How to use boost::split with boost::string_ref in boost 1.55

Question

The code:

#include <iostream>
#include <boost/algorithm/string.hpp>
#include <boost/utility/string_ref.hpp>

int main() 
{
    boost::string_ref str = "test_the_world";
    std::vector<boost::string_ref> strs;
    boost::split(strs, str, boost::is_any_of("_"), boost::token_compress_on);
    for (auto& v : strs)
    {
        std::cout << v << std::endl;
    }
    return 0;
}

Error:

1>C:\boost_1_55_0\boost/range/iterator_range_core.hpp(643): error C2665: 'boost::basic_string_ref<char,std::char_traits<char>>::basic_string_ref' : none of the 4 overloads could convert all the argument types
1>          C:\boost_1_55_0\boost/utility/string_ref.hpp(79): could be 'boost::basic_string_ref<char,std::char_traits<char>>::basic_string_ref(const charT *,boost::basic_string_ref<charT,std::char_traits<char>>::size_type)'
1>          with
1>          [
1>              charT=char
1>          ]
1>          while trying to match the argument list '(const char *, const char *)'
1>          C:\boost_1_55_0\boost/algorithm/string/detail/util.hpp(97) : see reference to function template instantiation 'SeqT boost::copy_range<SeqT,boost::iterator_range<const char *>>(const Range &)' being compiled
1>          with
1>          [
1>              SeqT=boost::basic_string_ref<char,std::char_traits<char>>
1>  ,            Range=boost::iterator_range<const char *>
1>          ]
1>          C:\boost_1_55_0\boost/algorithm/string/detail/util.hpp(96) : while compiling class template member function 'boost::basic_string_ref<char,std::char_traits<char>> boost::algorithm::detail::copy_iterator_rangeF<boost::basic_string_ref<char,std::char_traits<char>>,input_iterator_type>::operator ()(const boost::iterator_range<const char *> &) const'

Etc .. how can I make this work? Can't see why it shouldn't work?

Answer 1

Boost iter_split is well suited for this, although it naturally favours iterator_range instead of string_ref: Difference between boost::split vs boost::iter_split

using R = boost::iterator_range<std::string::const_iterator>;
using V = std::vector<R>;
V v;

for (auto&& r : iter_split(v, input, token_finder(is_any_of(";"))))
    std::cout << r << "\n";

If you insist you can adapt the result:

for (string_ref&& r : iter_split(v, input, token_finder(is_any_of(";"))) 
    | transformed([](R const& r){return boost::string_ref(&*r.begin(), r.size());})
    std::cout << r << "\n";

Or if you don't have c++11 support:

for (string_ref&& r : iter_split(v, input, token_finder(is_any_of(";"))) 
         | transformed(phx::construct<boost::string_ref>(&*phx::begin(_1), phx::size(_1))))
    std::cout << r << "\n";

Combine with copy_range<vector<string_ref> >() if you want a real vector of string_ref objects.

See it Live On Coliru

#include <boost/algorithm/string.hpp>
#include <boost/algorithm/string/split.hpp>
#include <boost/algorithm/string/predicate.hpp>
#include <boost/algorithm/string/finder.hpp>
#include <boost/range/adaptor/transformed.hpp>
#include <boost/phoenix.hpp>
#include <boost/phoenix/stl.hpp>
#include <boost/utility/string_ref.hpp>
#include <iostream>

using namespace boost::algorithm;
namespace phx = boost::phoenix;
using namespace phx::arg_names;
using boost::adaptors::transformed;

int main() {
    std::string input = "1;3;5;7";

    using boost::string_ref;
    using R = boost::iterator_range<std::string::const_iterator>;
    using V = std::vector<R>;
    V v;

    // just the iterator ranges:
    for (auto&& r : iter_split(v, input, token_finder(is_any_of(";"))))
        std::cout << r << "\n";

    // using a lambda to create the string_refs:
    for (string_ref&& r : iter_split(v, input, token_finder(is_any_of(";"))) 
            | transformed([](R const& r){return string_ref(&*r.begin(), r.size());}))
        std::cout << r << "\n";

    // c++03 version:
    for (string_ref&& r : iter_split(v, input, token_finder(is_any_of(";"))) 
            | transformed(phx::construct<string_ref>(&*phx::begin(_1), phx::size(_1))))
        std::cout << r << "\n";
}

Answer 2

Cheap hack: Specialize copy_range for string_ref, because it doesn't take an iterator pair.

namespace boost
{
    template <>
    inline string_ref
    copy_range<string_ref, iterator_range<char const*>>
      ( iterator_range<char const*> const& r )
    {
        return string_ref( begin(r),  end(r) - begin(r) );
    }
}

Demo.

However, there are certainly far better solutions to this, so I'll keep looking.