blob: 69576612452f1849c42d20fb6b7872835ae72157 [file] [log] [blame]
/*=============================================================================
Copyright (c) 2006 Joel de Guzman
Distributed under the Boost Software License, Version 1.0. (See accompanying
file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
Problem:
So... you have an input sequence I and a target vector R. You want to
copy I into R. But, I may have less elements than the result vector R.
For those elements not in R, you want them to be default constructed.
Here's a case:
I: list<double, std::string>
R: vector<double, std::string, int, short>
You want the elements at the right of I not in R (i.e. int, short)
default constructed. Those at the left, found in both I and R, you want
to simply copy from I.
Of course you want to be able to handle any type of I and R.
==============================================================================*/
// We'll use these containers as examples
#include <boost/fusion/container/list.hpp>
#include <boost/fusion/container/vector.hpp>
// For doing I/O
#include <boost/fusion/sequence/io.hpp>
// We'll use join and advance for processing
#include <boost/fusion/algorithm/transformation/join.hpp>
#include <boost/fusion/iterator/advance.hpp>
// The fusion <--> MPL link header
#include <boost/fusion/mpl.hpp>
// Same-o same-o
#include <iostream>
#include <string>
int
main()
{
using namespace boost::fusion;
using namespace boost;
// Let's specify our own tuple delimeters for nicer printing
std::cout << tuple_open('[');
std::cout << tuple_close(']');
std::cout << tuple_delimiter(", ");
// Here's your input sequence
typedef list<double, std::string> I;
I i(123.456, "Hello");
// Here's your output sequence. For now, it is just a typedef
typedef vector<double, std::string, int, short> R;
// Let's get the sizes of the sequences. Yeah, you already know that.
// But with templates, you are simply given, say, R and I, corresponding
// to the types of the sequences. You'll have to deal with it generically.
static int const r_size = result_of::size<R>::value;
static int const i_size = result_of::size<I>::value;
// Make sure that I has no more elements than R
// Be nice and catch obvious errors earlier rather than later.
// Without this assert, the mistake will still be caught by Fusion,
// but the error will point to somewhere really obscure.
BOOST_STATIC_ASSERT(i_size <= r_size);
// Let's get the begin and end iterator types of the output sequence
// There's no actual vector yet. We just want to know the types.
typedef result_of::begin<R>::type r_begin;
typedef result_of::end<R>::type r_end;
// Let's skip i_size elements from r_begin. Again, we just want to know the type.
typedef result_of::advance_c<r_begin, i_size>::type r_advance;
// Now, make MPL iterators from r_advance and r_end. Ditto, just types.
typedef mpl::fusion_iterator<r_advance> mpl_r_advance;
typedef mpl::fusion_iterator<r_end> mpl_r_end;
// Make an mpl::iterator_range from the MPL iterators we just created
// You guessed it! --just a type.
typedef mpl::iterator_range<mpl_r_advance, mpl_r_end> tail;
// Use join to join the input sequence and our mpl::iterator_range
// Our mpl::iterator_range is 'tail'. Here, we'll actually instantiate
// 'tail'. Notice that this is a flyweight object, typically just 1 byte
// in size -- it doesn't really hold any data, but is a fully conforming
// sequence nonetheless. When asked to return its elements, 'tail' returns
// each element default constructed. Breeds like a rabbit!
// Construct R from the joined sequences:
R r(join(i, tail()));
// Then finally, print the result:
std::cout << r << std::endl;
return 0;
}