How do I tell a wstring that a string I am feeding it is already a wstring?

Question

I am using a std::string as a text buffer. Then, I am sure the data contained in that buffer is UTF-16 (i.e. it is really a std::wstring). How can I coerce a std::string into a std::wstring? The std::string is a misnomer, the data is really a wstring.

Answer 1

Consider using a std::vector<char> instead of a std::string. It's the correct container when you want "a contiguous sequence of bytes."

With a std::vector source container, the code is rather straightforward, assuming you really just want to reinterpret the data (i.e., you really just want to treat the bytes as if they were a sequence of wchar_t):

std::vector<char> v = get_my_wstring_character_data();
if (v.size() % sizeof (wchar_t) != 0)
    throw std::runtime_error("Invalid wstring length");

std::wstring ws(reinterpret_cast<wchar_t*>(&v[0]), 
                reinterpret_cast<wchar_t*>(&v[0] + v.size()));

If your source is a std::string, this same approach will work if you can guarantee that the implementation of std::string you are using stores its characters contiguously. In practice, this is always the case.

Answer 2

There is no guarantee that std::wstring stores/interprets byte arrays as UTF-16 (although it happens to do that in Windows). Check out this question: std::wstring VS std::string

Therefore I would advise you to rethink the idea of constructing a std::wstring from a UTF-16 encoded byte array unless you are sure your application will only ever be compiled with MSVC.