mail_to_misp/mail_to_misp.py

258 lines
8.4 KiB
Python
Raw Normal View History

2017-05-24 11:02:28 +02:00
#!/usr/bin/python3
# -*- coding: utf-8 -*-
2017-04-27 13:58:49 +02:00
import urlmarker
2017-04-28 10:00:58 +02:00
import hashmarker
2017-04-27 13:58:49 +02:00
import sys
import re
from pyfaup.faup import Faup
from pymisp import PyMISP
from defang import refang
import dns.resolver
2017-05-23 15:09:25 +02:00
import email
from email.generator import Generator
2017-05-24 11:02:28 +02:00
import tempfile
2017-05-24 15:44:14 +02:00
import socket
2017-05-30 11:24:30 +02:00
import syslog
import ftfy
2017-05-31 16:47:08 +02:00
import os
configfile = os.path.basename(sys.argv[0]).split(".py")[0] + "_config"
config = __import__(configfile)
2017-05-30 11:24:30 +02:00
syslog.openlog(logoption=syslog.LOG_PID, facility=syslog.LOG_USER)
2017-05-24 15:44:14 +02:00
def is_valid_ipv4_address(address):
try:
socket.inet_pton(socket.AF_INET, address)
except AttributeError: # no inet_pton here, sorry
try:
socket.inet_aton(address)
except socket.error:
return False
return address.count('.') == 3
except socket.error: # not a valid address
return False
return True
def is_valid_ipv6_address(address):
try:
socket.inet_pton(socket.AF_INET6, address)
except socket.error: # not a valid address
return False
return True
2017-04-27 13:58:49 +02:00
debug = config.debug
2017-05-23 15:09:25 +02:00
stdin_used = False
2017-05-30 11:24:30 +02:00
email_subject = config.email_subject_prefix
2017-05-23 15:09:25 +02:00
try:
2017-05-24 15:50:05 +02:00
if not sys.stdin.isatty():
email_data = b''
mailcontent = sys.stdin.buffer.read().decode("utf-8", "ignore")
syslog.syslog(mailcontent)
2017-05-24 15:50:05 +02:00
msg = email.message_from_string(mailcontent)
mail_subject = msg.get('Subject').encode("utf-8", "ignore")
2017-05-24 15:50:05 +02:00
for part in msg.walk():
if part.get_content_charset() is None:
# This could probably be detected
charset = 'utf-8'
else:
charset = part.get_content_charset()
2017-05-24 15:50:05 +02:00
if part.get_content_maintype() == 'multipart':
continue
if part.get_content_maintype() == 'text':
part.set_charset(charset)
2017-05-24 15:50:05 +02:00
email_data += part.get_payload(decode=True)
email_subject += mail_subject
stdin_used = True
2017-05-23 15:09:25 +02:00
except Exception as e:
syslog.syslog("EXCEPTION")
syslog.syslog(str(e))
2017-05-23 15:09:25 +02:00
pass
2017-04-27 13:58:49 +02:00
try:
2017-05-23 15:09:25 +02:00
if not stdin_used:
email_data = sys.argv[1]
email_subject = sys.argv[2]
2017-04-27 13:58:49 +02:00
except:
if debug:
2017-05-30 11:24:30 +02:00
syslog.syslog("FATAL ERROR: Not all required input received")
2017-04-27 13:58:49 +02:00
sys.exit(1)
if debug:
syslog.syslog("Encoding of subject: {0}".format(ftfy.guess_bytes(email_subject)[1]))
syslog.syslog("Encoding of body: {0}".format(ftfy.guess_bytes(email_data)[1]))
email_data = ftfy.fix_text(email_data.decode("utf-8", "ignore"))
email_subject = ftfy.fix_text(email_subject.decode("utf-8", "ignore"))
2017-04-27 13:58:49 +02:00
if debug:
syslog.syslog(email_subject)
syslog.syslog(email_data)
2017-04-27 13:58:49 +02:00
misp_url = config.misp_url
misp_key = config.misp_key
misp_verifycert = config.misp_verifycert
