SELECT NVL(t2.partition_date,t3.partition_date) as partition_date
SELECT NVL(t2.partition_date,t3.partition_date) as partition_date
,NVL(t2.cl_id,t3.cl_id) as cl_id
,NVL(t2.cl_id,t3.cl_id) as cl_id
,NVL(t2.query,t3.query) as query
,NVL(t2.pv,0) as hexin
,NVL(t2.pv,0) as hexin
,NVL(t3.pv,0) as neirong
,NVL(t3.pv,0) as neirong
FROM
FROM
(
(--核心卡片点击
SELECT partition_date
SELECT partition_date
,params['query'] as query
,cl_id
,cl_id
,count(1) as pv
,count(1) as pv
FROM online.bl_hdfs_maidian_updates
FROM online.bl_hdfs_maidian_updates
...
@@ -261,11 +335,13 @@ for t in range(0, task_days):
...
@@ -261,11 +335,13 @@ for t in range(0, task_days):
or (action = 'on_click_card' AND params['card_content_type'] in ('service','hospital','doctor') AND page_name in ('search_result_more','search_result_welfare','search_result_hospital','search_result_doctor'))
or (action = 'on_click_card' AND params['card_content_type'] in ('service','hospital','doctor') AND page_name in ('search_result_more','search_result_welfare','search_result_hospital','search_result_doctor'))
or (action = 'on_click_button' AND params['button_name'] = 'check_plan' AND page_name = 'search_result_more'))
or (action = 'on_click_button' AND params['button_name'] = 'check_plan' AND page_name = 'search_result_more'))
GROUP BY partition_date
GROUP BY partition_date
,params['query']
,cl_id
,cl_id
)t2
)t2
FULL JOIN
FULL JOIN
(
(--内容卡片点击
SELECT partition_date
SELECT partition_date
,params['query'] as query
,cl_id
,cl_id
,count(1) as pv
,count(1) as pv
FROM online.bl_hdfs_maidian_updates
FROM online.bl_hdfs_maidian_updates
...
@@ -275,9 +351,11 @@ for t in range(0, task_days):
...
@@ -275,9 +351,11 @@ for t in range(0, task_days):
AND page_name in ('search_result_more','search_result_diary','search_result_post'))
AND page_name in ('search_result_more','search_result_diary','search_result_post'))
or (action = 'on_click_card' AND params['card_content_type'] in ('answer','diary') AND page_name in ('search_result_more','search_result_diary','search_result_question_answer')))
or (action = 'on_click_card' AND params['card_content_type'] in ('answer','diary') AND page_name in ('search_result_more','search_result_diary','search_result_question_answer')))
GROUP BY partition_date
GROUP BY partition_date
,params['query']
,cl_id
,cl_id
)t3
)t3
on t3.partition_date=t2.partition_date
on t3.partition_date=t2.partition_date
AND t3.query=t2.query
AND t3.cl_id=t2.cl_id
AND t3.cl_id=t2.cl_id
)t1
)t1
JOIN
JOIN
...
@@ -301,15 +379,90 @@ for t in range(0, task_days):
...
@@ -301,15 +379,90 @@ for t in range(0, task_days):
WHERE partition_date >= {yesterday_str}
WHERE partition_date >= {yesterday_str}
AND partition_date < {today_str}
AND partition_date < {today_str}
AND active_type in ('1','2','4')
AND active_type in ('1','2','4')
AND first_channel_source_type not in ('yqxiu1','yqxiu2','yqxiu3','yqxiu4','yqxiu5','mxyc1','mxyc2','mxyc3'
AND first_channel_source_type not like 'promotion\_jf\_%'
) mas
) mas
LATERAL VIEW explode(mas.channel) t2 AS channel
LATERAL VIEW explode(mas.channel) t2 AS channel
LATERAL VIEW explode(mas.device_os_type) t2 AS device_os_type
LATERAL VIEW explode(mas.device_os_type) t2 AS device_os_type
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
GROUP BY active_type,device_os_type,channel
LEFT JOIN
)t4
(
on t3.active_type=t4.active_type and t3.device_os_type = t4.device_os_type AND t3.channel = t4.channel group by t3.active_type , t3.device_os_type ,t3.channel
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>='${start_date}' AND partition_date<'${end_date}'
)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 ='')
and (dev.device_id is null or dev.device_id ='')
GROUP BY t1.partition_date,active_type,device_os_type,channel
)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
