0

我有一个表，其中包含更改维护服务状态的事件。这是一个数据样本：

id  date    event
234 2009-04-22 10:00:00 Service Request
234 2009-04-22 10:02:00 Service Approbation
234 2009-04-22 10:03:00 Service Asignation
234 2009-04-22 10:40:00 Service Fulfilled
...

在报告中，我需要显示在每个项目上花费的时间

id  time
234 40 minutes
235 37 minutes
...

基本上，我正在做这样的 DATEDIFF：

SELECT ...
, DATEDIFF(SECOND, 
(SELECT TOP 1 date FROM event_table ev WHERE ev.item = itm.id AND event = 1 ORDER BY date ), 
(SELECT TOP 1 date  FROM event_table ev WHERE ev.item = itm.id AND event = 4 ORDER BY date DESC)
        ), ...
FROM item_table itm
WHERE ...

现在我必须实现一个新的状态。延迟和重新激活服务，但在计算花费在该服务上的总时间时不必考虑这些时间。

id  date    event
234 2009-04-22 10:00:00 Service Request
234 2009-04-22 10:02:00 Service Approbation
234 2009-04-22 10:03:00 Service Asignation
234 2009-04-22 10:07:00 Service Delayed           -- new state
234 2009-04-22 10:37:00 Service Reactivated       -- new state
234 2009-04-22 10:40:00 Service Fulfilled
...

报告

id  time
234 10 minutes  -- Substract 30 minutes between Delay and Reactivation
235 26 minutes
...

对于任何项目，延迟和重新激活都可能发生多次。

我知道我可以用光标来做到这一点，但对我来说这似乎很老套，有没有办法在基于集合的方法上计算这个？

4

1 回答 1

1

我这样做的方法是首先找到整个工作的持续时间，减去所有延迟的持续时间......

查找延迟持续时间的代码：

SELECT
   [delay_start].event_id,
   SUM(DATEDIFF(second, [delay_start].date, [delay_finish].date))
FROM
   event_table AS [delay_start]
INNER JOIN
   event_table AS [delay_finish]
      ON  [delay_finish].event_id = [delay_start].event_id
      AND [delay_finish].date = (
                                 SELECT
                                    MIN(date)
                                 FROM
                                    event_table
                                 WHERE
                                    event_id = [delay_start].event_id
                                    AND date > [delay_start].date
                                    AND event = 'Service Reactivated'
                                )
WHERE
   [delay_start].event = 'Service Delayed'
   AND [delay_finish].event = 'Service Reactivated'
GROUP BY
   [delay_start].event_id

编辑：纠正了几个错别字...

于 2009-04-22T15:47:28.830 回答