AIL/ail-framework
7
0
Fork 0
mirror of https://github.com/ail-project/ail-framework.git synced 2024-09-20 07:58:29 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

Big cleanup, pep8

Browse source
This commit is contained in:
Raphaël Vinot 2014-08-14 17:55:18 +02:00
parent 715adb4546
commit 078c8ea836
25 changed files with 394 additions and 408 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

38
bin/Repartition_graph.py
View file

@ -1,14 +1,13 @@
#!/usr/bin/python2.7 #!/usr/bin/python2.7
# -*-coding:UTF-8 -* # -*-coding:UTF-8 -*
import redis, argparse, zmq, ConfigParser, time, os import redis
import argparse
import ConfigParser
from pubsublogger import publisher from pubsublogger import publisher
from packages import ZMQ_PubSub
import texttable
import numpy as np
import matplotlib.pyplot as plt import matplotlib.pyplot as plt
from pylab import *
def main(): def main():
"""Main Function""" """Main Function"""
@ -19,24 +18,20 @@ def main():
# SCRIPT PARSER # # SCRIPT PARSER #
parser = argparse.ArgumentParser( parser = argparse.ArgumentParser(
description = '''This script is a part of the Analysis Information description='''This script is a part of the Analysis Information Leak framework.''',
Leak framework.''', epilog='''''')
epilog = '''''')
parser.add_argument('-f', parser.add_argument('-f', type=str, metavar="filename", default="figure",
type = str, help='The absolute path name of the "figure.png"',
metavar = "filename", action='store')
default = "figure",
help = 'The absolute path name of the "figure.png"',
action = 'store')
args = parser.parse_args() args = parser.parse_args()
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Level_DB_Hashs", "host"), host=cfg.get("Redis_Level_DB_Hashs", "host"),
port = cfg.getint("Redis_Level_DB_Hashs", "port"), port=cfg.getint("Redis_Level_DB_Hashs", "port"),
db = cfg.getint("Redis_Level_DB_Hashs", "db")) db=cfg.getint("Redis_Level_DB_Hashs", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Graph" publisher.channel = "Graph"
@ -68,12 +63,11 @@ def main():
pastie_list.append(pastie) pastie_list.append(pastie)
pastebin_list.append(pastebin) pastebin_list.append(pastebin)
codepad_list.sort(reverse = True) codepad_list.sort(reverse=True)
pastie_list.sort(reverse = True) pastie_list.sort(reverse=True)
pastebin_list.sort(reverse = True) pastebin_list.sort(reverse=True)
total_list.sort(reverse = True)
total_list.sort(reverse=True)
plt.plot(codepad_list, 'b', label='Codepad.org') plt.plot(codepad_list, 'b', label='Codepad.org')
plt.plot(pastebin_list, 'g', label='Pastebin.org') plt.plot(pastebin_list, 'g', label='Pastebin.org')

14
bin/ZMQ_PubSub_Categ_Q.py
View file

@ -17,12 +17,14 @@ Requirements
*Should register to the Publisher "ZMQ_PubSub_Tokenize" *Should register to the Publisher "ZMQ_PubSub_Tokenize"
""" """
import redis, zmq, ConfigParser import redis
import ConfigParser
from pubsublogger import publisher from pubsublogger import publisher
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -32,9 +34,9 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Queuing" publisher.channel = "Queuing"
@ -44,12 +46,12 @@ def main():
subscriber_name = "categ" subscriber_name = "categ"
subscriber_config_section = "PubSub_Words" subscriber_config_section = "PubSub_Words"
Sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name) sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name)
# FUNCTIONS # # FUNCTIONS #
publisher.info("""Suscribed to channel {0}""".format(channel)) publisher.info("""Suscribed to channel {0}""".format(channel))
while True: while True:
Sub.get_and_lpush(r_serv) sub.get_and_lpush(r_serv)
if r_serv.sismember("SHUTDOWN_FLAGS", "Categ_Q"): if r_serv.sismember("SHUTDOWN_FLAGS", "Categ_Q"):
r_serv.srem("SHUTDOWN_FLAGS", "Categ_Q") r_serv.srem("SHUTDOWN_FLAGS", "Categ_Q")

58
bin/ZMQ_PubSub_Lines.py
View file

@ -26,13 +26,17 @@ Requirements
*Need the ZMQ_PubSub_Line_Q Module running to be able to work properly. *Need the ZMQ_PubSub_Line_Q Module running to be able to work properly.
""" """
import redis, argparse, zmq, ConfigParser, time import redis
from packages import Paste as P import argparse
import ConfigParser
import time
from packages import Paste
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from pubsublogger import publisher from pubsublogger import publisher
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -42,62 +46,58 @@ def main():
# SCRIPT PARSER # # SCRIPT PARSER #
parser = argparse.ArgumentParser( parser = argparse.ArgumentParser(
description = '''This script is a part of the Analysis Information description='''This script is a part of the Analysis Information Leak framework.''',
Leak framework.''', epilog='''''')
epilog = '''''')
parser.add_argument('-max', parser.add_argument('-max', type=int, default=500,
type = int, help='The limit between "short lines" and "long lines" (500)',
default = 500, action='store')
help = 'The limit between "short lines" and "long lines" (500)',
action = 'store')
args = parser.parse_args() args = parser.parse_args()
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Data_Merging", "host"), host=cfg.get("Redis_Data_Merging", "host"),
port = cfg.getint("Redis_Data_Merging", "port"), port=cfg.getint("Redis_Data_Merging", "port"),
db = cfg.getint("Redis_Data_Merging", "db")) db=cfg.getint("Redis_Data_Merging", "db"))
r_serv1 = redis.StrictRedis( r_serv1 = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
p_serv = r_serv.pipeline(False)
# LOGGING # # LOGGING #
publisher.channel = "Script" publisher.channel = "Script"
# ZMQ # # ZMQ #
#Subscriber # Subscriber
channel = cfg.get("PubSub_Global", "channel") channel = cfg.get("PubSub_Global", "channel")
subscriber_name = "line" subscriber_name = "line"
subscriber_config_section = "PubSub_Global" subscriber_config_section = "PubSub_Global"
#Publisher # Publisher
publisher_config_section = "PubSub_Longlines" publisher_config_section = "PubSub_Longlines"
publisher_name = "publine" publisher_name = "publine"
Sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name) sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name)
Pub = ZMQ_PubSub.ZMQPub(configfile, publisher_config_section, publisher_name) pub = ZMQ_PubSub.ZMQPub(configfile, publisher_config_section, publisher_name)
channel_0 = cfg.get("PubSub_Longlines", "channel_0") channel_0 = cfg.get("PubSub_Longlines", "channel_0")
channel_1 = cfg.get("PubSub_Longlines", "channel_1") channel_1 = cfg.get("PubSub_Longlines", "channel_1")
# FUNCTIONS # # FUNCTIONS #
publisher.info("""Lines script Subscribed to channel {0} and Start to publish tmp_string = "Lines script Subscribed to channel {} and Start to publish on channel {}, {}"
on channel {1}, {2}""".format(cfg.get("PubSub_Global", "channel"), publisher.info(tmp_string.format(
cfg.get("PubSub_Global", "channel"),
cfg.get("PubSub_Longlines", "channel_0"), cfg.get("PubSub_Longlines", "channel_0"),
cfg.get("PubSub_Longlines", "channel_1"))) cfg.get("PubSub_Longlines", "channel_1")))
while True: while True:
try: try:
message = Sub.get_msg_from_queue(r_serv1) message = sub.get_msg_from_queue(r_serv1)
if message != None: if message is not None:
PST = P.Paste(message.split(" ",-1)[-1]) PST = Paste.Paste(message.split(" ", -1)[-1])
else: else:
if r_serv1.sismember("SHUTDOWN_FLAGS", "Lines"): if r_serv1.sismember("SHUTDOWN_FLAGS", "Lines"):
r_serv1.srem("SHUTDOWN_FLAGS", "Lines") r_serv1.srem("SHUTDOWN_FLAGS", "Lines")
@ -113,13 +113,13 @@ def main():
PST.save_attribute_redis(r_serv, "p_nb_lines", lines_infos[0]) PST.save_attribute_redis(r_serv, "p_nb_lines", lines_infos[0])
PST.save_attribute_redis(r_serv, "p_max_length_line", lines_infos[1]) PST.save_attribute_redis(r_serv, "p_max_length_line", lines_infos[1])
r_serv.sadd("Pastes_Objects",PST.p_path) r_serv.sadd("Pastes_Objects", PST.p_path)
if lines_infos[1] >= args.max: if lines_infos[1] >= args.max:
msg = channel_0+" "+PST.p_path msg = channel_0+" "+PST.p_path
else: else:
msg = channel_1+" "+PST.p_path msg = channel_1+" "+PST.p_path
Pub.send_message(msg) pub.send_message(msg)
except IOError: except IOError:
print "CRC Checksum Error on : ", PST.p_path print "CRC Checksum Error on : ", PST.p_path
pass pass

