I have UTF-8 encoded NSData from windows server and I want to convert it to NSString for iPhone. Since data contains characters (like a degree symbol) which have different values on both platforms, how do I convert data to string?
 
    
    - 3,221
- 5
- 30
- 41
 
    
    - 6,586
- 6
- 26
- 35
- 
                    16UTF-8 is UTF-8 everywhere. Once it's UTF-8, there's no different values for different platforms. That's the whole point of it. – gnasher729 Apr 12 '14 at 11:27
7 Answers
If the data is not null-terminated, you should use -initWithData:encoding:
NSString* newStr = [[NSString alloc] initWithData:theData encoding:NSUTF8StringEncoding];
If the data is null-terminated, you should instead use -stringWithUTF8String: to avoid the extra \0 at the end.
NSString* newStr = [NSString stringWithUTF8String:[theData bytes]];
(Note that if the input is not properly UTF-8-encoded, you will get nil.)
Swift variant:
let newStr = String(data: data, encoding: .utf8)
// note that `newStr` is a `String?`, not a `String`.
If the data is null-terminated, you could go though the safe way which is remove the that null character, or the unsafe way similar to the Objective-C version above.
// safe way, provided data is \0-terminated
let newStr1 = String(data: data.subdata(in: 0 ..< data.count - 1), encoding: .utf8)
// unsafe way, provided data is \0-terminated
let newStr2 = data.withUnsafeBytes(String.init(utf8String:))
- 
                    5watch out!! if using stringWithUTF8String, don't pass it a NULL argument or it will throw an exception – JasonZ Jul 05 '12 at 15:57
- 
                    31MIND THIS: when using "stringWithUTF8String:" on a string that is not null-terminated, the result is unpredictable! – Berik Aug 01 '12 at 09:34
- 
                    2
- 
                    You can also drop the NS if using swift 2 making it: let newStr = String(data: data, encoding: NSUTF8StringEncoding) – Jeremiah Jan 20 '16 at 19:46
- 
                    I meet a crash: NSString *jsonStr = [NSString stringWithUTF8String:[jsonData bytes]]; And NSString* newStr = [[NSString alloc] initWithData:theData encoding:NSUTF8StringEncoding]; – ylgwhyh Nov 14 '16 at 16:15
- 
                    Dumb question probably (just learning Swift, don't know Objective-C at all): You show two variants for Objective-C depending on whether the input is null-terminated or not, but only one variant for Swift? – RenniePet Dec 06 '16 at 02:46
- 
                    1How do you know whether your NSData is null-terminated or not? See Tom Harrington's answer at: https://stackoverflow.com/questions/27935054/determine-if-utf-8-encoded-nsdata-contains-a-null-terminated-string. In my experience, one should not ever assume NSData is either null-terminated or not: it can differ from one transmission to the next, even from a known server. – Elise van Looij Jul 08 '18 at 14:44
- 
                    1@ElisevanLooij Thanks for the link. I'd argue that if the transmitted data could be randomly null-terminated or not the protocol is ill-defined. – kennytm Jul 09 '18 at 16:09
- 
                    1@kennytm No doubt you're right, but sometimes one has to row with the oars that one has. (Apologies for translating a Dutch saying) – Elise van Looij Jul 09 '18 at 17:11
You could call this method
+(id)stringWithUTF8String:(const char *)bytes.
- 
                    27Only if the data is null-terminated. Which it may not be (and, in fact, probably is not). – Ivan Vučica Mar 08 '13 at 10:33
- 
                    i don't know why on earth this would break on non-null-terminated strings seeing how the `NSData` knows how many bytes it has... – Claudiu Oct 01 '13 at 01:24
- 
                    5@Claudiu, you're not passing in an NSData object, you're passing it a (const char *) obtained with [data bytes], which is just a pointer, no size information. Hence the data block it points to must be null terminated. Check out the documentation, it says so explicitly. – jbat100 Oct 21 '13 at 11:26
- 
                    1@jbat100: Of course. I wasn't clear. I meant, given that it's possible to go from a non-null-terminated `NSData` to an `NSString` (see KennyTM's answer), I'm surprised there isn't a `+(id)stringWithUTF8Data:(NSData *)data` which just works. – Claudiu Oct 21 '13 at 15:36
- 
                    stringWithUTF8Data, hence most of us create a NSString+Foo category and create the method. – Cerniuk Oct 14 '17 at 10:42
I humbly submit a category to make this less annoying:
@interface NSData (EasyUTF8)
// Safely decode the bytes into a UTF8 string
- (NSString *)asUTF8String;
@end
and
@implementation NSData (EasyUTF8)
- (NSString *)asUTF8String {
    return [[NSString alloc] initWithData:self encoding:NSUTF8StringEncoding];    
}
@end
(Note that if you're not using ARC you'll need an autorelease there.)
Now instead of the appallingly verbose:
NSData *data = ...
[[NSString alloc] initWithData:data encoding:NSUTF8StringEncoding];
You can do:
NSData *data = ...
[data asUTF8String];
 
