带串联的动态枢轴

Question

应如何修改 SQL 以将文本“Name”连接到列标题和 2 个数据透视数据的名称字段中？

我有这样的桌子：

UmpID	第N	最后N	年龄
1	A	H	J
2	B	我	S
3	C	J	A
4	D	K	J
5	E	L	S
6	F	M	J
7	G	N	J

查询：

SET NoCount ON;

DECLARE @cols AS NVARCHAR(MAX), @sql AS NVARCHAR(MAX);

WITH GS AS(
SELECT LastN, Age, QUOTENAME(Row_Number() OVER(PARTITION BY Age ORDER BY Age, LastN)) As GrpSeq FROM Umpires),
DS AS
(SELECT DISTINCT GrpSeq FROM GS),
CS AS
(SELECT STRING_AGG(GrpSeq,',') AS G FROM DS)

SELECT @cols = G FROM CS

SELECT @sql = 'SELECT *
            FROM (SELECT LastN, Age, Row_Number() OVER(PARTITION BY Age ORDER BY Age, LastN) As GrpSeq FROM Umpires) AS Q
            PIVOT(Min(LastN) FOR GrpSeq IN(' + @cols + ')) AS P;';

EXEC sp_executesql @sql;

产生输出：

年龄	1	2	3	4
A	J
J	H	K	M	N
S	我	L

我想制作：

年龄	姓名1	姓名2	姓名3	姓名4
A	J、C
J	H、A	K、D	M、F	N、G
S	我，B	L、E

尝试简单串联内部 SELECT 中的字段并引用 PIVOT 中的别名字段来构建 @sql 字符串生成的错误：

为变量赋值的 select 语句不得与数据检索操作结合使用

尝试连接“名称”字符串只会触发：

“名称”附近的语法不正确

作为旁注，这在 Access 查询中很容易完成，尽管由于 DCount() 域聚合函数，对于非常大的数据集可能会非常慢：

TRANSFORM Max([LastN] & ", " & [FirstN]) AS FullName
SELECT Umpires.Age
FROM Umpires
GROUP BY Umpires.Age
PIVOT "Name" & DCount("*","Umpires","Age='" & [Age] & "' AND UmpID<=" & [UmpID]);

我还可以在静态 SQLServer 数据透视表中生成已知数量的列标题值：

WITH DS AS(
SELECT Age, LastN + ', ' + FirstN AS FN, 'Name' + CAST(Row_Number() OVER(PARTITION BY Age ORDER BY Age, LastN) AS VARCHAR) As GrpSeq FROM Umpires)

SELECT * FROM (SELECT Age, FN, GrpSeq FROM DS) AS Q PIVOT(Max(FN) FOR GrpSeq IN([Name1],[Name2],[Name3],[Name4])) P;

Answer 1

可能会发生这样的事情：

DECLARE @cols AS NVARCHAR(MAX), @cols2 NVARCHAR(MAX), @sql AS NVARCHAR(MAX);

SELECT  *
INTO    #data
FROM    (
    VALUES  (1, N'A', N'H', N'J')
    ,   (2, N'B', N'I', N'S')
    ,   (3, N'C', N'J', N'A')
    ,   (4, N'D', N'K', N'J')
    ,   (5, N'E', N'L', N'S')
    ,   (6, N'F', N'M', N'J')
    ,   (7, N'G', N'N', N'J')
) t (UmpID,FirstN,LastN,Age)

;WITH GS AS(
    SELECT  LastN, Age
    ,   Row_Number() OVER(PARTITION BY Age ORDER BY Age, LastN) AS GrpSeq
    FROM    #data
    )
,DS AS (
    SELECT  DISTINCT GrpSeq
    FROM    GS
    )
,CS AS (
    SELECT  STRING_AGG(QUOTENAME(GrpSeq),',') WITHIN GROUP(ORDER BY GrpSeq) AS G
    ,   STRING_AGG(CONCAT(QUOTENAME(GrpSeq), ' AS Name', GrpSeq),',') WITHIN GROUP(ORDER BY GrpSeq) AS G2
    FROM    DS
    )
SELECT  @cols = G
,   @cols2 = G2
FROM    CS

SELECT  @sql = 'SELECT Age, ' + @cols2 + N'
            FROM (SELECT CONCAT(LastN, '','', FirstN) AS LastN, Age, Row_Number() OVER(PARTITION BY Age ORDER BY Age, LastN) As GrpSeq FROM #data) AS Q
            PIVOT(Min(LastN) FOR GrpSeq IN(' + @cols + ')) AS P;';

EXEC    sp_executesql @sql;

我创建了第二个名为

@cols2

的变量，它执行

[1] AS Name1, ... [N] AS NameN

选择字符串，连接

CONCAT(LastN, '','', FirstN)

以创建名字 + 姓氏组合，然后执行您已经执行的操作。

我还在 STRING_AGG 中添加了“WITHIN GROUP”，以确保列的顺序正确。

条件聚合版本（上面合并的一些代码）：

;WITH GS AS(
    SELECT  LastN, Age
    ,   Row_Number() OVER(PARTITION BY Age ORDER BY Age, LastN) AS GrpSeq
    FROM    #data
    )
,DS AS (
    SELECT  DISTINCT GrpSeq
    FROM    GS
    )
,CS AS (
    SELECT  STRING_AGG(CONCAT('MAX(CASE WHEN GrpSeq = ', GrpSeq, ' THEN LastN END)', ' AS Name', GrpSeq),',') WITHIN GROUP(ORDER BY GrpSeq) AS G3
    FROM    DS
    )
SELECT  @cols = G3
FROM    CS

SELECT  @sql = '
   SELECT Age, ' + @cols + N'
   FROM (SELECT CONCAT(LastN, '','', FirstN) AS LastN, Age, Row_Number() OVER(PARTITION BY Age ORDER BY Age, LastN) As GrpSeq FROM #data) AS Q
   GROUP BY Age
'