14
bin/ZMQ_PubSub_Lines_Q.py
View file

@ -18,12 +18,14 @@ Requirements
""" """
import redis, zmq, ConfigParser import redis
import ConfigParser
from pubsublogger import publisher from pubsublogger import publisher
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -33,9 +35,9 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Queuing" publisher.channel = "Queuing"
@ -44,13 +46,13 @@ def main():
channel = cfg.get("PubSub_Global", "channel") channel = cfg.get("PubSub_Global", "channel")
subscriber_name = "line" subscriber_name = "line"
Sub = ZMQ_PubSub.ZMQSub(configfile,"PubSub_Global", channel, subscriber_name) sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Global", channel, subscriber_name)
# FUNCTIONS # # FUNCTIONS #
publisher.info("""Suscribed to channel {0}""".format(channel)) publisher.info("""Suscribed to channel {0}""".format(channel))
while True: while True:
Sub.get_and_lpush(r_serv) sub.get_and_lpush(r_serv)
if r_serv.sismember("SHUTDOWN_FLAGS", "Lines_Q"): if r_serv.sismember("SHUTDOWN_FLAGS", "Lines_Q"):
r_serv.srem("SHUTDOWN_FLAGS", "Lines_Q") r_serv.srem("SHUTDOWN_FLAGS", "Lines_Q")

27
bin/ZMQ_PubSub_Tokenize.py
View file

@ -21,13 +21,16 @@ Requirements
*Need the ZMQ_PubSub_Tokenize_Q Module running to be able to work properly. *Need the ZMQ_PubSub_Tokenize_Q Module running to be able to work properly.
""" """
import redis, zmq, ConfigParser, time import redis
from packages import Paste as P import ConfigParser
import time
from packages import Paste
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from pubsublogger import publisher from pubsublogger import publisher
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -37,9 +40,9 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Script" publisher.channel = "Script"
@ -49,12 +52,12 @@ def main():
subscriber_name = "tokenize" subscriber_name = "tokenize"
subscriber_config_section = "PubSub_Longlines" subscriber_config_section = "PubSub_Longlines"
#Publisher # Publisher
publisher_config_section = "PubSub_Words" publisher_config_section = "PubSub_Words"
publisher_name = "pubtokenize" publisher_name = "pubtokenize"
Sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name) sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name)
Pub = ZMQ_PubSub.ZMQPub(configfile, publisher_config_section, publisher_name) pub = ZMQ_PubSub.ZMQPub(configfile, publisher_config_section, publisher_name)
channel_0 = cfg.get("PubSub_Words", "channel_0") channel_0 = cfg.get("PubSub_Words", "channel_0")
@ -62,10 +65,10 @@ def main():
publisher.info("Tokeniser subscribed to channel {0}".format(cfg.get("PubSub_Longlines", "channel_1"))) publisher.info("Tokeniser subscribed to channel {0}".format(cfg.get("PubSub_Longlines", "channel_1")))
while True: while True:
message = Sub.get_msg_from_queue(r_serv) message = sub.get_msg_from_queue(r_serv)
print message print message
if message != None: if message is not None:
PST = P.Paste(message.split(" ",-1)[-1]) PST = Paste.Paste(message.split(" ", -1)[-1])
else: else:
if r_serv.sismember("SHUTDOWN_FLAGS", "Tokenize"): if r_serv.sismember("SHUTDOWN_FLAGS", "Tokenize"):
r_serv.srem("SHUTDOWN_FLAGS", "Tokenize") r_serv.srem("SHUTDOWN_FLAGS", "Tokenize")
@ -80,7 +83,7 @@ def main():
for word, score in PST._get_top_words().items(): for word, score in PST._get_top_words().items():
if len(word) >= 4: if len(word) >= 4:
msg = channel_0+' '+PST.p_path+' '+str(word)+' '+str(score) msg = channel_0+' '+PST.p_path+' '+str(word)+' '+str(score)
Pub.send_message(msg) pub.send_message(msg)
print msg print msg
else: else:
pass pass

14
bin/ZMQ_PubSub_Tokenize_Q.py
View file

@ -17,12 +17,14 @@ Requirements
*Should register to the Publisher "ZMQ_PubSub_Line" channel 1 *Should register to the Publisher "ZMQ_PubSub_Line" channel 1
""" """
import redis, zmq, ConfigParser import redis
import ConfigParser
from pubsublogger import publisher from pubsublogger import publisher
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -32,9 +34,9 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Queuing" publisher.channel = "Queuing"
@ -44,13 +46,13 @@ def main():
subscriber_name = "tokenize" subscriber_name = "tokenize"
subscriber_config_section = "PubSub_Longlines" subscriber_config_section = "PubSub_Longlines"
Sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name) sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name)
# FUNCTIONS # # FUNCTIONS #
publisher.info("""Suscribed to channel {0}""".format(channel)) publisher.info("""Suscribed to channel {0}""".format(channel))
while True: while True:
Sub.get_and_lpush(r_serv) sub.get_and_lpush(r_serv)
if r_serv.sismember("SHUTDOWN_FLAGS", "Tokenize_Q"): if r_serv.sismember("SHUTDOWN_FLAGS", "Tokenize_Q"):
r_serv.srem("SHUTDOWN_FLAGS", "Tokenize_Q") r_serv.srem("SHUTDOWN_FLAGS", "Tokenize_Q")

35
bin/ZMQ_Sub_Attributes.py
View file

@ -26,13 +26,16 @@ Requirements
*Need the ZMQ_PubSub_Line_Q Module running to be able to work properly. *Need the ZMQ_PubSub_Line_Q Module running to be able to work properly.
""" """
import redis, zmq, ConfigParser, time import redis
from packages import Paste as P import ConfigParser
import time
from packages import Paste
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from pubsublogger import publisher from pubsublogger import publisher
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -42,37 +45,35 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Data_Merging", "host"), host=cfg.get("Redis_Data_Merging", "host"),
port = cfg.getint("Redis_Data_Merging", "port"), port=cfg.getint("Redis_Data_Merging", "port"),
db = cfg.getint("Redis_Data_Merging", "db")) db=cfg.getint("Redis_Data_Merging", "db"))
r_serv1 = redis.StrictRedis( r_serv1 = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
p_serv = r_serv.pipeline(False)
# LOGGING # # LOGGING #
publisher.channel = "Script" publisher.channel = "Script"
# ZMQ # # ZMQ #
#Subscriber # Subscriber
channel = cfg.get("PubSub_Global", "channel") channel = cfg.get("PubSub_Global", "channel")
subscriber_name = "attributes" subscriber_name = "attributes"
subscriber_config_section = "PubSub_Global" subscriber_config_section = "PubSub_Global"
Sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name) sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name)
# FUNCTIONS # # FUNCTIONS #
publisher.info("""ZMQ Attribute is Running""") publisher.info("""ZMQ Attribute is Running""")
while True: while True:
try: try:
message = Sub.get_msg_from_queue(r_serv1) message = sub.get_msg_from_queue(r_serv1)
if message != None: if message is not None:
PST = P.Paste(message.split(" ",-1)[-1]) PST = Paste.Paste(message.split(" ", -1)[-1])
else: else:
if r_serv1.sismember("SHUTDOWN_FLAGS", "Attributes"): if r_serv1.sismember("SHUTDOWN_FLAGS", "Attributes"):
r_serv1.srem("SHUTDOWN_FLAGS", "Attributes") r_serv1.srem("SHUTDOWN_FLAGS", "Attributes")
@ -89,12 +90,12 @@ def main():
PST.save_attribute_redis(r_serv, "p_encoding", encoding) PST.save_attribute_redis(r_serv, "p_encoding", encoding)
PST.save_attribute_redis(r_serv, "p_language", language) PST.save_attribute_redis(r_serv, "p_language", language)
r_serv.sadd("Pastes_Objects",PST.p_path) r_serv.sadd("Pastes_Objects", PST.p_path)
PST.save_all_attributes_redis(r_serv) PST.save_all_attributes_redis(r_serv)
except IOError: except IOError:
print "CRC Checksum Failed on :", PST.p_path print "CRC Checksum Failed on :", PST.p_path
publisher.error('{0};{1};{2};{3};{4}'.format("Duplicate", PST.p_source, PST.p_date, PST.p_name, "CRC Checksum Failed" )) publisher.error('{0};{1};{2};{3};{4}'.format("Duplicate", PST.p_source, PST.p_date, PST.p_name, "CRC Checksum Failed"))
pass pass

14
bin/ZMQ_Sub_Attributes_Q.py
View file

@ -18,12 +18,14 @@ Requirements
""" """
import redis, zmq, ConfigParser import redis
import ConfigParser
from pubsublogger import publisher from pubsublogger import publisher
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -33,9 +35,9 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Queuing" publisher.channel = "Queuing"
@ -44,13 +46,13 @@ def main():
channel = cfg.get("PubSub_Global", "channel") channel = cfg.get("PubSub_Global", "channel")
subscriber_name = "attributes" subscriber_name = "attributes"
Sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Global", channel, subscriber_name) sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Global", channel, subscriber_name)
# FUNCTIONS # # FUNCTIONS #
publisher.info("""Suscribed to channel {0}""".format(channel)) publisher.info("""Suscribed to channel {0}""".format(channel))
while True: while True:
Sub.get_and_lpush(r_serv) sub.get_and_lpush(r_serv)
if r_serv.sismember("SHUTDOWN_FLAGS", "Attributes_Q"): if r_serv.sismember("SHUTDOWN_FLAGS", "Attributes_Q"):
r_serv.srem("SHUTDOWN_FLAGS", "Attributes_Q") r_serv.srem("SHUTDOWN_FLAGS", "Attributes_Q")