resolver = dns.resolver.Resolver(configure=False)
resolver.nameservers = config.nameservers
excludelist = config.excludelist
externallist = config.externallist
2017-05-29 17:06:46 +02:00
internallist = config.internallist
2017-04-28 10:00:58 +02:00
noidsflaglist = config.noidsflaglist
ignorelist = config.ignorelist
removelist = config.removelist
2017-04-27 13:58:49 +02:00
malwaretags = config.malwaretags
dependingtags = config.dependingtags
tlptag_default = config.tlptag_default
2017-05-17 09:51:47 +02:00
stopword = config.stopword
hash_only_tags = config.hash_only_tags
2017-05-29 15:35:56 +02:00
forward_identifiers = config.forward_identifiers
2017-05-31 15:45:53 +02:00
attach_original_mail = config.attach_original_mail
2017-04-27 13:58:49 +02:00
original_email_data = email_data
2017-04-27 13:58:49 +02:00
# Ignore lines in body of message
for ignoreline in ignorelist:
email_data = re.sub(ignoreline, "", email_data)
2017-04-27 13:58:49 +02:00
# Remove words from subject
for removeword in removelist:
email_subject = re.sub(removeword, "", email_subject)
2017-04-27 13:58:49 +02:00
def init(url, key):
return PyMISP(url, key, misp_verifycert, 'json')
# Evaluate classification
tlp_tag = tlptag_default
2017-04-27 13:58:49 +02:00
tlptags = config.tlptags
for tag in tlptags:
for alternativetag in tlptags[tag]:
if alternativetag in email_data.lower():
2017-04-27 13:58:49 +02:00
tlp_tag = tag
# Create the MISP event
misp = init(misp_url, misp_key)
new_event = misp.new_event(info=email_subject, distribution=0, threat_level_id=3, analysis=1)
2017-04-27 13:58:49 +02:00
misp.add_tag(new_event, tlp_tag)
2017-05-31 15:45:53 +02:00
if attach_original_mail and original_email_data:
misp.add_named_attribute(new_event, 'email-body', original_email_data, category='Payload delivery', to_ids=False)
2017-04-27 13:58:49 +02:00
# Add additional tags depending on others
for tag in dependingtags:
if tag in tlp_tag:
for dependingtag in dependingtags[tag]:
misp.add_tag(new_event, dependingtag)
2017-05-29 15:35:56 +02:00
## Prepare extraction of IOCs
# Limit the input if the stopword is found
2017-05-17 09:51:47 +02:00
email_data = email_data.split(stopword, 1)[0]
2017-05-29 15:35:56 +02:00
# Find the first forwarding message and use that content
position = 99999
t_email_data = email_data
for identifier in forward_identifiers:
new_position = email_data.find(identifier)
2017-05-29 17:06:46 +02:00
if new_position == -1:
new_position = position
2017-05-29 15:35:56 +02:00
if new_position < position:
t_before, t_split, t_email_data = email_data.partition(identifier)
position = new_position
email_data = t_email_data
# Refang email data
email_data = refang(email_data)
2017-05-29 15:35:56 +02:00
## Extract various IOCs
2017-05-24 15:44:14 +02:00
urllist = list()
urllist += re.findall(urlmarker.WEB_URL_REGEX, email_data)
2017-04-28 10:00:58 +02:00
urllist += re.findall(urlmarker.IP_REGEX, email_data)
2017-04-27 13:58:49 +02:00
if debug:
2017-05-30 11:24:30 +02:00
syslog.syslog(str(urllist))
2017-04-27 13:58:49 +02:00
# Init Faup
f = Faup()
2017-04-27 17:50:50 +02:00
# Add tags according to configuration
2017-04-27 13:58:49 +02:00
for malware in malwaretags:
if malware in email_subject.lower():
2017-04-27 13:58:49 +02:00
