对数据库行中的匹配条目进行计数，除非该行为空

Question

计算一行中的连续（特定）匹配的entrys，但是如果其他数据位于其间，则从头开始计数。

我的想法是查看某人在某个时间段内进行了多少次登录尝试（在我的情况下为15分钟）并相应采取行动。例如（简化表格）：

Row: User_id | Loginstatus | Timestamp 
    5   Attempt  most recent 
    5   Attempt  . 
    5   Logoff   . 
    5   Login   . 
    5   Attempt  . 
    5   Attempt  . 
    5   Attempt  . 
    5   Attempt  . 
    5   Logoff   . 
    5   Login   . 
    5   Logoff   . 
    5   Login   . 

我选择现在应出示：2

我走到这一步：

select count(Loginstatus) Result from Userlog 
where User_Id = 5 
and Loginstatus = 'ATTEMPT' 
and Timestamp > Systimestamp - Interval '15' Minute; 

但不幸的是计算所有的尝试。我很高兴的任何帮助，并感谢您提前

Answer 1

您可以使用行号的差异来确定相邻序列：

select user_id, count(*) as num_consecutive 
from (select ul.*, 
      row_number() over (partition by user_id order by timestamp) as seqnum_u, 
      row_number() over (partition by user_id, loginstatus order by timestamp) as seqnum_us 

     from Userlog ul 
     where User_Id = 5 and Timestamp > Systimestamp - Interval '15' Minute; 
    ) ul 
where Loginstatus = 'ATTEMPT' 
group by user_id, (seqnum_u - seqnum_us); 

如果你想要一个值，你可以添加：

order by num_consecutive desc 
fetch first 1 row only 

Answer 2

您需要使用GROUP BY：

SELECT [LoginStatus]=MAX(Loginstatus), [Count]=count(Loginstatus) 
FROM Userlog 
WHERE User_Id = 5 
AND Loginstatus = 'ATTEMPT' 
AND Timestamp > Systimestamp - Interval '15' Minute; 
GROUP BY (Loginstatus)