如何从返回numpy数组的c++调用python函数时访问数据

Question

我正在通过pybind 11从c++调用python函数。python函数返回一个数值数组，我想在c++中分析数值数组中的数据。

//the code for testing
#include <pybind11/embed.h> // everything needed for embedding
#include <pybind11/pybind11.h>
#include <pybind11/numpy.h>

int main()
{

py::scoped_interpreter guard{};

py::print("test python interpreter");
auto sys = py::module::import("sys");
sys.attr("path").attr("append")(
"<path for python module>");

auto hdf5 = py::module::import("reader_hdf5");

auto rdata = hdf5.attr("load_next_chunk")(0, 1);

// how to access the data from the return value ?

 py::array_t<int16_t> array(331776);
 array = rdata.cast<py::array_t<int16_t>>();
}

这里的代码有错误：

terminate called after throwing an instance of 'pybind11::error_already_set'
what():  TypeError: int() argument must be a string, a bytes-like object or a number, not 'NoneType'

Answer 1

对mutable_data()的调用可以工作

    py::array_t<int>rdata= hdf5.attr("load_next_chunk")(0, 1);

    int *carray = rdata.mutable_data();

然后内容应该存在于carray中。我不是pybind专家，但是这个解决方案是由pybind github forum提供的