def extract_features(file_path):
result = {}
try:
a = APK(file_path)
d = DalvikVMFormat(a.get_dex())
dx = Analysis(d)
vm = dvm.DalvikVMFormat(a.get_dex())
vmx = analysis.Analysis(vm)
d.set_vmanalysis(dx)
d.set_decompiler(DecompilerDAD(d, dx))
except Exception as e:
print e
return None
result['android_version_code'] = a.get_androidversion_code()
result['android_version_name'] = a.get_androidversion_name()
result['max_sdk'] = a.get_max_sdk_version()
result['min_sdk'] = a.get_min_sdk_version()
result['libraries'] = a.get_libraries()
result['filename'] = a.get_filename()
result['target_sdk'] = a.get_target_sdk_version()
result['md5'] = hashlib.md5(a.get_raw()).hexdigest()
result['sha256'] = hashlib.sha256(a.get_raw()).hexdigest()
result['permissions'] = a.get_permissions()
result['activities'] = a.get_activities()
result['providers'] = a.get_providers()
result['services'] = a.get_services()
result['strings'] = d.get_strings()
result['class_names'] = [c.get_name() for c in d.get_classes()]
result['method_names'] = [m.get_name() for m in d.get_methods()]
result['field_names'] = [f.get_name() for f in d.get_fields()]
# result['is_native_code'] = 1 if analysis.is_native_code(dx) else 0
result['is_obfuscation'] = 1 if analysis.is_ascii_obfuscation(d) else 0
# result['is_crypto_code'] = 1 if analysis.is_crypto_code(dx) else 0
# result['is_dyn_code'] = 1 if analysis.is_dyn_code(dx) else 0
# result['is_reflection_code'] = 1 if analysis.is_reflection_code(vmx) else 0
result['is_database'] = 1 if d.get_regex_strings(DB_REGEX) else 0
s_list = []
s_list.extend(result['class_names'])
s_list.extend(result['method_names'])
s_list.extend(result['field_names'])
result['entropy_rate'] = entropy_rate(s_list)
result['feature_vectors'] = {}
result['feature_vectors']['api_calls'] = []
for call in API_CALLS:
status = 1 if dx.get_method_by_name(".", call, ".") else 0
result['feature_vectors']['api_calls'].append(status)
result['feature_vectors']['permissions'] = []
for permission in PERMISSIONS:
status = 1 if permission in result['permissions'] else 0
result['feature_vectors']['permissions'].append(status)
result['feature_vectors']['special_strings'] = []
for word in SPECIAL_STRINGS:
status = 1 if d.get_regex_strings(word) else 0
result['feature_vectors']['special_strings'].append(status)
result['feature_vectors']['others'] = [
# result['is_reflection_code'],
# result['is_crypto_code'],
# result['is_native_code'],
result['is_obfuscation'],
result['is_database'],
# result['is_dyn_code']
]
return result
def extract_features(file_path):
result = {}
try:
a = APK(file_path)
d = DalvikVMFormat(a.get_dex())
dx = Analysis(d)
vm = dvm.DalvikVMFormat(a.get_dex())
#vmx = analysis.uVMAnalysis(vm)
d.set_vmanalysis(dx)
d.set_decompiler(DecompilerDAD(d, dx))
except:
return None
result['android_version_code'] = a.get_androidversion_code()
result['android_version_name'] = a.get_androidversion_name()
result['max_sdk'] = a.get_max_sdk_version()
result['min_sdk'] = a.get_min_sdk_version()
#result['libraries'] = a.get_libraries()
result['filename'] = a.get_filename()
result['target_sdk'] = a.get_target_sdk_version()
result['md5'] = hashlib.md5(a.get_raw()).hexdigest()
result['sha256'] = hashlib.sha256(a.get_raw()).hexdigest()
result['permissions'] = a.get_permissions()
result['activities'] = a.get_activities()
result['providers'] = a.get_providers()
result['services'] = a.get_services()
result['strings'] = d.get_strings()
result['class_names'] = [c.get_name() for c in d.get_classes()]
result['method_names'] = [m.get_name() for m in d.get_methods()]
result['field_names'] = [f.get_name() for f in d.get_fields()]
#result['is_native_code'] = 1 if analysis.is_native_code(dx) else 0
result['is_obfuscation'] = 1 if analysis.is_ascii_obfuscation(d) else 0
#result['is_crypto_code'] = 1 if analysis.is_crypto_code(dx) else 0
'''result['is_dyn_code'] = 1 if analysis.is_dyn_code(dx) else 0
result['is_reflection_code'] = 1 if analysis.is_reflection_code(vmx) else 0'''
result['is_database'] = 1 if d.get_regex_strings(DB_REGEX) else 0
arr = []
s = a.get_elements("action", "name")
for i in s:
arr.append(i)
result['intents'] = arr
s_list = []
s_list.extend(result['class_names'])
s_list.extend(result['method_names'])
s_list.extend(result['field_names'])
result['entropy_rate'] = entropy_rate(s_list)
