update

task/spark_test.py

@@ -154,7 +154,7 @@ 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   card_name,count(card_name)
+   SELECT   query,count(query)
 --             click.partition_date as partition_date
 --             ,active_type
 --             ,device_os_type
@@ -163,15 +163,8 @@ sql_search_ctr = r"""
 --             ,count(distinct click.cl_id) as all_search_click_uv_724
         FROM
         (
-            SELECT cl_id,partition_date,action,params['page_name'] as page_name,'' as input_type,app_version,params['card_name'] as card_name
-            FROM online.bl_hdfs_maidian_updates
-            WHERE partition_date >= '{start_date}'  
-              AND partition_date < '{end_date}'  
-              AND action = 'on_click_navbar_search'
-              AND (int(split(app_version,'\\.')[0]) = 7 AND int(split(app_version,'\\.')[1]) >= 24)
-
-            UNION all
-            SELECT cl_id,partition_date,action,params['page_name'] as page_name,params['input_type'] as input_type,app_version,params['card_name'] as card_name
+    
+            SELECT app_version,params['query'] as query
             FROM online.bl_hdfs_maidian_updates
             WHERE partition_date >= '{start_date}'  
               AND partition_date < '{end_date}'  
@@ -179,7 +172,7 @@ sql_search_ctr = r"""
               AND params['input_type'] = '详情页默认词'
 
             union all 
-            SELECT cl_id,partition_date,action,'home' as page_name,'首页-猜你喜欢' as input_type,app_version,params['card_name'] as card_name
+            SELECT 'home' as page_name,'首页-猜你喜欢' as input_type,app_version,params['card_name'] as query
             FROM online.bl_hdfs_maidian_updates
             WHERE partition_date >= '{start_date}'  
               AND partition_date < '{end_date}'  
@@ -190,7 +183,7 @@ sql_search_ctr = r"""
               --AND page_name='home' android的page_name为空
 
             union all 
-            SELECT cl_id,partition_date,action,params['page_name'] as page_name,'美购首页-大家都在搜' as input_type,app_version,params['card_name'] as card_name
+            SELECT params['query'] as query
             FROM online.bl_hdfs_maidian_updates
             WHERE partition_date >= '{start_date}'  
               AND partition_date < '{end_date}'  
@@ -200,13 +193,13 @@ sql_search_ctr = r"""
               AND params['in_page_pos']='大家都在搜'
 
             union all 
-            SELECT cl_id,partition_date,action,params['page_name'] as page_name,'高亮词' as input_type,app_version,params['card_name'] as card_name
+            SELECT params['card_name'] as query
             FROM online.bl_hdfs_maidian_updates
             WHERE partition_date >= '{start_date}'  
               AND partition_date < '{end_date}'  
               AND action = 'on_click_card'
               AND params['card_type'] ='highlight_word'
-        ) group by card_name
+        ) group by query
 """.format(start_date='20201024',end_date='20201025')
 
 print(sql_search_ctr)