class HTTPScanningEngine(ANPNodeEngine, MySQLCollectionEngine, Engine): # XXX do I need to specify Engine here?
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
# if set to True then we don't delete the work directories
self.keep_work_dir = False
# the location of the incoming http streams
self.bro_http_dir = os.path.join(saq.SAQ_HOME, self.config['bro_http_dir'])
# the list of streams (connection ids) that we need to process
self.stream_list = collections.deque()
# http whitelist
self.whitelist = None
# path to the whitelist file
self.whitelist_path = os.path.join(saq.SAQ_HOME, self.config['whitelist_path'])
@property
def name(self):
return 'http_scanner'
def initialize_collection(self, *args, **kwargs):
# before we start collecting, make sure that everything in our local directory
# has a matching entry in the workload database
# TODO
pass
super().initialize_collection(*args, **kwargs)
def anp_command_handler(self, anp, command):
"""Handle inbound ANP commands from remote http engines."""
if command.command == ANP_COMMAND_COPY_FILE:
anp.send_message(ANPCommandOK())
elif command.command == ANP_COMMAND_PROCESS:
self.add_sql_work_item(command.target)
anp.send_message(ANPCommandOK())
else:
self.default_command_handler(anp, command)
def get_next_stream(self):
"""Returns the next HTTP stream to be processed or None if nothing is available to be processed."""
# do we have a list yet?
if len(self.stream_list) == 0:
for file_name in os.listdir(self.bro_http_dir):
m = REGEX_CONNECTION_ID.match(file_name)
if m:
self.stream_list.append(m.group(1))
if len(self.stream_list) == 0:
return None
return self.stream_list.popleft()
def submit_stream(self, stream_prefix, node_id):
# submit http request files
logging.info("sending stream {}".format(stream_prefix))
source_files = [ os.path.join(self.bro_http_dir, '{}.request'.format(stream_prefix)),
os.path.join(self.bro_http_dir, '{}.request.entity'.format(stream_prefix)),
os.path.join(self.bro_http_dir, '{}.reply'.format(stream_prefix)),
os.path.join(self.bro_http_dir, '{}.reply.entity'.format(stream_prefix)),
os.path.join(self.bro_http_dir, '{}.ready'.format(stream_prefix)) ]
sent_files = []
for source_file in source_files:
if not os.path.exists(source_file):
continue
result = self.submit_command(ANPCommandCOPY_FILE(source_file, source_file), node_id)
if result is None:
# no servers available at the moment
return False
elif result.command == ANP_COMMAND_OK:
sent_files.append(source_file)
continue
elif result.command == ANP_COMMAND_ERROR:
raise RuntimeError("remote server returned error message: {}".fomrat(result.error_message))
else:
raise ValueError("got unexpected command {}".format(result))
# tell the remote system to process the files
result = self.submit_command(ANPCommandPROCESS(stream_prefix), node_id)
if result is None:
logging.warning("did not receive a response for PROCESS command on {}".format(stream_prefix))
return False
elif result.command == ANP_COMMAND_OK:
# if we get this far then all the files have been sent
for sent_file in sent_files:
try:
logging.info("removing {}".format(sent_file))
os.remove(sent_file)
except Exception as e:
logging.error("unable to delete {}: {}".format(sent_file, e))
elif result.command == ANP_COMMAND_ERROR:
logging.warning("remote server returned error message: {}".format(result.error_message))
return False
else:
logging.error("got unexpected command {}".format(result))
return False
def collect_client_mode(self):
while not self.collection_shutdown:
# gather extracted http files and submit them to the server node
stream_prefix = self.get_next_stream()
if stream_prefix is None:
# nothing to do right now...
logging.debug("no streams available to send")
return False
# do we have an anp node to send data to?
node_id = self.get_available_node()
if node_id is None:
logging.info("waiting for available ANP node...")
