T-SQL with关键字

Select字句在逻辑上是SQL语句最后进行处理的最后一步，所以，以下查询会发生错误：

SELECT
 YEAR(OrderDate) AS OrderYear,
 COUNT(DISTINCT CustomerID) AS NumCusts
FROM dbo.Orders
GROUP BY OrderYear;
因为group by是在Select之前进行的，那个时候orderYear这个列并没有形成。

如果要查询成功，可以像下面进行修改：

SELECT OrderYear, COUNT(DISTINCT CustomerID) AS NumCusts
FROM (SELECT YEAR(OrderDate) AS OrderYear, CustomerID
   FROM dbo.Orders) AS D
GROUP BY OrderYear;
还有一种很特殊的写法：

SELECT OrderYear, COUNT(DISTINCT CustomerID) AS NumCusts
FROM (SELECT YEAR(OrderDate), CustomerID
   FROM dbo.Orders) AS D(OrderYear, CustomerID)
GROUP BY OrderYear;
在作者眼里，他是非常喜欢这种写法的，因为更清晰，更明确，更便于维护。

在查询中使用参数定向产生一批结果，这个技巧没有什么好说的。

嵌套查询，在处理逻辑上是从里向外进行执行的。

多重引用，有可能你的SQL语句包含了多次从一个表进行查询后进行连接组合。比如你要比较每年的顾客数同先前年的顾客数的变化，所以你的查询就必须JOIN了2个相同的表的实例，这也是不可避免的。

Common Table Expressions (CTE)

CTE是在SQL2005新加入的一种表的表示类型。

它的定义如下：

WITH cte_name

AS

(

cte_query

)

outer_query_refferring to_cte_name;

注意：因为在标准的T-SQL语言中已经包含了WITH关键字，所以为了区分，CTE在语句的结尾加上了“;”作为停止符。

CTE实例一（结果集别名）

WITH C AS
(
 SELECT YEAR(OrderDate) AS OrderYear, CustomerID
 FROM dbo.Orders
)
SELECT OrderYear, COUNT(DISTINCT CustomerID) AS NumCusts
FROM C
GROUP BY OrderYear;
当然，作者本人有更推荐的写法：

WITH C(OrderYear, CustomerID) AS
(
 SELECT YEAR(OrderDate), CustomerID
 FROM dbo.Orders
)
SELECT OrderYear, COUNT(DISTINCT CustomerID) AS NumCusts
FROM C
GROUP BY OrderYear;
CTE实例二（多重CTEs）

WITH C1 AS
(
 SELECT YEAR(OrderDate) AS OrderYear, CustomerID
 FROM dbo.Orders
),
C2 AS
(
 SELECT OrderYear, COUNT(DISTINCT CustomerID) AS NumCusts
 FROM C1
 GROUP BY OrderYear
)
SELECT OrderYear, NumCusts
FROM C2
WHERE NumCusts > 70;
CTE实例三（多重引用）

WITH YearlyCount AS
(
 SELECT YEAR(OrderDate) AS OrderYear,
  COUNT(DISTINCT CustomerID) AS NumCusts
 FROM dbo.Orders
 GROUP BY YEAR(OrderDate)
)
SELECT Cur.OrderYear,
 Cur.NumCusts AS CurNumCusts, Prv.NumCusts AS PrvNumCusts,
 Cur.NumCusts - Prv.NumCusts AS Growth
FROM YearlyCount AS Cur
 LEFT OUTER JOIN YearlyCount AS Prv
  ON Cur.OrderYear = Prv.OrderYear + 1;
CTE实例四（修改数据）

1.把从customer表查询出来的结果，动态的组装进新表CustomersDups里：

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

WITH CrossCustomers AS
(
 SELECT 1 AS c, C1.*
 FROM dbo.Customers AS C1, dbo.Customers AS C2
)
SELECT ROW_NUMBER() OVER(ORDER BY c) AS KeyCol,
 CustomerID, CompanyName, ContactName, ContactTitle, Address,
 City, Region, PostalCode, Country, Phone, Fax
INTO dbo.CustomersDups
FROM CrossCustomers;

2.使用CTE移除数据，只保留CustomerDups表里同一CustomerID里KeyCol为最大的记录。

WITH JustDups AS
(
 SELECT * FROM dbo.CustomersDups AS C1
 WHERE KeyCol <
  (SELECT MAX(KeyCol) FROM dbo.CustomersDups AS C2
   WHERE C2.CustomerID = C1.CustomerID)
)
DELETE FROM JustDups;
CTE实例五（对象容器）

即提供了封装的能力，有利于组件化的编程。作者额外的提醒，CTE无法直接内嵌，但是可以通过把CTE封装进一个对象容器里并从一个外部的CTE里对这容器的数据进行查询而实现内嵌。

作者也说明了，使用CTEs在VIEW和UDFs里是没有什么价值的。

有个例子，如下:

CREATE VIEW dbo.VYearCnt
AS
WITH YearCnt AS
(
 SELECT YEAR(OrderDate) AS OrderYear,
  COUNT(DISTINCT CustomerID) AS NumCusts
 FROM dbo.Orders
 GROUP BY YEAR(Orde