从具有分组结果的Join中删除重复项

Question

我使用带有EF Core的SQL Server。由于缺少EF的GROUP BY支持，我运行了原始SQL查询。

所以，我有一个历史记录表

+------------+---------+------------------+
| HardwareId | StateId |    Timestamp     |
+------------+---------+------------------+
| Dev1       |       1 | 2019-10-11 11:00 |
| Dev2       |       2 | 2019-10-11 11:01 |
| Dev1       |       3 | 2019-10-11 11:09 |
| Dev2       |       1 | 2019-10-11 11:10 |
| Dev2       |       2 | 2019-10-11 11:10 |
| Dev1       |       3 | 2019-10-11 11:11 |
+------------+---------+------------------+

这里一些设备记录可以具有相同的时间戳。

现在，我想获取每个设备的最后状态：

+------------+---------+------------------+
| HardwareId | StateId |    Timestamp     |
+-----==-----+---------+------------------+
| Dev2       |       1 | 2019-10-11 11:10 |
| Dev1       |       3 | 2019-10-11 11:11 |
+------------+---------+------------------+

我跑步

SELECT H.TimeStamp, H.HardwareId, H.ErrorCode, SD.Description FROM History H
INNER JOIN 
(SELECT HardwareId, MAX(TimeStamp) LastDateTime from History 
GROUP BY HardwareId) AS LastStates 
ON H.TimeStamp = LastStates.LastDateTime
INNER JOIN StateDescription SD ON H.ErrorCode = SD.Id ORDER BY H.HardwareId

但是它给了我重复项（由于给定设备的时间戳重复，一种设备的多个最后状态。我希望选择这些状态中的任何一个）

+------------+---------+------------------+
| HardwareId | StateId |    Timestamp     |
+------------+---------+------------------+
| Dev2       |       1 | 2019-10-11 11:10 |
| Dev2       |       2 | 2019-10-11 11:10 |
| Dev1       |       3 | 2019-10-11 11:11 |
+------------+---------+------------------+

添加DISTINCT对其进行修复

SELECT DISTINCT H.TimeStamp, H.HardwareId, H.ErrorCode, SD.Description FROM History H
INNER JOIN 
(SELECT HardwareId, MAX(TimeStamp) LastDateTime from History 
GROUP BY HardwareId) AS LastStates 
ON H.TimeStamp = LastStates.LastDateTime
INNER JOIN StateDescription SD ON H.ErrorCode = SD.Id ORDER BY H.HardwareId

但是如果我想将记录ID （（H.Id）添加到最终结果中，我显然会再次得到重复。

  SELECT DISTINCT H.TimeStamp, H.Id, H.HardwareId, H.ErrorCode, SD.Description FROM History H
    INNER JOIN 
    (SELECT HardwareId, MAX(TimeStamp) LastDateTime from History 
    GROUP BY HardwareId) AS LastStates 
    ON H.TimeStamp = LastStates.LastDateTime
    INNER JOIN StateDescription SD ON H.ErrorCode = SD.Id ORDER BY H.HardwareId

如何获得每个设备的某种TOP（1）记录？

Answer 1

我只用row_number()：

select h.*
from (select h.*,
             row_number() over (partition by HardwareId order by timestamp desc) as seqnum
      from history h
     ) h
where seqnum = 1;

这将每天为每台设备精确选择一行。如果当天重复，它将返回一个任意值。如果要全部使用，请使用rank()代替row_number()。