def sensitive_func(index_name, ts):
bulk_action = []
query_body = {
'query': {
'match_all': {}
},
'size': 999,
}
res = es.search(index=index_name, doc_type='text',
body=query_body)['hits']['hits']
for r in res:
_id = r['_id']
uid = r['_source']['uid']
mid = ''
if r['_source'].has_key('mid'):
mid = r['_source']['mid']
text = ''
if r['_source'].has_key('text'):
text = r['_source']['text']
sensitive_info = get_sensitive_info(ts, mid, text)
sensitive_user = get_sensitive_user(ts, uid)
item = {
'sensitive_info': sensitive_info,
'sensitive_user': sensitive_user,
}
action = {'update': {'_id': _id}}
bulk_action.extend([action, {'doc': item}])
if bulk_action:
print es.bulk(bulk_action,
index=index_name,
doc_type='text',
timeout=600)
def sensitive_func(ts, text, uid):
sensitive_info = get_sensitive_info(ts, text)
sensitive_user = get_sensitive_user(ts, uid)
return sensitive_info, sensitive_user
def sensitive_func(ts, text, uid):
sensitive_info = get_sensitive_info(timestamp=ts, text=text)
sensitive_user = get_sensitive_user(timestamp=ts, uid=uid)
return sensitive_info, sensitive_user