66
bin/ZMQ_Sub_CreditCards.py
View file

@ -1,7 +1,10 @@
#!/usr/bin/env python2 #!/usr/bin/env python2
# -*-coding:UTF-8 -* # -*-coding:UTF-8 -*
import redis, zmq, ConfigParser, json, pprint, time import redis
from packages import Paste as P import ConfigParser
import pprint
import time
from packages import Paste
from packages import lib_refine from packages import lib_refine
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from pubsublogger import publisher from pubsublogger import publisher
@ -9,6 +12,7 @@ from pubsublogger import publisher
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -18,60 +22,58 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
r_serv1 = redis.StrictRedis( r_serv1 = redis.StrictRedis(
host = cfg.get("Redis_Data_Merging", "host"), host=cfg.get("Redis_Data_Merging", "host"),
port = cfg.getint("Redis_Data_Merging", "port"), port=cfg.getint("Redis_Data_Merging", "port"),
db = cfg.getint("Redis_Data_Merging", "db")) db=cfg.getint("Redis_Data_Merging", "db"))
p_serv = r_serv.pipeline(False)
# LOGGING # # LOGGING #
publisher.channel = "Script" publisher.channel = "Script"
# ZMQ # # ZMQ #
Sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Categ", "creditcard_categ", "cards") sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Categ", "creditcard_categ", "cards")
# FUNCTIONS # # FUNCTIONS #
publisher.info("Creditcard script subscribed to channel creditcard_categ") publisher.info("Creditcard script subscribed to channel creditcard_categ")
message = Sub.get_msg_from_queue(r_serv) message = sub.get_msg_from_queue(r_serv)
prec_filename = None prec_filename = None
creditcard_regex = "4[0-9]{12}(?:[0-9]{3})?" creditcard_regex = "4[0-9]{12}(?:[0-9]{3})?"
mastercard_regex = "5[1-5]\d{2}([\ \-]?)\d{4}\1\d{4}\1\d{4}" # mastercard_regex = "5[1-5]\d{2}([\ \-]?)\d{4}\1\d{4}\1\d{4}"
visa_regex = "4\d{3}([\ \-]?)\d{4}\1\d{4}\1\d{4}" # visa_regex = "4\d{3}([\ \-]?)\d{4}\1\d{4}\1\d{4}"
discover_regex = "6(?:011\d\d|5\d{4}|4[4-9]\d{3}|22(?:1(?:2[6-9]|[3-9]\d)|[2-8]\d\d|9(?:[01]\d|2[0-5])))\d{10}" # discover_regex = "6(?:011\d\d|5\d{4}|4[4-9]\d{3}|22(?:1(?:2[6-9]|[3-9]\d)|[2-8]\d\d|9(?:[01]\d|2[0-5])))\d{10}"
jcb_regex = "35(?:2[89]|[3-8]\d)([\ \-]?)\d{4}\1\d{4}\1\d{4}" # jcb_regex = "35(?:2[89]|[3-8]\d)([\ \-]?)\d{4}\1\d{4}\1\d{4}"
amex_regex = "3[47]\d\d([\ \-]?)\d{6}\1\d{5}" # amex_regex = "3[47]\d\d([\ \-]?)\d{6}\1\d{5}"
chinaUP_regex = "62[0-5]\d{13,16}" # chinaUP_regex = "62[0-5]\d{13,16}"
maestro_regex = "(?:5[0678]\d\d|6304|6390|67\d\d)\d{8,15}" # maestro_regex = "(?:5[0678]\d\d|6304|6390|67\d\d)\d{8,15}"
while True: while True:
if message != None: if message is not None:
channel, filename, word, score = message.split() channel, filename, word, score = message.split()
if prec_filename == None or filename != prec_filename: if prec_filename is None or filename != prec_filename:
Creditcard_set = set([]) creditcard_set = set([])
PST = P.Paste(filename) PST = Paste.Paste(filename)
for x in PST.get_regex(creditcard_regex): for x in PST.get_regex(creditcard_regex):
if lib_refine.is_luhn_valid(x): if lib_refine.is_luhn_valid(x):
Creditcard_set.add(x) creditcard_set.add(x)
PST.__setattr__(channel, creditcard_set)
PST.save_attribute_redis(r_serv1, channel, creditcard_set)
PST.__setattr__(channel, Creditcard_set) pprint.pprint(creditcard_set)
PST.save_attribute_redis(r_serv1, channel, Creditcard_set) to_print = 'CreditCard;{};{};{};'.format(PST.p_source, PST.p_date, PST.p_name)
if (len(creditcard_set) > 0):
pprint.pprint(Creditcard_set) publisher.critical('{}Checked {} valid number(s)'.format(to_print, len(creditcard_set)))
if (len(Creditcard_set) > 0):
publisher.critical('{0};{1};{2};{3};{4}'.format("CreditCard", PST.p_source, PST.p_date, PST.p_name,"Checked " + str(len(Creditcard_set))+" valid number(s)" ))
else: else:
publisher.info('{0};{1};{2};{3};{4}'.format("CreditCard", PST.p_source, PST.p_date, PST.p_name, "CreditCard related" )) publisher.info('{}CreditCard related'.format(to_print))
prec_filename = filename prec_filename = filename
@ -84,7 +86,7 @@ def main():
publisher.debug("Script creditcard is idling 1m") publisher.debug("Script creditcard is idling 1m")
time.sleep(60) time.sleep(60)
message = Sub.get_msg_from_queue(r_serv) message = sub.get_msg_from_queue(r_serv)
if __name__ == "__main__": if __name__ == "__main__":

14
bin/ZMQ_Sub_CreditCards_Q.py
View file

@ -1,12 +1,14 @@
#!/usr/bin/env python2 #!/usr/bin/env python2
# -*-coding:UTF-8 -* # -*-coding:UTF-8 -*
import redis, zmq, ConfigParser import redis
import ConfigParser
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from pubsublogger import publisher from pubsublogger import publisher
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -16,21 +18,21 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Queuing" publisher.channel = "Queuing"
# ZMQ # # ZMQ #
Sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Categ", "creditcard_categ", "cards") sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Categ", "creditcard_categ", "cards")
# FUNCTIONS # # FUNCTIONS #
publisher.info("""Suscribed to channel {0}""".format("creditcard_categ")) publisher.info("""Suscribed to channel {0}""".format("creditcard_categ"))
while True: while True:
Sub.get_and_lpush(r_serv) sub.get_and_lpush(r_serv)
if r_serv.sismember("SHUTDOWN_FLAGS", "Creditcards_Q"): if r_serv.sismember("SHUTDOWN_FLAGS", "Creditcards_Q"):
r_serv.srem("SHUTDOWN_FLAGS", "Creditcards_Q") r_serv.srem("SHUTDOWN_FLAGS", "Creditcards_Q")

