Error when decoding certain Base64 strings, but no

Question

To keep this simple, I'll only be encoding/decoding a single byte.

If I encode the byte 127, I get the base64 string "fw==" which can be successfully decoded back to byte 127. If, however, I encode a byte ≥ 128, then even though I can produce a base64 string without error (for example, byte 128 gives the string "gA=="), I get an error when I try to decode it.

Here's my code which can be copy-pasted into any Xcode playground to reproduce the problem:

func stringToByteArray(string: String) -> [UInt8] {
    var bytes: [UInt8] = [];
    for code in string.utf8 {
        bytes.append(UInt8(code));
    }
    return bytes;
}

func byteArrayToBase64(bytes: [UInt8]) -> String {
    let nsdata: NSData = NSData(bytes: bytes as [Byte], length: bytes.count)
    let base64Encoded: NSString = nsdata.base64EncodedStringWithOptions(NSDataBase64EncodingOptions(rawValue: 0));
    return String(base64Encoded);
}

func base64ToByteArray(base64String: String) -> [UInt8] {
    let nsdata: NSData = NSData(base64EncodedString: base64String, options: NSDataBase64DecodingOptions(rawValue: 0))!
    let base64Decoded: NSString = NSString(data: nsdata, encoding: NSUTF8StringEncoding)!
    return stringToByteArray(String(base64Decoded));
}

/* Replacing 127 with 128 below or greater produces an error */
var testString = byteArrayToBase64([127]);
base64ToByteArray(testString)

Answer 1

The problem is here:

let base64Decoded: NSString = NSString(data: nsdata, encoding: NSUTF8StringEncoding)!

You convert the decoded data to a string. This fails for [128] because that does not represent a valid UTF-8 sequence.

Here is a version that avoids the intermediate string:

func base64ToByteArray(base64String: String) -> [UInt8] {
    let nsdata: NSData = NSData(base64EncodedString: base64String, options: NSDataBase64DecodingOptions(rawValue: 0))!
    // Create array of the required size ...
    var bytes = [UInt8](count: nsdata.length, repeatedValue: 0)
    // ... and fill it with the data
    nsdata.getBytes(&bytes)
    return bytes
}

Remarks:

• options: NSDataBase64DecodingOptions(rawValue: 0) can be simplified to options: nil.
• There are some unnecessary type annotations and conversions in your code.
• Your function crashes if baseString is not a valid Base64 string. You could change it to return an optional.

Then it would look like this:

func byteArrayToBase64(bytes: [UInt8]) -> String {
    let nsdata = NSData(bytes: bytes, length: bytes.count)
    let base64Encoded = nsdata.base64EncodedStringWithOptions(nil);
    return base64Encoded;
}

func base64ToByteArray(base64String: String) -> [UInt8]? {
    if let nsdata = NSData(base64EncodedString: base64String, options: nil) {
        var bytes = [UInt8](count: nsdata.length, repeatedValue: 0)
        nsdata.getBytes(&bytes)
        return bytes
    }
    return nil // Invalid input
}

Example usage:

let testString = byteArrayToBase64([127, 128, 0, 130]);
println(testString) // Output: f4AAgg==
if let result = base64ToByteArray(testString) {
    println(result) // Output: [127, 128, 0, 130]
} else {
    println("failed")
}

---

Update for Swift 2 / Xcode 7:

func byteArrayToBase64(bytes: [UInt8]) -> String {
    let nsdata = NSData(bytes: bytes, length: bytes.count)
    let base64Encoded = nsdata.base64EncodedStringWithOptions([]);
    return base64Encoded;
}

func base64ToByteArray(base64String: String) -> [UInt8]? {
    if let nsdata = NSData(base64EncodedString: base64String, options: []) {
        var bytes = [UInt8](count: nsdata.length, repeatedValue: 0)
        nsdata.getBytes(&bytes, length: bytes.count)
        return bytes
    }
    return nil // Invalid input
}

let testString = byteArrayToBase64([127, 128, 0, 130]);
print(testString) // Output: f4AAgg==
if let result = base64ToByteArray(testString) {
    print(result) // Output: [127, 128, 0, 130]
} else {
    print("failed")
}