How to convert a unichar value to an NSString in Objective-C?

I've got an international character stored in a unichar variable. This character does not come from a file or url. The variable itself only stores an unsigned short(0xce91) which is in UTF-8 format and translates to the greek capital letter 'A'. I'm trying to put that character into an NSString variable but i fail miserably.

I've tried 2 different ways both of which unsuccessful:

unichar greekAlpha = 0xce91; //could have written greekAlpha = 'Α' instead.

NSString *theString = [NSString s开发者_开发知识库tringWithFormat:@"Greek Alpha: %C", greekAlpha];

No good. I get some weird chinese characters. As a sidenote this works perfectly with english characters.

Then I also tried this:

NSString *byteString = [[NSString alloc] initWithBytes:&greekAlpha
                                                length:sizeof(unichar)
                                              encoding:NSUTF8StringEncoding];

But this doesn't work either. I'm obviously doing something terribly wrong, but I don't know what. Can someone help me please ? Thanks!

---

unichar greekAlpha = 0x0391;
NSString* s = [NSString stringWithCharacters:&greekAlpha length:1];

And now you can incorporate that NSString into another in any way you like. Do note, however, that it is now legal to type a Greek alpha directly into an NSString literal.

---

Since 0xce91 is in the UTF-8 format and %C expects it to be in UTF-16 a simple solution like the one above won't work. For stringWithFormat:@"%C" to work you need to input 0x391 which is the UTF-16 unicode.

In order to create a string from the UTF-8 encoded unichar you need to first split the unicode into it's octets and then use initWithBytes:length:encoding.

unichar utf8char = 0xce91; 
char chars[2];
int len = 1;

if (utf8char > 127) {
    chars[0] = (utf8char >> 8) & (1 << 8) - 1;
    chars[1] = utf8char & (1 << 8) - 1; 
    len = 2;
} else {
    chars[0] = utf8char;
}

NSString *string = [[NSString alloc] initWithBytes:chars
                                            length:len 
                                          encoding:NSUTF8StringEncoding];

---

The above answer is great but doesn't account for UTF-8 characters longer than 16 bits, e.g. the ellipsis symbol - 0xE2,0x80,0xA6. Here's a tweak to the code:

if (utf8char > 65535) {
   chars[0] = (utf8char >> 16) & 255;
   chars[1] = (utf8char >> 8) & 255;
   chars[2] = utf8char & 255; 
   chars[3] = 0x00;
} else if (utf8char > 127) {
    chars[0] = (utf8char >> 8) & 255;
    chars[1] = utf8char & 255; 
    chars[2] = 0x00;
} else {
    chars[0] = utf8char;
    chars[1] = 0x00;
}
NSString *string = [[[NSString alloc] initWithUTF8String:chars] autorelease];

Note the different string initialisation method which doesn't require a length parameter.

---

Here is an algorithm for UTF-8 encoding on a single character:

if (utf8char<0x80){ 
    chars[0] = (utf8char>>0)  & (0x7F | 0x00);
    chars[1] = 0x00;
    chars[2] = 0x00;
    chars[3] = 0x00;
}
else if (utf8char<0x0800){
    chars[0] = (utf8char>>6)  & (0x1F | 0xC0);
    chars[1] = (utf8char>>0)  & (0x3F | 0x80);
    chars[2] = 0x00;
    chars[3] = 0x00;
}
else if (utf8char<0x010000) {
    chars[0] = (utf8char>>12) & (0x0F | 0xE0);
    chars[1] = (utf8char>>6)  & (0x3F | 0x80);
    chars[2] = (utf8char>>0)  & (0x3F | 0x80);
    chars[3] = 0x00;
}
else if (utf8char<0x110000) {
    chars[0] = (utf8char>>18) & (0x07 | 0xF0);
    chars[1] = (utf8char>>12) & (0x3F | 0x80);
    chars[2] = (utf8char>>6)  & (0x3F | 0x80);
    chars[3] = (utf8char>>0)  & (0x3F | 0x80);
}

---

The code above is the moral equivalent of unichar foo = 'abc';.

The problem is that 'Α' doesn't map to a single byte in the "execution character set" (I'm assuming UTF-8) which is "implementation-defined" in C99 §6.4.4.4 10:

The value of an integer character constant containing more than one character (e.g., 'ab'), or containing a character or escape sequence that does not map to a single-byte execution character, is implementation-defined.

One way is to make 'ab' equal to 'a'<<8|b. Some Mac/iOS system headers rely on this for things like OSType/FourCharCode/FourCC; the only one in iOS that comes to mind is CoreVideo pixel formats. This is, however, unportable.

If you really want a unichar literal, you can try L'A' (technically it's a wchar_t literal, but on OS X and iOS, wchar_t is typically UTF-16 so it'll work for things inside the BMP). However, it's far simpler to just use @"Α" (which works as long as you set the source character encoding correctly) or @"\u0391" (which has worked since at least the iOS 3 SDK).