从同一列中拆分时间并存储到两列中，然后在sql中查找这两列之间的差异

从同一列中拆分时间并存储到两列中，然后在sql server中查找这两列之间的差异

select SourceName,Active,EventTimeStamp,
case when Active=1
then EventTimeStamp END as ActiveTime,

case when Active=0
then EventTimeStamp END as InactiveTime

from dbo.AllEvent
where SourceName='R2_20'
order by SourceName 
;


sname   active  EventTimeStamp      Active      Inavtive
R2_20       1   14:51.9     14:51.9     NULL
R2_20       0   14:53.9     NULL        14:53.9
R2_20       1   15:05.9     15:05.9     NULL
R2_20       0   15:07.9     NULL        15:07.9
R2_20       1   15:15.9     15:15.9     NULL
R2_20       0   15:17.9     NULL        15:17.9
R2_20       1   15:26.0     15:26.0     NULL
R2_20       0   15:28.0     NULL        15:28.0
R2_20       1   15:36.0     15:36.0     NULL
R2_20       0   15:38.0     NULL        15:38.0

---

如果您使用的SQL Server版本支持此功能，请使用它。但这只是根据你提供的数据我的猜测

SELECT
  SourceName sname,
  Active active,
  EventTimeStamp start,
  [end],
  DATEDIFF(millisecond, EventTimeStamp, [end]) difference
FROM (
  SELECT
    *,
    LEAD(EventTimeStamp)
      OVER (PARTITION BY SourceName
            ORDER BY EventTimeStamp, Active DESC) [end]
  FROM AllEvents
) t
WHERE Active = 1;

但对于更复杂的数据集，以下查询可能更可取

WITH
  AllEvents AS (
    SELECT
      sn SourceName,
      CAST(a as INT) Active,
      CAST(ts as TIME) EventTimeStamp
    FROM (VALUES
      ('R2_20', 1, '00:14:51.9'),
      ('R2_20', 1, '00:14:52.9'),
      ('R2_20', 0, '00:14:53.9'),
      ('R2_20', 1, '00:15:05.9'),
      ('R2_20', 0, '00:15:07.9'),
      ('R2_20', 1, '00:15:15.9'),
      ('R2_20', 1, '00:15:16.9'),
      ('R2_20', 0, '00:15:17.5'),
      ('R2_20', 0, '00:15:17.9'),
      ('R2_20', 1, '00:15:26.0'),
      ('R2_20', 0, '00:15:28.0'),
      ('R2_20', 1, '00:15:36.0'),
      ('R2_20', 0, '00:15:37.0'),
      ('R2_20', 0, '00:15:38.0')
    ) t(sn, a, ts)
  ),
  a AS (
    SELECT
      *,
      ROW_NUMBER() OVER (PARTITION BY SourceName ORDER BY EventTimeStamp) -
        ROW_NUMBER() OVER (PARTITION BY SourceName, Active ORDER BY EventTimeStamp) g
    FROM AllEvents
  ),
  b AS (
    SELECT
      SourceName,
      Active,
      IIF(Active = 0, MAX(EventTimeStamp), MIN(EventTimeStamp)) start
    FROM a
    GROUP BY SourceName, Active, g
  ),
  c AS (
    SELECT
      *,
      LEAD(start) OVER (PARTITION BY SourceName ORDER BY start, Active DESC) [end],
      DATEDIFF(millisecond,
               start,
               LEAD(start) OVER (PARTITION BY SourceName
                                 ORDER BY start, Active DESC)) difference
    FROM b
  )
SELECT * FROM c WHERE Active = 1 ORDER BY start;

---

我在“开始”和“结束”列中存储的时间，现在我需要找到这两列之间的差异，您的预期输出要求您包括什么并解释逻辑？sname active event timestamp active Inavtive R2_20 1 14:51.9 14:51.9 NULL R2_20 0 14:53.9 NULL 14:53.9 R2_20 1 15:05.9 15:05.9 NULL R2_20 15:07.9 NULL 15:07.9请参见第1行，14:51.9（active=1）应该从14:53.9（inactive=0）中减去，并存储在Difference列中感谢您的快速响应，请参阅Eventtimestamp是存储同一列的时间，基于active和inactive，我必须将其分为两列，这是我使用发布的代码所做的，现在使用这些active和inactive列，我必须找到差异，在这种情况下，发生的是，活动（1）存储一些时间，在非活动列中存储null，活动列中的下一行存储null，非活动（0）列存储时间，所以我无法找到活动和非活动之间的差异