Comments on Cocoa with Love: Base64 encoding options on the Mac and iPhone

If you're getting that error message then the ...

2011-11-23T18:16:06.384-08:00

If you're getting that error message then the NSData+Base64.m file is not being linked into your build.

Good work, but when I call [imageData base64Encode...

2011-11-23T13:22:29.970-08:00

Good work, but when I call [imageData base64EncodedString] generate a error: *** Terminating app due to uncaught exception 'NSInvalidArgumentException', reason: '-[NSConcreteData base64EncodedString]: unrecognized selector sent to instance 0x91d2400'

I am not able to find file here

2011-08-18T00:50:14.803-07:00

I am not able to find file here

Good stuff!

2011-05-29T01:04:11.000-07:00

Good stuff!

If you need strings on both sides: - (NSString *)...

2010-12-13T17:52:20.000-08:00

If you need strings on both sides:

- (NSString *)base64Encode:(NSString *)plainText
{
NSLog(@"%@",plainText);
NSData *plainTextData = [plainText dataUsingEncoding:NSUTF8StringEncoding];
NSString *base64String = [plainTextData base64EncodedString];
NSLog(@"%@",base64String);
return base64String;
}

- (NSString *)base64Decode:(NSString *)base64String
{
NSData *plainTextData = [NSData dataFromBase64String:base64String];
NSString *plainText = [[NSString alloc] initWithData:plainTextData
encoding:NSUTF8StringEncoding];
NSLog(@"%@",plainText);
return plainText;
}

Thanks a lot. This saved my day.

2010-11-08T03:39:07.000-08:00

Thanks a lot. This saved my day.

To fix the clang static analyser warnings, you can...

2010-10-07T07:41:58.000-07:00

To fix the clang static analyser warnings, you can replace these lines

outputBuffer[j] = (accumulated[0] << 2) | (accumulated[1] >> 4);
outputBuffer[j + 1] = (accumulated[1] << 4) | (accumulated[2] >> 2);
outputBuffer[j + 2] = (accumulated[2] << 6) | accumulated[3];

by these lines:

if(accumulateIndex >= 2)
outputBuffer[j] = (accumulated[0] << 2) | (accumulated[1] >> 4);
if(accumulateIndex >= 3)
outputBuffer[j + 1] = (accumulated[1] << 4) | (accumulated[2] >> 2);
if(accumulateIndex >= 4)
outputBuffer[j + 2] = (accumulated[2] << 6) | accumulated[3];

The solution you mention would cause some base64 s...

2010-10-07T07:40:36.000-07:00

The solution you mention would cause some base64 strings to not be decoded correctly.
For example the string "PD94bWwgdmVyc2lvbj0iMS4wIiBlbmNvZGluZz0iVVRGLTgiPz48MTIzNDU+PC8xMjM0NT4=" would be decoded as "<12345></1234" instead of "<12345></12345>".

The solution would be to replace the lines:

outputBuffer[j] = (accumulated[0] << 2) | (accumulated[1] >> 4);
outputBuffer[j + 1] = (accumulated[1] << 4) | (accumulated[2] >> 2);
outputBuffer[j + 2] = (accumulated[2] << 6) | accumulated[3];

by these lines:

if(accumulateIndex >= 2)
outputBuffer[j] = (accumulated[0] << 2) | (accumulated[1] >> 4);
if(accumulateIndex >= 3)
outputBuffer[j + 1] = (accumulated[1] << 4) | (accumulated[2] >> 2);
if(accumulateIndex >= 4)
outputBuffer[j + 2] = (accumulated[2] << 6) | accumulated[3];

has anyone run the clang analyzer over this code? ...

2010-09-17T09:29:46.000-07:00

has anyone run the clang analyzer over this code? I get an error/warning on this line:

outputBuffer[j] = (accumulated[0] << 2) | (accumulated[1] >> 4);

which says: The left operand of '<<' is a garbage value.

Mark, Technically, if the input is bad (i.e. the ...

2010-08-22T20:43:41.000-07:00

Mark,

Technically, if the input is bad (i.e. the accumulated array is not filled correctly) then part or all of accumulated could be garbage.

It isn't a buffer overflow but bad data in may result in bad data out.

A good solution would be:

<pre>if (accumulateIndex != BASE64_UNIT_SIZE)
{
continue;
}</pre>

Immediately before the line you quoted. This will prevent accumulated being used if it is not correctly filled.

Hi Matt, Clang complains that in the following li...

2010-08-22T17:20:31.000-07:00

Hi Matt,

Clang complains that in the following line that accumulated[0] is garbage value.

outputBuffer[j] = (accumulated[0] << 2) | (accumulated[1] >> 4);

I assume that's because it can't determine which passes through the loop the if statement will be true. It seems to me there's a possibility that it could be garbage in some pathological scenario although unlikely. Initialising the array is a bit of overhead if it's unnecessary though.

Cheers.

I hadn't considered invalid input lengths (my ...

2010-03-30T18:59:07.000-07:00

I hadn't considered invalid input lengths (my brain is thinking about a different project right now but I think 7 bytes is an invalid length).

I agree that a buffer overrun -- even on invalid input -- is a major bad. I've altered the outputBufferSize as you suggested to round up, rather than down, and correctly account for this.

Thanks!

First of all, thanks for the very useful code. Se...

2010-03-30T07:19:29.000-07:00

First of all, thanks for the very useful code.

Secondly: It's a small point, but I think there's a potential for buffer overflow in NewBase64Decode().

The line:
<pre>size_t outputBufferSize = (length / BASE64_UNIT_SIZE) * BINARY_UNIT_SIZE;</pre>
doesn't anticipate irregularly-sized input buffers (e.g., a 7-byte input). AFAICT, if asked to process such a buffer, the code will write a few bytes beyond the bounds of outputBuffer.

Code like:
<pre>size_t outputBufferSize = ((length+BASE64_UNIT_SIZE-1) / BASE64_UNIT_SIZE) * BINARY_UNIT_SIZE;</pre>
will, I think, address the problem.