是否可以创建 std::make_shared 的 boost phoenix 惰性变体？我的意思是，使之类的事情成为可能
namespace p = boost::phoenix;
...
expr = custom_parser[_a=p::make_shared<Node>(_1,_2,_3)] >> ...
由于std::make_shared 的可变模板性质，无法使用BOOST_PHOENIX_ADAPT_FUNCTION。因此，如果可以编写一个，包装器可能应该是可变参数模板本身。
【问题讨论】：
标签： c++ boost boost-phoenix make-shared
如果你能多出一组括号：
namespace {
template <typename T>
struct make_shared_f
{
template <typename... A> struct result
{ typedef boost::shared_ptr<T> type; };
template <typename... A>
typename result<A...>::type operator()(A&&... a) const {
return boost::make_shared<T>(std::forward<A>(a)...);
}
};
template <typename T>
using make_shared_ = boost::phoenix::function<make_shared_f<T> >;
}
你可以像这样使用
typedef std::vector<int> IntVec;
auto LazyInts = make_shared_<IntVec>()(arg1, arg2);
// create a shared vector of 7 ints '42'
auto ints = LazyInts(7, 42);
for (auto i : *ints) std::cout << i << " ";
【讨论】：
您的phx::make_shared 与phx::construct 非常相似。通过稍微改变construct's header file你可以得到：
//START OF PHOENIX_MAKE_SHARED.HPP
#ifndef BOOST_PHOENIX_MAKE_SHARED_HPP
#define BOOST_PHOENIX_MAKE_SHARED_HPP
#include <boost/shared_ptr.hpp>
#include <boost/make_shared.hpp>
#include <boost/phoenix/core/limits.hpp>
#include <boost/phoenix/core/call.hpp>
#include <boost/phoenix/core/expression.hpp>
#include <boost/phoenix/core/meta_grammar.hpp>
#include <boost/phoenix/object/detail/target.hpp>
#include <boost/phoenix/support/iterate.hpp>
#include <boost/preprocessor/repetition/repeat_from_to.hpp>
#include <boost/preprocessor/repetition/enum_shifted_binary_params.hpp>
BOOST_PHOENIX_DEFINE_EXPRESSION_VARARG(
(boost)(phoenix)(make_shared)
, (proto::terminal<detail::target<proto::_> >)
(meta_grammar)
, BOOST_PHOENIX_COMPOSITE_LIMIT
)
namespace boost { namespace phoenix
{
struct make_shared_eval
{
template <typename Sig>
struct result;
template <typename This, typename A0,typename Context>
struct result<This(A0, Context)>
{
typedef boost::shared_ptr<typename detail::result_of::target<A0>::type > type;
};
template <typename Target, typename Context>
typename boost::shared_ptr<typename detail::result_of::target<Target>::type>
operator()(Target, Context const &) const
{
return boost::make_shared<typename detail::result_of::target<Target>::type>();
}
#define EVAL_a(_,N,__) \
BOOST_PP_COMMA_IF(BOOST_PP_DEC(N)) boost::phoenix::eval(a ## N, ctx)
#define GENERATE_MAKE_SHARED_EVAL(_,N,__) \
template <typename This, BOOST_PHOENIX_typename_A(N), typename Context> \
struct result<This(BOOST_PHOENIX_A(N), Context)> \
{ \
typedef boost::shared_ptr<typename detail::result_of::target<A0>::type > type; \
}; \
template <BOOST_PHOENIX_typename_A(N), typename Context> \
typename boost::shared_ptr<typename detail::result_of::target<A0>::type > \
operator()( \
A0 const& \
, BOOST_PP_ENUM_SHIFTED_BINARY_PARAMS( \
N \
, A \
, const& a \
) \
, Context const & ctx \
) const \
{ \
return \
