I took down this Gist due to concerns about the security of the encryption/decryption part of this code (see comments below).
Rob Napier (@rnapier) has created a publicly available class that provides similar AES encryption/decryption functionality at https://github.com/rnapier/RNCryptor.
@rnapier, thank you for the explanation.
Rob, thanks for your clarification what the problem with this code is. I took down the Gist (or the code files in it) and added a link to your class. I hope you're OK with that.
When I find the time I will fork your repository and add "encryptString" / "decryptString" methods using BASE64 encoding to it — something that was provided by this code.
Again, thanks for pointing out the weaknesses in this code collection.
– Michael
@gonecoding, what kind of API are you seeing for Base-64 encoding? From its name, it's not clear what encryptString: or decryptString: would take and return. My expectation would be that encryptString: would take an NSString and return an NSData. decryptString: is unclear. Perhaps decryptBase64:, which would take Base64 data and return an NSData. In any case, though it's challenging to see how that expands to the stream functions. If the Base64 data is in a file, can you use decryptFromURL:...? I would be thoughtful of what belongs in the cryptor and what doesn't. It's very easy to provide a function that does NSData <-> Base64 to wrap around the cryptor. And an NSStream that converts data to/from Base64 might be much more flexible than putting it directly into the cryptor.
Do you have a use specific case in mind? Fork and code is great; also feel free to open issues.
Some body could help me, I get two errors
Undefined symbols for architecture i386:
"_STAssertTrue", referenced from:
-[BrowserController testStream] in SomeFile.o
"_STAssertEqualObjects", referenced from:
-[BrowserController testStream] in SomeFile.o
ld: symbol(s) not found for architecture i386
clang: error: linker command failed with exit code 1 (use -v to see invocation)
Have you pulled the entire project, or just dropped the files into another project? The failed symbols are from the unit tests. You don't need those in a real project. If you want to build the unit tests, you need to link with SenTestingKit.framework (this comes with Xcode).
ok thanks. Now when I try to build in iOS 4.3 i got this error
dyld: lazy symbol binding failed: Symbol not found: _CCKeyDerivationPBKDF
and
dyld: Symbol not found: _CCKeyDerivationPBKDF
but it works fine in simulator 5.0
See the discussion here: RNCryptor/RNCryptor#22. RNCryptor is currently designed to be iOS5 only. I'm considering whether to modify that.
To @tqbf's comment, yes, this is terribly insecure. For a full discussion on the problems with this code, see http://robnapier.net/blog/aes-commoncrypto-564. I'm worried that this code seems to be very popular, leading to serious encryption problems within the iOS community. It's easy to use, which has led to broad usage, but it doesn't correctly generate the key (no reasonable KDF, and no salt), and it doesn't provide an IV.
It's good to see an easy-to-use AES implementation, but this code is not something people should use without careful understanding of what it does. It dramatically constricts the keyspace, and has significant vulnerabilities in the first block due to the lack of IV.
To @tudormunteanu's comment about OpenSSL, the issue is that there is no standard for AES data format. OpenSSL uses EVP_BytesToKey as their KDF. AES256Encrypt doesn't use a KDF at all (which is why the keyspace is so constricted). The standard is PBKDF2, but there is no standard for the number of rounds. Some implementations use bcrypt or scrypt, but there isn't any standard there. OpenSSL also has its own format for providing the IV that will not be compatible with this code.
To put code where my mouth is, I offer https://github.com/rnapier/RNCryptor, which I intend to address the issues I've raised here, and I hope is as easy to use as this category. I'm still cleaning up some of the error handling, but the main encryption/decryption + HMAC code should be solid and stable.