update

task/tractate_analysis_in_7000.py

@@ -107,9 +107,8 @@ one_week_age_str = (now + datetime.timedelta(days=-7)).strftime("%Y%m%d")
 device_id_dict = {}
 
 huidu_device_id_sql = r"""
-select t2.device_id from 
-(select first_device as device_id from online.ml_user_history_detail where partition_date = {today_str} and  substr(md5(first_device),-1) in ('8', '9', 'a', 'b')) t2
-
+select t1.device_id from 
+(select distinct(first_device) as device_id from online.ml_user_history_detail where partition_date = {today_str} and  substr(md5(first_device),-1) in ('8', '9', 'a', 'b') and last_active_date >= {last_30_day_str}) t2
  LEFT JOIN
     (
         select distinct device_id
@@ -176,7 +175,7 @@ select t2.device_id from
 WHERE spam_pv.device_id IS NULL
   and dev.device_id is null
   
-""".format(today_str=today_str)
+""".format(today_str=today_str,last_30_day_str=last_30_day_str)
 
 print(huidu_device_id_sql)
 huidu_device_id_df = spark.sql(huidu_device_id_sql)