update

task/search_strategy_d.py

@@ -75,33 +75,11 @@ for t in range(0, task_days):
     sql = """SELECT t1.query as query
             ,t1.device_os_type as device_type
             ,t1.active_type as active_type
-            ,t1.channel as channel_type
-            ,NVL(t1.search_pv,0) as 30_pv
-            ,NVL(t1.search_uv,0) as 30_uv
-            ,if(NVL(t3.search_uv,0) <> 0 ,concat(cast((NVL(t4.hexin_card_click_pv,0)/NVL(t3.search_uv,0)) as decimal(18,2)),'') , '-') as 1_core_pv_division_uv
-            ,if(NVL(t3.search_uv,0) <> 0 ,concat(cast((NVL(t4.neirong_card_click_pv,0)/NVL(t3.search_uv,0)) as decimal(18,2)),'') , '-') as 1_pv_division_uv
-    
-            ,if(NVL(t5.search_uv,0) <> 0 ,concat(cast((NVL(t6.hexin_card_click_pv,0)/NVL(t5.search_uv,0)) as decimal(18,2)),'') , '-') as 7_core_pv_division_uv
-            ,if(NVL(t5.search_uv,0) <> 0 ,concat(cast((NVL(t6.neirong_card_click_pv,0)/NVL(t5.search_uv,0)) as decimal(18,2)),'') , '-') as 7_pv_division_uv
-    
-            ,if(NVL(t1.search_uv,0) <> 0 ,concat(cast((NVL(t2.hexin_card_click_pv,0)/NVL(t1.search_uv,0)) as decimal(18,2)),'') , '-') as 30_core_pv_division_uv
-            ,if(NVL(t1.search_uv,0) <> 0 ,concat(cast((NVL(t2.neirong_card_click_pv,0)/NVL(t1.search_uv,0)) as decimal(18,2)),'') , '-') as 30_pv_division_uv
-    
-    
-            ,NVL(t2.hexin_card_click_pv,0) as 30_search_core_pv
-            ,NVL(t2.neirong_card_click_pv,0) as 30_search_pv
-    
+            ,t1.channel as channel_type  
             ,NVL(t3.search_pv,0) as 1_pv
             ,NVL(t3.search_uv,0) as 1_uv
-    
             ,NVL(t4.hexin_card_click_pv,0) as 1_search_core_pv
             ,NVL(t4.neirong_card_click_pv,0) as 1_search_pv
-    
-            ,NVL(t5.search_pv,0) as 7_pv
-            ,NVL(t5.search_uv,0) as 7_uv
-    
-            ,NVL(t6.hexin_card_click_pv,0) as 7_search_core_pv
-            ,NVL(t6.neirong_card_click_pv,0) as 7_search_pv
     FROM
     (
         SELECT query,active_type,device_os_type,channel,search_pv,search_uv
@@ -634,8 +612,17 @@ for t in range(0, task_days):
     order by 30_pv desc,query""".format(last_30_day_str=last_30_day_str, today_str=today_str,
                                                  yesterday_str=yesterday_str, one_week_age_str=one_week_age_str)
     device_df = spark.sql(sql)
+    device_df.createOrReplaceTempView("data_table")
+
+
+    collects_sql = """
+    SELECT *,if(NVL(sum(1_uv),0) <> 0 ,concat(cast((NVL(sum(1_search_core_pv),0)/NVL(sum(1_uv),0)) as decimal(18,2)),'') , '-') as 1_core_pv_division_uv 
+    ,if(NVL(sum(1_uv),0) <> 0 ,concat(cast((NVL(sum(1_search_pv),0)/NVL(sum(1_uv),0)) as decimal(18,2)),'') , '-') as 1_pv_division_uv
+    FROM data_table GROUP BY device_type,active_type,channel_type
+    """
+    finnal_df = spark.sql(collects_sql)
 
-    device_df.show(1, False)
-    sql_res = device_df.collect()
+    finnal_df.show(1, False)
+    sql_res = finnal_df.collect()
     for res in sql_res:
         print(res)