如何将行转换为列

Question

我有一张桌子 Columns

以及Response保存所有数据的第二个表。

现在我想创建一个 SQL 视图，其中的结果应该是这样的

我尝试使用枢轴

select UserId ,FromDate, ToDate, Project, Comment
from
(
  select R.UserId ,R.Text , C.ColumnName
  from [Columns] C
  INNER JOIN Response R ON C.Id=R.ColumnId
) d
pivot
(
  max(Text)
  for ColumnName in (FromDate, ToDate, Project, Comment)
) piv;

但这对我不起作用，我也提到了这个/sf/ask/1102152971/但无法实现它。任何想法如何在 SQL 视图中实现相同的目标？

表格脚本：

CREATE TABLE [dbo].[Columns](
    [Id] [bigint] IDENTITY(1,1) NOT NULL,
    [Name] [nvarchar](1000) NULL,
    [IsActive] [bit] NULL,
 CONSTRAINT [PK_Columns] PRIMARY KEY CLUSTERED 
(
    [Id] ASC
)WITH (PAD_INDEX  = OFF, STATISTICS_NORECOMPUTE  = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS  = ON, ALLOW_PAGE_LOCKS  = ON) ON [PRIMARY]
) ON [PRIMARY]

GO

insert into [Columns] values('FromDate',1)
insert into [Columns] values('ToDate',1)
insert into [Columns] values('Project',1)
insert into [Columns] values('Comment',1)

CREATE TABLE [dbo].[Response](
    [Id] [bigint] IDENTITY(1,1) NOT NULL,
    [UserId] [bigint]  NOT NULL,
    [ColumnId] [bigint]  NOT NULL,
    [Text] [nvarchar](max) NULL,
    [IsActive] [bit] NULL,
    CONSTRAINT [PK_Response] PRIMARY KEY CLUSTERED 
(
    [Id] ASC
)WITH (PAD_INDEX  = OFF, STATISTICS_NORECOMPUTE  = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS  = ON, ALLOW_PAGE_LOCKS  = ON) ON [PRIMARY]
) ON [PRIMARY]

GO
insert into [Response] values(1,1,'1/1/2012',1)
insert into [Response] values(1,2,'1/2/2012',1)
insert into [Response] values(1,3,'p1',1)
insert into [Response] values(1,4,'c1',1)
insert into [Response] values(2,1,'1/1/2013',1)
insert into [Response] values(2,2,'1/2/2013',1)
insert into [Response] values(2,3,'p2',1)
insert into [Response] values(2,4,'c2',1)
insert into [Response] values(2,1,'1/1/2014',1)
insert into [Response] values(2,2,'1/2/2014',1)
insert into [Response] values(2,3,'p3',1)
insert into [Response] values(2,4,'c3',1)
insert into [Response] values(3,1,'1/1/2015',1)
insert into [Response] values(3,2,'1/2/2015',1)
insert into [Response] values(3,3,'p4',1)
insert into [Response] values(3,4,'c4',1)

Answer 1

您没有解释为什么您的 PIVOT 查询对您不起作用，尽管通常不难猜测您的Response样本是否足以代表该表中的数据。您对用户 2 有两组响应，但您的查询仅选择一组。

由于您的预期输出显示您希望查询返回两组，您需要教它区分同一用户的不同组答案。一种方法是使用 ROW_NUMBER 分析函数：

SELECT
  UserId,
  FromDate,
  ToDate,
  Project,
  Comment
FROM
  (
    SELECT
      r.UserId,
      r.Text,
      c.ColumnName,
      SetNo = ROW_NUMBER() OVER (PARTITION BY r.UserId, r.ColumnId ORDER BY r.Id)
    FROM
      dbo.Columns AS c
      INNER JOIN dbo.Response AS r ON c.Id = r.ColumnId
  ) AS derived
  PIVOT
  (
    MAX(Text)
    FOR ColumnName IN (FromDate, ToDate, Project, Comment)
  ) AS p
;

上面的查询做了一些假设：

• 每组答案都包含相同的一组 ColumnId；
• 相关响应排列在一起（基于 的顺序Id），或者至少，如果一个FromDate响应晚于另一个FromDate响应，则其相关ToDate响应也晚于ToDate与前一个 相关的响应FromDate。

如果真实数据中存在特殊情况（例如某些集合可能不完整），则可能需要不同的方法。