Python using C++ library with Pybin11
Python using C++ library with Pybin11
0 Why pybind11?
上文介绍了Py-Boost的搭建和使用Py-Boost build on win10 with MSVC，这篇文章介绍的pybind11与Boost::Python一样可以作为C++ => Python的桥梁。不同的是Boost太重了，引用pybin11官方说法"Boost is an enormously large and complex suite of utility libraries that works with almost every C++ compiler in existence", 这带来的代价就是，Boost::Python里有很多不必要、不实用的代码，因此pybind11这类轻量级（仅头文件库只需要大约 4K 行代码）就有了用武之地。
1 build environment
- 安装pybind11
git clone https://github.com/pybind/pybind11.git
- 构建开发文件夹
在pybind11同级目录下，构建如果工程文件结构
- root
- src
main.cpp
*.cpp
- inc
*.h
- pybind11
CMakeLists.txt
- 设置CMakeLists.txt
cmake_minimum_required(VERSION 3.4...3.18)
project(example)
# C++ 17
set(CMAKE_CXX_STANDARD 17)
# include
set(INCLUDE_H ${CMAKE_CURRENT_SOURCE_DIR}/inc)
include_directories(${INCLUDE_H})
# src
aux_source_directory(${CMAKE_CURRENT_SOURCE_DIR}/src DIR_SRCS)
# pybind11
add_subdirectory(pybind11) # 包含子目录 pybind11 https://github.com/pybind/pybind11.git
pybind11_add_module(example ${DIR_SRCS})
# version setup
set(EXAMPLE_VERSION_INFO 1.0) # 设定版本号
target_compile_definitions(example
PRIVATE VERSION_INFO=${EXAMPLE_VERSION_INFO})
2 C++开发 - Python使用
以下全部代码见examplePybind
2.1 say hello
say hello是pybind11官方文档中一个hello world测试案例
#include <pybind11/pybind11.h>
#define STRINGIFY(x) #x
#define MACRO_STRINGIFY(x) STRINGIFY(x)
namespace py = pybind11;
int add(int i, int j) // add func, pybind hello world.
{
return i + j;
}
PYBIND11_MODULE(example, m) {
m.doc() = R"pbdoc(
Pybind11 example plugin
-----------------------
.. currentmodule:: cmake_example
.. autosummary::
:toctree: _generate
add
subtract
)pbdoc"; // 说明文档
m.def("add", &add, R"pbdoc(
Add two numbers
Some other explanation about the add function.
)pbdoc"); // 绑定 add 函数
#ifdef VERSION_INFO // 版本号
m.attr("__version__") = MACRO_STRINGIFY(VERSION_INFO);
#else
m.attr("__version__") = "dev";
#endif
}
2.2 build py lib
采用CMake方式build
- windows
mkdir build
cd build
cmake "Visual Studio 15 2022" -A x64 ..
cmake --build . --config Release -- /maxcpucount:8
- osx
mkdir build
cd build
cmake ..
