NSArray from NSCharacterSet

The following code creates an array containing all characters of a given character set. It works also for characters outside of the "basic multilingual plane" (characters > U+FFFF, e.g. U+10400 DESERET CAPITAL LETTER LONG I).

NSCharacterSet *charset = [NSCharacterSet uppercaseLetterCharacterSet];
NSMutableArray *array = [NSMutableArray array];
for (int plane = 0; plane <= 16; plane++) {
    if ([charset hasMemberInPlane:plane]) {
        UTF32Char c;
        for (c = plane << 16; c < (plane+1) << 16; c++) {
            if ([charset longCharacterIsMember:c]) {
                UTF32Char c1 = OSSwapHostToLittleInt32(c); // To make it byte-order safe
                NSString *s = [[NSString alloc] initWithBytes:&c1 length:4 encoding:NSUTF32LittleEndianStringEncoding];
                [array addObject:s];
            }
        }
    }
}

For the uppercaseLetterCharacterSet this gives an array of 1467 elements. But note that characters > U+FFFF are stored as UTF-16 surrogate pair in NSString, so for example U+10400 actually is stored in NSString as 2 characters "\uD801\uDC00".

Swift 2 code can be found in other answers to this question. Here is a Swift 3 version, written as an extension method:

extension CharacterSet {
    func allCharacters() -> [Character] {
        var result: [Character] = []
        for plane: UInt8 in 0...16 where self.hasMember(inPlane: plane) {
            for unicode in UInt32(plane) << 16 ..< UInt32(plane + 1) << 16 {
                if let uniChar = UnicodeScalar(unicode), self.contains(uniChar) {
                    result.append(Character(uniChar))
                }
            }
        }
        return result
    }
}

Example:

let charset = CharacterSet.uppercaseLetters
let chars = charset.allCharacters()
print(chars.count) // 1521
print(chars) // ["A", "B", "C", ... "]

(Note that some characters may not be present in the font used to display the result.)


Inspired by Satachito answer, here is a performant way to make an Array from CharacterSet using bitmapRepresentation:

extension CharacterSet {
    func characters() -> [Character] {
        // A Unicode scalar is any Unicode code point in the range U+0000 to U+D7FF inclusive or U+E000 to U+10FFFF inclusive.
        return codePoints().compactMap { UnicodeScalar($0) }.map { Character($0) }
    }
    
    func codePoints() -> [Int] {
        var result: [Int] = []
        var plane = 0
        // following documentation at https://developer.apple.com/documentation/foundation/nscharacterset/1417719-bitmaprepresentation
        for (i, w) in bitmapRepresentation.enumerated() {
            let k = i % 0x2001
            if k == 0x2000 {
                // plane index byte
                plane = Int(w) << 13
                continue
            }
            let base = (plane + k) << 3
            for j in 0 ..< 8 where w & 1 << j != 0 {
                result.append(base + j)
            }
        }
        return result
    }
}

Example for uppercaseLetters

let charset = CharacterSet.uppercaseLetters
let chars = charset.characters()
print(chars.count) // 1733
print(chars) // ["A", "B", "C", ... "]

Example for discontinuous planes

let charset = CharacterSet(charactersIn: "𝚨󌞑")
let codePoints = charset.codePoints()
print(codePoints) // [120488, 837521]

Performances

Very good depending on the data/usage: this solution built in release with bitmapRepresentation seems 2 to 10 times faster than Martin R's solution with contains or Oliver Atkinson's solution with longCharacterIsMember.

Be sure to compare depending on your own needs: performances are best compared in a non-debug build; so avoid comparing performances in a Playground.