How to call c++ in-class functions (whose arguments are float pointers) in python with pybind11

The error you are seeing is because the C++ function myFunc expects a float* pointer, but the Python code is passing a ctypes.POINTER(ctypes.c_float) object. You need to cast the ctypes.POINTER(ctypes.c_float) object to a float* pointer before passing it to the myFunc function.

You can do this by changing the my_obj.my_func call in the Python code to the following:

my_obj.my_func(arr.ctypes.data_as(pybind11.c_float_p), arr.size)

Here, pybind11.c_float_p is a convenience typedef for float*. By using this typedef, you can avoid having to manually cast the ctypes.POINTER(ctypes.c_float) object to a float* pointer.

With this change, your Python code should work correctly and print the updated array.

The solution was to define a class to WRAP as follows.

#include <pybind11/pybind11.h>
#include <pybind11/numpy.h>
#include <pybind11/stl.h>

namespace py = pybind11;

class MyClass {
public:
  void myFunc(float *arr, int size) {
    for (int i = 0; i < size; i++) {
      arr[i] += 1.0f;
    }
  }
};

class MyClassWrapper {
private:
    MyClass my_py_class;
public:
    void myPyFunc(py::array_t<float> input_array, int size) {
        // Get a pointer to the underlying data
        float *data = static_cast<float *>(input_array.request().ptr);
        
        my_py_class.myFunc(data, size);
    }
};

PYBIND11_MODULE(example, m) {
    py::class_<MyClassWrapper>(m, "MyClassWrapper")
      .def(py::init<>())
      .def("myPyFunc", [](MyClassWrapper& wrapper, py::array_t<float> input_array, int size){
          wrapper.myPyFunc(input_array, size);
      });
}

The code for the caller is as follows.

import numpy as np
import sys
sys.path.append( "./build" )
import example

# Create an instance of MyClass
my_obj = example.MyClassWrapper()

# Create a NumPy array
arr = np.array([1.0, 2.0, 3.0], dtype=np.float32)

# Call myFunc with the NumPy array
my_obj.myPyFunc(arr, arr.size)

# Print the updated array
print(arr)  #[2. 3. 4.]