boost::make_shared<typename detail::result_of::target<A0>::type>( \
BOOST_PP_REPEAT_FROM_TO( \
1 \
, N \
, EVAL_a, _ \
) \
); \
}
BOOST_PP_REPEAT_FROM_TO(2,BOOST_PHOENIX_COMPOSITE_LIMIT,GENERATE_MAKE_SHARED_EVAL,_)
#undef EVAL_a
#undef GENERATE_MAKE_SHARED_EVAL
};
template <typename Dummy>
struct default_actions::when<rule::make_shared, Dummy>
: call<make_shared_eval, Dummy>
{};
template <typename T>
inline
typename expression::make_shared<detail::target<T> >::type const
make_shared()
{
return
expression::
make_shared<detail::target<T> >::
make(detail::target<T>());
}
#define GENERATE_MAKE_SHARED(_,N,__) \
template <typename T, BOOST_PHOENIX_typename_A(N)> \
inline \
typename expression::make_shared<detail::target<T>, BOOST_PHOENIX_A(N)>::type const \
make_shared(BOOST_PHOENIX_A_const_ref_a(N)) \
{ \
return \
expression:: \
make_shared<detail::target<T>, BOOST_PHOENIX_A(N)>:: \
make(detail::target<T>(), BOOST_PHOENIX_a(N)); \
}
BOOST_PP_REPEAT_FROM_TO(1,BOOST_PHOENIX_COMPOSITE_LIMIT,GENERATE_MAKE_SHARED,_)
#undef GENERATE_MAKE_SHARED
}}
#endif
//END OF PHOENIX_MAKE_SHARED.HPP
#include <iostream>
#include <string>
#include <vector>
#define BOOST_SPIRIT_USE_PHOENIX_V3
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/phoenix.hpp>
namespace phx=boost::phoenix;
namespace qi=boost::spirit::qi;
struct foo
{
foo():foo1(1),foo2(1.1){}
foo(int foo1_):foo1(foo1_),foo2(2.2){}
foo(int foo1_, double foo2_):foo1(foo1_),foo2(foo2_){}
int foo1;
double foo2;
};
int main()
{
//using phx::arg_names::arg1;
//using phx::arg_names::arg2;
//boost::shared_ptr<foo> var1 = phx::make_shared<foo>()();
//boost::shared_ptr<foo> var2 = phx::make_shared<foo>(arg1)(2);
//boost::shared_ptr<foo> var3 = phx::make_shared<foo>(arg1,arg2)(3,3.3);
//std::cout << "var1: " << var1->foo1 << ", " << var1->foo2 << std::endl;
//std::cout << "var2: " << var2->foo1 << ", " << var2->foo2 << std::endl;
//std::cout << "var3: " << var3->foo1 << ", " << var3->foo2 << std::endl;
std::string test = "default,2,3-3.3";
qi::rule<std::string::const_iterator,boost::shared_ptr<foo>()> awful_rule =
qi::lit("default")[qi::_val=phx::make_shared<foo>()]
| (qi::int_ >> !qi::lit('-'))[qi::_val=phx::make_shared<foo>(qi::_1)]
| (qi::int_ >> qi::lit('-') >> qi::double_)[qi::_val=phx::make_shared<foo>(qi::_1,qi::_2)]
;
qi::rule<std::string::const_iterator,std::vector<boost::shared_ptr<foo> >()> parser = awful_rule%qi::lit(',');
std::string::const_iterator iter=test.begin(), end=test.end();
std::vector<boost::shared_ptr<foo> > foo_vector;
bool result = qi::parse(iter, end, parser, foo_vector);
if(result && (iter==end) )
{
std::cout << "Success:" << std::endl;
for(std::size_t cont=0; cont < foo_vector.size(); ++cont)
std::cout << foo_vector[cont]->foo1 << ", " << foo_vector[cont]->foo2 << std::endl;
}
else
{
std::cout << "Failure. Unparsed: " << std::string(iter,end) << std::endl;
}
}
【讨论】：