2019-11-08 15:27:55 +00:00
|
|
|
#!/usr/bin/env python3
|
|
|
|
# -*-coding:UTF-8 -*
|
|
|
|
|
|
|
|
import os
|
|
|
|
import sys
|
|
|
|
import redis
|
|
|
|
|
|
|
|
|
|
|
|
sys.path.append(os.path.join(os.environ['AIL_BIN'], 'packages'))
|
|
|
|
import Item
|
|
|
|
|
|
|
|
|
|
|
|
import ConfigLoader
|
|
|
|
|
|
|
|
config_loader = ConfigLoader.ConfigLoader()
|
|
|
|
r_serv_metadata = config_loader.get_redis_conn("ARDB_Metadata")
|
|
|
|
config_loader = None
|
|
|
|
|
2019-11-12 16:08:52 +00:00
|
|
|
def get_decoded_item_type(sha1_string):
|
|
|
|
'''
|
|
|
|
Retun the estimed type of a given decoded item.
|
|
|
|
|
|
|
|
:param sha1_string: sha1_string
|
|
|
|
'''
|
|
|
|
return r_serv_metadata.hget('metadata_hash:{}'.format(sha1_string), 'estimated_type')
|
|
|
|
|
2019-11-14 16:05:58 +00:00
|
|
|
def get_decoded_metadata(sha1_string):
|
|
|
|
metadata_dict = {}
|
|
|
|
metadata_dict['first_seen'] = r_serv_metadata.hget('metadata_hash:{}'.format(sha1_string), 'first_seen')
|
|
|
|
metadata_dict['last_seen'] = r_serv_metadata.hget('metadata_hash:{}'.format(sha1_string), 'last_seen')
|
|
|
|
return metadata_dict
|
|
|
|
|
2019-11-13 09:20:11 +00:00
|
|
|
def get_decoded_items_list(sha1_string):
|
2019-11-08 15:27:55 +00:00
|
|
|
return r_serv_metadata.zrange('nb_seen_hash:{}'.format(sha1_string), 0, -1)
|
|
|
|
|
|
|
|
def get_item_decoded(item_id):
|
|
|
|
'''
|
|
|
|
Retun all decoded item of a given item id.
|
|
|
|
|
|
|
|
:param item_id: item id
|
|
|
|
'''
|
|
|
|
res = r_serv_metadata.smembers('hash_paste:{}'.format(item_id))
|
|
|
|
if res:
|
|
|
|
return list(res)
|
|
|
|
else:
|
|
|
|
return []
|
|
|
|
|
|
|
|
def get_domain_decoded_item(domain):
|
|
|
|
'''
|
|
|
|
Retun all decoded item of a given domain.
|
|
|
|
|
|
|
|
:param domain: crawled domain
|
|
|
|
'''
|
|
|
|
res = r_serv_metadata.smembers('hash_domain:{}'.format(domain))
|
|
|
|
if res:
|
|
|
|
return list(res)
|
|
|
|
else:
|
|
|
|
return []
|
|
|
|
|
2019-11-12 16:08:52 +00:00
|
|
|
def get_decoded_domain_item(sha1_string):
|
|
|
|
'''
|
|
|
|
Retun all domain of a given decoded item.
|
|
|
|
|
|
|
|
:param sha1_string: sha1_string
|
|
|
|
'''
|
|
|
|
res = r_serv_metadata.smembers('domain_hash:{}'.format(sha1_string))
|
|
|
|
if res:
|
|
|
|
return list(res)
|
|
|
|
else:
|
|
|
|
return []
|
|
|
|
|
2019-11-13 15:46:28 +00:00
|
|
|
def get_decoded_correlated_object(sha1_string, correlation_objects=[]):
|
|
|
|
'''
|
|
|
|
Retun all correlation of a given sha1.
|
|
|
|
|
|
|
|
:param sha1_string: sha1
|
|
|
|
:type sha1_string: str
|
|
|
|
|
|
|
|
:return: a dict of all correlation for a given sha1
|
|
|
|
:rtype: dict
|
|
|
|
'''
|
|
|
|
if correlation_objects is None:
|
|
|
|
correlation_objects = Correlation.get_all_correlation_objects()
|
|
|
|
decoded_correlation = {}
|
|
|
|
for correlation_object in correlation_objects:
|
|
|
|
if correlation_object == 'paste':
|
|
|
|
res = get_decoded_items_list(sha1_string)
|
|
|
|
elif correlation_object == 'domain':
|
|
|
|
res = get_decoded_domain_item(sha1_string)
|
|
|
|
else:
|
|
|
|
res = None
|
|
|
|
if res:
|
|
|
|
decoded_correlation[correlation_object] = res
|
|
|
|
return decoded_correlation
|
|
|
|
|
2019-11-08 15:27:55 +00:00
|
|
|
def save_domain_decoded(domain, sha1_string):
|
|
|
|
r_serv_metadata.sadd('hash_domain:{}'.format(domain), sha1_string) # domain - hash map
|
|
|
|
r_serv_metadata.sadd('domain_hash:{}'.format(sha1_string), domain) # hash - domain ma
|