Python 中的 C++ 扩展“DLL 加载失败”

Question

对于一个学校项目，我需要将英特尔 Realsense（3D 摄像头）与 Python 连接，以便将其与 RoboDK 和 OpenCV 一起使用。我正在使用 MVS 2019 执行此操作。将 superfastcode2 (C++) 和 RealsenseDistanceV3 (Python) 设置为 64 位。

我按照本教程创建了一个 C++ 扩展，效果很好。

现在我将<librealsense2\rs.hpp> 包含到module.cpp 代码中，一切都可以编译并且运行良好。但是，仅处于发布模式。在调试模式下，编译时出现以下错误：

Error   LNK2019 unresolved external symbol __imp__invalid_parameter referenced in function "void * __cdecl std::_Allocate_manually_vector_aligned<struct std::_Default_allocate_traits>(unsigned __int64)" (??$_Allocate_manually_vector_aligned@U_Default_allocate_traits@std@@@std@@YAPEAX_K@Z)   superfastcode2  C:\Users\Gebruiker\source\repos\RealsenseDistanceV3\superfastcode2\module.obj

Error   LNK2019 unresolved external symbol __imp__CrtDbgReport referenced in function "void * __cdecl std::_Allocate_manually_vector_aligned<struct std::_Default_allocate_traits>(unsigned __int64)" (??$_Allocate_manually_vector_aligned@U_Default_allocate_traits@std@@@std@@YAPEAX_K@Z)    superfastcode2  C:\Users\Gebruiker\source\repos\RealsenseDistanceV3\superfastcode2\module.obj

在 C++ 代码中使用 Realsense 库中的函数 (rs2::pipeline p;) 时，会出现以下错误。 DLL load failed while importing superfastcode2: Kan opgegeven module niet vinden（英文：cannot find module）。

这是合乎逻辑的，因为在查看 superfastcode2.log 时会显示：

     Creating library C:\Users\Gebruiker\source\repos\RealsenseDistanceV3\Release\superfastcode2.lib and object C:\Users\Gebruiker\source\repos\RealsenseDistanceV3\Release\superfastcode2.exp
module.obj : error LNK2001: unresolved external symbol __imp__PyFloat_AsDouble
module.obj : error LNK2001: unresolved external symbol __imp__PyFloat_FromDouble
module.obj : error LNK2001: unresolved external symbol __imp__PyModule_Create2
C:\Users\Gebruiker\source\repos\RealsenseDistanceV3\Release\superfastcode2.pyd : fatal error LNK1120: 3 unresolved

Module.cpp（来自 superfastcode2）：

#include <Python.h>
#include <Windows.h>
#include <cmath>
#include <librealsense2\rs.hpp> // Include RealSense Cross Platform API
#include <iostream>             // for cout

const double e = 2.7182818284590452353602874713527;

double sinh_impl(double x) {
    return (1 - pow(e, (-2 * x))) / (2 * pow(e, -x));
}

double cosh_impl(double x) {
    return (1 + pow(e, (-2 * x))) / (2 * pow(e, -x));
}

double askRealsenseDistance()
{
    return 5.25;
}

void connectRealsense()
{
    rs2::pipeline p;
}


PyObject* tanh_impl(PyObject*, PyObject* o) {
    double x = PyFloat_AsDouble(o);
    double tanh_x = sinh_impl(x) / cosh_impl(x);
    return PyFloat_FromDouble(tanh_x);
}

PyObject* askRealsenseDistance_impl(PyObject*, PyObject* o) {
    //double distance = askRealsenseDistance();
    double distance = PyFloat_AsDouble(o)/100;
    connectRealsense();
    return PyFloat_FromDouble(distance);
}


static PyMethodDef superfastcode2_methods[] = {
    // The first property is the name exposed to Python, fast_tanh, the second is the C++
    // function name that contains the implementation.
    { "fast_tanh", (PyCFunction)tanh_impl, METH_O, nullptr },
    { "askRealsenseDistance_py", (PyCFunction)askRealsenseDistance_impl, METH_O, nullptr },

    // Terminate the array with an object containing nulls.
{ nullptr, nullptr, 0, nullptr }
};



static PyModuleDef superfastcode2_module = {
    PyModuleDef_HEAD_INIT,
    "superfastcode2",                        // Module name to use with Python import statements
    "Provides some functions, but faster",  // Module description
    0,
    superfastcode2_methods                   // Structure that defines the methods of the module
};

PyMODINIT_FUNC PyInit_superfastcode2() {
    return PyModule_Create(&superfastcode2_module);
}

还有 RealsenseDitanceV3.py（来自 RealsenseDistanceV3）：

from itertools import islice
from random import random
from time import perf_counter
from superfastcode2 import fast_tanh
from superfastcode2 import askRealsenseDistance_py

COUNT = 500000  # Change this value depending on the speed of your computer
DATA = list(islice(iter(lambda: (random() - 0.5) * 3.0, None), COUNT))

e = 2.7182818284590452353602874713527


def sinh(x):
    return (1 - (e ** (-2 * x))) / (2 * (e ** -x))


def cosh(x):
    return (1 + (e ** (-2 * x))) / (2 * (e ** -x))


def tanh(x):
    tanh_x = sinh(x) / cosh(x)
    return tanh_x


def test(fn, name):

    start = perf_counter()
    result = fn(DATA)
    duration = perf_counter() - start
    print('{} took {:.3f} seconds\n\n'.format(name, duration))

    for d in result:
        assert -1 <= d <= 1, " incorrect values"


#if __name__ == "__main__":
    #print('Running benchmarks with COUNT = {}'.format(COUNT))

    #test(lambda d: [tanh(x) for x in d], '[tanh(x) for x in d] (Python implementation)')

    #test(lambda d: [fast_tanh(x) for x in d], '[fast_tanh(x) for x in d] (CPython C++ extension)')

number = 8050
print('send: {} to cpp which divides it by 100'.format(number))
output = askRealsenseDistance_py(number)
print('received from cpp: {}'.format(output))

为了澄清，如果rs2::pipeline p; 行或connectRealsense(); 行从module.cpp 中删除，一切正常。不知何故，我认为 Python.h (CPython) 库无法识别（或其他）Realsense C++ 函数。

希望您有一些建议，在此先感谢！

Answer 1

找到了解决办法！！ 您需要从输出库中的 Realsense 库中复制 .dll 文件。在我发现之后，这似乎很合乎逻辑，但首先我认为 MVS 会使 .dll 文件本身（来自整个代码）在 Python 中使用.事实证明，情况并非如此，而是 .dll 文件仅来自 Realsenselibrary。感谢大家的帮助！