converting wide string to narrow and vice-versa

convert_example.cpp

#include <locale>
#include <string>
#include <stdexcept>

std::wstring
widen(std::string const& str, std::codecvt<wchar_t, char, std::mbstate_t> const& cvt)
{
  if(str.empty())
    return std::wstring();

  std::mbstate_t mb = std::mbstate_t();

  std::wstring wstr(4 + str.size(), L'\0');

  const char* from_beg = &str[0];
  const char* from_end = from_beg + str.size();
  const char* from_nxt;
  wchar_t* to_beg = &wstr[0];
  wchar_t* to_end = to_beg + wstr.size();
  wchar_t* to_nxt;

  switch(cvt.in(mb, from_beg, from_end, from_nxt, to_beg, to_end, to_nxt))
    {
      case std::codecvt_base::error:
        throw std::runtime_error("string conversion error in widen()");
      case std::codecvt_base::partial:
        throw std::runtime_error("string conversion partial in widen()");
      case std::codecvt_base::noconv:
        wstr.resize(from_end - from_beg, L'\0');
        for(size_t i = 0ul; i != str.size(); ++i)
          wstr[i] = static_cast<wchar_t>(str[i]);
        break;
      case std::codecvt_base::ok:
        wstr.resize(to_nxt - to_beg);
        break;
      default:
        throw std::runtime_error("internal error in widen()");
    }

  return wstr;
}

std::string
narrow(std::wstring const& wstr, std::codecvt<wchar_t, char, std::mbstate_t> const& cvt)
{
  if(wstr.empty())
    return std::string();

  std::mbstate_t mb = std::mbstate_t();

  std::string str(4 + wstr.size() * cvt.max_length(), '\0');

  const wchar_t* from_beg = &wstr[0];
  const wchar_t* from_end = from_beg + wstr.size();
  const wchar_t* from_nxt;
  char* to_beg = &str[0];
  char* to_end = to_beg + str.size();
  char* to_nxt;

  switch(cvt.out(mb, from_beg, from_end, from_nxt, to_beg, to_end, to_nxt))
    {
      case std::codecvt_base::error:
        throw std::runtime_error("string conversion error in narrow()");
      case std::codecvt_base::partial:
        throw std::runtime_error("string conversion partial in narrow()");
      case std::codecvt_base::noconv:
        str.resize(from_end - from_beg, L'\0');
        for(size_t i = 0ul; i != wstr.size(); ++i)
          str[i] = static_cast<char>(wstr[i]);
        break;
      case std::codecvt_base::ok:
        str.resize(to_nxt - to_beg);
        break;
      default:
        throw std::runtime_error("internal error in narrow()");
    }

  return str;
}

#include <iostream>
#include <iomanip>

int main()
{
  typedef std::codecvt<wchar_t,char,std::mbstate_t> codecvt_t;
  std::string s = "\xB3"; // polish l with stroke in latin-2
  std::wstring w = widen(s, std::use_facet<codecvt_t>(std::locale("pl_PL.ISO-8859-2")));
  std::string u = narrow(w, std::use_facet<codecvt_t>(std::locale("")));
  std::cout << u << std::endl;
  return 0;
}