return False
try:
self.submit_stream(stream_prefix, node_id)
except Exception as e:
logging.error("unable to submit stream {}: {}".format(stream_prefix, e))
report_exception()
def collect_local_mode(self):
# gather extracted files and just process them
stream_prefix = self.get_next_stream()
if stream_prefix:
self.add_work_item(stream_prefix)
return True
return False
def collect_server_mode(self):
# in server mode we just process our local workload
return MySQLCollectionEngine.collect(self)
def process(self, stream_prefix):
# process the .ready file
# file format is as follows
#
# C7kebl1wNwKQ1qOPck.1.ready
# time = 1537467014.49546
# interrupted = F
# finish_msg = message ends normally
# body_length = 433994
# content_gap_length = 0
# header_length = 494
#
details = {
HTTP_DETAILS_REQUEST: [],
HTTP_DETAILS_REPLY: [],
HTTP_DETAILS_READY: [],
}
base_path = os.path.join(self.bro_http_dir, stream_prefix)
# the ready file contains stream summary info
ready_path = '{}.ready'.format(base_path)
# http request headers
request_path = '{}.request'.format(base_path)
# http request content (POST content for example)
request_entity_path = '{}.request.entity'.format(base_path)
# http response headers
reply_path = '{}.reply'.format(base_path)
# http response content
reply_entity_path = '{}.reply.entity'.format(base_path)
# make sure we have at least the files we expect (summary, and request headers)
for path in [ ready_path, request_path ]:
if not os.path.exists(path):
logging.error("missing expected file {}".format(path))
return False
# parse the ready file
stream_time = None
interrupted = False
content_gap_length = 0
with open(ready_path, 'r') as fp:
for line in fp:
details[HTTP_DETAILS_READY].append(line.strip())
key, value = [_.strip() for _ in line.split(' = ')]
if key == 'time':
stream_time = datetime.datetime.fromtimestamp(float(value))
elif key == 'interrupted':
interrupted = value == 'T'
elif key == 'content_gap_length':
content_gap_length = int(value)
# parse the request
request_headers = [] # of tuples of key, value
request_headers_lookup = {} # key = key.lower()
with open(request_path, 'r') as fp:
request_ipv4 = fp.readline().strip()
request_method = fp.readline().strip()
request_original_uri = fp.readline().strip()
request_unescaped_uri = fp.readline().strip()
request_version = fp.readline().strip()
logging.info("processing {} ipv4 {} method {} uri {}".format(stream_prefix, request_ipv4,
request_method, request_original_uri))
details[HTTP_DETAILS_REQUEST].append(request_ipv4)
details[HTTP_DETAILS_REQUEST].append(request_method)
details[HTTP_DETAILS_REQUEST].append(request_original_uri)
details[HTTP_DETAILS_REQUEST].append(request_unescaped_uri)
details[HTTP_DETAILS_REQUEST].append(request_version)
for line in fp:
details[HTTP_DETAILS_REQUEST].append(line.strip())
key, value = [_.strip() for _ in line.split('\t')]
request_headers.append((key, value))
request_headers_lookup[key.lower()] = value
# parse the response if it exists
reply_headers = [] # of tuples of key, value
reply_headers_lookup = {} # key = key.lower()
reply_version = None
reply_code = None
reply_reason = None
reply_ipv4 = None
reply_port = None
if os.path.exists(reply_path):
with open(reply_path, 'r') as fp:
first_line = fp.readline()
details[HTTP_DETAILS_REPLY].append(first_line)
reply_ipv4, reply_port = [_.strip() for _ in first_line.split('\t')]
reply_port = int(reply_port)
reply_version = fp.readline().strip()
reply_code = fp.readline().strip()
reply_reason = fp.readline().strip()
details[HTTP_DETAILS_REPLY].append(reply_version)
details[HTTP_DETAILS_REPLY].append(reply_code)
details[HTTP_DETAILS_REPLY].append(reply_reason)
for line in fp:
details[HTTP_DETAILS_REPLY].append(line.strip())
key, value = [_.strip() for _ in line.split('\t')]
reply_headers.append((key, value))
reply_headers_lookup[key.lower()] = value
self.root = RootAnalysis()
self.root.uuid = str(uuid.uuid4())
self.root.storage_dir = os.path.join(self.collection_dir, self.root.uuid[0:3], self.root.uuid)
self.root.initialize_storage()
self.root.tool = 'ACE - Bro HTTP Scanner'
self.root.tool_instance = self.hostname
self.root.alert_type = 'http'
self.root.description = 'BRO HTTP Scanner Detection - {} {}'.format(request_method, request_original_uri)
self.root.event_time = datetime.datetime.now() if stream_time is None else stream_time
self.root.details = details
self.root.add_observable(F_IPV4, request_ipv4)
if reply_ipv4:
self.root.add_observable(F_IPV4, reply_ipv4)
self.root.add_observable(F_IPV4_CONVERSATION, create_ipv4_conversation(request_ipv4, reply_ipv4))
if 'host' in request_headers_lookup:
self.root.add_observable(F_FQDN, request_headers_lookup['host'])
uri = request_original_uri[:]
if 'host' in request_headers_lookup:
# I don't think we'll ever see https here as that gets parsed as a different protocol in bro
# we should only be seeing HTTP traffic
uri = '{}://{}{}{}'.format('https' if reply_port == 443 else 'http',
request_headers_lookup['host'],
# if the default port is used then leave it out, otherwise include it in the url
'' if reply_port == 80 else ':{}'.format(reply_port),
uri)
self.root.add_observable(F_URL, uri)
if request_original_uri != request_unescaped_uri:
uri = request_unescaped_uri[:]
if 'host' in request_headers_lookup:
uri = '{}:{}'.format(request_headers_lookup['host'], uri)
self.root.add_observable(F_URL, uri)
# move all the files into the work directory and add them as file observables
shutil.move(ready_path, self.root.storage_dir)
self.root.add_observable(F_FILE, os.path.basename(ready_path))
shutil.move(request_path, self.root.storage_dir)
self.root.add_observable(F_FILE, os.path.basename(request_path))
if os.path.exists(request_entity_path):
shutil.move(request_entity_path, self.root.storage_dir)
self.root.add_observable(F_FILE, os.path.basename(request_entity_path))
if os.path.exists(reply_path):
shutil.move(reply_path, self.root.storage_dir)
self.root.add_observable(F_FILE, os.path.basename(reply_path))
if os.path.exists(reply_entity_path):
shutil.move(reply_entity_path, self.root.storage_dir)
self.root.add_observable(F_FILE, os.path.basename(reply_entity_path))
try:
self.root.save()
except Exception as e:
logging.error("unable to save {}: {}".format(self.root, e))
report_exception()
