df=df.groupby("time").count()
index = df.index
lock.acquire()
if z=="malware":
for i in index:
malware_dict[i] = malware_dict.get(i,0)+int(df[df.index==i].iloc[0].sha256)
else:
for i in index:
benign_dict[i] = benign_dict.get(i,0)+int(df[df.index==i],iloc[0].sha256)
lock.release()
if __name__ == '__main__':
lock = Lock()
for z in [malware_path,benign_path]:
pool = Pool(16)
for i in range(16):
for j in range(16):
for k in range(16):
current_dir = z +'/'.join([hex(i),hex(j),hex(k)])+'/'
pool.appy_async(func=get_year_count,args=(current_dir,))
pool.close()
pool.join()
if z == malware_path:
malware_dict['type'] = ['malware' for i in range(len(malware_dict['year'])]
else:
benign_dict['type'] = ['benign' for i in range(len(benign_dict['year'])]
total = pd.concat([pd.DataFrame(malware_dict),pd.DataFrame(benign_dict)])
total.to_csv("/data/total_list.csv",index=False)