46
bin/ZMQ_Sub_Curve.py
View file

@ -21,7 +21,9 @@ Requirements
*Need the ZMQ_PubSub_Tokenize_Q Module running to be able to work properly. *Need the ZMQ_PubSub_Tokenize_Q Module running to be able to work properly.
""" """
import redis, argparse, zmq, ConfigParser, time import redis
import ConfigParser
import time
from packages import Paste as P from packages import Paste as P
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from pubsublogger import publisher from pubsublogger import publisher
@ -29,6 +31,7 @@ from packages import lib_words
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -36,30 +39,16 @@ def main():
cfg = ConfigParser.ConfigParser() cfg = ConfigParser.ConfigParser()
cfg.read(configfile) cfg.read(configfile)
# SCRIPT PARSER #
parser = argparse.ArgumentParser(
description = '''This script is a part of the Analysis Information
Leak framework.''',
epilog = '''''')
parser.add_argument('-l',
type = str,
default = "../files/list_categ_files",
help = 'Path to the list_categ_files (../files/list_categ_files)',
action = 'store')
args = parser.parse_args()
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
r_serv1 = redis.StrictRedis( r_serv1 = redis.StrictRedis(
host = cfg.get("Redis_Level_DB", "host"), host=cfg.get("Redis_Level_DB", "host"),
port = cfg.get("Redis_Level_DB", "port"), port=cfg.get("Redis_Level_DB", "port"),
db = 0) db=0)
# LOGGING # # LOGGING #
publisher.channel = "Script" publisher.channel = "Script"
@ -69,7 +58,7 @@ def main():
subscriber_name = "curve" subscriber_name = "curve"
subscriber_config_section = "PubSub_Words" subscriber_config_section = "PubSub_Words"
Sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name) sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name)
# FUNCTIONS # # FUNCTIONS #
publisher.info("Script Curve subscribed to channel {0}".format(cfg.get("PubSub_Words", "channel_0"))) publisher.info("Script Curve subscribed to channel {0}".format(cfg.get("PubSub_Words", "channel_0")))
@ -78,24 +67,23 @@ def main():
csv_path = cfg.get("Directories", "wordtrending_csv") csv_path = cfg.get("Directories", "wordtrending_csv")
wordfile_path = cfg.get("Directories", "wordsfile") wordfile_path = cfg.get("Directories", "wordsfile")
paste_words = [] message = sub.get_msg_from_queue(r_serv)
message = Sub.get_msg_from_queue(r_serv)
prec_filename = None prec_filename = None
while True: while True:
if message != None: if message is not None:
channel, filename, word, score = message.split() channel, filename, word, score = message.split()
if prec_filename == None or filename != prec_filename: if prec_filename is None or filename != prec_filename:
PST = P.Paste(filename) PST = P.Paste(filename)
lib_words.create_curve_with_word_file(r_serv1, csv_path, wordfile_path, int(PST.p_date.year), int(PST.p_date.month)) lib_words.create_curve_with_word_file(r_serv1, csv_path, wordfile_path, int(PST.p_date.year), int(PST.p_date.month))
prec_filename = filename prec_filename = filename
prev_score = r_serv1.hget(word.lower(), PST.p_date) prev_score = r_serv1.hget(word.lower(), PST.p_date)
print prev_score print prev_score
if prev_score != None: if prev_score is not None:
r_serv1.hset(word.lower(), PST.p_date, int(prev_score) + int(score)) r_serv1.hset(word.lower(), PST.p_date, int(prev_score) + int(score))
else: else:
r_serv1.hset(word.lower(), PST.p_date, score) r_serv1.hset(word.lower(), PST.p_date, score)
#r_serv.expire(word,86400) #1day # r_serv.expire(word,86400) #1day
else: else:
if r_serv.sismember("SHUTDOWN_FLAGS", "Curve"): if r_serv.sismember("SHUTDOWN_FLAGS", "Curve"):
@ -107,7 +95,7 @@ def main():
print "sleepin" print "sleepin"
time.sleep(1) time.sleep(1)
message = Sub.get_msg_from_queue(r_serv) message = sub.get_msg_from_queue(r_serv)
if __name__ == "__main__": if __name__ == "__main__":

14
bin/ZMQ_Sub_Curve_Q.py
View file

@ -17,12 +17,14 @@ Requirements
*Should register to the Publisher "ZMQ_PubSub_Tokenize" *Should register to the Publisher "ZMQ_PubSub_Tokenize"
""" """
import redis, zmq, ConfigParser import redis
import ConfigParser
from pubsublogger import publisher from pubsublogger import publisher
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -32,9 +34,9 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Queuing" publisher.channel = "Queuing"
@ -44,12 +46,12 @@ def main():
subscriber_name = "curve" subscriber_name = "curve"
subscriber_config_section = "PubSub_Words" subscriber_config_section = "PubSub_Words"
Sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name) sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name)
# FUNCTIONS # # FUNCTIONS #
publisher.info("""Suscribed to channel {0}""".format(channel)) publisher.info("""Suscribed to channel {0}""".format(channel))
while True: while True:
Sub.get_and_lpush(r_serv) sub.get_and_lpush(r_serv)
if r_serv.sismember("SHUTDOWN_FLAGS", "Curve_Q"): if r_serv.sismember("SHUTDOWN_FLAGS", "Curve_Q"):
r_serv.srem("SHUTDOWN_FLAGS", "Curve_Q") r_serv.srem("SHUTDOWN_FLAGS", "Curve_Q")

76
bin/ZMQ_Sub_Duplicate.py
View file

@ -12,15 +12,18 @@ Requirements:
""" """
import redis, zmq, ConfigParser, time, datetime, pprint, time, os import redis
from packages import Paste as P import ConfigParser
import os
import time
from packages import Paste
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from datetime import date
from pubsublogger import publisher from pubsublogger import publisher
from pybloomfilter import BloomFilter from pybloomfilter import BloomFilter
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -31,25 +34,24 @@ def main():
# REDIS # # REDIS #
# DB QUEUE ( MEMORY ) # DB QUEUE ( MEMORY )
r_Q_serv = redis.StrictRedis( r_Q_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
r_serv_merge = redis.StrictRedis( r_serv_merge = redis.StrictRedis(
host = cfg.get("Redis_Data_Merging", "host"), host=cfg.get("Redis_Data_Merging", "host"),
port = cfg.getint("Redis_Data_Merging", "port"), port=cfg.getint("Redis_Data_Merging", "port"),
db = cfg.getint("Redis_Data_Merging", "db")) db=cfg.getint("Redis_Data_Merging", "db"))
# REDIS # # REDIS #
# DB OBJECT & HASHS ( DISK ) # DB OBJECT & HASHS ( DISK )
dico_redis = {} dico_redis = {}
for year in xrange(2013, 2015): for year in xrange(2013, 2015):
for month in xrange(0,16): for month in xrange(0, 16):
dico_redis[str(year)+str(month).zfill(2)] = redis.StrictRedis( dico_redis[str(year)+str(month).zfill(2)] = redis.StrictRedis(
host = cfg.get("Redis_Level_DB", "host"), host=cfg.get("Redis_Level_DB", "host"),
port = year, port=year,
db = month) db=month)
# LOGGING # # LOGGING #
publisher.channel = "Script" publisher.channel = "Script"
@ -59,7 +61,7 @@ def main():
subscriber_name = "duplicate" subscriber_name = "duplicate"
subscriber_config_section = "PubSub_Global" subscriber_config_section = "PubSub_Global"
Sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name) sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name)
# FUNCTIONS # # FUNCTIONS #
publisher.info("""Script duplicate subscribed to channel {0}""".format(cfg.get("PubSub_Global", "channel"))) publisher.info("""Script duplicate subscribed to channel {0}""".format(cfg.get("PubSub_Global", "channel")))
@ -75,10 +77,10 @@ def main():
x = time.time() x = time.time()
message = Sub.get_msg_from_queue(r_Q_serv) message = sub.get_msg_from_queue(r_Q_serv)
if message != None: if message is not None:
path = message.split(" ",-1)[-1] path = message.split(" ", -1)[-1]
PST = P.Paste(path) PST = Paste.Paste(path)
else: else:
publisher.debug("Script Attribute is idling 10s") publisher.debug("Script Attribute is idling 10s")
time.sleep(10) time.sleep(10)
@ -91,17 +93,17 @@ def main():
PST._set_p_hash_kind("md5") PST._set_p_hash_kind("md5")
#Assignate the correct redis connexion # Assignate the correct redis connexion
r_serv1 = dico_redis[PST.p_date.year + PST.p_date.month] r_serv1 = dico_redis[PST.p_date.year + PST.p_date.month]
#Creating the bloom filter name: bloomyyyymm # Creating the bloom filter name: bloomyyyymm
bloomname = 'bloom' + PST.p_date.year + PST.p_date.month bloomname = 'bloom' + PST.p_date.year + PST.p_date.month
bloompath = cfg.get("Directories", "bloomfilters") bloompath = cfg.get("Directories", "bloomfilters")
filebloompath = bloompath + bloomname filebloompath = bloompath + bloomname
#datetime.date(int(PST.p_date.year),int(PST.p_date.month),int(PST.p_date.day)).timetuple().tm_yday % 7 # datetime.date(int(PST.p_date.year),int(PST.p_date.month),int(PST.p_date.day)).timetuple().tm_yday % 7
if os.path.exists(filebloompath): if os.path.exists(filebloompath):
bloom = BloomFilter.open(filebloompath) bloom = BloomFilter.open(filebloompath)
@ -117,40 +119,40 @@ def main():
r_serv1.set(index, PST.p_path) r_serv1.set(index, PST.p_path)
r_serv1.sadd("INDEX", index) r_serv1.sadd("INDEX", index)
#For each bloom filter # For each bloom filter
opened_bloom = [] opened_bloom = []
for bloo in r_Q_serv.smembers("bloomlist"): for bloo in r_Q_serv.smembers("bloomlist"):
#Opening blooms # Opening blooms
opened_bloom.append(BloomFilter.open(bloo)) opened_bloom.append(BloomFilter.open(bloo))
# For each hash of the paste # For each hash of the paste
for hash in PST._get_hash_lines(min = 5, start = 1, jump = 0): for hash in PST._get_hash_lines(min=5, start=1, jump=0):
nb_hash_current += 1 nb_hash_current += 1
#Adding the hash in Redis & limiting the set # Adding the hash in Redis & limiting the set
if r_serv1.scard(hash) <= set_limit: if r_serv1.scard(hash) <= set_limit:
r_serv1.sadd(hash, index) r_serv1.sadd(hash, index)
r_serv1.sadd("HASHS", hash) r_serv1.sadd("HASHS", hash)
#Adding the hash in the bloom of the month # Adding the hash in the bloom of the month
bloom.add(hash) bloom.add(hash)
#Go throught the Database of the bloom filter (of the month) # Go throught the Database of the bloom filter (of the month)
for bloo in opened_bloom: for bloo in opened_bloom:
if hash in bloo: if hash in bloo:
db = bloo.name[-6:] db = bloo.name[-6:]
#Go throught the Database of the bloom filter (of the month) # Go throught the Database of the bloom filter (of the month)
r_serv_bloom = dico_redis[db] r_serv_bloom = dico_redis[db]
#set of index paste: set([1,2,4,65]) # set of index paste: set([1,2,4,65])
hash_current = r_serv_bloom.smembers(hash) hash_current = r_serv_bloom.smembers(hash)
#removing itself from the list # removing itself from the list
hash_current = hash_current - set([index]) hash_current = hash_current - set([index])
# if the hash is present at least in 1 files (already processed) # if the hash is present at least in 1 files (already processed)
if len(hash_current) != 0: if len(hash_current) != 0:
hash_dico[hash] = hash_current hash_dico[hash] = hash_current
#if there is data in this dictionnary # if there is data in this dictionnary
if len(hash_dico) != 0: if len(hash_dico) != 0:
super_dico[index] = hash_dico super_dico[index] = hash_dico
else: else:
@ -159,12 +161,11 @@ def main():
########################################################################################### ###########################################################################################
#if there is data in this dictionnary # if there is data in this dictionnary
if len(super_dico) != 0: if len(super_dico) != 0:
# current = current paste, phash_dico = {hash: set, ...} # current = current paste, phash_dico = {hash: set, ...}
occur_dico = {} occur_dico = {}
for current, phash_dico in super_dico.items(): for current, phash_dico in super_dico.items():
nb_similar_hash = len(phash_dico)
# phash = hash, pset = set([ pastes ...]) # phash = hash, pset = set([ pastes ...])
for phash, pset in hash_dico.items(): for phash, pset in hash_dico.items():
@ -180,17 +181,18 @@ def main():
dupl.append((paste, percentage)) dupl.append((paste, percentage))
# Creating the object attribute and save it. # Creating the object attribute and save it.
to_print = 'Duplicate;{};{};{};'.format(PST.p_source, PST.p_date, PST.p_name)
if dupl != []: if dupl != []:
PST.__setattr__("p_duplicate", dupl) PST.__setattr__("p_duplicate", dupl)
PST.save_attribute_redis(r_serv_merge, "p_duplicate", dupl) PST.save_attribute_redis(r_serv_merge, "p_duplicate", dupl)
publisher.info('{0};{1};{2};{3};{4}'.format("Duplicate", PST.p_source, PST.p_date, PST.p_name,"Detected " + str(len(dupl)))) publisher.info('{}Detected {}'.format(to_print, len(dupl)))
y = time.time() y = time.time()
publisher.debug('{0};{1};{2};{3};{4}'.format("Duplicate", PST.p_source, PST.p_date, PST.p_name, "Processed in "+str(y-x)+ " sec" )) publisher.debug('{}Processed in {} sec'.format(to_print, y-x))
except IOError: except IOError:
print "CRC Checksum Failed on :", PST.p_path print "CRC Checksum Failed on :", PST.p_path
publisher.error('{0};{1};{2};{3};{4}'.format("Duplicate", PST.p_source, PST.p_date, PST.p_name, "CRC Checksum Failed" )) publisher.error('{}CRC Checksum Failed'.format(to_print))
pass pass
if __name__ == "__main__": if __name__ == "__main__":

