Problems converting wide string to multibyte in c-CodePudding

I'm trying to convert a path from a wstring to a string. But when I do it outputs garbage. Is there a way to get these japanese characters into a string?

wstring foo = L"C:\\projects\\サービス\\b2";
char bar[256] = { 0 };
string baz( bar );

int len = WideCharToMultiByte( CP_UTF8, 0, foo.c_str(), -1, NULL, 0, 0, 0 );
WideCharToMultiByte( CP_UTF8, 0, foo.c_str(), -1, &bar[0], len, 0, 0 );

when debugging on windows (vs2019) baz contains: "C:\projects\ã‚µãƒ¼ãƒ“ã‚¹\b2"

CodePudding user response：

As @Codo explained it was actually the correct data. And as @Richard Critten tried to explain, I needed to check the hex, not just the decimals.

For others encountering similar problems I wanted to do this so that later I could test the string to see if the directory exists. Here's how I did that:

Since I was using stat, it didn't like the "unicode" string. So I had to convert back to a wstring using MultiByteToWideChar and use the wide string version: _wstat.

So assuming path is actually wchars:

int wchars_num = MultiByteToWideChar( CP_UTF8, 0, path, -1, NULL, 0 );
wchar_t *wpath = new wchar_t[wchars_num];
MultiByteToWideChar( CP_UTF8, 0, path, -1, wpath, wchars_num );

struct _stat64i32 wInfo;
statRC = _wstat( wpath, &wInfo );
if ( statRC != 0 )
{
    if ( errno == ENOENT ) { return false; } // something along the path does not exist
    if ( errno == ENOTDIR ) { return false; } // something in path prefix is not a dir
    return false;
}
else if ( wInfo.st_mode & S_IFDIR )
{
    return true;
}

CodePudding user response：

You are converting properly, but not viewing the string correctly in the debugger. The default when viewing the char* in Visual Studio is to assume an ANSI encoding specific to the localized version of Windows used. In Western European/US Windows, that would be Windows-1252.

At least as of Visual Studio 2015, use the s8 format code in the Watch Window to view UTF-8-encoded strings:

main() with OP code viewing variables with and without s8 format code in watch window

See Format specifiers in the debugger

CodePudding user response：

You are not allocating any memory for the target string. You are creating the string from an empty char[] first, THEN you are converting into that same char[], which will NOT update the string automatically.

You don't need the char[] at all. Try this instead:

wstring foo = L"C:\\projects\\サービス\\b2";
string baz;

int len = WideCharToMultiByte( CP_UTF8, 0, foo.c_str(), foo.size(), NULL, 0, 0, 0 );
baz.resize(len);
WideCharToMultiByte( CP_UTF8, 0, foo.c_str(), foo.size(), &baz[0]/*or: baz.data()*/, len, 0, 0 );