Character array to wide string?

[Sephiroth]

I am still working on my project which will be reading some old data from some old DOS files. The data stored there is naturally, char*. Once I read in my character array, how do I assign this to a wstring since my application is UNICODE?

Here is my current solution:

wchar_t* Class::Function(char *pName)
{
  //I verify the pointer and such first, then do the below
  this->_Name.assign(pName, (pName + strlen(pName));
  return this->_Name.c_str();
}

Am I on the right track here?

[Disch]

The root questions are:

1) How is the source data encoded (UTF-8 seems unlikely if this is old DOS files... but I guess it depends on HOW old).
2) Do you care about preserving anything beyond the ASCII set?
3) How is the dest data to be encoded (I would assume UTF-16?). Given that you said "UNICODE" I'm assuming this is on Windows and you want UTF-16.


WinAPI provides a method (MultiByteToWideChar) which can convert pretty much any codepage, as well as UTF-8, to UTF-16.


Or... if you don't care and all you care about is the basic ASCII set... it's a straight 1:1 copy (just converting a 1 byte character to a 2+ byte character):

void copyToWide(wchar_t* dst, const char* src)
{
    while(*src)
    {
        *dst = *src;
        ++dst;
        ++src;
    }
    *dst = 0;
}

EDIT: your use of assign is actually simpler than my approach. Hah. That will work just fine if you don't care about anything beyond normal ASCII.

[Sephiroth]

Yes I am on Windows using wchar_t which is 16bit, but I will also be using this on Linux where wchar_t is 32bit. The good news? I am only READING the data on both systems so there will be no new writing at this point. If I decide to write data and share it, I will write it as 16bit for Windows and easily read that into the Linux equivalent.

Now it is old DOS, not UTF8. I am assuming old ASCII only, and I do not plan on converting it to any other languages at this point. That would be a project in itself, so the ASCII set is fine with me. I can jump any conversion hurdles down the road.

Finally, it looks as though my call to the "assign" method is correct? I am asking because this will be a library and as such I will not be able to test it until after the program which uses this library is also to a certain stage in development.

[Disch]

Finally, it looks as though my call to the "assign" method is correct?

Yes.

Though I would question your use of wchar_t in this library, as chars are typically much easier to work with. Case in point... you just mentioned that wchar_t is different sizes on different platforms... which makes it more difficult to write portable code.

[Sephiroth]

Yes, but everything is UNICODE now. Plus it will allow me to target other languages and such in the future if I need to. Just because it is easy doesn't mean it is right, after all. I honestly don't know anybody not coding UNICODE apps anymore, and I enjoy learning as I go with the new UNICODE stuff.

[Disch]

UTF-8 is unicode... and can be contained in a normal char array. Or a normal std::string.

In fact.. IIRC, the API for filesystem I/O on linux platforms all accept UTF-8 strings (const char*).

[Sephiroth]

I know you can use UTF8 with a char array, but I have never come across a UTF8 file that wasn't using some other form of storage. Most of the time it is wchar_t. This particular file was out about the same time Windows 95 was released, so no worries of UTF8 there. If it was a Linux system, maybe. Linux always seems to be ahead of the game.

[Cubbi]

The data stored there is naturally, char*. Once I read in my character array, how do I assign this to a wstring

Why read into a char array in the first place? Open the file as std::wifstream, and you can getline or operator>> straight into a std::wstring.

[Disch]

I know you can use UTF8 with a char array, but I have never come across a UTF8 file that wasn't using some other form of storage.

Different strokes I guess. I've found that UTF-8 is more common pretty much everywhere. Especially in places that had to be 'upgraded' while still maintaining backward compatibility (switching from ANSI to UTF-8 is much easier than switching to UTF-16).

Posix API is an example .. none of it from what I've seen uses UTF-16. Any kind of binary file with embedded comments (zip, png, etc). In fact the only place I can think of where I've seen UTF-16 in widespread use is in WinAPI.

But whatever... it's your code and you can do what you want. =) Don't let me bully you.

[Sephiroth]

I just wanted to answer Cubbi. I have to read the file into a char array due to the file being binary. It contains 3D object data as well as ANSI names for said objects.