update

task/search_strategy_d.py

@@ -91,64 +91,72 @@ for t in range(0, task_days):
 
     print("-------------------------------")
 
-    sql = r"""SELECT
-            ,t3.device_os_type as device_type
+    sql = r"""
+   SELECT    t3.device_os_type as device_type
             ,t3.active_type as active_type
             ,t3.channel as channel_type
             ,NVL(t3.search_pv,0) as pv
             ,NVL(t3.search_uv,0) as uv
+            ,NVL(t4.hexin_card_click_pv,0) as search_core_pv
+            ,NVL(t4.neirong_card_click_pv,0) as search_pv
     FROM
     (
-        SELECT active_type,device_os_type,channel,search_pv,search_uv FROM
+        SELECT query,active_type,device_os_type,channel,search_pv,search_uv
+        FROM
         (
-            SELECT active_type,device_os_type,channel
+            SELECT query,active_type,device_os_type,channel
                     ,count(t1.cl_id) as search_pv
                     ,count(distinct t1.cl_id) as search_uv
             FROM
             (
-                SELECT partition_date,cl_id
+                SELECT partition_date
+                        ,params['query'] as query
+                        ,cl_id
                 FROM online.bl_hdfs_maidian_updates
-                WHERE partition_date >= '{yesterday_str}'
-                AND partition_date < '{today_str}'
+                WHERE partition_date >= {yesterday_str} 
+                AND partition_date < {today_str}
                 AND action in ('do_search','search_result_click_search')
-
+    
                 UNION ALL
-                SELECT cl_id
+                SELECT partition_date,params['query'] as query,cl_id
                 FROM online.bl_hdfs_maidian_updates
-                WHERE partition_date >= '{yesterday_str}'
-                AND partition_date < '{today_str}'
+                WHERE partition_date >= {yesterday_str} 
+                AND partition_date < {today_str}
                 AND action = 'on_click_card'
                 AND params['page_name']='search_home'
-
+    
                 UNION ALL
                 SELECT partition_date
+                        ,params['card_name'] as query
                         ,cl_id
                 FROM online.bl_hdfs_maidian_updates
-                WHERE partition_date >= '{yesterday_str}'
-                AND partition_date < '{today_str}'
+                WHERE partition_date >= {yesterday_str} 
+                AND partition_date < {today_str}
                 AND action = 'on_click_card'
                 AND params['in_page_pos']='猜你喜欢'
                 AND params['tab_name']='精选'
                 AND params['card_type']='search_word'
-
-
+               
+    
                 UNION ALL
                 SELECT partition_date
+                        ,params['card_name'] as query
                         ,cl_id
                 FROM online.bl_hdfs_maidian_updates
-                WHERE partition_date >= '{yesterday_str}'
-                AND partition_date < '{today_str}'
+                WHERE partition_date >= {yesterday_str} 
+                AND partition_date < {today_str}
                 AND action = 'on_click_card'
                 AND page_name='welfare_home'
                 AND params['card_type'] ='search_word'
                 AND params['in_page_pos']='大家都在搜'
-
+    
                 UNION ALL
                 SELECT partition_date
+                        ,params['card_name'] as query
                         ,cl_id
                 FROM online.bl_hdfs_maidian_updates
-                WHERE partition_date >= '{yesterday_str}'
-                AND partition_date < '{today_str}'
+                WHERE partition_date >= {yesterday_str} 
+                AND partition_date < {today_str}
                 AND int(split(app_version,'\\.')[1]) >= 27
                 AND action='on_click_card'
                 AND params['card_type']='highlight_word'
@@ -161,18 +169,18 @@ for t in range(0, task_days):
                     SELECT
                     partition_date,m.device_id
                     ,array(device_os_type ,'合计') as device_os_type
-                    ,array(case WHEN active_type = '4'  THEN '老活'
-                          WHEN active_type  in ('1','2')  then '新增' END ,'合计') as active_type
+                    ,array(case WHEN active_type = '4'  THEN '老活' 
+                          WHEN active_type  in ('1','2')  then '新增' END ,'合计') as active_type 
                     ,array(CASE WHEN is_ai_channel = 'true' THEN 'AI'  ELSE '其他' END , '合计') as channel
                     FROM online.ml_device_day_active_status m
-                        LEFT JOIN
+                        LEFT JOIN 
                         (SELECT code,is_ai_channel,partition_day
                          FROM DIM.DIM_AI_CHANNEL_ZP_NEW
-                         WHERE partition_day>= '{yesterday_str}'
-                         AND partition_day < '{today_str}') tmp
+                         WHERE partition_day>= {yesterday_str}  
+                         AND partition_day < {today_str}) tmp
                         ON  m.partition_date=tmp.partition_day AND first_channel_source_type=code
-                    WHERE partition_date >= '{yesterday_str}'
-                    AND partition_date < '{today_str}'
+                    WHERE partition_date >= {yesterday_str} 
+                    AND partition_date < {today_str}
                     AND active_type in ('1','2','4')
                 ) mas
                 LATERAL VIEW explode(mas.channel) t2 AS channel
@@ -180,9 +188,7 @@ for t in range(0, task_days):
                 LATERAL VIEW explode(mas.active_type) t2 AS active_type
             )t2
             on t1.cl_id=t2.device_id AND t1.partition_date = t2.partition_date
-    GROUP BY active_type,device_os_type,channel
-        )t
-    )t3
+            GROUP BY active_type,device_os_type,channel
    """.format(today_str=today_str, yesterday_str=yesterday_str, )
     device_df = spark.sql(sql)
     device_df.show(1, False)