make -j8
在build/Release/文件夹下可以找到example.pyd或者example.so文件
2.3 python import
>>> import example
>>> print(example.add(1, 2))
>>> 3
2.4 template function
当然仅仅只是跑通官方测试案例还是不过瘾的，下面对模板函数进行构建
main.cpp
template <typename T> // 模板函数
T mul(T x, T y)
{
return x * y;
}
PYBIND11_MODULE(example, m) {
m.doc() = R"pbdoc(
Pybind11 example plugin
-----------------------
.. currentmodule:: cmake_example
.. autosummary::
:toctree: _generate
add
subtract
)pbdoc";
m.def("mul", mul<int>, py::arg("x")=1, py::arg("y")=2); // 模板函数 int
m.def("mul", mul<float>, py::arg("x")=2.0, py::arg("y")=3.0); // 模板函数 float
m.def("mul", mul<double>, py::arg("x")=3.0, py::arg("y")=4.0); // 模板函数 double
#ifdef VERSION_INFO
m.attr("__version__") = MACRO_STRINGIFY(VERSION_INFO);
#else
m.attr("__version__") = "dev";
#endif
}
examplePybind.py
import example
print("example version: ", example.__version__)
def test_func():
print("mul() = {}".format(example.mul())) # 默认参数 int x, y = 1, 2
print("mul(12, 3) = {}".format(example.mul(12, 3))) # int x, y = 12, 3
print("mul(12.0, 2.0) = {}".format(example.mul(12.0, 2.0))) # float x, y = 12.0, 2.0
if __name__ == "__main__":
test_func()
结果
mul() = 2
mul(12, 3) = 36
mul(12.0, 2.0) = 24.0
2.5 template class
更进一步的，模板 + 类 + 分文件编写
分别支持int和float类型的微型数值计算的类。
numeric.h
#include <pybind11/pybind11.h>
#define FLT_MAX 3.402823466e+38F
#define FLT_MIN 1.175494351e-38F
template <typename T>
class SimpleNum // 类声明
{
public:
T m_x;
SimpleNum();
SimpleNum(T x): m_x(x) {}
T Add(T y);
T Sub(T y);
T Mul(T y);
T Div(T y);
void Set(T x);
T Get();
private:
bool _IsZero(T y);
};
numeric.cpp
#include "../inc/numeric.h"
namespace py = pybind11;
template <typename T>
T SimpleNum<T>::Add(T y) // 具体实现
{
return this->m_x + y;
}
template <typename T>
T SimpleNum<T>::Sub(T y)
{
return this->m_x - y;
}
template <typename T>
T SimpleNum<T>::Mul(T y)
{
return this->m_x * y;
}
template <typename T>
T SimpleNum<T>::Div(T y)
{
if (_IsZero(y))
{
return FLT_MAX;
}
return this->m_x / y;
}
template <typename T>
void SimpleNum<T>::Set(T x)
{
this->m_x = x;
}
template <typename T>
T SimpleNum<T>::Get()
{
return this->m_x;
}
template <typename T>
bool SimpleNum<T>::_IsZero(T y)
{
return y == 0;
}
// 就地绑定, 会在main.cpp对class_ext也进行绑定
void class_ext(py::module &m) {
py::class_<SimpleNum<int> > (m, "SimpleNum")
.def(py::init<int>())
.def_readwrite("x", &SimpleNum<int>::m_x) // 并不建议直接访问成员变量x, 安全起见用Set(), Get()
.def("Add", &SimpleNum<int>::Add)
.def("Sub", &SimpleNum<int>::Sub)
.def("Mul", &SimpleNum<int>::Mul)
.def("Div", &SimpleNum<int>::Div)
.def("Set", &SimpleNum<int>::Set)
.def("Get", &SimpleNum<int>::Get);
py::class_<SimpleNum<float> > (m, "SimpleNumFLT")
.def(py::init<float>())
.def_readwrite("x", &SimpleNum<float>::m_x) // 并不建议直接访问成员变量x, 安全起见用Set(), Get()
.def("Add", &SimpleNum<float>::Add)
.def("Sub", &SimpleNum<float>::Sub)
.def("Mul", &SimpleNum<float>::Mul)
.def("Div", &SimpleNum<float>::Div)