    
    - 224,032
- 165
- 485
- 680
The Swift version from String to Data and back to String:
Xcode 10.1 • Swift 4.2.1
extension Data {
    var string: String? {
        return String(data: self, encoding: .utf8)
    }
}
extension StringProtocol {
    var data: Data {
        return Data(utf8)
    }
}
extension String {
    var base64Decoded: Data? {
        return Data(base64Encoded: self)
    }
}
Playground
let string = "Hello World"                                  // "Hello World"
let stringData = string.data                                // 11 bytes
let base64EncodedString = stringData.base64EncodedString()  // "SGVsbG8gV29ybGQ="
let stringFromData = stringData.string                      // "Hello World"
let base64String = "SGVsbG8gV29ybGQ="
if let data = base64String.base64Decoded {
    print(data)                                    //  11 bytes
    print(data.base64EncodedString())              // "SGVsbG8gV29ybGQ="
    print(data.string ?? "nil")                    // "Hello World"
}
let stringWithAccent = "Olá Mundo"                          // "Olá Mundo"
print(stringWithAccent.count)                               // "9"
let stringWithAccentData = stringWithAccent.data            // "10 bytes" note: an extra byte for the acute accent
let stringWithAccentFromData = stringWithAccentData.string  // "Olá Mundo\n"
 
    
    - 229,809
- 59
- 489
- 571
Sometimes, the methods in the other answers don't work. In my case, I'm generating a signature with my RSA private key and the result is NSData. I found that this seems to work:
Objective-C
NSData *signature;
NSString *signatureString = [signature base64EncodedStringWithOptions:0];
Swift
let signatureString = signature.base64EncodedStringWithOptions(nil)
 
    
    - 6,790
- 3
- 34
- 46
- 
                    
- 
                    1@DarshanKunjadiya: **Objective-C**: `[[NSData alloc] initWithBase64EncodedString:signatureString options:0]`; **Swift**: `NSData(base64EncodedString: str options: nil)` – mikeho Feb 11 '15 at 19:09
Just to summarize, here's a complete answer, that worked for me.
My problem was that when I used
[NSString stringWithUTF8String:(char *)data.bytes];
The string I got was unpredictable:  Around 70% it did contain the expected value, but too often it resulted with Null or even worse: garbaged at the end of the string.
After some digging I switched to
[[NSString alloc] initWithBytes:(char *)data.bytes length:data.length encoding:NSUTF8StringEncoding];
And got the expected result every time.
 
    
    - 1,582
- 2
- 14
- 30
With Swift 5, you can use String's init(data:encoding:) initializer in order to convert a Data instance into a String instance using UTF-8. init(data:encoding:) has the following declaration:
init?(data: Data, encoding: String.Encoding)
Returns a
Stringinitialized by converting given data into Unicode characters using a given encoding.
The following Playground code shows how to use it:
import Foundation
let json = """
{
"firstName" : "John",
"lastName" : "Doe"
}
"""
let data = json.data(using: String.Encoding.utf8)!
let optionalString = String(data: data, encoding: String.Encoding.utf8)
print(String(describing: optionalString))
/*
 prints:
 Optional("{\n\"firstName\" : \"John\",\n\"lastName\" : \"Doe\"\n}")
*/
 
    
    - 89,880
- 29
- 256
- 218
 
     
     
    