// sample-utf8-bom.cpp: non-ASCII text in a source file saved as UTF-8
// WITH a byte order mark (the bytes EF BB BF before these comments).
// Free sample from 99helpers.com/file-types/cpp
//
// Why the BOM matters: GCC and Clang read source as UTF-8 either way, but
// Microsoft's compiler reads a file without a BOM in the system code page,
// unless you pass /utf-8. With the BOM, MSVC knows the file is UTF-8.
//     c++ -Wall -Wextra -std=c++20 sample-utf8-bom.cpp -o sample-utf8 && ./sample-utf8
//     cl /W4 /EHsc /std:c++20 sample-utf8-bom.cpp
// (/utf-8 is still worth adding on MSVC: it also makes the program's
// strings UTF-8, whatever the system code page. On Windows, run
// `chcp 65001` first so the console shows UTF-8 output.)
#include <cstddef>
#include <iostream>
#include <string_view>
#include <type_traits>

// In C++20, u8"..." literals have their own character type, char8_t, which
// iostreams can't print directly. Plain "..." literals in a UTF-8 file stay
// char, which is why this program uses them.
static_assert(std::is_same_v<decltype(u8"é"), const char8_t(&)[3]>);

// std::string_view::size() counts bytes, like strlen. Counting the bytes
// that don't start with the bits 10 counts code points instead.
constexpr std::size_t code_points(std::string_view text)
{
    std::size_t n = 0;
    for (unsigned char c : text) {
        if ((c & 0xC0) != 0x80) {
            ++n;
        }
    }
    return n;
}

int main()
{
    constexpr std::string_view greetings[] = {
        "Hello", "Grüß Gott", "Bonjour, ça va ?", "Привет", "Γειά σου",
        "שלום",  "مرحبا",     "こんにちは",       "你好",   "👋🙂",
    };
    static_assert(code_points("こんにちは") == 5);

    for (auto text : greetings) {
        std::cout << text.size() << " bytes, " << code_points(text)
                  << " code points: " << text << '\n';
    }
}