I am trying to wrap in Python with SWIG some C++17 code that uses std::variant.
I've found this answer ( https://stackoverflow.com/a/58513139 ) about wrapping boost::variant and I managed to adapt the code so that it works with std::variant instead. However, according to the specification of the answer, the code should work so that 'anywhere a C++ function takes a boost::variant we should transparently accept any of the types the variant can hold for that function argument'. This requirement seems to only be satisfied when a const reference to a std::variant is used. For example if in my c++ file dummy.cpp I have
void foo(std::variant<int, double> value)
{
std::cout << "foo" << std::endl;
}
void bar(const std::variant<int, double>& value)
{
std::cout << "bar" << std::endl;
}
then in my dummy.i
%include "variant.i" # My customization of boost_variant.i
%std_variant(DummyVariant, int, double); # Creating the variant type for SWIG/python
when I try to use it from a Python script the following works
from dummy import bar
bar(5)
this, however, does not:
from dummy import foo
foo(5)
I get: TypeError: in method 'foo', argument 1 of type 'std::variant< int,double >'.
Does anybody have an idea of what is missing in my variant.i file to make this works as expected? This is the file I have:
%{
#include <variant>
static PyObject *this_module = NULL;
%}
%init %{
// We need to "borrow" a reference to this for our typemaps to be able to look up the right functions
this_module = m; // borrow should be fine since we can only get called when our module is loaded right?
// Wouldn't it be nice if $module worked *anywhere*
%}
#define FE_0(...)
#define FE_1(action,a1) action(0,a1)
#define FE_2(action,a1,a2) action(0,a1); action(1,a2)
#define FE_3(action,a1,a2,a3) action(0,a1); action(1,a2); action(2,a3)
#define FE_4(action,a1,a2,a3,a4) action(0,a1); action(1,a2); action(2,a3); action(3,a4)
#define FE_5(action,a1,a2,a3,a4,a5) action(0,a1); action(1,a2); action(2,a3); action(3,a4); action(4,a5)
#define GET_MACRO(_1,_2,_3,_4,_5,NAME,...) NAME
%define FOR_EACH(action,...)
GET_MACRO(__VA_ARGS__, FE_5, FE_4, FE_3, FE_2, FE_1, FE_0)(action,__VA_ARGS__)
%enddef
#define in_helper(num,type) const type & convert_type ## num () { return std::get<type>(*$self); }
#define constructor_helper(num,type) variant(const type&)
%define %std_variant(Name, ...)
%rename(Name) std::variant<__VA_ARGS__>;
namespace std {
struct variant<__VA_ARGS__> {
variant();
variant(const std::variant<__VA_ARGS__>&);
FOR_EACH(constructor_helper, __VA_ARGS__);
int index();
%extend {
FOR_EACH(in_helper, __VA_ARGS__);
}
};
}
%typemap(out) std::variant<__VA_ARGS__> {
// Make our function output into a PyObject
PyObject *tmp = SWIG_NewPointerObj(&$1, $&1_descriptor, 0); // Python does not own this object...
// Pass that temporary PyObject into the helper function and get another PyObject back in exchange
const std::string func_name = "convert_type" + std::to_string($1.index());
$result = PyObject_CallMethod(tmp, func_name.c_str(), "");
Py_DECREF(tmp);
}
%typemap(in) const std::variant<__VA_ARGS__>& (PyObject *tmp=NULL) {
// I don't much like having to "guess" the name of the make_variant we want to use here like this...
// But it's hard to support both -builtin and regular modes and generically find the right code.
PyObject *helper_func = PyObject_GetAttrString(this_module, "new_" #Name );
assert(helper_func);
// TODO: is O right, or should it be N?
tmp = PyObject_CallFunction(helper_func, "O", $input);
Py_DECREF(helper_func);
if (!tmp) SWIG_fail; // An exception is already pending
// TODO: if we cared, we chould short-circuit things a lot for the case where our input really was a variant object
const int res = SWIG_ConvertPtr(tmp, (void**)&$1, $1_descriptor, 0);
if (!SWIG_IsOK(res)) {
SWIG_exception_fail(SWIG_ArgError(res), "Variant typemap failed, not sure if this can actually happen");
}
}
%typemap(freearg) const std::variant<__VA_ARGS__>& %{
Py_DECREF(tmp$argnum);
%}
%enddef
which is taken almost verbatim from Flexo's answer, save a few changes to make it work with std::variant instead of boost::variant.
The easiest approach to avoid the use of templates on your interface and create an application binary interface (ABI). In your case this corresponds to creating an interface dummy.i
%module dummy
%{
#include "dummy.h"
void mywrap_foo_int(int i) {
std::variant<int, double> tmp;
tmp = i;
foo(tmp);
}
void mywrap_foo_double(double d) {
std::variant<int, double> tmp;
tmp = d;
foo(tmp);
}
%}
%include "variant.i"
%std_variant(DummyVariant, int, double);
%include "dummy.h"
Soon, you will encounter templates and nested templates. This can be done in a generic way like you have already done, which handles on the signature for bar
. The task of wrapping basically corresponds to this.
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.