def derive(strings):
values = set()
observables = set()
for string in iterify(strings):
if string:
try:
t = Observable.guess_type(string)
observable = t(value=string)
observable.normalize()
observables.add(observable)
values.add(observable.value)
except ObservableValidationError:
values.add(string)
new = []
for observable in observables:
for a in analyzers.get(observable.__class__, []):
new.extend([
n for n in a.analyze_string(observable.value)
if n and n not in values
])
if len(new) == 0:
return values, values
else:
_, extended = derive(new + list(values))
return values, extended
def post(self):
q = request.get_json(silent=True)
params = q.pop("params", {})
observables = []
for o in q["observables"]:
try:
obs = Observable.guess_type(o['value'])(value=o['value'])
obs.clean()
observables.append(obs.value)
# Save observables & eventual tags to database
if params.get('save_query', False):
obs = obs.save()
obs.tag(o.get("tags", []))
obs.add_source("query")
except ObservableValidationError:
continue
# match observables with known indicators
data = match_observables([o for o in observables])
# find related observables (eg. URLs for domain, etc.)
# related_observables = [obs.get_related() for obs in observables]
# data = self.match_observables(related_observable)
#
# we need to find a way to degrade the "confidence" in
# hits obtained from related observables
return render(data, "analysis.html")
def derive(observables):
"""Indicate that the module needs a specific attribute to work properly.
This function is only useful in abstract modules, in order to make sure
that modules that inherit from this class correctly defines needed class
attributes.
Args:
variables: a string or an array of strings containing the name of
needed class attributes.
Raises:
ModuleInitializationError: One of the needed attributes is not
correctly defined.
"""
new = []
observables = list(iterify(observables))
for i, observable in enumerate(observables):
try:
t = Observable.guess_type(observable)
temp = t(value=observable)
temp.clean()
observable = temp.value
observables[i] = observable
for a in analyzers.get(t, []):
new.extend([n for n in a.analyze_string(observable) if n and n not in observables])
except ObservableValidationError:
pass
if len(new) == 0:
return observables
else:
return observables + derive(new)
def _make_threat_nodes(threat, context, tags):
# extract Url and Hash info
threats = dict()
if threat['threatStatus'] != 'active':
# FIXME, clear out false positive ?
log.warning("threatStatus %s for threat %s",
threat['threatStatus'], threat['threatID'])
log.debug(pprint.pformat(threat))
return None
log.debug('_make_threat_nodes for threat %s', threat['threatID'])
# threattype, classification
# url, phish: url leads to phishing page (threat is url)
# url, malware: url leads to malware download (threat is url, threatid is maybe sha256)
# attachment, malware: attachement is malware (threat is sha256)
# spam, url
if threat['threatType'] == 'url':
if threat['classification'] == 'phish':
pass # just keep the url
elif threat['classification'] == 'malware':
# get url and hash
threats['attachment'] = threat
elif threat['classification'] == 'spam':
log.info('URL threat - ignore classification %s',
threat['classification'])
else:
log.error('Type: url, Unsupported classification %s',
threat['classification'])
log.debug(pprint.pformat(threat))
return None
threats['url'] = threat
elif threat['threatType'] == 'attachment':
if threat['classification'] == 'malware':
threats['attachment'] = threat
else:
log.error('Type: attachment, Unsupported classification %s',
threat['classification'])
log.debug(pprint.pformat(threat))
return None
else:
log.error('Unsupported threatType %s classification %s',
threat['threatType'], threat['classification'])
log.debug(pprint.pformat(threat))
return None
# FIXME check if they exist already.
# if they do, do not parse the threat a second time ?
threat_nodes = []
if 'url' in threats:
#Proofpoint sometimes supplies a hostname marked as a Url.
#this relies on Yeti to determine the type/class and add act appropriately
threat_nodes.append(
Observable.guess_type(threats['url']['threat']).get_or_create(
value=threats['url']['threat'], context=[context]))
if 'attachment' in threats:
threat_nodes.append(
Hash.get_or_create(value=threats['attachment']['threatID'],
context=[context]))
for o in threat_nodes:
o.tag([t['name'] for t in tags])
return threat_nodes
def derive(observables):
"""Indicate that the module needs a specific attribute to work properly.