.def("Set", &SimpleNum<float>::Set)
.def("Get", &SimpleNum<float>::Get);
}
main.cpp
void class_ext(py::module &m); // 分文件绑定，先声明，其他源文件会定义
PYBIND11_MODULE(example, m) {
m.doc() = R"pbdoc(
Pybind11 example plugin
-----------------------
.. currentmodule:: cmake_example
.. autosummary::
:toctree: _generate
add
subtract
)pbdoc";
class_ext(m); // 绑定class
#ifdef VERSION_INFO
m.attr("__version__") = MACRO_STRINGIFY(VERSION_INFO);
#else
m.attr("__version__") = "dev";
#endif
}
examplePybind.py
import example
print("example version: ", example.__version__)
def test_class():
num = example.SimpleNum(3) # 类 -> 对象
print("num = {}".format(num.x))
print("{} + 4 = {}".format(num.x, num.Add(4))) # 成员函数 Add
print("{} - 4 = {}".format(num.x, num.Sub(4))) # 成员函数 Sub
print("{} * 4 = {}".format(num.x, num.Mul(4))) # 成员函数 Mul
print("{} / 4 = {}".format(num.x, num.Div(4))) # 成员函数 Div
print("num.Set(99)")
num.Set(99) # 成员函数 设置成员变量x
print("num.Get(): {}".format(num.Get())) # Get函数 获取成员变量x
numFlt = example.SimpleNumFLT(30.0) # 类 -> 对象，float类型
print("numFlt = {}".format(numFlt.x))
print("{} + 4.0 = {}".format(numFlt.x, numFlt.Add(4.0)))
print("{} - 4.0 = {}".format(numFlt.x, numFlt.Sub(4.0)))
print("{} * 4.0 = {}".format(numFlt.x, numFlt.Mul(4.0)))
print("{} / 4.0 = {}".format(numFlt.x, numFlt.Div(4.0)))
print("numFlt.Set(199.0)")
numFlt.Set(199.0)
print("numFlt.Get(): {}".format(numFlt.Get()))
if __name__ == "__main__":
test_class()
结果
num = 3
3 + 4 = 7
3 - 4 = -1
3 * 4 = 12
3 / 4 = 0
num.Set(99)
num.Get(): 99
numFlt = 30.0
30.0 + 4.0 = 34.0
30.0 - 4.0 = 26.0
30.0 * 4.0 = 120.0
30.0 / 4.0 = 7.5
numFlt.Set(199.0)
numFlt.Get(): 199.0
2.6 C++ vector - Python list
C++中的vector与python list实际上可以做一个对应
实现一个针对数组的平方函数，支持int和float类型
vectorTest.h
#include <pybind11/pybind11.h>
#include <pybind11/stl.h>
#include <pybind11/complex.h>
#include <pybind11/functional.h>
#include <pybind11/chrono.h>
#include <vector>
using namespace std;
namespace py = pybind11;
template <typename T>
vector<T> vecSquare(vector<T> vec);
vectorTest.cpp
#include "../inc/vectorTest.h"
template <typename T>
vector<T> vecSquare(vector<T> vec)
{
vector<T> ret;
for (int i=0; i < vec.size(); i++)
{
ret.emplace_back(vec[i] * vec[i]);
}
return ret;
}
// 就地绑定, 会在main.vector_ext
void vector_ext(py::module &m) {
m.def("Square", &vecSquare<int>);
m.def("Square", &vecSquare<float>);
m.def("Square", &vecSquare<double>);
}
examplePybind.py
def test_vector():
b = [1, 2, 3]
print("b^2 = {}".format(example.Square(b)))
c = [4.0, 5.0, 6.0]
print("c^2 = {}".format(example.Square(c)))
if __name__ == "__main__":
test_vector()
结果
b^2 = [1, 4, 9]
c^2 = [16.0, 25.0, 36.0]
标签:
C++开发
【推荐】国内首个AI IDE，深度理解中文开发场景，立即下载体验Trae
【推荐】编程新体验，更懂你的AI，立即体验豆包MarsCode编程助手
【推荐】抖音旗下AI助手豆包，你的智能百科全书，全免费不限次数
【推荐】轻量又高性能的 SSH 工具 IShell：AI 加持，快人一步
· 震惊！C++程序真的从main开始吗？99%的程序员都答错了
· 【硬核科普】Trae如何「偷看」你的代码？零基础破解AI编程运行原理
· 单元测试从入门到精通
· 上周热点回顾（3.3-3.9）
· winform 绘制太阳，地球，月球 运作规律