On Sun, 2007-05-06 at 11:39 -0400, David Clymer wrote: > I'm using pycrypto's AES module, and am attempting to automate the > padding of the plaintext (the size of the text to be encrypted must be a > multiple of 16). However, my padding scheme sometimes fails to calculate > the necessary padding correctly, and I'm not sure why. > > Using the text 'foo bar', it fails, but if I do 'foo bars' it works. > Randomized testing fails sometimes and not others. > > Any pointers as to what I might be doing wrong (code attached)?
Nevermind. After I walked away from it for a bit, the mistake was
obvious. I was using the modulus of the text length by the block size
directly, rather than the block size minus that number:
--- /home/david/Desktop/test.py 2007-05-06 13:38:52.000000000 -0400
+++ test.py 2007-05-06 13:39:38.000000000 -0400
@@ -9,12 +9,12 @@
def __init__(self, text, key_phrase):
key = SHA256.new(key_phrase)
self.aes = AES.new(key.digest())
- #self.encrypted_text = self.aes.encrypt(self._pad(text))
+ self.encrypted_text = self.aes.encrypt(self._pad(text))
def _unpad(self, txt):
"""Remove padding from the given text"""
for x in xrange(len(txt) - self.aes.block_size,len(txt)):
- if x == ord(txt[x]):
+ if len(txt[x:]) == ord(txt[x]):
if txt[x-1:] + self._gen_pad(txt[x-1:]):
return txt[:x]
return txt
@@ -25,7 +25,7 @@
def _gen_pad(self, txt):
"""Generate padding for the given plaintext"""
- pad_len = (len(txt)) % self.aes.block_size
+ pad_len = self.aes.block_size - (len(txt) %
self.aes.block_size)
if pad_len > 0:
return chr(pad_len) * pad_len
return chr(self.aes.block_size) * self.aes.block_size
--
gpg-key: http://www.zettazebra.com/files/key.gpg
signature.asc
Description: This is a digitally signed message part
-- http://mail.python.org/mailman/listinfo/python-list