This function is only useful in abstract modules, in order to make sure
that modules that inherit from this class correctly defines needed class
attributes.
Args:
variables: a string or an array of strings containing the name of
needed class attributes.
Raises:
ModuleInitializationError: One of the needed attributes is not
correctly defined.
"""
new = []
for observable in iterify(observables):
try:
t = Observable.guess_type(observable)
for a in analyzers.get(t, []):
new.extend([
n for n in a.analyze_string(observable)
if n and n not in observables
])
except ObservableValidationError:
pass
if len(new) == 0:
return observables
else:
return derive(new + observables)
def derive(strings):
values = set()
observables = set()
for string in iterify(strings):
if string:
try:
t = Observable.guess_type(string)
observable = t(value=string)
observable.normalize()
observables.add(observable)
values.add(observable.value)
except ObservableValidationError:
values.add(string)
new = []
for observable in observables:
for a in analyzers.get(observable.__class__, []):
new.extend([
n for n in a.analyze_string(observable.value)
if n and n not in values
])
if len(new) == 0:
return values, values
else:
_, extended = derive(new + list(values))
return values, extended
def each(url):
try:
host = ProcessUrl.analyze_string(url.value)[0]
h = Observable.guess_type(host).get_or_create(value=host)
h.add_source("analytics")
Link.connect(src=url, dst=h)
except ObservableValidationError:
logging.error("An error occurred when trying to add {} to the database".format(host))
def each(url):
try:
host = ProcessUrl.analyze_string(url.value)[0]
h = Observable.guess_type(host).get_or_create(value=host)
h.add_source("analytics")
url.active_link_to(h, "hostname", "ProcessUrl", clean_old=False)
return h
except ObservableValidationError:
logging.error("An error occurred when trying to add {} to the database".format(host))
def each(url):
try:
host = ProcessUrl.analyze_string(url.value)[0]
h = Observable.guess_type(host).get_or_create(value=host)
h.add_source("analytics")
url.active_link_to(h, "hostname", "ProcessUrl", clean_old=False)
return h
except ObservableValidationError:
logging.error(
"An error occurred when trying to add {} to the database".
format(host))
def derive(observables):
if isinstance(observables, (str, unicode)):
observables = [observables]
new = []
for observable in observables:
t = Observable.guess_type(observable)
for a in analyzers.get(t, []):
new.extend([n for n in a.analyze_string(observable) if n and n not in observables])
if len(new) == 0:
return observables
else:
return derive(new + observables)
def derive(observables):
new = []
for observable in iterify(observables):
try:
t = Observable.guess_type(observable)
for a in analyzers.get(t, []):
new.extend([n for n in a.analyze_string(observable) if n and n not in observables])
except ObservableValidationError:
pass
if len(new) == 0:
return observables
else:
return derive(new + observables)
import sys
import logging
from os import path
YETI_ROOT = path.normpath(path.dirname(path.dirname(path.abspath(__file__))))
sys.path.append(YETI_ROOT)
from core.analytics import OneShotAnalytics
from core.scheduling import Scheduler
from core.observables import Observable
if __name__ == '__main__':
logging.basicConfig(level=logging.DEBUG)
Scheduler()
if len(sys.argv) == 1:
print("Re-run using a analytic name as argument")
for f in OneShotAnalytics.objects():
print(" {}".format(f.name))
if len(sys.argv) > 1:
name = sys.argv[1]
f = OneShotAnalytics.objects.get(name=name)
print("Running {}...".format(f.name))
observable = Observable.guess_type(
sys.argv[2]).get_or_create(value=sys.argv[2])
f.analyze(observable, {})
def match_observables(observables, save_matches=False, fetch_neighbors=True):
# Remove empty observables
observables, extended_query = derive(observables)
observables = list(observables)
data = {
"matches": [],
"unknown": set(observables),
"entities": {},
"known": [],
"neighbors": [],
}
# add to "known"
for o in Observable.objects(value__in=list(extended_query)):
data['known'].append(o.info())
del_from_set(data['unknown'], o.value)
if fetch_neighbors:
for link, node in (o.incoming()):
if isinstance(node, Observable):
if (link.src.value not in extended_query or link.dst.value
not in extended_query) and node.tags:
data['neighbors'].append((link.info(), node.info()))
for nodes in o.neighbors("Entity").values():
for l, node in nodes:
# add node name and link description to indicator
node_data = {
"entity": node.type,
"name": node.name,
"link_description": l.description
}
# uniquely add node information to related entitites
ent = data['entities'].get(node.name, node.info())
if 'matches' not in ent:
ent['matches'] = {"observables": []}
if 'observables' not in ent['matches']:
ent['matches']['observables'] = []
info = node.info()
o_info = o.info()
info['matched_observable'] = {
"value": o_info['value'],
"tags": [t['name'] for t in o_info['tags']],
"human_url": o_info['human_url'],
"url": o_info['url'],
"context": o_info['context']
}
if info not in ent['matches']['observables']:
ent['matches']['observables'].append(info)
data['entities'][node.name] = ent
# add to "matches"
for o, i in Indicator.search(extended_query):
if save_matches:
o = Observable.add_text(o)
else:
o = Observable.guess_type(o)(value=o)
try:
o.validate()
except ObservableValidationError:
pass
try:
o = Observable.objects.get(value=o.value)
except Exception:
pass
match = i.info()
match.update({
"observable": o.info() if o.id else o.value,
"related": [],
"suggested_tags": set()
})
for nodes in i.neighbors("Entity").values():
for l, node in nodes:
# add node name and link description to indicator
node_data = {
"entity": node.type,
"name": node.name,
"link_description": l.description
}
match["related"].append(node_data)
# uniquely add node information to related entitites
ent = data['entities'].get(node.name, node.info())
if 'matches' not in ent:
ent['matches'] = {"indicators": []}
if 'indicators' not in ent['matches']:
ent['matches']['indicators'] = []
info = i.info()
info['matched_observable'] = o.value
if info not in ent['matches']['indicators']:
ent['matches']['indicators'].append(info)
data['entities'][node.name] = ent
o_tags = o.get_tags()
for tag in node.generate_tags():
if tag not in o_tags:
match["suggested_tags"].add(tag)
data["matches"].append(match)
data['entities'] = data['entities'].values()
return data
def match_observables(observables, save_matches=False):
# Remove empty observables
observables = [observable for observable in observables if observable]
extended_query = set(observables) | set(derive(observables))
added_entities = set()
data = {
"matches": [],
"unknown": set(observables),
"entities": [],
"known": [],
"neighbors": [],
}
# add to "known"
for o in Observable.objects(value__in=list(extended_query)):
data['known'].append(o.info())
del_from_set(data['unknown'], o.value)
for link, node in (o.incoming()):
if isinstance(node, Observable):
if (link.src.value not in extended_query or link.dst.value not in extended_query) and node.tags:
data['neighbors'].append((link.info(), node.info()))
# add to "matches"
for o, i in Indicator.search(extended_query):
del_from_set(data["unknown"], o)
if save_matches:
o = Observable.add_text(o)
else:
o = Observable.guess_type(o)(value=o)
o.validate()
try:
o = Observable.objects.get(value=o.value)
except Exception:
pass
match = i.info()
match.update({"observable": o.info(), "related": [], "suggested_tags": set()})
for nodes in i.neighbors("Entity").values():
for l, node in nodes:
# add node name and link description to indicator
node_data = {"entity": node.type, "name": node.name, "link_description": l.description}
match["related"].append(node_data)
# uniquely add node information to related entitites
if node.name not in added_entities:
nodeinfo = node.info()
nodeinfo['type'] = node.type
data["entities"].append(nodeinfo)
added_entities.add(node.name)
o_tags = o.get_tags()
[match["suggested_tags"].add(tag) for tag in node.generate_tags() if tag not in o_tags]
data["matches"].append(match)
del_from_set(data["unknown"], o.value)
return data
def match_observables(observables, save_matches=False, fetch_neighbors=True):
# Remove empty observables
observables = [refang(observable) for observable in observables if observable]
extended_query = set(observables) | set(derive(observables))
data = {
"matches": [],
"unknown": set(observables),
"entities": {},
"known": [],
"neighbors": [],
}
# add to "known"
for o in Observable.objects(value__in=list(extended_query)):
data['known'].append(o.info())
del_from_set(data['unknown'], o.value)
if fetch_neighbors:
for link, node in (o.incoming()):
if isinstance(node, Observable):
if (link.src.value not in extended_query or link.dst.value not in extended_query) and node.tags:
data['neighbors'].append((link.info(), node.info()))
for nodes in o.neighbors("Entity").values():
for l, node in nodes:
# add node name and link description to indicator
node_data = {"entity": node.type, "name": node.name, "link_description": l.description}
# uniquely add node information to related entitites
ent = data['entities'].get(node.name, node.info())
if 'matches' not in ent:
ent['matches'] = {"observables": []}
if 'observables' not in ent['matches']:
ent['matches']['observables'] = []
info = node.info()
o_info = o.info()
info['matched_observable'] = {
"value": o_info['value'],
"tags": [t['name'] for t in o_info['tags']],
"human_url": o_info['human_url'],
"url": o_info['url']
}
if info not in ent['matches']['observables']:
ent['matches']['observables'].append(info)
data['entities'][node.name] = ent
# add to "matches"
for o, i in Indicator.search(extended_query):
if save_matches:
o = Observable.add_text(o)
else:
o = Observable.guess_type(o)(value=o)
try:
o.validate()
except ObservableValidationError:
pass
try:
o = Observable.objects.get(value=o.value)
except Exception:
pass
match = i.info()
match.update({"observable": o.info(), "related": [], "suggested_tags": set()})
for nodes in i.neighbors("Entity").values():
for l, node in nodes:
# add node name and link description to indicator
node_data = {"entity": node.type, "name": node.name, "link_description": l.description}
match["related"].append(node_data)
# uniquely add node information to related entitites
ent = data['entities'].get(node.name, node.info())
if 'matches' not in ent:
ent['matches'] = {"indicators": []}
if 'indicators' not in ent['matches']:
ent['matches']['indicators'] = []
info = i.info()
info['matched_observable'] = o.value
if info not in ent['matches']['indicators']:
ent['matches']['indicators'].append(info)
data['entities'][node.name] = ent
o_tags = o.get_tags()
[match["suggested_tags"].add(tag) for tag in node.generate_tags() if tag not in o_tags]
data["matches"].append(match)
data['entities'] = data['entities'].values()
return data
def _make_threat_nodes(threat, context, tags):
# extract Url and Hash info
threats = dict()
if threat['threatStatus'] != 'active':
# FIXME, clear out false positive ?
log.warning(
"threatStatus %s for threat %s", threat['threatStatus'],
threat['threatID'])
log.debug(pprint.pformat(threat))
return None
log.debug('_make_threat_nodes for threat %s', threat['threatID'])
# threattype, classification
# url, phish: url leads to phishing page (threat is url)
# url, malware: url leads to malware download (threat is url, threatid is maybe sha256)
# attachment, malware: attachement is malware (threat is sha256)
# spam, url
if threat['threatType'] == 'url':
if threat['classification'] == 'phish':
pass # just keep the url
elif threat['classification'] == 'malware':
# get url and hash
threats['attachment'] = threat
elif threat['classification'] == 'spam':
log.info(
'URL threat - ignore classification %s',
threat['classification'])
else:
log.error(
'Type: url, Unsupported classification %s',
threat['classification'])
log.debug(pprint.pformat(threat))
return None
threats['url'] = threat
elif threat['threatType'] == 'attachment':
if threat['classification'] == 'malware':
threats['attachment'] = threat
else:
log.error(
'Type: attachment, Unsupported classification %s',
threat['classification'])
log.debug(pprint.pformat(threat))
return None
else:
log.error(
'Unsupported threatType %s classification %s',
threat['threatType'], threat['classification'])
log.debug(pprint.pformat(threat))
return None
# FIXME check if they exist already.
# if they do, do not parse the threat a second time ?
threat_nodes = []
if 'url' in threats:
#Proofpoint sometimes supplies a hostname marked as a Url.
#this relies on Yeti to determine the type/class and add act appropriately
threat_nodes.append(
Observable.guess_type(threats['url']['threat']).get_or_create(
value=threats['url']['threat'], context=[context]))
if 'attachment' in threats:
threat_nodes.append(
Hash.get_or_create(
value=threats['attachment']['threatID'], context=[context]))
for o in threat_nodes:
o.tag([t['name'] for t in tags])
return threat_nodes