result['feature_vectors'] = {}
# Search for the presence of api calls in a given apk
result['feature_vectors']['api_calls'] = []
for call in API_CALLS:
status = 1 if dx.get_method(call) else 0
result['feature_vectors']['api_calls'].append(status)
# Search for the presence of permissions in a given apk
result['feature_vectors']['permissions'] = []
for permission in PERMISSIONS:
status = 1 if permission in result['permissions'] else 0
result['feature_vectors']['permissions'].append(status)
#Search for the presence of intents in a given apk
result['feature_vectors']['intents'] = []
n = len(INTENTS)
m = len(result['intents'])
for i in range(n):
stri = INTENTS[i]
flg = False
for j in range(m):
if stri in result['intents'][j]:
flg = True
break
if flg:
status = 1
else:
status = 0
result['feature_vectors']['intents'].append(status)
#Check for special strings in code
result['feature_vectors']['special_strings'] = []
for word in SPECIAL_STRINGS:
status = 1 if d.get_regex_strings(word) else 0
result['feature_vectors']['special_strings'].append(status)
return result
def extract_features(file_path):
result = {}
try:
a = APK(file_path)
d = DalvikVMFormat(a.get_dex())
dx = VMAnalysis(d)
vm = dvm.DalvikVMFormat(a.get_dex())
vmx = analysis.uVMAnalysis(vm)
d.set_vmanalysis(dx)
d.set_decompiler(DecompilerDAD(d, dx))
except:
return None
result['android_version_code'] = a.get_androidversion_code()
result['android_version_name'] = a.get_androidversion_name()
result['max_sdk'] = a.get_max_sdk_version()
result['min_sdk'] = a.get_min_sdk_version()
result['libraries'] = a.get_libraries()
result['filename'] = a.get_filename()
result['target_sdk'] = a.get_target_sdk_version()
result['md5'] = hashlib.md5(a.get_raw()).hexdigest()
result['sha256'] = hashlib.sha256(a.get_raw()).hexdigest()
result['permissions'] = a.get_permissions()
result['activities'] = a.get_activities()
result['providers'] = a.get_providers()
result['services'] = a.get_services()
#result['strings'] = d.get_strings()
#result['class_names'] = [c.get_name() for c in d.get_classes()]
#result['method_names'] = [m.get_name() for m in d.get_methods()]
#result['field_names'] = [f.get_name() for f in d.get_fields()]
class_names = [c.get_name() for c in d.get_classes()]
method_names = [m.get_name() for m in d.get_methods()]
field_names = [ f.get_name() for f in d.get_fields()]
result['is_native_code'] = 1 if analysis.is_native_code(dx) else 0
result['is_obfuscation'] = 1 if analysis.is_ascii_obfuscation(d) else 0
result['is_crypto_code'] = 1 if analysis.is_crypto_code(dx) else 0
result['is_dyn_code'] = 1 if analysis.is_dyn_code(dx) else 0
result['is_reflection_code'] = 1 if analysis.is_reflection_code(vmx) else 0
result['is_database'] = 1 if d.get_regex_strings(DB_REGEX) else 0
s_list = []
#s_list.extend(result['class_names'])
#s_list.extend(result['method_names'])
#s_list.extend(result['field_names'])
s_list.extend(class_names)
s_list.extend(method_names)
s_list.extend(method_names)
result['entropy_rate'] = entropy_rate(s_list)
result['feature_vectors'] = {}
# Search for the presence of api calls in a given apk
result['feature_vectors']['api_calls'] = []
for call in API_CALLS:
status = 1 if dx.tainted_packages.search_methods(".", call, ".") else 0
result['feature_vectors']['api_calls'].append(status)
# Search for the presence of permissions in a given apk
result['feature_vectors']['permissions'] = []
for permission in PERMISSIONS:
status = 1 if permission in result['permissions'] else 0
result['feature_vectors']['permissions'].append(status)
result['feature_vectors']['special_strings'] = []
for word in SPECIAL_STRINGS:
status = 1 if d.get_regex_strings(word) else 0
result['feature_vectors']['special_strings'].append(status)
opt_seq = []
for m in d.get_methods():
for i in m.get_instructions():
opt_seq.append(i.get_name())
optngramlist = [tuple(opt_seq[i:i+NGRAM]) for i in xrange(len(opt_seq) - NGRAM)]
optngram = Counter(optngramlist)
optcodes = dict()
tmpCodes = dict(optngram)
#for k,v in optngram.iteritems():
# if v>=NGRAM_THRE:
#optcodes[str(k)] = v
# optcodes[str(k)] = 1
tmpCodes = sorted(tmpCodes.items(),key =lambda d:d[1],reverse=True)
for value in tmpCodes[:NGRAM_THRE]:
optcodes[str(value[0])] = 1
result['feature_vectors']['opt_codes'] = optcodes
return result
