Ucryptolib mode CBC

[PatLub85]

Hi,
I am trying to encrypt and decrypt a text file from the example https://hwwong168.wordpress.com/2019/09/25/esp32-micropython-implementation-of-cryptographic/ and I have an error.
My program:

import uos
from ucryptolib import aes

MODE_CBC = 2
BLOCK_SIZE = 16
f = open('code.txt', 'r')
content = f.read()
f.close()
 
# key size must be 16 or 32
# key = uos.urandom(32)
key = b'I_am_32bytes=256bits_key_padding'
# Padding plain text with space 
pad = (BLOCK_SIZE - len(content)) % BLOCK_SIZE
print ('pad=',pad,' ', 'lenght',len(content))
content = content + "-"*pad


###################################################
#             AES CBC Cryptographic               #
###################################################

# Generate iv with HW random generator 
iv = uos.urandom(BLOCK_SIZE)
cipher = aes(key,MODE_CBC,iv)
 
ct_bytes = iv + cipher.encrypt(content)
print ('AES-CBC encrypted:', ct_bytes)
 
iv = ct_bytes[:BLOCK_SIZE]
cipher = aes(key,MODE_CBC,iv)
decrypted = cipher.decrypt(ct_bytes)[BLOCK_SIZE:]
print('AES-CBC decrypted:', decrypted)

The error

pad= 13   lenght 2643
Traceback (most recent call last):
  File "<stdin>", line 27, in <module>
ValueError: blksize % 16

I had fun completing the text file with a number of complete 16-byte blocks and I got the same error

pad= 0   lenght 2656
Traceback (most recent call last):
  File "<stdin>", line 27, in <module>
ValueError: blksize % 16

Could someone help me to understand what's wrong?

[PatLub85]

Hi,
I found the origin of the problem. In my code.txt file, some data had accents (french).
I removed them and it works.
Is there an encoding that would allow to manage accents?

[rkompass]

It might be that you need f = open('code.txt', 'r', encoding='utf-8') instead of just f = open('code.txt', 'r').
So that len(content) gives the lenght in bytes, which you need for correct encryption.

Strings with characters outside ASCII are usually given as unicode, which is supported in MP.
The string length given by len() is not the length of bytes of those strings.
A demonstration:

s = "Página Não Encontrada!"
print(f's: {s}, len(s): {len(s)}')
s_enc = s.encode("utf-8")
print(f's_enc: {s_enc}, len(s_enc): {len(s_enc)}')
s_enc_dec = s_enc.decode("utf-8")
print(f's_enc_dec: {s_enc_dec}, len(s_enc_dec): {len(s_enc_dec)}')

which prints:

s: Página Não Encontrada!, len(s): 22
s_enc: b'P\xc3\xa1gina N\xc3\xa3o Encontrada!', len(s_enc): 24
s_enc_dec: Página Não Encontrada!, len(s_enc_dec): 22

and you see that the accented characters are represent by two bytes.
In general Unicode utf-8 coding may represent characters by 1, 2 or 4 bytes.

I have no experience with cryptography in Python, but I see from the code you presented that the content string is padded with '-' characters. So it seems plausible that the padding won't be correct, as it is done based on len(content) which is not reporting the actual length in bytes (with accented characters).
Perhaps a simple replacement by len(content.encode("utf-8")) would also work?

[mendenm]

since the attempt is to read from a file, wouldn't just opening it in mode 'rb' work to get a bytes item to encrypt? It seems going through the encoding process is redundant, isn't it?

[PatLub85]

Hi,
Kompass has found the origin of the problem, and Mendenm has the solution. Indeed, you just have to open the text file in 'rb'
Thank you both