return False
# has the destination host been whitelisted?
try:
if self.whitelist is None:
self.whitelist = BrotexWhitelist(self.whitelist_path)
self.whitelist.load_whitelist()
else:
self.whitelist.check_whitelist()
if 'host' in request_headers_lookup and request_headers_lookup['host']:
if self.whitelist.is_whitelisted_fqdn(request_headers_lookup['host']):
logging.debug("stream {} whitelisted by fqdn {}".format(stream_prefix, request_headers_lookup['host']))
return
except Exception as e:
logging.error("whitelist check failed for {}: {}".format(stream_prefix, e))
report_exception()
# now analyze the file
try:
self.analyze(self.root)
except Exception as e:
logging.error("analysis failed for {}: {}".format(path, e))
report_exception()
def post_analysis(self, root):
if self.should_alert(self.root):
self.root.submit()
self.cancel_analysis()
def cleanup(self, work_item):
if not self.root:
return
if self.root.delayed:
return
if not self.keep_work_dir:
logging.debug("deleting {}".format(self.root.storage_dir))
self.root.delete()
def submit():
if KEY_ANALYSIS not in request.values:
abort(
Response(
"missing {} field (see documentation)".format(KEY_ANALYSIS),
400))
r = json.loads(request.values[KEY_ANALYSIS])
# the specified company needs to match the company of this node
# TODO eventually we'll have a single node that serves API to all configured companies
if KEY_COMPANY_NAME in r and r[KEY_COMPANY_NAME] != saq.CONFIG['global'][
'company_name']:
abort(
Response(
"wrong company {} (are you sending to the correct system?)".
format(r[KEY_COMPANY_NAME]), 400))
if KEY_DESCRIPTION not in r:
abort(
Response("missing {} field in submission".format(KEY_DESCRIPTION),
400))
root = RootAnalysis()
root.uuid = str(uuid.uuid4())
# does the engine use a different drive for the workload?
analysis_mode = r[
KEY_ANALYSIS_MODE] if KEY_ANALYSIS_MODE in r else saq.CONFIG['engine'][
'default_analysis_mode']
if analysis_mode != ANALYSIS_MODE_CORRELATION:
root.storage_dir = workload_storage_dir(root.uuid)
else:
root.storage_dir = storage_dir_from_uuid(root.uuid)
root.initialize_storage()
try:
root.analysis_mode = r[
KEY_ANALYSIS_MODE] if KEY_ANALYSIS_MODE in r else saq.CONFIG[
'engine']['default_analysis_mode']
root.company_id = saq.CONFIG['global'].getint('company_id')
root.tool = r[KEY_TOOL] if KEY_TOOL in r else 'api'
root.tool_instance = r[
KEY_TOOL_INSTANCE] if KEY_TOOL_INSTANCE in r else 'api({})'.format(
request.remote_addr)
root.alert_type = r[KEY_TYPE] if KEY_TYPE in r else saq.CONFIG['api'][
'default_alert_type']
root.description = r[KEY_DESCRIPTION]
root.event_time = LOCAL_TIMEZONE.localize(datetime.datetime.now())
if KEY_EVENT_TIME in r:
try:
root.event_time = parse_event_time(r[KEY_EVENT_TIME])
except ValueError as e:
abort(
Response(
"invalid event time format for {} (use {} format)".
format(r[KEY_EVENT_TIME],
event_time_format_json_tz), 400))
root.details = r[KEY_DETAILS] if KEY_DETAILS in r else {}
# go ahead and allocate storage
# XXX use temp dir instead...
if KEY_TAGS in r:
for tag in r[KEY_TAGS]:
root.add_tag(tag)
# add the observables
if KEY_OBSERVABLES in r:
for o in r[KEY_OBSERVABLES]:
# check for required fields
for field in [KEY_O_TYPE, KEY_O_VALUE]:
if field not in o:
abort(
Response(
"an observable is missing the {} field".format(
field), 400))
o_type = o[KEY_O_TYPE]
o_value = o[KEY_O_VALUE]
o_time = None
if KEY_O_TIME in o:
try:
o_time = parse_event_time(o[KEY_O_TIME])
except ValueError:
abort(
Response(
"an observable has an invalid time format {} (use {} format)"
.format(o[KEY_O_TIME],
event_time_format_json_tz), 400))
observable = root.add_observable(o_type,
o_value,
o_time=o_time)
if KEY_O_TAGS in o:
for tag in o[KEY_O_TAGS]:
observable.add_tag(tag)
if KEY_O_DIRECTIVES in o:
for directive in o[KEY_O_DIRECTIVES]:
# is this a valid directive?
if directive not in VALID_DIRECTIVES:
abort(
Response(
"observable {} has invalid directive {} (choose from {})"
.format('{}:{}'.format(o_type,
o_value), directive,
','.join(VALID_DIRECTIVES)), 400))
observable.add_directive(directive)
if KEY_O_LIMITED_ANALYSIS in o:
for module_name in o[KEY_O_LIMITED_ANALYSIS]:
observable.limit_analysis(module_name)
# save the files to disk and add them as observables of type file
for f in request.files.getlist('file'):
logging.debug("recording file {}".format(f.filename))
#temp_dir = tempfile.mkdtemp(dir=saq.CONFIG.get('api', 'incoming_dir'))
#_path = os.path.join(temp_dir, secure_filename(f.filename))
try:
#if os.path.exists(_path):
#logging.error("duplicate file name {}".format(_path))
#abort(400)
#logging.debug("saving file to {}".format(_path))
#try:
#f.save(_path)
#except Exception as e:
#logging.error("unable to save file to {}: {}".format(_path, e))
#abort(400)
full_path = os.path.join(root.storage_dir, f.filename)
try:
dest_dir = os.path.dirname(full_path)
if not os.path.isdir(dest_dir):
try:
os.makedirs(dest_dir)
except Exception as e:
logging.error(
"unable to create directory {}: {}".format(
dest_dir, e))
abort(400)
logging.debug("saving file {}".format(full_path))
f.save(full_path)
# add this as a F_FILE type observable
root.add_observable(
F_FILE,
os.path.relpath(full_path, start=root.storage_dir))
except Exception as e:
logging.error(
"unable to copy file from {} to {} for root {}: {}".
format(_path, full_path, root, e))
abort(400)
except Exception as e:
logging.error("unable to deal with file {}: {}".format(f, e))
report_exception()
abort(400)
#finally:
#try:
#shutil.rmtree(temp_dir)
#except Exception as e:
#logging.error("unable to delete temp dir {}: {}".format(temp_dir, e))
try:
if not root.save():
logging.error("unable to save analysis")
abort(
Response(
"an error occured trying to save the alert - review the logs",
400))
# if we received a submission for correlation mode then we go ahead and add it to the database
if root.analysis_mode == ANALYSIS_MODE_CORRELATION:
ALERT(root)
# add this analysis to the workload
root.schedule()
except Exception as e:
logging.error("unable to sync to database: {}".format(e))
report_exception()
abort(
Response(
"an error occured trying to save the alert - review the logs",
400))
return json_result({'result': {'uuid': root.uuid}})
except Exception as e:
logging.error("error processing submit: {}".format(e))
report_exception()
try:
if os.path.isdir(root.storage_dir):
logging.info("removing failed submit dir {}".format(
root.storage_dir))
shutil.rmtree(root.storage_dir)
except Exception as e2:
logging.error("unable to delete failed submit dir {}: {}".format(
root.storage_dir, e))
raise e
def process(self, binary_path):
logging.debug("processing {0}".format(binary_path))
analysis_start_time = datetime.datetime.now()
# load the JSON acquired from Carbon Black
try:
with open('{0}.json'.format(binary_path), 'r') as fp:
binary_json = json.load(fp)
except Exception as e:
logging.error(
"unable to parse JSON from Carbon Black for {}: {}".format(
binary_path, str(e)))
return
# we have to copy the file into the new storage directory for it to be analyzed
# we use the file name that Carbon Black saw on the endpoint
try:
file_name = binary_json['observed_filename'][-1]
except Exception as e:
logging.error(
"cannot determine file name for {}".format(binary_path))
file_name = 'unknown'
# we need to figure out if this is a path from a Windows machine or a Unix machine
# so we count the number of backslashes and forward slashes
# it's a hack but it should work 99.9% of the time
if file_name.count('\\') > file_name.count('/'):
logging.debug("{0} appears to be a windows path".format(file_name))
file_name = ntpath.basename(file_name)
else:
logging.debug("{0} appears to be a unix path".format(file_name))
file_name = os.path.basename(file_name)
# figure out when this binary arrived to the carbon black server
# some times the time does not have the .%fZ at the end for some reason
time_stamp_format = "%Y-%m-%dT%H:%M:%SZ"
if '.' in binary_json['server_added_timestamp']:
time_stamp_format = "%Y-%m-%dT%H:%M:%S.%fZ"
event_time = datetime.datetime.strptime(
binary_json['server_added_timestamp'],
time_stamp_format).replace(tzinfo=pytz.utc)
event_time = pytz.timezone('US/Eastern').normalize(event_time)
# create the root analysis object
root = RootAnalysis()
# set all of the properties individually
# XXX fix me
# it looks like the construction logic doesn't quite work here
# when loading from the arguments to the constructor, the internal
# variables with leading underscores get set rather than the properties
# representing the database columns it was designed that way to allow the
# JSON stuff to work correctly, so I'll need to revisit that later
root.tool = 'ACE - Carbon Black Binary Analysis'
root.tool_instance = socket.gethostname()
root.alert_type = 'carbon_black_binary'
root.description = 'Carbon Black binary {0}'.format(file_name)
root.event_time = event_time
root.details = binary_json
# XXX database.Alert does not automatically create this
root.uuid = str(uuid.uuid4())
# we use a temporary directory while we process the file
root.storage_dir = os.path.join(self.work_dir, root.uuid[0:3],
root.uuid)
root.initialize_storage()
logging.debug("using storage directory {0} for {1}".format(
root.storage_dir, binary_path))
dest_path = os.path.join(root.storage_dir, file_name)
try:
shutil.copy(binary_path, dest_path)
except Exception as e:
logging.error("unable to copy {0} to {1}: {2}".format(
binary_path, dest_path, str(e)))
report_exception()
return
# note that the path is relative to the storage directory
root.add_observable(F_FILE, file_name)
