ail-framework/bin/modules/Decoder.py

174 lines
5.8 KiB
Python
Raw Normal View History

2018-07-19 14:52:09 +00:00
#!/usr/bin/env python3
# -*-coding:UTF-8 -*
"""
Decoder module
Dectect Binary and decode it
"""
##################################
# Import External packages
##################################
2018-07-19 14:52:09 +00:00
import time
import os
import redis
import base64
from hashlib import sha1
import magic
import json
import datetime
import re
import signal
import sys
2018-07-19 14:52:09 +00:00
sys.path.append(os.environ['AIL_BIN'])
##################################
# Import Project packages
##################################
from modules.abstract_module import AbstractModule
2018-07-19 14:52:09 +00:00
from Helper import Process
from packages import Item
from lib import ConfigLoader
from lib import Decoded
2018-07-19 14:52:09 +00:00
# # TODO: use regex_helper
2018-07-19 14:52:09 +00:00
class TimeoutException(Exception):
pass
def timeout_handler(signum, frame):
raise TimeoutException
signal.signal(signal.SIGALRM, timeout_handler)
class Decoder(AbstractModule):
"""
Decoder module for AIL framework
"""
2018-07-19 14:52:09 +00:00
# TODO to lambda expr
def hex_decoder(self, hexStr):
#hexStr = ''.join( hex_string.split(" ") )
return bytes(bytearray([int(hexStr[i:i+2], 16) for i in range(0, len(hexStr), 2)]))
2018-07-19 14:52:09 +00:00
# TODO to lambda expr
def binary_decoder(self, binary_string):
return bytes(bytearray([int(binary_string[i:i+8], 2) for i in range(0, len(binary_string), 8)]))
2018-07-19 14:52:09 +00:00
# TODO to lambda expr
def base64_decoder(self, base64_string):
return base64.b64decode(base64_string)
2018-07-19 14:52:09 +00:00
def __init__(self):
super(Decoder, self).__init__()
2018-07-19 14:52:09 +00:00
serv_metadata = ConfigLoader.ConfigLoader().get_redis_conn("ARDB_Metadata")
2018-07-19 14:52:09 +00:00
regex_binary = '[0-1]{40,}'
#regex_hex = '(0[xX])?[A-Fa-f0-9]{40,}'
regex_hex = '[A-Fa-f0-9]{40,}'
regex_base64 = '(?:[A-Za-z0-9+/]{4}){2,}(?:[A-Za-z0-9+/]{2}[AEIMQUYcgkosw048]=|[A-Za-z0-9+/][AQgw]==)'
2018-07-19 14:52:09 +00:00
cmp_regex_binary = re.compile(regex_binary)
cmp_regex_hex = re.compile(regex_hex)
cmp_regex_base64 = re.compile(regex_base64)
2018-07-19 14:52:09 +00:00
# map decoder function
self.decoder_function = {'binary':self.binary_decoder,'hexadecimal':self.hex_decoder, 'base64':self.base64_decoder}
2018-07-19 14:52:09 +00:00
hex_max_execution_time = self.process.config.getint("Hex", "max_execution_time")
binary_max_execution_time = self.process.config.getint("Binary", "max_execution_time")
base64_max_execution_time = self.process.config.getint("Base64", "max_execution_time")
2018-07-19 14:52:09 +00:00
# list all decoder with regex,
decoder_binary = {'name': 'binary', 'regex': cmp_regex_binary, 'encoded_min_size': 300, 'max_execution_time': binary_max_execution_time}
decoder_hexadecimal = {'name': 'hexadecimal', 'regex': cmp_regex_hex, 'encoded_min_size': 300, 'max_execution_time': hex_max_execution_time}
decoder_base64 = {'name': 'base64', 'regex': cmp_regex_base64, 'encoded_min_size': 40, 'max_execution_time': base64_max_execution_time}
2018-07-19 14:52:09 +00:00
self.decoder_order = [ decoder_base64, decoder_binary, decoder_hexadecimal, decoder_base64]
2018-07-19 14:52:09 +00:00
for decoder in self.decoder_order:
serv_metadata.sadd('all_decoder', decoder['name'])
2018-07-25 08:47:36 +00:00
# Waiting time in secondes between to message proccessed
self.pending_seconds = 1
2018-07-19 14:52:09 +00:00
# Send module state to logs
self.redis_logger.info(f'Module {self.module_name} initialized')
2018-07-19 14:52:09 +00:00
def compute(self, message):
2018-07-19 14:52:09 +00:00
obj_id = Item.get_item_id(message)
2018-07-19 14:52:09 +00:00
# Extract info from message
content = Item.get_item_content(obj_id)
date = Item.get_item_date(obj_id)
2018-07-19 14:52:09 +00:00
for decoder in self.decoder_order: # add threshold and size limit
2018-07-19 14:52:09 +00:00
# max execution time on regex
signal.alarm(decoder['max_execution_time'])
2018-07-19 14:52:09 +00:00
try:
encoded_list = decoder['regex'].findall(content)
2018-07-19 14:52:09 +00:00
except TimeoutException:
encoded_list = []
self.process.incr_module_timeout_statistic() # add encoder type
self.redis_logger.debug(f"{obj_id} processing timeout")
2018-07-19 14:52:09 +00:00
continue
else:
signal.alarm(0)
if(len(encoded_list) > 0):
content = self.decode_string(content, message, date, encoded_list, decoder['name'], decoder['encoded_min_size'])
def decode_string(self, content, item_id, item_date, encoded_list, decoder_name, encoded_min_size):
find = False
for encoded in encoded_list:
if len(encoded) >= encoded_min_size:
decoded_file = self.decoder_function[decoder_name](encoded)
find = True
sha1_string = sha1(decoded_file).hexdigest()
mimetype = Decoded.get_file_mimetype(decoded_file)
if not mimetype:
self.redis_logger.debug(item_id)
self.redis_logger.debug(sha1_string)
print(item_id)
print(sha1_string)
raise Exception('Invalid mimetype')
Decoded.save_decoded_file_content(sha1_string, decoded_file, item_date, mimetype=mimetype)
Decoded.save_item_relationship(sha1_string, item_id)
Decoded.create_decoder_matadata(sha1_string, item_id, decoder_name)
#remove encoded from item content
content = content.replace(encoded, '', 1)
self.redis_logger.debug(f'{item_id} : {decoder_name} - {mimetype}')
print(f'{item_id} : {decoder_name} - {mimetype}')
if(find):
self.set_out_item(decoder_name, item_id)
return content
def set_out_item(self, decoder_name, item_id):
self.redis_logger.warning(f'{decoder_name} decoded')
# Send to duplicate
self.send_message_to_queue(item_id, 'Duplicate')
# Send to Tags
msg = f'infoleak:automatic-detection="{decoder_name}";{item_id}'
self.send_message_to_queue(msg, 'Tags')
if __name__ == '__main__':
module = Decoder()
module.run()