to_ck/clear_up.py.template
2021-08-16 18:07:38 +08:00

49 lines
1.3 KiB
Plaintext

"""
ÇåÀí²âÊÔÊý¾Ý
"""
import json
from settings import settings
game = ''
db = settings.GAME
server_list_url = f'http://gametools.legu.cc/?app=api&act=getServerList&game={game}'
import pandas as pd
from clickhouse_driver import Client
client = Client(**settings.CK_CONFIG)
df = pd.read_json(server_list_url)
df = df[~df['hostname'].isin(['119.3.89.14', '119.3.105.109'])]
serverid = tuple((str(i) for i in df['serverid'].to_list()))
sql = f"""select `#account_id`, `#event_time`, `#event_name`,`#os`,`svrindex`, count() as n
from {db}.event
where
`#event_time`>addDays(now('UTC'),-3) and (
lower(`#os`) = 'windows'
or svrindex not in {serverid}
)
group by `#account_id`, `#event_time`, `#event_name`,`#os`,`svrindex`
having n = 1 limit 2000"""
data, columns = client.execute(
sql, columnar=True, with_column_types=True
)
if not data:
exit(0)
data_df = pd.DataFrame(
{col[0]: d for d, col in zip(data, columns)}
)
data_df.drop('n', axis=1, inplace=True)
data_df['sign'] = -1
data_df['#event_time'] = data_df['#event_time'].apply(lambda x: x.strftime('%Y-%m-%d %H:%M:%S'))
insert_sql = f'INSERT INTO {db}.event FORMAT JSONEachRow '
insert_sql = insert_sql + '\n'.join([json.dumps(item) for item in data_df.T.to_dict().values()])
client.execute(insert_sql)