SQL 按行到列的数据透视组

Question

所以我有一个 table 用于我想每天计算的司机活动。示例数据：

Name	StartDateAndTime	EndDateAndTime	StartText	EndText	WorkState
Jim	2021-09-09 06:28:16	2021-09-09 07:28:16	"StartPlace0"	"EndPlace0"	"Pause"
Jim	2021-09-09 07:28:16	2021-09-09 08:28:16	"StartPlace1"	"EndPlace1"	"Loading"
Jim	2021-09-09 08:28:16	2021-09-09 09:28:16	"StartPlace2"	"EndPlace2"	"Driving"
Jim	2021-09-09 10:28:16	2021-09-09 11:28:16	"StartPlace4"	"EndPlace4"	"Loading"
Jim	2021-10-09 09:28:16	2021-10-09 10:28:16	"StartPlace5"	"EndPlace5"	"Driving"
Jim	2021-10-09 10:28:16	2021-10-09 11:28:16	"StartPlace6"	"EndPlace6"	"Loading"
Jim	2021-10-09 11:28:16	2021-10-09 14:28:16	"StartPlace7"	"EndPlace7"	"Driving"

期望的输出：（我只有 3 种可能的工作状态）

Name	StartDateAndTime	EndDateAndTime	StartText	EndText	Loading	Driving
Jim	2021-09-09 06:28:16	2021-09-09 11:28:16	"StartPlace0"	"EndPlace4"	00:02:00	00:01:00
Jim	2021-10-09 09:28:16	2021-09-09 14:28:16	"StartPlace5"	"EndPlace7"	00:01:00	00:04:00

Pause	DayPerformed
00:01:00	2021-09-09
00:00:00	2021-10-09

到目前为止我的看法：（摘要）

 --CREATE view vwWorkingTimesPerDay as
select  
   w.DriverName, 
   convert(date, w.StartDateAndTime) as DayPerformed,
   MIN(w.StartDateAndTime) as StartTime,
   MAX(w.EndDateAndTime) as EndTime,
   StartPositionText,
   EndPositionText

from  ( 
  SELECT *,
  StartPositionText = FIRST_VALUE(StartPosText) OVER (PARTITION BY w.DriverName, convert(date, 
  w.StartDateAndTime)
        ORDER BY w.StartDateAndTime ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING),
  EndPositionText   = LAST_VALUE(EndPosText) OVER (PARTITION BY w.DriverName, convert(date, 
  w.StartDateAndTime)
        ORDER BY w.StartDateAndTime ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING)
FROM webfleet.tblWorkingTimes w) w

  group by w.DriverName, 
     convert(date, w.StartDateAndTime),
     StartPositionText,
     EndPositionText,

为了显示原始 table 中的持续时间，我使用了：

right('00' + convert(varchar, datepart(hour, dateadd(s, sum(datediff(second, 
    w.StartDateAndTime, w.EndDateAndTime)), 0))), 2) + ':'
 + right('00' + convert(varchar, datepart(minute, dateadd(s, sum(datediff(second, 
  w.StartDateAndTime, w.EndDateAndTime)), 0))), 2) + ':'
  + right('00' + convert(varchar, datepart(second, dateadd(s, sum(datediff(second, 
  w.StartDateAndTime, w.EndDateAndTime)), 0))), 2) 
   as Duration

我试过这篇文章https://www.sqlshack.com/multiple-options-to-transposing-rows-into-columns

但我没有足够的经验来使用这些选项。我想要的数据顺序并不重要，如果有更好的方法来计算和显示持续时间，我很高兴听到。

Answer 1

这需要几个步骤

我们使用CAST AS date来获取没有时间的日期
ROW_NUMBER和LEAD帮助我们识别每个分组的开始和结束
然后简单地按 Name 和日期
并使用条件聚合得到我们想要的结果
要获得时间值，我们将秒数加到 0:00:00 时间

SELECT
  Name,
  DayPerformed,
  StartDateAndTime = MIN(StartDateAndTime),
  EndDateAndTime = MAX(EndDateAndTime),
  StartText = MIN(CASE WHEN rn = 1 THEN StartText END),
  EndText = MIN(CASE WHEN nxt IS NULL THEN EndText END),
  Loading = DATEADD(ms,
      ISNULL(SUM(CASE WHEN WorkState = 'Loading' THEN DATEDIFF(ms, StartDateAndTime, EndDateAndTime) END), 0),
      CAST('0:00:00' AS time)),
  Driving = DATEADD(ms, 
      ISNULL(SUM(CASE WHEN WorkState = 'Driving' THEN DATEDIFF(ms, StartDateAndTime, EndDateAndTime) END), 0),
      CAST('0:00:00' AS time)),
  Pause   = DATEADD(ms,
      ISNULL(SUM(CASE WHEN WorkState = 'Pause'   THEN DATEDIFF(ms, StartDateAndTime, EndDateAndTime) END), 0), 
      CAST('0:00:00' AS time))
FROM (
    SELECT *,
      rn = ROW_NUMBER() OVER (PARTITION BY Name, DayPerformed ORDER BY StartDateAndTime),
      nxt = LEAD(WorkState) OVER (PARTITION BY Name, DayPerformed ORDER BY StartDateAndTime)
    FROM tblWorkingTimes wt
    CROSS APPLY (VALUES (CAST(StartDateAndTime AS date))) v(DayPerformed)
) wt
GROUP BY
  Name,
  DayPerformed;

要按月汇总，只需将 DayPerformed 替换为

CROSS APPLY (VALUES (EOMONTH(StartDateAndTime))) v(DayPerformed)

db<>fiddle

SQL 按行到列的数据透视组

SQL Pivot group by rows to columns

sql

sql-server

pivot