Skip to content

M
meta_base_code
Commit a8259c6d authored by litaolemo's avatar litaolemo
Browse files
Options

update

parent 982295ac
Hide whitespace changes
Inline Side-by-side
Showing with 6 additions and 129 deletions
task/search_strategy_d.py
View file @ a8259c6d
...@@ -255,70 +255,10 @@ FROM ...@@ -255,70 +255,10 @@ FROM
)t2 )t2
on t1.cl_id=t2.device_id AND t1.partition_date = t2.partition_date on t1.cl_id=t2.device_id AND t1.partition_date = t2.partition_date
LEFT JOIN LEFT JOIN distinct_device_id_view
( on t1.partition_date=distinct_device_id_view.partition_date and t1.cl_id=distinct_device_id_view.device_id
SELECT DISTINCT device_id
FROM ml.ml_d_ct_dv_devicespam_d --去除机构刷单设备，即作弊设备（浏览和曝光事件去除）
WHERE partition_day={yesterday_str}
UNION ALL
SELECT DISTINCT device_id
FROM dim.dim_device_user_staff --去除内网用户
)spam_pv
on spam_pv.device_id=t1.cl_id
LEFT JOIN
(
SELECT partition_date,device_id
FROM
(--找出user_id当天活跃的第一个设备id
SELECT user_id,partition_date,
if(size(device_list) > 0, device_list [ 0 ], '') AS device_id
FROM online.ml_user_updates
WHERE partition_date>='${{yesterday_str}}' AND partition_date<'${{today_str}}'
)t1
JOIN
( --医生账号
SELECT distinct user_id
FROM online.tl_hdfs_doctor_view
WHERE partition_date = {yesterday_str}
--马甲账号/模特用户
UNION ALL
SELECT user_id
FROM ml.ml_c_ct_ui_user_dimen_d
WHERE partition_day = {yesterday_str}
AND (is_puppet = 'true' or is_classifyuser = 'true')
UNION ALL
--公司内网覆盖用户
select distinct user_id
from dim.dim_device_user_staff
UNION ALL
--登陆过医生设备
SELECT distinct t1.user_id
FROM
(
SELECT user_id, v.device_id as device_id
FROM online.ml_user_history_detail
LATERAL VIEW EXPLODE(device_history_list) v AS device_id
WHERE partition_date = {yesterday_str}
)t1
JOIN
(
SELECT device_id
FROM online.ml_device_history_detail
WHERE partition_date = {yesterday_str}
AND is_login_doctor = '1'
)t2
ON t1.device_id = t2.device_id
)t2
on t1.user_id=t2.user_id
group by partition_date,device_id
)dev
on t1.partition_date=dev.partition_date and t1.cl_id=dev.device_id
WHERE (spam_pv.device_id IS NULL or spam_pv.device_id ='') WHERE (spam_pv.device_id IS NULL or spam_pv.device_id ='')
and (dev.device_id is null or dev.device_id ='') and (distinct_device_id_view.device_id is null or distinct_device_id_view.device_id ='')
GROUP BY t1.partition_date,t2.active_type,device_os_type,channel GROUP BY t1.partition_date,t2.active_type,device_os_type,channel
)t )t
)t3 )t3
...@@ -408,76 +348,13 @@ LEFT JOIN ...@@ -408,76 +348,13 @@ LEFT JOIN
LATERAL VIEW explode(mas.active_type) t2 AS active_type LATERAL VIEW explode(mas.active_type) t2 AS active_type
)dev )dev
on t1.cl_id=dev.device_id and t1.partition_date = dev.partition_date on t1.cl_id=dev.device_id and t1.partition_date = dev.partition_date
LEFT JOIN LEFT JOIN distinct_device_id_view
( on t1.partition_date=distinct_device_id_view.partition_date and t1.cl_id=distinct_device_id_view.device_id
SELECT DISTINCT device_id
FROM ml.ml_d_ct_dv_devicespam_d --去除机构刷单设备，即作弊设备（浏览和曝光事件去除）
WHERE partition_day={yesterday_str}
UNION ALL
SELECT DISTINCT device_id
FROM dim.dim_device_user_staff --去除内网用户
)spam_pv
on spam_pv.device_id=t1.cl_id
LEFT JOIN
(
SELECT partition_date,device_id
FROM
(--找出user_id当天活跃的第一个设备id
SELECT user_id,partition_date,
if(size(device_list) > 0, device_list [ 0 ], '') AS device_id
FROM online.ml_user_updates
WHERE partition_date>='${{yesterday_str}}' AND partition_date<'${{today_str}}'
)t1
JOIN
( --医生账号
SELECT distinct user_id
FROM online.tl_hdfs_doctor_view
WHERE partition_date = {yesterday_str}
--马甲账号/模特用户
UNION ALL
SELECT user_id
FROM ml.ml_c_ct_ui_user_dimen_d
WHERE partition_day = {yesterday_str}
AND (is_puppet = 'true' or is_classifyuser = 'true')
UNION ALL
--公司内网覆盖用户
select distinct user_id
from dim.dim_device_user_staff
UNION ALL
--登陆过医生设备
SELECT distinct t1.user_id
FROM
(
SELECT user_id, v.device_id as device_id
FROM online.ml_user_history_detail
LATERAL VIEW EXPLODE(device_history_list) v AS device_id
WHERE partition_date = {yesterday_str}
)t1
JOIN
(
SELECT device_id
FROM online.ml_device_history_detail
WHERE partition_date = {yesterday_str}
AND is_login_doctor = '1'
)t2
ON t1.device_id = t2.device_id
)t2
on t1.user_id=t2.user_id
group by t1.partition_date,device_id
)dev1
on t1.partition_date=dev1.partition_date and t1.cl_id=dev1.device_id
WHERE (spam_pv.device_id IS NULL or spam_pv.device_id ='') WHERE (spam_pv.device_id IS NULL or spam_pv.device_id ='')
and (dev1.device_id is null or dev1.device_id ='') and (distinct_device_id_view.device_id is null or distinct_device_id_view.device_id ='')
GROUP BY t1.partition_date,active_type,device_os_type,channel GROUP BY t1.partition_date,active_type,device_os_type,channel
)t4 )t4
on t3.partition_date=t4.partition_date and t3.active_type=t4.active_type and t3.device_os_type = t4.device_os_type AND t3.channel = t4.channel on t3.partition_date=t4.partition_date and t3.active_type=t4.active_type and t3.device_os_type = t4.device_os_type AND t3.channel = t4.channel
""".format(today_str=today_str, yesterday_str=yesterday_str, ) """.format(today_str=today_str, yesterday_str=yesterday_str, )
device_df = spark.sql(sql) device_df = spark.sql(sql)
device_df.show(1, False) device_df.show(1, False)
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment