mirror of
https://github.com/ail-project/ail-framework.git
synced 2024-09-20 07:58:29 +00:00
1379ef705a
AIL is a modular framework to analyse potential information leak from unstructured data source like pastes from Past ebin or similar services. AIL framework is flexible and can be extended to support other functionalities to mine sen sitive information
89 lines
2.6 KiB
Python
Executable file
89 lines
2.6 KiB
Python
Executable file
#!/usr/bin/env python2
|
|
# -*-coding:UTF-8 -*
|
|
"""
|
|
The ZMQ_PubSub_Lines Module
|
|
============================
|
|
|
|
This module is consuming the Redis-list created by the ZMQ_PubSub_Tokenize_Q Module.
|
|
|
|
It tokenize the content of the paste and publish the result in the following format:
|
|
channel_name+' '+/path/of/the/paste.gz+' '+tokenized_word+' '+scoring
|
|
|
|
..seealso:: Paste method (_get_top_words)
|
|
|
|
..note:: Module ZMQ_Something_Q and ZMQ_Something are closely bound, always put
|
|
the same Subscriber name in both of them.
|
|
|
|
Requirements
|
|
------------
|
|
|
|
*Need running Redis instances. (Redis)
|
|
*Need the ZMQ_PubSub_Tokenize_Q Module running to be able to work properly.
|
|
|
|
"""
|
|
import redis, zmq, ConfigParser, time
|
|
from packages import Paste as P
|
|
from packages import ZMQ_PubSub
|
|
from pubsublogger import publisher
|
|
|
|
configfile = './packages/config.cfg'
|
|
|
|
def main():
|
|
"""Main Function"""
|
|
|
|
# CONFIG #
|
|
cfg = ConfigParser.ConfigParser()
|
|
cfg.read(configfile)
|
|
|
|
# REDIS #
|
|
r_serv = redis.StrictRedis(
|
|
host = cfg.get("Redis_Queues", "host"),
|
|
port = cfg.getint("Redis_Queues", "port"),
|
|
db = cfg.getint("Redis_Queues", "db"))
|
|
|
|
# LOGGING #
|
|
publisher.channel = "Script"
|
|
|
|
# ZMQ #
|
|
channel = cfg.get("PubSub_Longlines", "channel_1")
|
|
subscriber_name = "tokenize"
|
|
subscriber_config_section = "PubSub_Longlines"
|
|
|
|
#Publisher
|
|
publisher_config_section = "PubSub_Words"
|
|
publisher_name = "pubtokenize"
|
|
|
|
Sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name)
|
|
Pub = ZMQ_PubSub.ZMQPub(configfile, publisher_config_section, publisher_name)
|
|
|
|
channel_0 = cfg.get("PubSub_Words", "channel_0")
|
|
|
|
# FUNCTIONS #
|
|
publisher.info("Tokeniser subscribed to channel {0}".format(cfg.get("PubSub_Longlines", "channel_1")))
|
|
|
|
while True:
|
|
message = Sub.get_msg_from_queue(r_serv)
|
|
print message
|
|
if message != None:
|
|
PST = P.Paste(message.split(" ",-1)[-1])
|
|
else:
|
|
if r_serv.sismember("SHUTDOWN_FLAGS", "Tokenize"):
|
|
r_serv.srem("SHUTDOWN_FLAGS", "Tokenize")
|
|
print "Shutdown Flag Up: Terminating"
|
|
publisher.warning("Shutdown Flag Up: Terminating.")
|
|
break
|
|
publisher.debug("Tokeniser is idling 10s")
|
|
time.sleep(10)
|
|
print "sleepin"
|
|
continue
|
|
|
|
for word, score in PST._get_top_words().items():
|
|
if len(word) >= 4:
|
|
msg = channel_0+' '+PST.p_path+' '+str(word)+' '+str(score)
|
|
Pub.send_message(msg)
|
|
print msg
|
|
else:
|
|
pass
|
|
|
|
if __name__ == "__main__":
|
|
main()
|