# the endpoints are stored as an array of host names optionally appended with a pipe and count
# I assume the number of times that executable has executed on that host?
for endpoint in binary_json['endpoint']:
if '|' in endpoint:
endpoint = endpoint[:endpoint.index('|')]
root.add_observable(F_HOSTNAME, endpoint)
for file_path in binary_json['observed_filename']:
root.add_observable(F_FILE_PATH, file_path)
# now analyze the file
try:
self.analyze(root)
except Exception as e:
logging.error("analysis failed for {0}: {1}".format(
binary_path, str(e)))
report_exception()
logging.info("completed {0} analysis time {1}".format(
binary_path,
datetime.datetime.now() - analysis_start_time))
def _create_analysis(url, reprocess, details, db, c):
assert isinstance(url, str)
assert isinstance(reprocess, bool)
assert isinstance(details, dict)
sha256_url = hash_url(url)
if reprocess:
# if we're reprocessing the url then we clear any existing analysis
# IF the current analysis has completed
# it's OK if we delete nothing here
execute_with_retry("""DELETE FROM cloudphish_analysis_results
WHERE sha256_url = UNHEX(%s) AND status = 'ANALYZED'""",
(sha256_url,), commit=True)
# if we're at this point it means that when we asked the database for an entry from cloudphish_analysis_results
# it was empty, OR, we cleared existing analysis
# however, we could have multiple requests coming in at the same time for the same url
# so we need to take that into account here
# first we'll generate our analysis uuid we're going to use
_uuid = str(uuid.uuid4())
# so first we try to insert it
try:
execute_with_retry(db, c, ["""INSERT INTO cloudphish_analysis_results ( sha256_url, uuid, insert_date )
VALUES ( UNHEX(%s), %s, NOW() )""",
"""INSERT INTO cloudphish_url_lookup ( sha256_url, url )
VALUES ( UNHEX(%s), %s )"""],
[(sha256_url, _uuid),
(sha256_url, url)], commit=True)
except pymysql.err.IntegrityError as e:
# (<class 'pymysql.err.IntegrityError'>--(1062, "Duplicate entry
# if we get a duplicate key entry here then it means that an entry was created between when we asked
# and now
if e.args[0] != 1062:
raise e
# so just return that one that was already created
return get_cached_analysis(url)
# at this point we've inserted an entry into cloudphish_analysis_results for this url
# now at it's processing to the workload
root = RootAnalysis()
root.uuid = _uuid
root.storage_dir = storage_dir_from_uuid(root.uuid)
root.initialize_storage()
root.analysis_mode = ANALYSIS_MODE_CLOUDPHISH
# this is kind of a kludge but,
# the company_id initially starts out as whatever the default is for this node
# later, should the analysis turn into an alert, the company_id changes to whatever
# is stored as the "d" field in the KEY_DETAILS_CONTEXT
root.company_id = saq.COMPANY_ID
root.tool = 'ACE - Cloudphish'
root.tool_instance = saq.SAQ_NODE
root.alert_type = ANALYSIS_TYPE_CLOUDPHISH
root.description = 'ACE Cloudphish Detection - {}'.format(url)
root.event_time = datetime.datetime.now()
root.details = {
KEY_DETAILS_URL: url,
KEY_DETAILS_SHA256_URL: sha256_url,
# this used to be configurable but it's always true now
KEY_DETAILS_ALERTABLE: True,
KEY_DETAILS_CONTEXT: details, # <-- optionally contains the source company_id
}
url_observable = root.add_observable(F_URL, url)
if url_observable:
url_observable.add_directive(DIRECTIVE_CRAWL)
root.save()
root.schedule()
return get_cached_analysis(url)
def process(self, stream_archive_path):
logging.info("processing {}".format(stream_archive_path))
# format of file name is connection_id.stream_type.tar
stream_archive_file = os.path.basename(stream_archive_path)
connection_id, stream_type, _ = stream_archive_file.split('.')