14
bin/ZMQ_Sub_Duplicate_Q.py
View file

@ -1,12 +1,14 @@
#!/usr/bin/env python2 #!/usr/bin/env python2
# -*-coding:UTF-8 -* # -*-coding:UTF-8 -*
import redis, zmq, ConfigParser import redis
import ConfigParser
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from pubsublogger import publisher from pubsublogger import publisher
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -16,22 +18,22 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Queuing" publisher.channel = "Queuing"
# ZMQ # # ZMQ #
channel = cfg.get("PubSub_Global", "channel") channel = cfg.get("PubSub_Global", "channel")
Sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Global", channel, "duplicate") sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Global", channel, "duplicate")
# FUNCTIONS # # FUNCTIONS #
publisher.info("""Suscribed to channel {0}""".format(channel)) publisher.info("""Suscribed to channel {0}""".format(channel))
while True: while True:
Sub.get_and_lpush(r_serv) sub.get_and_lpush(r_serv)
if r_serv.sismember("SHUTDOWN_FLAGS", "Duplicate_Q"): if r_serv.sismember("SHUTDOWN_FLAGS", "Duplicate_Q"):
r_serv.srem("SHUTDOWN_FLAGS", "Duplicate_Q") r_serv.srem("SHUTDOWN_FLAGS", "Duplicate_Q")

38
bin/ZMQ_Sub_Indexer.py
View file

@ -9,17 +9,20 @@ The ZMQ_Sub_Indexer modules is fetching the list of files to be processed
and index each file with a full-text indexer (Whoosh until now). and index each file with a full-text indexer (Whoosh until now).
""" """
import redis, zmq, ConfigParser, time import redis
from packages import Paste as P import ConfigParser
import time
from packages import Paste
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from pubsublogger import publisher from pubsublogger import publisher
from whoosh.index import create_in, exists_in, open_dir from whoosh.index import create_in, exists_in, open_dir
from whoosh.fields import * from whoosh.fields import Schema, TEXT, ID
import os import os
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -29,19 +32,17 @@ def main():
# Redis # Redis
r_serv1 = redis.StrictRedis( r_serv1 = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
# Indexer configuration - index dir and schema setup # Indexer configuration - index dir and schema setup
indexpath = cfg.get("Indexer", "path") indexpath = cfg.get("Indexer", "path")
indexertype = cfg.get("Indexer", "type") indexertype = cfg.get("Indexer", "type")
if indexertype == "whoosh": if indexertype == "whoosh":
schema = Schema(title=TEXT(stored=True), path=ID(stored=True,unique=True), content=TEXT) schema = Schema(title=TEXT(stored=True), path=ID(stored=True, unique=True), content=TEXT)
if not os.path.exists(indexpath): if not os.path.exists(indexpath):
os.mkdir(indexpath) os.mkdir(indexpath)
if not exists_in(indexpath): if not exists_in(indexpath):
ix = create_in(indexpath, schema) ix = create_in(indexpath, schema)
else: else:
@ -51,22 +52,22 @@ def main():
publisher.channel = "Script" publisher.channel = "Script"
# ZMQ # # ZMQ #
#Subscriber # Subscriber
channel = cfg.get("PubSub_Global", "channel") channel = cfg.get("PubSub_Global", "channel")
subscriber_name = "indexer" subscriber_name = "indexer"
subscriber_config_section = "PubSub_Global" subscriber_config_section = "PubSub_Global"
Sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name) sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, channel, subscriber_name)
# FUNCTIONS # # FUNCTIONS #
publisher.info("""ZMQ Indexer is Running""") publisher.info("""ZMQ Indexer is Running""")
while True: while True:
try: try:
message = Sub.get_msg_from_queue(r_serv1) message = sub.get_msg_from_queue(r_serv1)
if message != None: if message is not None:
PST = P.Paste(message.split(" ",-1)[-1]) PST = Paste.Paste(message.split(" ", -1)[-1])
else: else:
if r_serv1.sismember("SHUTDOWN_FLAGS", "Indexer"): if r_serv1.sismember("SHUTDOWN_FLAGS", "Indexer"):
r_serv1.srem("SHUTDOWN_FLAGS", "Indexer") r_serv1.srem("SHUTDOWN_FLAGS", "Indexer")
@ -75,16 +76,19 @@ def main():
publisher.debug("Script Indexer is idling 10s") publisher.debug("Script Indexer is idling 10s")
time.sleep(1) time.sleep(1)
continue continue
docpath = message.split(" ",-1)[-1] docpath = message.split(" ", -1)[-1]
paste = PST.get_p_content() paste = PST.get_p_content()
print "Indexing :", docpath print "Indexing :", docpath
if indexertype == "whoosh": if indexertype == "whoosh":
indexwriter = ix.writer() indexwriter = ix.writer()
indexwriter.update_document(title=unicode(docpath, errors='ignore'),path=unicode(docpath, errors='ignore'),content=unicode(paste, errors='ignore')) indexwriter.update_document(
title=unicode(docpath, errors='ignore'),
path=unicode(docpath, errors='ignore'),
content=unicode(paste, errors='ignore'))
indexwriter.commit() indexwriter.commit()
except IOError: except IOError:
print "CRC Checksum Failed on :", PST.p_path print "CRC Checksum Failed on :", PST.p_path
publisher.error('{0};{1};{2};{3};{4}'.format("Duplicate", PST.p_source, PST.p_date, PST.p_name, "CRC Checksum Failed" )) publisher.error('Duplicate;{};{};{};CRC Checksum Failed'.format(PST.p_source, PST.p_date, PST.p_name))
pass pass

