Dynamic UTI decoding

decode-dyn-uti.swift

extension String
{
    /// Creates a string by decoding a sequence of UTF-8 code units.
    init?<S: Sequence>(utf8 input: S) where S.Iterator.Element == UTF8.CodeUnit {
        var chars: [Character] = []
        var decoder = UTF8()
        var iter = input.makeIterator()
        loop: while true {
            switch decoder.decode(&iter) {
            case .scalarValue(let scalar): chars.append(Character(scalar))
            case .emptyInput: break loop
            case .error: return nil
            }
        }
        self = String(chars)
    }
}


/// Decodes a dynamically-generated Uniform Type Identifier for inspection purposes. (**NOT FOR PRODUCTION USE!**)
/// Many, many thanks to http://alastairs-place.net/blog/2012/06/06/utis-are-better-than-you-think-and-heres-why/
///
/// <https://developer.apple.com/library/ios/documentation/FileManagement/Conceptual/understanding_utis/understand_utis_conc/understand_utis_conc.html#//apple_ref/doc/uid/TP40001319-CH202-BCGCDHIJ>
func decodeDynUTI(_ uti: String) -> String?
{
    let vec = Array("abcdefghkmnpqrstuvwxyz0123456789")
    
    let encoded = Array(uti).suffix(from: 5)
    var result: [UInt8] = []
    var decoded = 0
    var decodedBits = 0
    for char in encoded {
        // Each encoded character represents 5 bits (by its position in the length-32 vector).
        guard let pos = vec.index(of: char) else {
            print("unrecognized encoded character '\(char)'")
            return nil
        }
        decoded = (decoded << 5) | pos
        decodedBits += 5
        
        // Every 8 decoded bits represent a UTF-8 code unit.
        if decodedBits >= 8 {
            let extra = decodedBits - 8
            result.append(UInt8(decoded >> extra))
            decoded &= (1 << extra) - 1
            decodedBits = extra
        }
    }
    
    if decoded != 0 {
        print("\(decodedBits) leftover bits: \(decoded)")
        return nil
    }
    
    return String(utf8: result)
}


// Examples

decodeDynUTI("dyn.ah62d4rv4gu8yc6durvwwa3xmrvw1gkdusm1044pxqyuha2pxsvw0e55bsmwca7d3sbwu")  // "?0=6:4=Apple files promise pasteboard type"
decodeDynUTI("dyn.ah62d4rv4gu8y6y4usm1044pxqzb085xyqz1hk64uqm10c6xenv61a3k") // "?0=6:4=NSPromiseContentsPboardType"
decodeDynUTI("dyn.ah62d46dzqm0gw23srf4gn5m4ge8068dytf2gn3dgrf0gn25tr34gn5xysr2ge55bsmwhk8puqzsdy4xuq6") // "?0=public.item:1=nxtypedfilecontentspboardtype\:jpg"
decodeDynUTI("dyn.ah62d4uv4ge804550") // "?0=B:1=mov"
decodeDynUTI("dyn.agq80c6durvy0g2pyrf106p5rsa4a") // "3=application/mp4"
decodeDynUTI("dyn.ah62d4rv4gu8ykzwynr11n6xdqzngn8dyn3y0n74qqf1gs7pbq7wza2xtqf3gkzd3sbwu") // "?0=6:4=DVTSourceTextViewLanguagePboardType"
decodeDynUTI("dyn.ah62d4rv4gu8ykzwynr11n6xdqzngn8dyn3y0n74xqr11a3nqqf1gs7pbq7wyg55ssvw1u7cuqm10c6xenv61a3k") // "?0=6:4=DVTSourceTextViewScopeLanguageContextPboardType"
decodeDynUTI("dyn.ah62d4rv4gu8ykzwynr11n6xdqzngn8dyn3y0n74msra1kuwtmvkge55bsmwfk8puqy") // "?0=6:4=DVTSourceTextViewIsAtBOLPboardType"
decodeDynUTI("dyn.ah62d4rv4gk81n65yru") // "?0=6:2=ustl"
decodeDynUTI("dyn.ah62d4rv4gk81g7d3ru") // "?0=6:2=styl"