个性化push指标 测试

task/ctr_push_strategy_test.py

@@ -75,20 +75,38 @@ def getSpark():
 def getSql(startDay,endDay):
     sql = """
     --push接收数据
-        select partition_date,push_type,count(*)
-                    from online.tl_hdfs_push2_new_view --增量表
-                    where partition_date >= {startDay} and partition_date<{endDay} and push_type in ('101','102','111','112')
-                    group by partition_date,push_type
-                    
-        union all
-        
-        SELECT partition_date,regexp_replace(labels['event'], '\\s+', '') AS push_type,count(*)
-        FROM online.tl_hdfs_push2_task_view
-        WHERE partition_date >= '20210618' and partition_date<'20210619'
-        AND labels['event_type'] = 'push'
-        group by  partition_date,regexp_replace(labels['event'], '\\s+', '')
+        SELECT partition_day
+                ,push_type
+                ,count(a.device_id)
+                ,count(distinct a.device_id)
+        FROM
+        (
+            select partition_day
+                    ,device_id,msg_id
+            from bl.bl_et_bg_trackingpushlog_inc_d
+            where partition_day=regexp_replace(DATE_SUB(current_date,1) ,'-','')
+            group by partition_day,device_id,msg_id
+        )a
+        JOIN
+        (
+            select msg_id,push_type,time_stamp
+            from online.tl_hdfs_push2_new_view --增量表
+            where partition_date=regexp_replace(DATE_SUB(current_date,1) ,'-','')
+            group by msg_id,push_type,time_stamp
+
+            union all
+
+            SELECT msg_id,regexp_replace(labels['event'], '\\s+', '') AS push_type,time_stamp
+            FROM online.tl_hdfs_push2_task_view
+            WHERE partition_date=regexp_replace(DATE_SUB(current_date,1) ,'-','')
+            AND labels['event_type'] = 'push'
+            group by  msg_id,regexp_replace(labels['event'], '\\s+', ''),time_stamp
+        )c
+        on a.msg_id=c.msg_id
+        group by partition_day
+                ,push_type
     
-    """.format(startDay=startDay,endDay=endDay)
+    """
     return sql
 
 if __name__ == '__main__':