use padding option on AES.decrypt results in endless loading

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Just use CryptoJS.AES.decrypt(message, key, { iv: iv, padding: 
CryptoJS.pad.ZeroPadding });

What is the expected output? What do you see instead?
Indeed specifying padding at decrypt is useless, but throw an error or just 
ignore the unnecessary option would be better than endless loading.

What version of the product are you using? On what operating system?
Tested on Mac OS X with XAMPP 1.8.3-3 (Apache Web Server) and Chrome 
35.0.1916.114.
Used crypto-js 3.1.2 aes.js and pad-zeropadding-min.js

Original issue reported on code.google.com by johannes...@gmail.com on 27 May 2014 at 2:49

[GoogleCodeExporter]

During working on I noticed that my description is mostly wrong.
In fact using padding on decrypt isn't useless. And just using this option 
doesn't reproduces the error at all.
I still trying to figure out what the actual problem is... just a bit 
over-worked.

Original comment by johannes...@gmail.com on 27 May 2014 at 5:09

[GoogleCodeExporter]

Ok, my big mistake was to pass an encoding object created by 
CryptoJS.enc.Hex.parse() to decrypt method. But the method requires as 
parameter an object with a "ciphertext" property:
CryptoJS.AES.decrypt({ ciphertext: encrypted_message }, key)
This is not well documented.

However I think there is a problem with the ZeroPadding option. If you pass 
something without a "ciphertext" property to the decrypt function and specify 
"padding: CryptoJS.pad.ZeroPadding" the script do endless processing. Use of 
"NoPadding" or default padding just results in an empty decryption object.

Original comment by johannes...@gmail.com on 27 May 2014 at 9:43

[GoogleCodeExporter]

I tried to mention this behavior in the Cipher Input section 
(https://code.google.com/p/crypto-js/#The_Cipher_Input). "For the ciphertext, 
the cipher algorithms accept either strings or instances of 
CryptoJS.lib.CipherParams. A CipherParams object represents a collection of 
parameters such as the IV, a salt, and the raw ciphertext itself."

I'd imagine that many parts, not just the padding, wouldn't work correctly if 
it received the wrong kind of input, but that would have to fall under the 
umbrella of "undefined behavior." Loosely typed languages such as JavaScript 
will let you pass in any random thing, so unfortunately that leaves it entirely 
up to you to make sure you pass in the right thing.

Original comment by Jeff.Mott.OR on 27 May 2014 at 11:42

• Changed state: Invalid