std::codecvt_utf8

Следующий пример демонстрирует разницу между преобразованиями UCS-2/UTF-8 и UTF-16/UTF-8: третий символ в строке не является допустимым символом UCS-2.

#include <codecvt>
#include <cstdint>
#include <iostream>
#include <locale>
#include <string>

int main()
{
    // UTF-8 data. The character U+1d10b, musical sign segno, does not fit in UCS-2
    std::string utf8 = "z\u6c34\U0001d10b";
    
    // the UTF-8 / UTF-16 standard conversion facet
    std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t> utf16conv;
    std::u16string utf16 = utf16conv.from_bytes(utf8);
    std::cout << "UTF-16 conversion produced " << utf16.size() << " code units:\n"
              << std::showbase << std::hex;
    for (char16_t c : utf16)
        std::cout << static_cast<std::uint16_t>(c) << ' ';
    
    // the UTF-8 / UCS-2 standard conversion facet
    std::wstring_convert<std::codecvt_utf8<char16_t>, char16_t> ucs2conv;
    try
    {
        std::u16string ucs2 = ucs2conv.from_bytes(utf8);
    }
    catch(const std::range_error& e)
    {
        std::u16string ucs2 = ucs2conv.from_bytes(utf8.substr(0, ucs2conv.converted()));
        std::cout << "\nUCS-2 failed after producing " << std::dec << ucs2.size()
                  << " characters:\n" << std::showbase << std::hex;
        for (char16_t c : ucs2)
            std::cout << static_cast<std::uint16_t>(c) << ' ';
        std::cout << '\n';
    }
}

Вывод:

UTF-16 conversion produced 4 code units:
0x7a 0x6c34 0xd834 0xdd0b
UCS-2 failed after producing 2 characters:
0x7a 0x6c34