyzer_context_sample for example)
• Statically linked, no external dependencies (except Clang instance on machine)
• Special macro definitions to hide unnecessary code
• Template specializations reporting
• Anonymous registration without changes in third party code
• Runtime constexpr C++ code evaluation with results extraction

Current limitations

Project focused on work around C/C++ headers (C++ especially). Feel free to fork project & add support of anything what you want :)

Third Party libraries

• LLVM - our main backend of C++ analysis
• Boost.Python - python support & process launcher
• fmt - string formatter
• googletest - for internal unit testing
• pytest - for python side unit testing