Sql server 将两行减法到列中多次，并链接回标题

我有一些可怕的数据，我需要把它们变成有意义的东西。我意识到这不是最好的方式来表明我已经很久没有使用StackOverflow了，我也不确定语法

我试着写一些查询，但我真的不知道从哪里开始这样的查询，所以任何帮助都将不胜感激，谢谢

我有一个链接到详细信息表的标题表。在details表中，我有3组2条记录，其中我想得到两列之间的差异

标题表

headerId
1
2

detailid|headerId|name|totalElapsedMs
1|1|Request1|100
2|1|Response1|1000
3|1|Request2|1100
4|1|Response2|1800
5|1|Request3|2000
6|1|Response3|2600

Declare @YourTable Table ([detailid] int,[headerId] int,[name] varchar(50),[totalElapsedMs] int)
Insert Into @YourTable Values 
 (1,1,'Request1',100)
,(2,1,'Response1',1000)
,(3,1,'Request2',1100)
,(4,1,'Response2',1800)
,(5,1,'Request3',2000)
,(6,1,'Response3',2600)

Select *
 From  (
        Select HeaderId
              ,Item  = concat('Request',replace(replace(name,'Request',''),'Response',''),'ElapsedMs')
              ,Value = [totalElapsedMs]*IIF(left(name,3)='Res',1,-1)
            From @YourTable
        ) src
 Pivot (sum(Value) for Item in ([Request1ElapsedMs],[Request2ElapsedMs],[Request3ElapsedMs]) ) pvt

明细表

headerId
1
2

detailid|headerId|name|totalElapsedMs
1|1|Request1|100
2|1|Response1|1000
3|1|Request2|1100
4|1|Response2|1800
5|1|Request3|2000
6|1|Response3|2600

Declare @YourTable Table ([detailid] int,[headerId] int,[name] varchar(50),[totalElapsedMs] int)
Insert Into @YourTable Values 
 (1,1,'Request1',100)
,(2,1,'Response1',1000)
,(3,1,'Request2',1100)
,(4,1,'Response2',1800)
,(5,1,'Request3',2000)
,(6,1,'Response3',2600)

Select *
 From  (
        Select HeaderId
              ,Item  = concat('Request',replace(replace(name,'Request',''),'Response',''),'ElapsedMs')
              ,Value = [totalElapsedMs]*IIF(left(name,3)='Res',1,-1)
            From @YourTable
        ) src
 Pivot (sum(Value) for Item in ([Request1ElapsedMs],[Request2ElapsedMs],[Request3ElapsedMs]) ) pvt

结果

HeaderId    Request1ElapsedMs   Request2ElapsedMs   Request3ElapsedMs
1           900                 700                 600

我想减去彼此匹配的行，然后将它们向上旋转到标题行，如下所示

headerId|Request1ElapsedMs|Request2ElapsedMs|Request3ElapsedMs
1|900|700|600

---

我认为在Detail表上使用PIVOT操作符来获得所需的输出非常简单。假设@Detail是您的源表，类似于：

SELECT
    HeaderID
,   COALESCE( [Response1], 0 ) - COALESCE( [Request1], 0 ) AS Request1ElapsedMs
,   COALESCE( [Response2], 0 ) - COALESCE( [Request2], 0 ) AS Request2ElapsedMs
,   COALESCE( [Response3], 0 ) - COALESCE( [Request3], 0 ) AS Request3ElapsedMs

FROM    
    (SELECT HeaderID, Name, TotalElapsedMs FROM @Detail) AS Detail
PIVOT (SUM( TotalElapsedMs ) FOR Name IN ( [Request1], [Response1], [Request2], [Response2], [Request3], [Response3] )) AS PivotedData

---

有许多枢轴和/或动态枢轴的示例

这里唯一的技巧是翻转标志，然后聚合

示例

headerId
1
2

detailid|headerId|name|totalElapsedMs
1|1|Request1|100
2|1|Response1|1000
3|1|Request2|1100
4|1|Response2|1800
5|1|Request3|2000
6|1|Response3|2600

Declare @YourTable Table ([detailid] int,[headerId] int,[name] varchar(50),[totalElapsedMs] int)
Insert Into @YourTable Values 
 (1,1,'Request1',100)
,(2,1,'Response1',1000)
,(3,1,'Request2',1100)
,(4,1,'Response2',1800)
,(5,1,'Request3',2000)
,(6,1,'Response3',2600)

Select *
 From  (
        Select HeaderId
              ,Item  = concat('Request',replace(replace(name,'Request',''),'Response',''),'ElapsedMs')
              ,Value = [totalElapsedMs]*IIF(left(name,3)='Res',1,-1)
            From @YourTable
        ) src
 Pivot (sum(Value) for Item in ([Request1ElapsedMs],[Request2ElapsedMs],[Request3ElapsedMs]) ) pvt

返回

HeaderId    Request1ElapsedMs   Request2ElapsedMs   Request3ElapsedMs
1           900                 700                 600

如果有帮助的话，子查询将“馈送”枢轴生成

HeaderId    Item                Value
1           Request1ElapsedMs   -100
1           Request1ElapsedMs   1000
1           Request2ElapsedMs   -1100
1           Request2ElapsedMs   1800
1           Request3ElapsedMs   -2000
1           Request3ElapsedMs   2600