14
bin/ZMQ_Sub_Indexer_Q.py
View file

@ -12,12 +12,14 @@ handling the indexing process of the files seen.
""" """
import redis, zmq, ConfigParser import redis
import ConfigParser
from pubsublogger import publisher from pubsublogger import publisher
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -27,9 +29,9 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Queuing" publisher.channel = "Queuing"
@ -38,7 +40,7 @@ def main():
channel = cfg.get("PubSub_Global", "channel") channel = cfg.get("PubSub_Global", "channel")
subscriber_name = "indexer" subscriber_name = "indexer"
Sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Global", channel, subscriber_name) sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Global", channel, subscriber_name)
publisher.info("""Suscribed to channel {0}""".format(channel)) publisher.info("""Suscribed to channel {0}""".format(channel))
@ -46,7 +48,7 @@ def main():
# will get the data from the global ZMQ queue and buffer it in Redis. # will get the data from the global ZMQ queue and buffer it in Redis.
while True: while True:
Sub.get_and_lpush(r_serv) sub.get_and_lpush(r_serv)
if r_serv.sismember("SHUTDOWN_FLAGS", "Indexer_Q"): if r_serv.sismember("SHUTDOWN_FLAGS", "Indexer_Q"):
r_serv.srem("SHUTDOWN_FLAGS", "Indexer_Q") r_serv.srem("SHUTDOWN_FLAGS", "Indexer_Q")

43
bin/ZMQ_Sub_Longlines.py
View file

@ -1,43 +0,0 @@
#!/usr/bin/env python2
# -*-coding:UTF-8 -*
import redis, zmq, ConfigParser
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_default", "host"),
port = cfg.getint("Redis_default", "port"),
db = args.db)
p_serv = r_serv.pipeline(False)
# LOGGING #
publisher.channel = "Script"
# ZMQ #
channel = cfg.get("PubSub_Longlines", "channel_0")
Sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Longlines", channel)
# FUNCTIONS #
publisher.info("Longlines ubscribed to channel {0}".format(cfg.get("PubSub_Longlines", "channel_0")))
while True:
PST = P.Paste(Sub.get_message().split(" ", -1)[-1])
r_serv.sadd("Longlines", PST.p_mime)
PST.save_in_redis(r_serv, PST.p_mime)
if __name__ == "__main__":
main()

39
bin/ZMQ_Sub_Mails.py
View file

@ -1,7 +1,10 @@
#!/usr/bin/env python2 #!/usr/bin/env python2
# -*-coding:UTF-8 -* # -*-coding:UTF-8 -*
import redis, zmq, ConfigParser, json, pprint, time import redis
import ConfigParser
import pprint
import time
import dns.exception import dns.exception
from packages import Paste as P from packages import Paste as P
from packages import lib_refine from packages import lib_refine
@ -10,6 +13,7 @@ from pubsublogger import publisher
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -19,40 +23,40 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
r_serv1 = redis.StrictRedis( r_serv1 = redis.StrictRedis(
host = cfg.get("Redis_Data_Merging", "host"), host=cfg.get("Redis_Data_Merging", "host"),
port = cfg.getint("Redis_Data_Merging", "port"), port=cfg.getint("Redis_Data_Merging", "port"),
db = cfg.getint("Redis_Data_Merging", "db")) db=cfg.getint("Redis_Data_Merging", "db"))
r_serv2 = redis.StrictRedis( r_serv2 = redis.StrictRedis(
host = cfg.get("Redis_Cache", "host"), host=cfg.get("Redis_Cache", "host"),
port = cfg.getint("Redis_Cache", "port"), port=cfg.getint("Redis_Cache", "port"),
db = cfg.getint("Redis_Cache", "db")) db=cfg.getint("Redis_Cache", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Script" publisher.channel = "Script"
# ZMQ # # ZMQ #
Sub = ZMQ_PubSub.ZMQSub(configfile,"PubSub_Categ", "mails_categ", "emails") sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Categ", "mails_categ", "emails")
# FUNCTIONS # # FUNCTIONS #
publisher.info("Suscribed to channel mails_categ") publisher.info("Suscribed to channel mails_categ")
message = Sub.get_msg_from_queue(r_serv) message = sub.get_msg_from_queue(r_serv)
prec_filename = None prec_filename = None
email_regex = "[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,6}" email_regex = "[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,6}"
while True: while True:
try: try:
if message != None: if message is not None:
channel, filename, word, score = message.split() channel, filename, word, score = message.split()
if prec_filename == None or filename != prec_filename: if prec_filename is None or filename != prec_filename:
PST = P.Paste(filename) PST = P.Paste(filename)
MX_values = lib_refine.checking_MX_record(r_serv2, PST.get_regex(email_regex)) MX_values = lib_refine.checking_MX_record(r_serv2, PST.get_regex(email_regex))
@ -62,10 +66,11 @@ def main():
PST.save_attribute_redis(r_serv1, channel, (MX_values[0], list(MX_values[1]))) PST.save_attribute_redis(r_serv1, channel, (MX_values[0], list(MX_values[1])))
pprint.pprint(MX_values) pprint.pprint(MX_values)
to_print = 'Mails;{};{};{};Checked {} e-mail(s)'.format(PST.p_source, PST.p_date, PST.p_name, MX_values[0])
if MX_values[0] > 10: if MX_values[0] > 10:
publisher.warning('{0};{1};{2};{3};{4}'.format("Mails", PST.p_source, PST.p_date, PST.p_name,"Checked "+ str(MX_values[0])+ " e-mails" )) publisher.warning(to_print)
else: else:
publisher.info('{0};{1};{2};{3};{4}'.format("Mails", PST.p_source, PST.p_date, PST.p_name,"Checked "+ str(MX_values[0])+ " e-mail(s)" )) publisher.info(to_print)
prec_filename = filename prec_filename = filename
else: else:
@ -77,7 +82,7 @@ def main():
publisher.debug("Script Mails is Idling 10s") publisher.debug("Script Mails is Idling 10s")
time.sleep(10) time.sleep(10)
message = Sub.get_msg_from_queue(r_serv) message = sub.get_msg_from_queue(r_serv)
except dns.exception.Timeout: except dns.exception.Timeout:
print "dns.exception.Timeout" print "dns.exception.Timeout"
pass pass

14
bin/ZMQ_Sub_Mails_Q.py
View file

@ -1,12 +1,14 @@
#!/usr/bin/env python2 #!/usr/bin/env python2
# -*-coding:UTF-8 -* # -*-coding:UTF-8 -*
import redis, zmq, ConfigParser import redis
import ConfigParser
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from pubsublogger import publisher from pubsublogger import publisher
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -16,21 +18,21 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Queuing" publisher.channel = "Queuing"
# ZMQ # # ZMQ #
Sub = ZMQ_PubSub.ZMQSub(configfile,"PubSub_Categ", "mails_categ", "emails") sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Categ", "mails_categ", "emails")
# FUNCTIONS # # FUNCTIONS #
publisher.info("""Suscribed to channel {0}""".format("mails_categ")) publisher.info("""Suscribed to channel {0}""".format("mails_categ"))
while True: while True:
Sub.get_and_lpush(r_serv) sub.get_and_lpush(r_serv)
if r_serv.sismember("SHUTDOWN_FLAGS", "Mails_Q"): if r_serv.sismember("SHUTDOWN_FLAGS", "Mails_Q"):
r_serv.srem("SHUTDOWN_FLAGS", "Mails_Q") r_serv.srem("SHUTDOWN_FLAGS", "Mails_Q")

