计算 SQL 中每个类别经过的时间

Question

我需要一个设备跟进项目的帮助。

我有一个包含 3 列（设备名称、状态、状态更改日期（DATETIME 格式））的 SQL 表。

EQUIPMENT	STATUS	CHANGEDATE
EQUIPMENT-1	QUALIFICATION	2020-06-30 09:37:42
EQUIPMENT-1	WAIT REPAIR	2020-06-30 16:29:20
EQUIPMENT-1	UP	2020-07-27 14:19:33
EQUIPMENT-1	ENGINEERING	2020-09-18 15:25:01
EQUIPMENT-1	UP	2020-09-20 17:31:53

这个想法是在2个固定日期之间确定每个状态下每个设备的经过时间。

例如，我想知道2020-07-01和2020-10-01之间所有状态下EQUIPMENT-1的经过时间表格结果是这样的

STATUS	ELAPSED TIME (in days)
WAIT REPAIR	26,60
UP	63,31 (10,27 + 53,05)
ENGINEERING	2,09

今天我有一个 C# 代码来计算这些经过的时间，但它很慢...... 所以我想知道用 SQL 查询替换这个过程是否容易。

感谢您的帮助，

Answer 1

我认为你想要lead() 和聚合：

select equipment, status,
       sum(datediff(minute,
                    changedate,
                    coalesce(next_changedate, '2020-10-01')
                   ) / (24 * 60.0)
          ) as decimal_days
from (select t.*,
             lead(changedate) over (partition by equipment order by changedate)
      from t
      where changedate >= '2020-07-01' and changedate < '2020-10-01'
     ) t
group by equipment, status;

编辑：

如果您也需要初始时间：

select equipment, status,
       sum(datediff(minute,
                    (case when changedate < '2020-07-01' then '2020-07-01' else changeddate end),
                    coalesce(next_changedate, '2020-10-01')
                   ) / (24 * 60.0)
          ) as decimal_days
from (select t.*,
             lead(changedate) over (partition by equipment order by changedate) as next_changedate
      from t
     ) t
where changedate >= '2020-07-01' and changedate < '2020-10-01' or
      next_changedate >= '2020-07-01' and next_changedate < '2020-10-01' or
group by equipment, status;