如何在pybind11中调用Python内置方法（如open）？

Got it, I totally get why you'd want to call Python's built-in open directly from C++ via pybind11 instead of wrapping it in a separate Python file—keeps everything self-contained and plays to pybind11's strength of bridging C++ and Python natively. Here's a step-by-step breakdown with code examples:

Core Concept #

Python's built-in functions (like open, print, len) live in the __builtins__ module. Pybind11 lets you import this module directly, grab the open function as a Python object, and invoke it just like you would in Python.

Full Code Example #

Here's a complete pybind11 module function that demonstrates opening a file, reading its content, and handling basic errors:

#include <pybind11/pybind11.h>
#include <string>

namespace py = pybind11;

PYBIND11_MODULE(builtin_demo, m) {
    m.def("read_file_with_python_open", [](const std::string& filename) -> std::string {
        try {
            // 1. Import the __builtins__ module
            py::module_ builtins = py::module_::import("__builtins__");
            
            // 2. Get the built-in open function
            py::object open_func = builtins.attr("open");
            
            // 3. Call open() just like in Python: open(filename, mode)
            py::object file_obj = open_func(filename, "r");
            
            // 4. Call read() on the file object
            py::str file_content = file_obj.attr("read")();
            
            // 5. Close the file (or use a try-finally to ensure cleanup)
            file_obj.attr("close")();
            
            // Convert Python string to C++ string and return
            return file_content.cast<std::string>();
        } catch (const py::error_already_set& e) {
            // Propagate Python exceptions back to the Python caller
            throw;
        }
    }, "Read a file using Python's built-in open function");
}

Key Details #

• Importing __builtins__: This module contains all of Python's default built-in functions and types, so you don't need any extra imports in Python or C++.
• Calling the open function: The open_func object acts exactly like Python's open—you pass arguments (filename, mode) just as you would in Python code.
• Working with the file object: Once you have the file object, you can call its methods (like read(), write(), close()) using py::object.attr("method_name")().
• Error handling: Wrapping the code in a try-catch block for py::error_already_set ensures that Python exceptions (like FileNotFoundError) are propagated back to the Python runtime instead of crashing your C++ code.

Alternative: Using Python's Context Manager (with Statement) #

If you want to mimic Python's with open(...) as f: pattern to automatically close the file, you can use pybind11's py::with helper (available in newer versions):

// Inside the try block:
py::object file_obj = open_func(filename, "r");
return py::with(file_obj, [&](py::object f) {
    py::str file_content = f.attr("read")();
    return file_content.cast<std::string>();
});

This ensures the file is closed automatically when the with block exits, even if an error occurs.

内容的提问来源于stack exchange，提问作者Markyroson