46
bin/ZMQ_Sub_Onion.py
View file

@ -21,13 +21,17 @@ Requirements
*Need the ZMQ_Sub_Onion_Q Module running to be able to work properly. *Need the ZMQ_Sub_Onion_Q Module running to be able to work properly.
""" """
import redis, zmq, ConfigParser, json, pprint, time import redis
from packages import Paste as P import ConfigParser
import pprint
import time
from packages import Paste
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from pubsublogger import publisher from pubsublogger import publisher
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -37,20 +41,14 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
r_serv1 = redis.StrictRedis( r_serv1 = redis.StrictRedis(
host = cfg.get("Redis_Data_Merging", "host"), host=cfg.get("Redis_Data_Merging", "host"),
port = cfg.getint("Redis_Data_Merging", "port"), port=cfg.getint("Redis_Data_Merging", "port"),
db = cfg.getint("Redis_Data_Merging", "db")) db=cfg.getint("Redis_Data_Merging", "db"))
r_serv2 = redis.StrictRedis(
host = cfg.get("Redis_Cache", "host"),
port = cfg.getint("Redis_Cache", "port"),
db = cfg.getint("Redis_Cache", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Script" publisher.channel = "Script"
@ -61,40 +59,40 @@ def main():
# FUNCTIONS # # FUNCTIONS #
publisher.info("Script subscribed to channel onion_categ") publisher.info("Script subscribed to channel onion_categ")
# Getting the first message from redis.
#Getting the first message from redis.
message = Sub.get_msg_from_queue(r_serv) message = Sub.get_msg_from_queue(r_serv)
prec_filename = None prec_filename = None
#Thanks to Faup project for this regex # Thanks to Faup project for this regex
# https://github.com/stricaud/faup # https://github.com/stricaud/faup
url_regex = "([a-zA-Z0-9\.\-]+(\:[a-zA-Z0-9\.&%\$\-]+)*@)*((25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[1-9])\.(25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[1-9]|0)\.(25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[1-9]|0)\.(25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[0-9])|localhost|([a-zA-Z0-9\-]+\.)*[a-zA-Z0-9\-]+\.(com|edu|gov|int|mil|net|org|biz|arpa|info|name|pro|aero|coop|museum|onion|[a-zA-Z]{2}))(\:[0-9]+)*(/($|[a-zA-Z0-9\.\,\?\'\\\+&%\$#\=~_\-]+))*" url_regex = "([a-zA-Z0-9\.\-]+(\:[a-zA-Z0-9\.&%\$\-]+)*@)*((25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[1-9])\.(25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[1-9]|0)\.(25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[1-9]|0)\.(25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[0-9])|localhost|([a-zA-Z0-9\-]+\.)*[a-zA-Z0-9\-]+\.(com|edu|gov|int|mil|net|org|biz|arpa|info|name|pro|aero|coop|museum|onion|[a-zA-Z]{2}))(\:[0-9]+)*(/($|[a-zA-Z0-9\.\,\?\'\\\+&%\$#\=~_\-]+))*"
while True: while True:
if message != None: if message is not None:
channel, filename, word, score = message.split() channel, filename, word, score = message.split()
# "For each new paste" # "For each new paste"
if prec_filename == None or filename != prec_filename: if prec_filename is None or filename != prec_filename:
domains_list = [] domains_list = []
PST = P.Paste(filename) PST = Paste.Paste(filename)
for x in PST.get_regex(url_regex): for x in PST.get_regex(url_regex):
#Extracting url with regex # Extracting url with regex
credential, subdomain, domain, host, tld, port, resource_path, query_string, f1, f2, f3, f4 = x credential, subdomain, domain, host, tld, port, resource_path, query_string, f1, f2, f3, f4 = x
if f1 == "onion": if f1 == "onion":
domains_list.append(domain) domains_list.append(domain)
#Saving the list of extracted onion domains. # Saving the list of extracted onion domains.
PST.__setattr__(channel, domains_list) PST.__setattr__(channel, domains_list)
PST.save_attribute_redis(r_serv1, channel, domains_list) PST.save_attribute_redis(r_serv1, channel, domains_list)
pprint.pprint(domains_list) pprint.pprint(domains_list)
print PST.p_path print PST.p_path
to_print = 'Onion;{};{};{};'.format(PST.p_source, PST.p_date, PST.p_name)
if len(domains_list) > 0: if len(domains_list) > 0:
publisher.warning('{0};{1};{2};{3};{4}'.format("Onion", PST.p_source, PST.p_date, PST.p_name,"Detected " + str(len(domains_list))+" .onion(s)" )) publisher.warning('{}Detected {} .onion(s)'.format(to_print, len(domains_list)))
else: else:
publisher.info('{0};{1};{2};{3};{4}'.format("Onion", PST.p_source, PST.p_date, PST.p_name, "Onion related" )) publisher.info('{}Onion related'.format(to_print))
prec_filename = filename prec_filename = filename

14
bin/ZMQ_Sub_Onion_Q.py
View file

@ -17,12 +17,14 @@ Requirements
*Should register to the Publisher "ZMQ_PubSub_Categ" *Should register to the Publisher "ZMQ_PubSub_Categ"
""" """
import redis, zmq, ConfigParser import redis
import ConfigParser
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from pubsublogger import publisher from pubsublogger import publisher
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -32,21 +34,21 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Queuing" publisher.channel = "Queuing"
# ZMQ # # ZMQ #
Sub = ZMQ_PubSub.ZMQSub(configfile,"PubSub_Categ", "onion_categ", "tor") sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Categ", "onion_categ", "tor")
# FUNCTIONS # # FUNCTIONS #
publisher.info("""Suscribed to channel {0}""".format("onion_categ")) publisher.info("""Suscribed to channel {0}""".format("onion_categ"))
while True: while True:
Sub.get_and_lpush(r_serv) sub.get_and_lpush(r_serv)
if r_serv.sismember("SHUTDOWN_FLAGS", "Onion_Q"): if r_serv.sismember("SHUTDOWN_FLAGS", "Onion_Q"):
r_serv.srem("SHUTDOWN_FLAGS", "Onion_Q") r_serv.srem("SHUTDOWN_FLAGS", "Onion_Q")

64
bin/ZMQ_Sub_Urls.py
View file

@ -1,8 +1,11 @@
#!/usr/bin/env python2 #!/usr/bin/env python2
# -*-coding:UTF-8 -* # -*-coding:UTF-8 -*
import redis, zmq, ConfigParser, json, pprint, time import redis
import ConfigParser
import pprint
import time
import dns.exception import dns.exception
from packages import Paste as P from packages import Paste
from packages import lib_refine from packages import lib_refine
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from pubsublogger import publisher from pubsublogger import publisher
@ -15,6 +18,7 @@ import ipaddress
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -24,62 +28,61 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
r_serv1 = redis.StrictRedis( r_serv1 = redis.StrictRedis(
host = cfg.get("Redis_Data_Merging", "host"), host=cfg.get("Redis_Data_Merging", "host"),
port = cfg.getint("Redis_Data_Merging", "port"), port=cfg.getint("Redis_Data_Merging", "port"),
db = cfg.getint("Redis_Data_Merging", "db")) db=cfg.getint("Redis_Data_Merging", "db"))
r_serv2 = redis.StrictRedis( r_serv2 = redis.StrictRedis(
host = cfg.get("Redis_Cache", "host"), host=cfg.get("Redis_Cache", "host"),
port = cfg.getint("Redis_Cache", "port"), port=cfg.getint("Redis_Cache", "port"),
db = cfg.getint("Redis_Cache", "db")) db=cfg.getint("Redis_Cache", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Script" publisher.channel = "Script"
# ZMQ # # ZMQ #
#Subscriber # Subscriber
subscriber_name = "urls" subscriber_name = "urls"
subscriber_config_section = "PubSub_Categ" subscriber_config_section = "PubSub_Categ"
#Publisher # Publisher
publisher_config_section = "PubSub_Url" publisher_config_section = "PubSub_Url"
publisher_name = "adress" publisher_name = "adress"
pubchannel = cfg.get("PubSub_Url", "channel") pubchannel = cfg.get("PubSub_Url", "channel")
#Country to log as critical # Country to log as critical
cc_critical = cfg.get("PubSub_Url", "cc_critical") cc_critical = cfg.get("PubSub_Url", "cc_critical")
Sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, "web_categ", subscriber_name) sub = ZMQ_PubSub.ZMQSub(configfile, subscriber_config_section, "web_categ", subscriber_name)
Pub = ZMQ_PubSub.ZMQPub(configfile, publisher_config_section, publisher_name) pub = ZMQ_PubSub.ZMQPub(configfile, publisher_config_section, publisher_name)
# FUNCTIONS # # FUNCTIONS #
publisher.info("Script URL subscribed to channel web_categ") publisher.info("Script URL subscribed to channel web_categ")
message = Sub.get_msg_from_queue(r_serv) message = sub.get_msg_from_queue(r_serv)
prec_filename = None prec_filename = None
url_regex = "(http|https|ftp)\://([a-zA-Z0-9\.\-]+(\:[a-zA-Z0-9\.&%\$\-]+)*@)*((25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[1-9])\.(25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[1-9]|0)\.(25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[1-9]|0)\.(25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[0-9])|localhost|([a-zA-Z0-9\-]+\.)*[a-zA-Z0-9\-]+\.(com|edu|gov|int|mil|net|org|biz|arpa|info|name|pro|aero|coop|museum|[a-zA-Z]{2}))(\:[0-9]+)*(/($|[a-zA-Z0-9\.\,\?\'\\\+&%\$#\=~_\-]+))*" url_regex = "(http|https|ftp)\://([a-zA-Z0-9\.\-]+(\:[a-zA-Z0-9\.&%\$\-]+)*@)*((25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[1-9])\.(25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[1-9]|0)\.(25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[1-9]|0)\.(25[0-5]|2[0-4][0-9]|[0-1]{1}[0-9]{2}|[1-9]{1}[0-9]{1}|[0-9])|localhost|([a-zA-Z0-9\-]+\.)*[a-zA-Z0-9\-]+\.(com|edu|gov|int|mil|net|org|biz|arpa|info|name|pro|aero|coop|museum|[a-zA-Z]{2}))(\:[0-9]+)*(/($|[a-zA-Z0-9\.\,\?\'\\\+&%\$#\=~_\-]+))*"
while True: while True:
try: try:
if message != None: if message is not None:
channel, filename, word, score = message.split() channel, filename, word, score = message.split()
if prec_filename == None or filename != prec_filename: if prec_filename is None or filename != prec_filename:
domains_list = [] domains_list = []
PST = P.Paste(filename) PST = Paste.Paste(filename)
client = ip2asn() client = ip2asn()
for x in PST.get_regex(url_regex): for x in PST.get_regex(url_regex):
scheme, credential, subdomain, domain, host, tld, port, resource_path, query_string, f1, f2, f3, f4 = x scheme, credential, subdomain, domain, host, tld, port, resource_path, query_string, f1, f2, f3, f4 = x
domains_list.append(domain) domains_list.append(domain)
msg = pubchannel + " " + str(x) msg = pubchannel + " " + str(x)
Pub.send_message(msg) pub.send_message(msg)
publisher.debug('{0} Published'.format(x)) publisher.debug('{0} Published'.format(x))
if f1 == "onion": if f1 == "onion":
@ -95,28 +98,29 @@ def main():
continue continue
try: try:
l = client.lookup(socket.inet_aton(ip),qType='IP') l = client.lookup(socket.inet_aton(ip), qType='IP')
except ipaddress.AddressValueError: except ipaddress.AddressValueError:
continue continue
cc = getattr(l,'cc') cc = getattr(l, 'cc')
asn = getattr(l,'asn') asn = getattr(l, 'asn')
# EU is not an official ISO 3166 code (but used by RIPE # EU is not an official ISO 3166 code (but used by RIPE
# IP allocation) # IP allocation)
if cc is not None and cc != "EU": if cc is not None and cc != "EU":
print hostl,asn,cc,pycountry.countries.get(alpha2=cc).name print hostl, asn, cc, pycountry.countries.get(alpha2=cc).name
if cc == cc_critical: if cc == cc_critical:
# FIXME: That's going to fail.
publisher.warning('{0};{1};{2};{3};{4}'.format("Url", PST.p_source, PST.p_date, PST.p_name, "Detected " + str(A_values[0]) + " " + hostl + " " + cc)) publisher.warning('{0};{1};{2};{3};{4}'.format("Url", PST.p_source, PST.p_date, PST.p_name, "Detected " + str(A_values[0]) + " " + hostl + " " + cc))
else: else:
print hostl,asn,cc print hostl, asn, cc
A_values = lib_refine.checking_A_record(r_serv2, domains_list) A_values = lib_refine.checking_A_record(r_serv2, domains_list)
if A_values[0] >= 1: if A_values[0] >= 1:
PST.__setattr__(channel, A_values) PST.__setattr__(channel, A_values)
PST.save_attribute_redis(r_serv1, channel, (A_values[0],list(A_values[1]))) PST.save_attribute_redis(r_serv1, channel, (A_values[0], list(A_values[1])))
pprint.pprint(A_values) pprint.pprint(A_values)
publisher.info('{0};{1};{2};{3};{4}'.format("Url", PST.p_source, PST.p_date, PST.p_name, "Checked " + str(A_values[0]) + " URL" )) publisher.info('{0};{1};{2};{3};{4}'.format("Url", PST.p_source, PST.p_date, PST.p_name, "Checked " + str(A_values[0]) + " URL"))
prec_filename = filename prec_filename = filename
else: else:
@ -128,7 +132,7 @@ def main():
publisher.debug("Script url is Idling 10s") publisher.debug("Script url is Idling 10s")
time.sleep(10) time.sleep(10)
message = Sub.get_msg_from_queue(r_serv) message = sub.get_msg_from_queue(r_serv)
except dns.exception.Timeout: except dns.exception.Timeout:
print "dns.exception.Timeout", A_values print "dns.exception.Timeout", A_values
pass pass

