C#3.0入门系列(十)-之Join操作
本节讲join操作。我们知道,T-sql中,有三种最基本的join,inner join, left join, 和right join。 而dlinq并不支持right join。道理很简单,right join以right表为基础,left表中没有对应记录的,将以null值填充。而dlinq以left表做为主表创建对象。如果一个对象为null,你如何获取它的其他的属性呢?
在C# 3.0入门系列(四)-之Select操作一文中,我们提到了query expression首先会被翻译成标准的API, 而dlinq在join操作中,一共为我们提供了三个API.它们是Join, SelectMany和GroupJoin
Join
在101 的sample中,并没有join的例子。当一个query expression 有join字句时,而没有into字句,它将会被翻译成join方法。如,以Customers为主表,Orders为子表,用CustomerID 做关联进行join操作。
from c in db.Customers
join o in db.Orders on c.CustomerID equals o.CustomerID
select new { c.CustomerID, o.EmployeeID };
该语句所产生的T-sql语句为
FROM [Customers] AS [t0]
INNER JOIN [Orders] AS [t1] ON [t0].[CustomerID] = [t1].[CustomerID]
SelectMany
在101sample中,给了4个SelectMany的例子。会被翻译成SelectMany需要满足2个条件。1,query语句中没有join和into,2,必须出现EntitySet。 关于EntitySet,请参考C#3.0进阶系列(一)-从映射讲起
先看第一个例子
from c in db.Customers
from o in c.Orders
where c.City == "London"
select o;
[Association(Name="Order_OrderDetail", Storage="_OrderDetails", OtherKey="OrderID")]
[Association(Name="Order_OrderDetail", Storage="_Order", ThisKey="OrderID", IsForeignKey=true)]
所以,你就不用担心,dlinq是否知道该按那个键进行关联。有兴趣的朋友,可以自己修改这里的OtherKey和ThisKey的值,看看翻译的T-sql语句是否变了。
第二个例子
from p in db.Products
where p.Supplier.Country == "USA" && p.UnitsInStock == 0
select p;
FROM [dbo].[Products] AS [t0]
LEFT OUTER JOIN [dbo].[Suppliers] AS [t1] ON [t1].[SupplierID] = [t0].[SupplierID]
WHERE ([t1].[Country] = @p0) AND ([t0].[UnitsInStock] = @p1)
-- @p0: Input String (Size = 3; Prec = 0; Scale = 0) [USA]
-- @p1: Input Int32 (Size = 0; Prec = 0; Scale = 0) [0]
-- Context: SqlProvider(Sql2005) Model: AttributedMetaModel Build: 2.0.20612.0
from e in db.Employees
from et in e.EmployeeTerritories
where e.City == "Seattle"
select new {e.FirstName, e.LastName, et.Territory.TerritoryDescription};
FROM [dbo].[Employees] AS [t0]
CROSS JOIN [dbo].[EmployeeTerritories] AS [t1]
INNER JOIN [dbo].[Territories] AS [t2] ON [t2].[TerritoryID] = [t1].[TerritoryID]
WHERE ([t0].[City] = @p0) AND ([t1].[EmployeeID] = [t0].[EmployeeID])
-- @p0: Input String (Size = 7; Prec = 0; Scale = 0) [Seattle]
-- Context: SqlProvider(Sql2005) Model: AttributedMetaModel Build: 2.0.20612.0
from e1 in db.Employees
from e2 in e1.Employees
where e1.City == e2.City
select new {
FirstName1 = e1.FirstName, LastName1 = e1.LastName,
FirstName2 = e2.FirstName, LastName2 = e2.LastName,
e1.City
};
FROM [dbo].[Employees] AS [t0], [dbo].[Employees] AS [t1]
WHERE ([t0].[City] = [t1].[City]) AND ([t1].[ReportsTo] = [t0].[EmployeeID])
-- Context: SqlProvider(Sql2005) Model: AttributedMetaModel Build: 2.0.20612.0
GroupJoin
当dlinq语句中,有join而且还有into时,它会被翻译为GroupJoin.我们先来看第一个例子。
from c in db.Customers
join o in db.Orders on c.CustomerID equals o.CustomerID into orders
select new {c.ContactName, OrderCount = orders.Count()};
本系列曾在C#3.0入门系列(八)-之GroupBy操作 一文中,第一次谈到到into。into的概念是对其结果进行重新命名。为什么需要重新命名呢?我们以本例为例。One To Many的关系中,左边是one,它每条记录叫做c(from c in db.Customers),右边是many,其每条记录叫做o ( join o in db.Orders ),每对应左边的一个c,都会有一组o,那这一组o,就叫做orders,也就是说,我们把一组o命名为orders,这就是into用途。(和groupby中类似)。这也就是为什么在select语句中,orders可以调用聚合函数Count。
