Cryptanalytic attacks on E-Safenet encryption
Switch branches/tags
Nothing to show
Clone or download
Latest commit a404106 Jul 14, 2016

README.md

E-Safenet

This GitHub repository contains files that assist in cryptanalytic attacks on E-Safenet encryption. Several attacks were developed that may partially or fully recover E-Safenet encryption keys.

  • Known-plaintext attack
  • Probable-plaintext attack
    • Against source code files
    • Against binary files
  • Ciphertext-only attack

All specifics and attacks are document in the research paper on E-Safenet encryption.

Python scripts

The python scripts provided can be used to encrypt and decrypt using the E-Safenet encryption, or to extract encryption keys.

Two main files are available:

  • esafenet.py: command-line interface to known-plaintext and probable-plaintext attacks
  • esafenet_gui.py: GUI interface for the ciphertext-only attack
Setup

Prior to using these scripts, the simplelzo1x module has to be compiled first. This module provides an interface to the LZO v1.00 compression library.

cd simplelzo1x && sudo python setup.py install

More information about the library can be found in the README file in the simplelzo1x directory.

esafenet.py

usage: esafenet.py [-h] [--infile INFILE] [--key KEY] [--outfile OUTFILE]
                   [--infolder INFOLDER] [--outfolder OUTFOLDER]
                   [--comp_file COMP_FILE] [--type pattern_type]
                   [--language text_pattern_language]
                   action

E-safenet encryption/decryption/key generation

positional arguments:
  action                Action to perform
                        Should be one of ['encrypt', 'decrypt', 'encrypt_folder', 
                        'decrypt_folder', 'keygen', 'findkey', 'pattern_decrypt']

optional arguments:
  -h, --help            show this help message and exit
  --infile INFILE       Input file
  --key KEY             Key file
  --outfile OUTFILE     Output file
  --infolder INFOLDER   Input folder
  --outfolder OUTFOLDER
                        Output folder
  --comp_file COMP_FILE
                        Plaintext comparison file used by findkey
  --type pattern_type   Type for pattern decrypt (binary or text)
  --language text_pattern_language
                        Language for text pattern decrypt (C, PHP or CS)
Examples
  • Recovering the encryption key of a binary file (probable-plaintext attack):
$ python esafenet.py pattern_decrypt --type binary --infile encrypted.xls --outfile key.dat
Decryption: key written to key.dat (4 0-bytes)
  • Decrypting an E-Safenet file using a provided key:
$ python esafenet.py decrypt --infile encrypted.xls --key key.dat --outfile decrypted.xls
Decryption: 153400 bytes written to decrypted.xls
  • Recovering the key using the known-plaintext attack:
$ python esafenet.py findkey --infile encrypted.xls --comp_file decrypted.xls --outfile key.dat
Succes: key written to key.dat
  • Recovering the encryption key of source code files (probable-plaintext attack, C#):
$ python esafenet.py pattern_decrypt --type text --infolder srcfiles --outfolder /tmp --language CS --outfile key.dat
Match found!! ...
Troubleshooting

If you get errors/crashes, they are probably caused by the LZO compression library. The first 512 bytes of an E-Safenet encrypted file are compressed. When using a wrong key, decompression may fail and lead to a crash. You can temporarily disable decompression of the first block by changing the plain_header variable in esafenet.py to an empty string:

             plain_header = ""
#            plain_header = simplelzo1x.decompress(decr_header)

esafenet_gui.py

The GUI app esafenet_gui.py can be used for the ciphertext-only attack. More information about this attack can be found in the research paper.

  1. menu -> Open folder or file, select an E-Safenet file, or a folder containing only E-Safenet files encrypted with the same key.
  2. menu -> Analyze, analyzes the files, tries to maximize plaintext in the file(s), as described in the report.

Note: The analyze step may take some time (15s for 200kB on my 5y/o laptop, displaying results in thhe grid takes even longer...)

Results are displayed as-is, this program is not complete. Feel free to do with it as you see fit.

COA tool

CPLEX model

For the mathematical implementation of the ciphertext-only attack, cplex_coa.mod provides a CPLEX model for the Binary Integer Programming problem that represents the maximization of printable characters in an E-Safenet encrypted document.

Credits

The code was released under the GPLv2 license.