if connection_id is None:
logging.error("unable to determine connection_id from {}".format(
stream_archive_file))
if stream_type is None:
logging.error("unable to determine stream type from {}".format(
stream_archive_file))
# we use a temporary directory while we process the file
storage_dir = os.path.relpath(os.path.join(self.work_dir,
connection_id),
start=saq.SAQ_HOME)
logging.debug("using storage_dir {} for {}".format(
storage_dir, stream_archive_file))
# XXX there seems to be an occasional issue where tar fails on the client side
# need to investigate and fix that side of it
# but this works for now
_count = 0
_new_storage_dir = storage_dir
while os.path.exists(_new_storage_dir):
_new_storage_dir = '{}.{}'.format(storage_dir, _count)
_count += 1
storage_dir = _new_storage_dir
try:
root = RootAnalysis(
storage_dir=storage_dir,
tool='ACE - Brotex',
tool_instance=socket.gethostname(),
alert_type='brotex - {} - v2'.format(stream_type),
desc='Brotex {} Stream Detection - {}'.format(
stream_type.upper(), connection_id),
event_time=datetime.datetime.now())
root.initialize_storage()
dest_path = os.path.join(storage_dir, stream_archive_file)
shutil.copy(stream_archive_path, dest_path)
file_observable = root.add_observable(
F_FILE, os.path.relpath(dest_path, start=root.storage_dir))
if file_observable:
file_observable.add_directive(DIRECTIVE_ARCHIVE)
file_observable.limited_analysis = [
BrotexSMTPPackageAnalyzer.__name__,
BrotexHTTPPackageAnalyzer.__name__,
BrotexSMTPStreamArchiveAction.__name__
]
# now analyze the file
try:
self.analyze(root)
except Exception as e:
logging.error("analysis failed for {}: {}".format(
dest_path, e))
report_exception()
except Exception as e:
logging.error("unable to process {}: {}".format(
stream_archive_file, e))
report_exception()
finally:
try:
# delete original stream file
os.remove(stream_archive_path)
except Exception as e:
logging.error("unable to delete file {0}: {1}".format(
stream_archive_path, str(e)))
report_exception()
# delete the work directory
if os.path.isdir(root.storage_dir):
try:
shutil.rmtree(root.storage_dir)
except Exception as e:
logging.error("unable to delete {}: {}".format(
root.storage_dir, e))
report_exception()
def post_smtp_analysis(self, root):
from saq.modules.email import EmailAnalysis, SMTPStreamAnalysis, \
BrotexSMTPPackageAnalysis, \
KEY_ENVELOPES_MAIL_FROM, KEY_ENVELOPES_RCPT_TO
# get the paths to the email scanning system
#email_scanner_dir = saq.CONFIG['engine_email_scanner']['collection_dir']
email_scanner_dir = self.collection_dir
# create a new analysis root for each email analysis we found
for analysis in root.all_analysis:
if not isinstance(analysis, EmailAnalysis) or not analysis.email:
continue
env_mail_from = None
env_rcpt_to = None
connection_id = None
# the observable for this EmailAnalysis will be a file
email_file = analysis.observable
if email_file.type != F_FILE:
logging.warning(
"the observable for {} should be F_FILE but it is {}".
format(analysis, email_file.type))
else:
# this will be either an rfc822 file generated by the SMTPStreamAnalysis module
# (which will have the envelope information)
# OR it is a "broken stream" file, which does not
stream_analysis = [
a for a in root.all_analysis
if isinstance(a, SMTPStreamAnalysis)
and email_file in a.observables
]
if len(stream_analysis) > 1:
logging.error("there should not be more than one of these")
elif len(stream_analysis) == 1:
stream_analysis = stream_analysis[0]
logging.debug(
"detected stream analysis for {}".format(email_file))
# get the MAIL FROM and RCPT TO from this
if not analysis.env_mail_from:
if email_file.value in stream_analysis.envelopes:
analysis.env_mail_from = stream_analysis.envelopes[
email_file.value][KEY_ENVELOPES_MAIL_FROM]
if not analysis.env_rcpt_to:
if email_file.value in stream_analysis.envelopes:
analysis.env_rcpt_to = stream_analysis.envelopes[
email_file.value][KEY_ENVELOPES_RCPT_TO]
# get the original brotex package file that the stream came from
stream_package = stream_analysis.observable
# get the BrotexSMTPPackageAnalysis for this stream package so we can get the connection id
package_analysis = [
a for a in root.all_analysis
if isinstance(a, BrotexSMTPPackageAnalysis)
and stream_package in a.observables
]
if len(package_analysis) > 1:
logging.error(
"there should not be more than one of these!")
elif len(package_analysis) == 1:
package_analysis = package_analysis[0]
connection_id = package_analysis.connection_id
# if we could not find the stream, we will want to find the brotex smtp package so we can have the connection id
package_analysis = [
a for a in root.all_analysis
if isinstance(a, BrotexSMTPPackageAnalysis)
and email_file in a.observables
]
if len(package_analysis) > 1:
logging.error(
"there should not be more than one of these!")
elif len(package_analysis) == 1:
package_analysis = package_analysis[0]
connection_id = package_analysis.connection_id
subroot = RootAnalysis()
subroot.company_name = root.company_name
subroot.tool = root.tool
subroot.tool_instance = root.tool_instance
subroot.alert_type = root.alert_type
subroot.description = 'Brotex SMTP Stream Detection - '
if analysis.decoded_subject:
subroot.description += '{} '.format(analysis.decoded_subject)
elif analysis.subject:
subroot.description += '{} '.format(analysis.subject)
else:
subroot.description += '(no subject) '
if analysis.env_mail_from:
subroot.description += 'From {} '.format(
normalize_email_address(analysis.env_mail_from))
elif analysis.mail_from:
subroot.description += 'From {} '.format(
normalize_email_address(analysis.mail_from))
if analysis.env_rcpt_to:
if len(analysis.env_rcpt_to) == 1:
subroot.description += 'To {} '.format(
analysis.env_rcpt_to[0])
else:
subroot.description += 'To ({} recipients) '.format(
len(analysis.env_rcpt_to))
elif analysis.mail_to:
if isinstance(analysis.mail_to,
list): # XXX I think this *has* to be a list
if len(analysis.mail_to) == 1:
subroot.description += 'To {} '.format(
analysis.mail_to[0])
else:
subroot.description += 'To ({} recipients) '.format(
len(analysis.mail_to))
else:
subroot.description += 'To {} '.format(
analysis.mail_to)
subroot.event_time = root.event_time
subroot.details = analysis.details
subroot.details['connection_id'] = connection_id
subroot.uuid = str(uuid.uuid4())
# we use a temporary directory while we process the file
subroot.storage_dir = os.path.join(email_scanner_dir,
subroot.uuid[0:3], subroot.uuid)
subroot.initialize_storage()
# copy the original file
src_path = os.path.join(root.storage_dir,
analysis.observable.value)
dest_path = os.path.join(subroot.storage_dir,
analysis.observable.value)
subroot.add_observable(
F_FILE, os.path.relpath(dest_path, start=subroot.storage_dir))
# so the EmailAnalysis that will trigger on the RFC822 file (or whatever you have)
# will *not* have the envelope headers
# so we do that here in the main alert
env_mail_from = None
if analysis.env_mail_from:
# this is to handle this: <*****@*****.**> SIZE=80280
# XXX assuming there can be no spaces in an email address
env_mail_from = analysis.env_mail_from.split(' ', 1)
env_mail_from = env_mail_from[0]