for tag in malwaretags[malware]:
misp.add_tag(new_event, tag)
2017-04-28 10:00:58 +02:00
# Extract and add hashes
hashlist_md5 = re.findall(hashmarker.MD5_REGEX, email_data)
hashlist_sha1 = re.findall(hashmarker.SHA1_REGEX, email_data)
hashlist_sha256 = re.findall(hashmarker.SHA256_REGEX, email_data)
for h in hashlist_md5:
misp.add_hashes(new_event, md5=h)
for h in hashlist_sha1:
misp.add_hashes(new_event, sha1=h)
for h in hashlist_sha256:
misp.add_hashes(new_event, sha256=h)
if (len(hashlist_md5) > 0) or (len(hashlist_sha1) > 0) or (len(hashlist_sha256) > 0):
for tag in hash_only_tags:
misp.add_tag(new_event, tag)
2017-04-27 13:58:49 +02:00
# Add IOCs and expanded information to MISP
for entry in urllist:
2017-04-28 10:00:58 +02:00
ids_flag = True
2017-04-27 13:58:49 +02:00
f.decode(entry)
domainname = f.get_domain().decode('utf-8', 'ignore')
hostname = f.get_host().decode('utf-8', 'ignore')
2017-04-27 13:58:49 +02:00
if debug:
syslog.syslog(domainname)
2017-04-27 13:58:49 +02:00
if domainname not in excludelist:
2017-05-29 17:06:46 +02:00
if domainname in internallist:
misp.add_named_attribute(new_event, 'link', entry, category='Internal reference', to_ids=False, distribution=0)
elif domainname in externallist:
2017-05-08 15:47:47 +02:00
misp.add_named_attribute(new_event, 'link', entry, category='External analysis', to_ids=False)
2017-04-27 13:58:49 +02:00
else:
2017-04-28 10:00:58 +02:00
if (domainname in noidsflaglist) or (hostname in noidsflaglist):
ids_flag = False
2017-04-27 13:58:49 +02:00
if debug:
2017-05-30 11:24:30 +02:00
syslog.syslog(str(entry))
2017-05-24 15:44:14 +02:00
if hostname:
2017-05-24 16:39:21 +02:00
if is_valid_ipv4_address(entry):
misp.add_url(new_event, entry, category='Network activity', to_ids=False)
else:
misp.add_url(new_event, entry, category='Network activity', to_ids=ids_flag)
2017-04-27 13:58:49 +02:00
if debug:
syslog.syslog(hostname)
2017-05-24 15:44:14 +02:00
port = f.get_port()
comment = ""
if port:
comment = "on port: " + str(port)
if is_valid_ipv4_address(hostname):
misp.add_ipdst(new_event, hostname, comment=comment, category='Network activity', to_ids=False)
2017-05-24 15:44:14 +02:00
else:
misp.add_hostname(new_event, hostname, comment=comment, category='Network activity', to_ids=ids_flag)
2017-05-24 15:44:14 +02:00
try:
for rdata in dns.resolver.query(hostname, 'A'):
2017-05-24 15:44:14 +02:00
if debug:
2017-05-30 11:24:30 +02:00
syslog.syslog(str(rdata))
misp.add_ipdst(new_event, rdata.to_text(), category='Network activity', to_ids=False, comment=hostname)
2017-05-24 15:44:14 +02:00
except Exception as e:
if debug:
2017-05-31 16:47:08 +02:00
syslog.syslog(str(e))
2017-05-24 11:02:28 +02:00
# Try to add attachments
if stdin_used:
for part in msg.walk():
if part.get_content_maintype() == 'multipart':
continue
if part.get_content_maintype() != 'text':
filename = part.get_filename()
_, output_path = tempfile.mkstemp()
output = open(output_path, 'wb')
output.write(part.get_payload(decode=True))
misp.upload_sample(filename, output_path, new_event, distribution=None, to_ids=True, category=None, comment=None, info='My Info', analysis=None, threat_level_id=None)
2017-05-24 11:02:28 +02:00
output.close()
syslog.syslog("Job finished.")