T-SQL 枢轴？从行值创建表列的可能性

Posted 2023-02-19

【中文标题】T-SQL 枢轴？从行值创建表列的可能性

【英文标题】：T-SQL Pivot? Possibility of creating table columns from row values

【发布时间】：2011-02-24 18:04:33

【问题描述】：

实际上是否可以旋转 T-SQL (2005) 以便（为了论证）第一列的行的值成为输出表列的标题？

我意识到这并不是 PIVOT 的真正用途，但这正是我所需要的 - 能够请求预先不知道列的表，因为它们已作为值输入到表中。

即使是 hack 也不错，tbh。

【参考方案1】：

Itzik Ben-Gan 关于如何构建动态 PIVOT 的示例，我强烈推荐他的 Inside Microsoft SQL Server 2008: T-SQL Programming 书

-- Creating and Populating the Orders Table
USE tempdb;
GO

IF OBJECT_ID('dbo.Orders') IS NOT NULL
DROP TABLE dbo.Orders;
GO

CREATE TABLE dbo.Orders
(
orderid   int        NOT NULL PRIMARY KEY NONCLUSTERED,
orderdate datetime   NOT NULL,
empid     int        NOT NULL,
custid    varchar(5) NOT NULL,
qty       int        NOT NULL
);

CREATE UNIQUE CLUSTERED INDEX idx_orderdate_orderid
ON dbo.Orders(orderdate, orderid);

INSERT INTO dbo.Orders(orderid, orderdate, empid, custid, qty)
VALUES(30001, '20020802', 3, 'A', 10);
INSERT INTO dbo.Orders(orderid, orderdate, empid, custid, qty)
VALUES(10001, '20021224', 1, 'A', 12);
INSERT INTO dbo.Orders(orderid, orderdate, empid, custid, qty)
VALUES(10005, '20021224', 1, 'B', 20);
INSERT INTO dbo.Orders(orderid, orderdate, empid, custid, qty)
VALUES(40001, '20030109', 4, 'A', 40);
INSERT INTO dbo.Orders(orderid, orderdate, empid, custid, qty)
VALUES(10006, '20030118', 1, 'C', 14);
INSERT INTO dbo.Orders(orderid, orderdate, empid, custid, qty)
VALUES(20001, '20030212', 2, 'B', 12);
INSERT INTO dbo.Orders(orderid, orderdate, empid, custid, qty)
VALUES(40005, '20040212', 4, 'A', 10);
INSERT INTO dbo.Orders(orderid, orderdate, empid, custid, qty)
VALUES(20002, '20040216', 2, 'C', 20);
INSERT INTO dbo.Orders(orderid, orderdate, empid, custid, qty)
VALUES(30003, '20040418', 3, 'B', 15);
INSERT INTO dbo.Orders(orderid, orderdate, empid, custid, qty)
VALUES(30004, '20020418', 3, 'C', 22);
INSERT INTO dbo.Orders(orderid, orderdate, empid, custid, qty)
VALUES(30007, '20020907', 3, 'D', 30);
GO

-- Static PIVOT
SELECT *
FROM (SELECT custid, YEAR(orderdate) AS orderyear, qty
FROM dbo.Orders) AS D
PIVOT(SUM(qty) FOR orderyear IN([2002],[2003],[2004])) AS P;
GO

-- Dynamic PIVOT
DECLARE @T AS TABLE(y INT NOT NULL PRIMARY KEY);

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

-- Construct the column list for the IN clause
-- e.g., [2002],[2003],[2004]
SET @cols = STUFF(
(SELECT N',' + QUOTENAME(y) AS [text()]
FROM (SELECT DISTINCT YEAR(orderdate) AS y FROM dbo.Orders) AS Y
ORDER BY y
FOR XML PATH('')),
1, 1, N'');

-- Construct the full T-SQL statement
-- and execute dynamically
SET @sql = N'SELECT *
FROM (SELECT custid, YEAR(orderdate) AS orderyear, qty
FROM dbo.Orders) AS D
PIVOT(SUM(qty) FOR orderyear IN(' + @cols + N')) AS P;';

EXEC sp_executesql @sql;
GO

【讨论】：

倒数第二行只需要 EXEC (@sql)

@JohnnyBones 你为什么会这样想？ EXEC sp_executesql @sql 是正确的语法。如果您尝试EXEC (@sql)，它将失败，因为它将尝试使用该文本查找存储过程。

【参考方案2】：

一个稍微好一点的pivot查询如下：

-- Static PIVOT
WITH PivotData AS
(
SELECT custid, YEAR(orderdate) AS orderyear, qty
FROM dbo.Orders
)
SELECT custid, [2002], [2003], [2004]
FROM PivotData
PIVOT(SUM(qty) FOR orderyear IN([2002],[2003],[2004])) AS P;

与派生表相比，我更喜欢公用表表达式 (CTE) 的样式，因为我认为它更容易理解。 Itzik 也是如此，他在他的书 Querying Microsoft SQL Server 2012 中推荐了这种风格。