# is this not the empty indicator?
if env_mail_from != '<>':
env_mail_from = normalize_email_address(env_mail_from)
subroot.add_observable(F_EMAIL_ADDRESS, env_mail_from)
if analysis.env_rcpt_to:
for address in analysis.env_rcpt_to:
address = normalize_email_address(address)
if address:
subroot.add_observable(F_EMAIL_ADDRESS, address)
if env_mail_from:
subroot.add_observable(
F_EMAIL_CONVERSATION,
create_email_conversation(
env_mail_from, address))
try:
subroot.save()
except Exception as e:
logging.error("unable to save {}: {}".format(alert, e))
report_exception()
continue
# TODO also add the stream and update any envelopment headers and stuff
try:
logging.debug("copying {} to {}".format(src_path, dest_path))
shutil.copy(src_path, dest_path)
except Exception as e:
logging.error("unable to copy {} to {}: {}".format(
src_path, dest_path, e))
report_exception()
continue
# submit the path to the database of the email scanner for analysis
try:
submit_sql_work_item('EMAIL', subroot.storage_dir)
except Exception as e:
logging.error("unable to add work item: {}".format(e))
report_exception()
continue
def post_http_analysis(self, root):
from saq.modules.http import BrotexHTTPPackageAnalysis, \
KEY_TIME, \
KEY_SRC_IP, \
KEY_SRC_PORT, \
KEY_DEST_IP, \
KEY_DEST_PORT, \
KEY_METHOD, \
KEY_HOST, \
KEY_URI, \
KEY_REFERRER, \
KEY_USER_AGENT, \
KEY_STATUS_CODE, \
KEY_FILES
# get the paths to the http scanning system
#http_scanner_dir = saq.CONFIG['engine_http_scanner']['collection_dir']
http_scanner_dir = self.collection_dir
analysis = None
for a in root.all_analysis:
if isinstance(a, BrotexHTTPPackageAnalysis) and a.requests:
analysis = a
break
# this can happen if the request was whitelisted
if analysis:
for request in analysis.requests:
subroot = RootAnalysis()
subroot.company_name = root.company_name
subroot.tool = root.tool
subroot.tool_instance = root.tool_instance
subroot.alert_type = root.alert_type
subroot.description = "Brotex HTTP Stream Detection - "
if request[KEY_HOST]:
subroot.description += " {} ".format(request[KEY_HOST])
if request[KEY_DEST_IP]:
subroot.description += " ({}) ".format(
request[KEY_DEST_IP])
if request[KEY_URI]:
# don't want to show all the fragments and query params
try:
parts = urlparse(request[KEY_URI])
subroot.description += parts.path
except Exception as e:
logging.warning("unable to parse {}: {}".format(
request[KEY_URI], e))
subroot.description += request[KEY_URI]
subroot.event_time = root.event_time
subroot.details = request
subroot.uuid = str(uuid.uuid4())
# we use a temporary directory while we process the file
subroot.storage_dir = os.path.join(http_scanner_dir,
subroot.uuid[0:3],
subroot.uuid)
subroot.initialize_storage()
if request[KEY_SRC_IP]:
subroot.add_observable(F_IPV4, request[KEY_SRC_IP])
if request[KEY_DEST_IP]:
subroot.add_observable(F_IPV4, request[KEY_DEST_IP])
if request[KEY_SRC_IP] and request[KEY_DEST_IP]:
subroot.add_observable(
F_IPV4_CONVERSATION,
create_ipv4_conversation(request[KEY_SRC_IP],
request[KEY_DEST_IP]))
if request[KEY_HOST]:
subroot.add_observable(F_FQDN, request[KEY_HOST])
if request[KEY_URI]:
subroot.add_observable(F_URL, request[KEY_URI])
if request[KEY_REFERRER]:
subroot.add_observable(F_URL, request[KEY_REFERRER])
for file_path in request[KEY_FILES]:
src_path = os.path.join(root.storage_dir, file_path)
dest_path = os.path.join(subroot.storage_dir,
os.path.basename(file_path))
try:
shutil.copy(src_path, dest_path)
except Exception as e:
logging.error("unable to copy {} to {}: {}".format(
src_path, dest_path, e))
report_exception()
subroot.add_observable(
F_FILE,
os.path.basename(file_path)) # already relative
try:
subroot.save()
except Exception as e:
logging.error("unable to save {}: {}".format(alert, e))
report_exception()
continue
# submit the path to the database of the email scanner for analysis
try:
submit_sql_work_item(
'HTTP', subroot.storage_dir) # XXX hard coded constant
except:
# failure is already logged inside the call
continue
def process(self, process):
logging.debug("processing json")
analysis_start_time = datetime.datetime.now()
try:
file_path = process['cmdline'].split('"')[-2]
except:
logging.error("cannot determine file path for {}".format(
