update

task/spark_test.py

@@ -154,36 +154,108 @@ print(huidu_device_id_sql)
 huidu_device_id_df = spark.sql(huidu_device_id_sql)
 huidu_device_id_df.createOrReplaceTempView("dev_view")
 sql_search_ctr = r"""
-   SELECT   query,count(query)
---             click.partition_date as partition_date
---             ,active_type
---             ,device_os_type
---             ,channel
---             ,count(click.cl_id) as all_search_click_pv_724
---             ,count(distinct click.cl_id) as all_search_click_uv_724
+SELECT query,search_pv,search_uv
 	FROM
 	(
+		SELECT query,
+				,count(t1.cl_id) as search_pv
+				,count(distinct t1.cl_id) as search_uv
+		FROM
+		(
+			SELECT partition_date
+					,params['query'] as query
+					,cl_id
+			FROM online.bl_hdfs_maidian_updates
+			WHERE partition_date >= start_date 
+            AND partition_date < end_date
+			AND ((action = 'do_search' AND params['input_type']<>'everyone_watch') or action='search_result_click_search')
 
-            SELECT params['query'] as query
+			UNION ALL
+            SELECT partition_date,params['query'] as query,cl_id
             FROM online.bl_hdfs_maidian_updates
-            WHERE partition_date >= '{start_date}'  
-              AND partition_date < '{end_date}'  
+            WHERE partition_date >= start_date 
+          	AND partition_date < end_date
           	AND action = 'do_search'
-              AND params['input_type'] = '详情页默认词'
+            and params['input_type']='everyone_watch'
+            and params['tab']='精选'
+            and page_name='home'
+            AND params['query'] not in ('AI测颜值','AI测肤质')  --这两个词不跳转搜索结果页 
 
-            union all 
-            SELECT params['card_name'] as query
+			UNION ALL
+            SELECT partition_date,params['query'] as query,cl_id
             FROM online.bl_hdfs_maidian_updates
-            WHERE partition_date >= '{start_date}'  
-              AND partition_date < '{end_date}'  
+            WHERE partition_date >= start_date 
+          	AND partition_date < end_date
+          	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 >= start_date 
+            AND partition_date < end_date
 			AND action = 'on_click_card'
 			AND params['in_page_pos']='猜你喜欢'
-              AND params['tab_name']='精选'
+			--AND params['tab_name']='精选'
 			AND params['card_type']='search_word'
+			AND params['card_name'] not in ('AI测颜值','AI测肤质')  --这两个词不跳转搜索结果页 
 			--AND page_name='home' android的page_name为空
 
+			UNION ALL
+			SELECT partition_date
+					,params['card_name'] as query
+					,cl_id
+			FROM online.bl_hdfs_maidian_updates
+			WHERE partition_date >= start_date 
+            AND partition_date < end_date
+			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'
 
-        ) group by query
+			UNION ALL
+			SELECT partition_date
+					,params['card_name'] as query
+					,cl_id
+			FROM online.bl_hdfs_maidian_updates
+			WHERE partition_date >= start_date 
+            AND partition_date < end_date
+			AND int(split(app_version,'\\.')[1]) >= 27
+			AND action='on_click_card'
+			AND params['card_type']='highlight_word'
+		)t1
+		JOIN
+		(
+			SELECT partition_date,device_id,t2.active_type,t2.channel,t2.device_os_type
+            FROM
+            (
+                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 is_ai_channel = 'true' THEN 'AI'  ELSE '其他' END , '合计') as channel
+                FROM online.ml_device_day_active_status m
+                	LEFT JOIN 
+	                (SELECT code,is_ai_channel,partition_day
+	                 FROM DIM.DIM_AI_CHANNEL_ZP_NEW
+	                 WHERE partition_day>= start_date  
+	                 AND partition_day < end_date) tmp
+	                ON  m.partition_date=tmp.partition_day AND first_channel_source_type=code
+                WHERE partition_date >= start_date 
+            	AND partition_date < end_date
+                AND active_type in ('1','2','4')
+            ) mas
+            LATERAL VIEW explode(mas.channel) t2 AS channel
+            LATERAL VIEW explode(mas.device_os_type) t2 AS device_os_type
+            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 query
+		)
 """.format(start_date='20201024',end_date='20201025')
 
 print(sql_search_ctr)