def createRiskHashes(malSource, googleSource, ignore_domain_list=[]):
# ignore_domain_list is a list of domains which will be ignored
# (won't appear as malware/copyright domains) for evaluation flow
global malwareDomains
global copyrightDomains
clear_risk_hashes(
) #clear malwareDomains copyrightDomains dicts (might be full from optional prev run)
malware_file_path = '/home/michal/malware_dict.csv'
copyright_file_path = '/home/michal/copyright_dict.csv'
if os.path.exists(malware_file_path):
malwareDomains = gm.readDict(malware_file_path)
copyrightDomains = gm.readDict(copyright_file_path)
else: # First time- the files don't exist
createDomainsHashFromFile('malwareDomains', malSource)
#DEBUG: writeHashToFile('malwareDomains','/home/michal/Desktop/malwareDomainsForTest')
gm.saveDict(malware_file_path, malwareDomains)
createDomainsHashFromFile('copyrightDomains', googleSource)
#DEBUG: writeHashToFile('copyrightDomains','/home/michal/Desktop/copyrightDomainsForTest')
gm.saveDict(copyright_file_path, copyrightDomains)
if len(ignore_domain_list): #for evaluation cases- this won't be zero
for d in ignore_domain_list:
if d in malwareDomains:
malwareDomains.pop(d)
if d in copyrightDomains:
copyrightDomains.pop(d)
return
def create_clean_domains_file(src,dest):
d = {}
with open(src,"r") as f:
for line in f:
d.setdefault(trx.getDomainFromRequestedSite(line.rstrip()),None)
gm.saveDict(dest, d)
return
def plotting_fine_tuning():
aFile = '/home/michal/SALSA_files/tmp/real_run/salsa_a_dict_pickle'
hFile = '/home/michal/SALSA_files/tmp/real_run/salsa_h_dict_pickle'
a = gm.read_object_from_file(aFile)
h = gm.read_object_from_file(hFile)
gm.saveDict('/home/michal/SALSA_files/tmp/real_run/salsa_a_dict', a)
gm.saveDict('/home/michal/SALSA_files/tmp/real_run/salsa_h_dict', h)
print max(a.values())
gm.histogram_of_dict(a)
return
def writeGraphToFile(self,filePath):
graph_as_dict=nx.to_dict_of_dicts(self.dGraph)
print graph_as_dict
gm.saveDict(filePath, graph_as_dict)
return
def writeGraphToFile(self, filePath):
graph_as_dict = nx.to_dict_of_dicts(self.dGraph)
print graph_as_dict
gm.saveDict(filePath, graph_as_dict)
return
def writeDomainRiskDictToFile(self,fileName):
gm.saveDict(fileName, self.DRD)
return
def writeTransDictToFile(self,fileName):
gm.saveDict(fileName, self.TD)
return
def writeDomainRiskDictToFile(self,fileName):
gm.saveDict(fileName, self.DRD)
return
def writeTransDictToFile(self,fileName):
gm.saveDict(fileName, self.TD)
return