process['cmdline']))
file_path = 'unknown'
try:
file_name = file_path.split('\\')[-1]
except:
logging.error(
"cannot determine file name for {}".format(file_path))
file_name = 'unknown'
# figure out when this binary arrived to the carbon black server
# some times the time does not have the .%fZ at the end for some reason
time_stamp_format = "%Y-%m-%dT%H:%M:%SZ"
if '.' in process['start']:
time_stamp_format = "%Y-%m-%dT%H:%M:%S.%fZ"
event_time = datetime.datetime.strptime(
process['start'], time_stamp_format).replace(tzinfo=pytz.utc)
event_time = pytz.timezone('US/Eastern').normalize(event_time)
# create the root analysis object
root = RootAnalysis()
# set all of the properties individually
# XXX fix me
# it looks like the construction logic doesn't quite work here
# when loading from the arguments to the constructor, the internal
# variables with leading underscores get set rather than the properties
# representing the database columns it was designed that way to allow the
# JSON stuff to work correctly, so I'll need to revisit that later
root.tool = 'ACE - Carbon Black Internet Office File Analysis'
root.tool_instance = socket.gethostname()
root.alert_type = 'carbon_black_internet_office_file'
root.description = 'Carbon Black Internet Office File {0}'.format(
file_name)
root.event_time = event_time
root.details = process
# XXX database.Alert does not automatically create this
root.uuid = str(uuid.uuid4())
# we use a temporary directory while we process the file
root.storage_dir = os.path.join(self.work_dir, root.uuid[0:3],
root.uuid)
root.initialize_storage()
# note that the path is relative to the storage directory
fl_observable = root.add_observable(
F_FILE_LOCATION,
create_file_location(process['hostname'], file_path))
if fl_observable: fl_observable.add_directive(DIRECTIVE_COLLECT_FILE)
root.add_observable(F_FILE_PATH, file_path)
root.add_observable(F_FILE_NAME, file_name)
root.add_observable(F_HOSTNAME, process['hostname'])
# now analyze the file
try:
self.analyze(root)
except Exception as e:
logging.error("analysis failed for {}: {}".format(
process['id'], e))
report_exception()
logging.info("completed {} analysis time {}".format(
process['id'],
datetime.datetime.now() - analysis_start_time))
def process(self, work_item):
url, alertable, details = work_item
# any other result means we should process it
logging.info("processing url {} (alertable {})".format(url, alertable))
#logging.debug("details = {}".format(details))
sha256_url = hash_url(url)
# create or update our analysis entry
with get_db_connection('cloudphish') as db:
c = db.cursor()
c.execute(
"""UPDATE analysis_results SET status = %s WHERE sha256_url = UNHEX(%s)""",
(STATUS_ANALYZING, sha256_url))
db.commit()
root = RootAnalysis()
# create a temporary storage directory for this work
root.tool = 'ACE - Cloudphish'
root.tool_instance = self.location
root.alert_type = 'cloudphish'
root.description = 'ACE Cloudphish Detection - {}'.format(url)
root.event_time = datetime.datetime.now()
root.uuid = str(uuid.uuid4())
root.storage_dir = os.path.join(self.work_dir, root.uuid[0:2],
root.uuid)
root.initialize_storage()
if 'i' in details:
root.company_name = details['i']
if 'd' in details:
root.company_id = details['d']
root.details = {
KEY_DETAILS_URL: url,
KEY_DETAILS_SHA256_URL: sha256_url,
KEY_DETAILS_ALERTABLE: alertable,
KEY_DETAILS_CONTEXT: details,
}
url_observable = root.add_observable(F_URL, url)
if url_observable is None:
logging.error("request for invalid url received: {}".format(url))
return
url_observable.add_directive(DIRECTIVE_CRAWL)
# the "details context" can also contain observables
for key in root.details[KEY_DETAILS_CONTEXT].keys():
if key in VALID_OBSERVABLE_TYPES:
root.add_observable(key,
root.details[KEY_DETAILS_CONTEXT][key])
try:
self.analyze(root)
except Exception as e:
logging.error("analysis failed for {}: {}".format(url, e))
report_exception()
with get_db_connection('cloudphish') as db:
c = db.cursor()
c.execute(
"""UPDATE analysis_results SET
result = %s,
status = %s,
http_result_code = NULL,
http_message = NULL,
sha256_content = NULL
WHERE sha256_url = UNHEX(%s)""",
(SCAN_RESULT_ERROR, STATUS_ANALYZED, sha256_url))
db.commit()
return