CppSharp generates the wrong code

Question

TL;DR : CppSharp generates the wrong binding code for a particular C++ method signature. How do I use TypeMaps and or passes to fix this? Is there documentation or examples of type transformations I can use?

I'm using CppSharp to generate C# bindings for OBS studio, in order to write a plugin for it.

When generating code from OBS.h using the following generator:

class OBSBindingsGenerator: ILibrary
{
    public void Postprocess(Driver driver, ASTContext ctx)
    {
    }

    public void Preprocess(Driver driver, ASTContext ctx)
    {
    }

    public void Setup(Driver driver)
    {
        var options = driver.Options;
        options.GeneratorKind = CppSharp.Generators.GeneratorKind.CSharp;
        var module = options.AddModule("OBS");
        module.IncludeDirs.Add(@"path\to\libobs");
        module.Headers.Add("obs.h");
    }

    public void SetupPasses(Driver driver)
    {
    }
}

class Program
{
    static void Main(string[] args)
    {
        ConsoleDriver.Run(new OBSBindingsGenerator());
    }
}

I find that it generates:

    [SuppressUnmanagedCodeSecurity]
    [DllImport("OBS", CallingConvention = global::System.Runtime.InteropServices.CallingConvention.Cdecl,
        EntryPoint="video_format_get_parameters")]
    [return: MarshalAs(UnmanagedType.I1)]
    internal static extern bool VideoFormatGetParameters_0(global::OBS.VideoColorspace color_space, global::OBS.VideoRangeType range, float matrix, float min_range, float max_range);

and the corresponding public method

public static bool VideoFormatGetParameters(global::OBS.VideoColorspace color_space, global::OBS.VideoRangeType range, float[] matrix, float[] min_range, float[] max_range)
{
    ... // Code snipped for brevity
    fixed (float* __ptr4 = max_range)
    {
        var __arg4 = new global::System.IntPtr(__ptr4);
        // See here that __arg4 is an IntPtr argument and is being passed to a float parameter
        var __ret = __Internal.VideoFormatGetParameters_0(color_space, range, __arg2, __arg3, __arg4);
        return __ret;
    }
}

This obviously does not compile. The C++ definition is:

EXPORT bool video_format_get_parameters(enum video_colorspace color_space,
    enum video_range_type range, float matrix[16],
    float min_range[3], float max_range[3]);

In hand-written PInvoke, this can be fixed in many ways (declare a struct with the right number of float members, or a [MarshalAs(UnmanagedType.LPArray, SizeConst = 3)] and a float[] passthrough or declare the PInvoke parameter as an IntPtr).

My question then is: How do I use CppSharp and its code transformation tools (TypeMap? Pass? What?) to achieve any of the correct translations? I could fix this as a hack, but since I interop with my C++ libraries I'd prefer to learn the right way to do this with CppSharp (learn to fish).

Thanks in advance!

Answer 1

请尝试使用最新的CppSharp版本（0.8.14+），其中包含与数组参数有关的修复程序。