AIL-framework/bin/modules/Phone.py

85 lines
2.6 KiB
Python
Raw Normal View History

2018-05-04 13:53:29 +02:00
#!/usr/bin/env python3
2016-02-05 19:58:21 +01:00
# -*-coding:UTF-8 -*
2016-02-05 19:58:21 +01:00
"""
The Phone Module
================
This module is consuming the Redis-list created by the Categ module.
It apply phone number regexes on item content and warn if above a threshold.
2016-02-05 19:58:21 +01:00
"""
2021-04-02 09:52:05 +02:00
##################################
# Import External packages
##################################
import os
2016-02-05 19:58:21 +01:00
import re
import sys
import time
import phonenumbers
2021-04-02 09:52:05 +02:00
sys.path.append(os.environ['AIL_BIN'])
2021-04-02 09:52:05 +02:00
##################################
# Import Project packages
##################################
from modules.abstract_module import AbstractModule
from packages.Item import Item
2016-02-05 19:58:21 +01:00
# # TODO: # FIXME: improve regex / filter false positives
2021-04-02 09:52:05 +02:00
class Phone(AbstractModule):
"""
Phone module for AIL framework
"""
2016-02-05 19:58:21 +01:00
# regex to find phone numbers, may raise many false positives (shalt thou seek optimization, upgrading is required)
2021-04-02 09:52:05 +02:00
# reg_phone = re.compile(r'(\+\d{1,4}(\(\d\))?\d?|0\d?)(\d{6,8}|([-/\. ]{1}\d{2,3}){3,4})')
REG_PHONE = re.compile(r'(\+\d{1,4}(\(\d\))?\d?|0\d?)(\d{6,8}|([-/\. ]{1}\(?\d{2,4}\)?){3,4})')
def __init__(self):
super(Phone, self).__init__()
# Waiting time in secondes between to message proccessed
self.pending_seconds = 1
def compute(self, message):
item = Item(message)
content = item.get_content()
# List of the regex results in the Item, may be null
2021-04-02 09:52:05 +02:00
results = self.REG_PHONE.findall(content)
# If the list is greater than 4, we consider the Item may contain a list of phone numbers
2021-04-02 09:52:05 +02:00
if len(results) > 4:
self.redis_logger.debug(results)
self.redis_logger.warning(f'{item.get_id()} contains PID (phone numbers)')
2021-04-02 09:52:05 +02:00
msg = f'infoleak:automatic-detection="phone-number";{item.get_id()}'
self.send_message_to_queue(msg, 'Tags')
2021-04-02 09:52:05 +02:00
# Send to duplicate
self.send_message_to_queue(item.get_id(), 'Duplicate')
2021-04-02 09:52:05 +02:00
stats = {}
for phone_number in results:
try:
x = phonenumbers.parse(phone_number, None)
country_code = x.country_code
if stats.get(country_code) is None:
stats[country_code] = 1
else:
stats[country_code] = stats[country_code] + 1
except:
pass
for country_code in stats:
if stats[country_code] > 4:
self.redis_logger.warning(f'{item.get_id()} contains Phone numbers with country code {country_code}')
2021-04-02 09:52:05 +02:00
2016-02-05 19:58:21 +01:00
if __name__ == '__main__':
2021-04-02 09:52:05 +02:00
module = Phone()
module.run()