C++ code to convert UTF-16 encoding to UTF-8 encoding.

utf16-to-utf8.cpp

std::string
to_utf8(const std::u16string& input)
{
#if 0
    // TODO: try using std::c16rtomb()
#else
    std::string output;
    char32_t codepoint;
    for (auto i = input.begin(); i != input.end(); ++i) {
        auto ic = *i;
        if (ic <= 0xd7ff)
            codepoint = ic;
        else {
            char32_t high_surrogate = ic;
            if ((high_surrogate & 0xfc00) != 0xd800)
                break; // error: invalid high surrogate

            if (++i == input.end())
                break; // error: reached the end too soon.
            char32_t low_surrogate = *i;
            if ((low_surrogate & 0xfc00) != 0xdc00)
                break; // error: invalid low surrogate

            codepoint =
                ((high_surrogate & 0x3ff) << 10)
                | (low_surrogate & 0x3ff);
        }

        if (!codepoint)
            break; // null terminator

        // now encode as UTF-8
        char8_t oc;
        if (codepoint <= 0x7f) {
            output.push_back(codepoint);
        } else if (codepoint <= 0x7ff) {
            oc = 0xc0 | (codepoint >> 6);
            output.push_back(oc);
            oc = 0x80 | ((codepoint >> 0) & 0x3f);
            output.push_back(oc);
        } else if (codepoint <= 0xffff) {
            oc = 0xe0 | (codepoint >> 12);
            output.push_back(oc);
            oc = 0x80 | ((codepoint >> 6) & 0x3f);
            output.push_back(oc);
            oc = 0x80 | ((codepoint >> 0) & 0x3f);
            output.push_back(oc);
        } else {
            oc = 0xf0 | (codepoint >> 18);
            output.push_back(oc);
            oc = 0x80 | ((codepoint >> 12) & 0x3f);
            output.push_back(oc);
            oc = 0x80 | ((codepoint >> 6) & 0x3f);
            output.push_back(oc);
            oc = 0x80 | ((codepoint >> 0) & 0x3f);
            output.push_back(oc);
        }
    }
#endif
    return output;
}