C++ string to wstring prints out incorrectly, cant get unicode path

Question

#include <iostream>
#include <Windows.h>
#include <locale>
#include <string>
#include <codecvt>
typedef wchar_t* LPWSTR, *PWSTR;

template <typename Facet>
struct deletable_facet : Facet
{
    using Facet::Facet;
};

int main(int argc, char *argv[])
{
    std::cout << argv[0] << std::endl;

    std::wstring_convert<std::codecvt_utf8_utf16<wchar_t>> converter;
    //std::wcout << converter.from_bytes(argv[0]) << std::endl; // range error


    std::wstring_convert<deletable_facet<std::codecvt<wchar_t, char, std::mbstate_t>>> conv;
    std::wstring ns = conv.from_bytes(argv[0]);
    std::wcout << ns << std::endl;

    wchar_t filename[MAX_PATH];
    //GetModuleFileName(NULL,filename,MAX_PATH); // cant convert wstring_t* to char*
    GetModuleFileNameW(NULL,filename,MAX_PATH);
    std::wcout << filename << std::endl;


    getchar();
    return 0;
}

Output:

 C:\Users\luka\Desktop\ⁿ?icΣ\unicode.exe
 C:\Users\luka\Desktop\ⁿ?icΣ\unicode.exe
 C:\Users\luka\Desktop\ⁿ

Actual name of the folder is üлicä

Ive been trying many many different ways for about 2 hours now, and as far as ive seen people suggested GetModuleFileName , but as you can see that returns a conversion error (typedef wchar_t* LPWSTR, *PWSTR; isnt fixing it).

So is there any way to to get the current folder path in unicode , and get the rest of the input arguments to unicode (non-latin characters)

Answer 1

The usage for GetModuleFileName is correct. You should see the expected result with MessageBoxW(0, filename, 0, 0);

The problem is in printing L"üлicä" on Windows console.

Try printing "üлicä" on the console:

int main(int argc, char *argv[])
{
    DWORD count;
    std::wstring str = GetCommandLineW() + (std::wstring)L"\n";
    WriteConsoleW(GetStdHandle(STD_OUTPUT_HANDLE), str.c_str(), str.size(), &count, 0);
    MessageBoxW(0, str.c_str(), 0, 0);

    wchar_t filename[MAX_PATH];
    GetModuleFileNameW(0, filename, MAX_PATH);
    WriteConsoleW(GetStdHandle(STD_OUTPUT_HANDLE), filename, wcslen(filename), &count, 0);
    return 0;
}

In Visual Studio you can also use _setmode to enable usage of std::wcout/std::wcin

You also have optional entry point wmain(int argc, wchar_t *argv[]) which provides argv in UTF16 encoding.

The main entry point provides argv in ANSI encoding (not UTF8 encoding). ANSI can loose information, unlike Unicode.

Answer 2

This probably is related not to the program but the console, I suggest you try to output into a file and check if the encoding is correct.

You can do that using freopen:

int main(int argc, char *argv[]){ freopen("output-file-name.txt", "w", stdout); /*rest of code*/ }

If problem persists, try using visual studio along with _setmode(..., _O_U16TEXT) just before using wcout as described here: https://stackoverflow.com/a/9051543/9541897

Answer 3

Here's an example that works with Windows. You'll have to find the right compiler/linker settings to support wmain on MinGW, but it will work. _setmode enables writing Unicode directly to the terminal, and should work as long as the font supports the characters. In my example I use some Chinese, which my font supports:

#include <Windows.h>
#include <iostream>
#include "fcntl.h"
#include "io.h"

int wmain(int argc, wchar_t* argv[])
{
    _setmode(_fileno(stdout), _O_U16TEXT);
    std::wcout << argv[0] << std::endl;

    wchar_t filename[MAX_PATH];
    GetModuleFileNameW(NULL,filename,MAX_PATH);
    std::wcout << filename << std::endl;

    return 0;
}

Output:

马克.exe
C:\üлicä\马克.exe

Answer 4

Why are you typedefing LPWSTR and PWSTR manually? windows.h already handles that for you.

In any case, as @n.m. said in comments, the arguments for main() are NOT encoded in UTF-8 on Windows, so converting non-ASCII characters using a UTF8->UTF16 converter will not produce the correct output. Use the Win32 MultiByteToWideChar() function instead to convert the arguments, using CP_ACP as the codepage to convert from. Or, use wmain() instead, which provides arguments as wchar_t* instead of as char*.

That will get you the data you want. Then, you just need to deal with the issue of Unicode output to the console. As other answers point out, the Windows console does not support UTF-16 output via std::wcout by default, so you have to jump through some additional hoops to make it work correctly (there are many other questions on StackOverflow about that issue).