Move 2 engineDoFinal methods from OpenSSLCipher.java to its subclasses. #1264
Conversation
|
Awesome, thanks! LGTM but I'll take another look with fresh eyes in the morning. |
prbprbprb
left a comment
prbprbprb
left a comment
There was a problem hiding this comment.
I think we're lacking some tests around zero length inputs but overall LGTM apart from one untested branch noted below.
|
|
||
| final int bytesWritten; | ||
| if (inputLen > 0) { | ||
| bytesWritten = updateInternal(input, inputOffset, inputLen, output, outputOffset, |
There was a problem hiding this comment.
No tests seem to hit this branch, could you add one, please.
There was a problem hiding this comment.
Done. The Cipher interface has many similar methods, and depending on how exactly it is called, a different method of the implementation is executed. I've now changed the tests of Cipher so that it tries a range of different call patterns. I think this should give a better coverage.
| } | ||
|
|
||
| /** Concatenates the given arrays into a single array.*/ | ||
| byte[] concatArrays(byte[]... arrays) { |
There was a problem hiding this comment.
nit: We already have ArrayUtils.concat for the two array case. Maybe move this method into ArrayUtils for the other cases (it's fairly small).
| byte[] concatArrays(byte[]... arrays) { | ||
| int length = 0; | ||
| for (byte[] array : arrays) { | ||
| if (array == null) { |
There was a problem hiding this comment.
maybe throw on null? But you can certainly skip zero length arrays.
There was a problem hiding this comment.
Actually don't worry about fixing the nits :)
2 participants
Each of the three subclasses needs a slightly different implementation of these methods. So it's better to have the implementation here.