14
bin/ZMQ_Sub_Urls_Q.py
View file

@ -1,12 +1,14 @@
#!/usr/bin/env python2 #!/usr/bin/env python2
# -*-coding:UTF-8 -* # -*-coding:UTF-8 -*
import redis, zmq, ConfigParser import redis
import ConfigParser
from packages import ZMQ_PubSub from packages import ZMQ_PubSub
from pubsublogger import publisher from pubsublogger import publisher
configfile = './packages/config.cfg' configfile = './packages/config.cfg'
def main(): def main():
"""Main Function""" """Main Function"""
@ -16,21 +18,21 @@ def main():
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
# LOGGING # # LOGGING #
publisher.channel = "Queuing" publisher.channel = "Queuing"
# ZMQ # # ZMQ #
Sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Categ", "web_categ", "urls") sub = ZMQ_PubSub.ZMQSub(configfile, "PubSub_Categ", "web_categ", "urls")
# FUNCTIONS # # FUNCTIONS #
publisher.info("""Suscribed to channel {0}""".format("web_categ")) publisher.info("""Suscribed to channel {0}""".format("web_categ"))
while True: while True:
Sub.get_and_lpush(r_serv) sub.get_and_lpush(r_serv)
if r_serv.sismember("SHUTDOWN_FLAGS", "Urls_Q"): if r_serv.sismember("SHUTDOWN_FLAGS", "Urls_Q"):
r_serv.srem("SHUTDOWN_FLAGS", "Urls_Q") r_serv.srem("SHUTDOWN_FLAGS", "Urls_Q")

8
bin/indexer_lookup.py
View file

@ -12,13 +12,13 @@
import ConfigParser import ConfigParser
import argparse import argparse
import sys
import gzip import gzip
def readdoc(path=None): def readdoc(path=None):
if path is None: if path is None:
return False return False
f = gzip.open (path, 'r') f = gzip.open(path, 'r')
return f.read() return f.read()
configfile = '../packages/config.cfg' configfile = '../packages/config.cfg'
@ -40,8 +40,8 @@ argParser.add_argument('-s', action='append', help='search similar documents')
args = argParser.parse_args() args = argParser.parse_args()
from whoosh import index from whoosh import index
from whoosh.fields import * from whoosh.fields import Schema, TEXT, ID
import whoosh
schema = Schema(title=TEXT(stored=True), path=ID(stored=True), content=TEXT) schema = Schema(title=TEXT(stored=True), path=ID(stored=True), content=TEXT)
ix = index.open_dir(indexpath) ix = index.open_dir(indexpath)

32
var/www/Flask_server.py
View file

@ -1,9 +1,10 @@
#!/usr/bin/env python2 #!/usr/bin/env python2
# -*-coding:UTF-8 -* # -*-coding:UTF-8 -*
import redis, ConfigParser, json import redis
from datetime import date, datetime, time import ConfigParser
from flask import Flask, request, render_template, jsonify import json
from flask import Flask, render_template, jsonify
import flask import flask
# CONFIG # # CONFIG #
@ -12,18 +13,19 @@ cfg.read('../../bin/packages/config.cfg')
# REDIS # # REDIS #
r_serv = redis.StrictRedis( r_serv = redis.StrictRedis(
host = cfg.get("Redis_Queues", "host"), host=cfg.get("Redis_Queues", "host"),
port = cfg.getint("Redis_Queues", "port"), port=cfg.getint("Redis_Queues", "port"),
db = cfg.getint("Redis_Queues", "db")) db=cfg.getint("Redis_Queues", "db"))
r_serv_log = redis.StrictRedis( r_serv_log = redis.StrictRedis(
host = cfg.get("Redis_Log", "host"), host=cfg.get("Redis_Log", "host"),
port = cfg.getint("Redis_Log", "port"), port=cfg.getint("Redis_Log", "port"),
db = cfg.getint("Redis_Log", "db")) db=cfg.getint("Redis_Log", "db"))
app = Flask(__name__, static_url_path='/static/') app = Flask(__name__, static_url_path='/static/')
def event_stream(): def event_stream():
pubsub = r_serv_log.pubsub() pubsub = r_serv_log.pubsub()
pubsub.psubscribe("Script" + '.*') pubsub.psubscribe("Script" + '.*')
@ -32,30 +34,34 @@ def event_stream():
if msg['type'] == 'pmessage' and level != "DEBUG": if msg['type'] == 'pmessage' and level != "DEBUG":
yield 'data: %s\n\n' % json.dumps(msg) yield 'data: %s\n\n' % json.dumps(msg)
@app.route("/_logs") @app.route("/_logs")
def logs(): def logs():
return flask.Response(event_stream(), mimetype="text/event-stream") return flask.Response(event_stream(), mimetype="text/event-stream")
@app.route("/_stuff", methods = ['GET']) @app.route("/_stuff", methods=['GET'])
def stuff(): def stuff():
row1 = [] row1 = []
for queue in r_serv.smembers("queues"): for queue in r_serv.smembers("queues"):
row1.append((queue, r_serv.llen(queue))) row1.append((queue, r_serv.llen(queue)))
return jsonify(row1=row1) return jsonify(row1=row1)
@app.route("/") @app.route("/")
def index(): def index():
row = [] row = []
for queue in r_serv.smembers("queues"): for queue in r_serv.smembers("queues"):
row.append((queue, r_serv.llen(queue))) row.append((queue, r_serv.llen(queue)))
return render_template("index.html", queues_name = row) return render_template("index.html", queues_name=row)
@app.route("/monitoring/") @app.route("/monitoring/")
def monitoring(): def monitoring():
for queue in r_serv.smembers("queues"): for queue in r_serv.smembers("queues"):
return render_template("Queue_live_Monitoring.html",last_value = queue) return render_template("Queue_live_Monitoring.html", last_value=queue)
@app.route("/wordstrending/") @app.route("/wordstrending/")
def wordstrending(): def wordstrending():
@ -63,4 +69,4 @@ def wordstrending():
if __name__ == "__main__": if __name__ == "__main__":
app.run(host='0.0.0.0' ,port=7000, threaded=True) app.run(host='0.0.0.0', port=7000, threaded=True)