在这个例子中,翻译的t-sql为
SELECT COUNT(*)
FROM [dbo].[Orders] AS [t1]
WHERE [t0].[CustomerID] = [t1].[CustomerID]
) AS [value]
FROM [dbo].[Customers] AS [t0]
-- Context: SqlProvider(Sql2005) Model: AttributedMetaModel Build: 2.0.20612.0
dlinq很聪明,直接用其内欠的t-sql返回值作为字段值。
第二个例子
var q =
from c in db.Customers
join o in db.Orders on c.CustomerID equals o.CustomerID into ords
join e in db.Employees on c.City equals e.City into emps
select new {c.ContactName, ords=ords.Count(), emps=emps.Count()};
三个表联合查询。在其join语句后,紧跟着又是一个join.只是表多了些,并没有太多新鲜的东西。
第三个例子
from e in db.Employees
join o in db.Orders on e equals o.Employee into ords
from o in ords.DefaultIfEmpty()
select new {e.FirstName, e.LastName, Order = o};
Left Out Join的标准写法。以Employees为左表,Orders 为右,Orders 表中为空时,填冲null值。在将join的结果重命名后,再使用DefaultEmpty()函数,对其再次查询。大家需要注意的时,其最后的结果中有个Order,因为from o in ords.DefaultIfEmpty() 是对ords组再一次遍历,所以,最后结果中的Order并不是一个集合。但是,如果没有from o in ords.DefaultIfEmpty() 这句,最后的select语句写成select new { e.FirstName, e.LastName, Order = ords }的话,那Order就是一个集合
上例翻译的T-sql 为
FROM [dbo].[Employees] AS [t0]
LEFT OUTER JOIN (
SELECT 1 AS [test], [t1].[OrderID], [t1].[CustomerID], [t1].[EmployeeID], [t1].[OrderDate], [t1].[RequiredDate], [t1].[ShippedDate], [t1].[ShipVia], [t1].[Freight], [t1].[ShipName], [t1].[ShipAddress], [t1].[ShipCity], [t1].[ShipRegion], [t1].[ShipPostalCode], [t1].[ShipCountry]
FROM [dbo].[Orders] AS [t1]
) AS [t2] ON [t0].[EmployeeID] = [t2].[EmployeeID]
-- Context: SqlProvider(Sql2005) Model: AttributedMetaModel Build: 2.0.20612.0
第四个例子,let语句
from c in db.Customers
join o in db.Orders on c.CustomerID equals o.CustomerID into ords
let z = c.City + c.Country
from o in ords
select new {c.ContactName, o.OrderID, z};
let语句有点类似into,也是个重命名的概念。需要提醒大家的是,let只要是放在第一个from后,select语句前就是符合语法的。上面的语句和下面这条是等价的。
from c in db.Customers
let z = c.City + c.Country
join o in db.Orders on c.CustomerID equals o.CustomerID into ords
from o in ords
select new { c.ContactName, o.OrderID, z };
其产生的T-sql均为:
FROM (
SELECT [t0].[City] + [t0].[Country] AS [value], [t0].[CustomerID], [t0].[ContactName]
FROM [dbo].[Customers] AS [t0]
) AS [t1]
CROSS JOIN [dbo].[Orders] AS [t2]
WHERE [t1].[CustomerID] = [t2].[CustomerID]
-- Context: SqlProvider(Sql2005) Model: AttributedMetaModel Build: 2.0.20612.0
它也应该和下面的语句等价,但其翻译的T-sql语句稍微有所不同。
from c in db.Customers
join o in db.Orders on c.CustomerID equals o.CustomerID into ords
from o in ords
let z = c.City + c.Country
select new { c.ContactName, o.OrderID, z };
有兴趣的朋友可以研究下,其产生的T-sql 为
FROM (
SELECT [t0].[City] + [t0].[Country] AS [value], [t0].[CustomerID], [t0].[ContactName], [t1].[OrderID], [t1].[CustomerID] AS [CustomerID2]
FROM [Customers] AS [t0], [Orders] AS [t1]
) AS [t2]
WHERE [t2].[CustomerID] = [t2].[CustomerID2]
-- Context: SqlProvider(Sql2005) Model: AttributedMetaModel Build: 2.0.20612.0
第五个例子为composite key.
from o in db.Orders
from p in db.Products
join d in db.OrderDetails
on new {o.OrderID, p.ProductID} equals new {d.OrderID, d.ProductID}
into details
from d in details
select new {o.OrderID, p.ProductID, d.UnitPrice};
这里,它使用三个表,并且用匿名类来表示它们之间的关系。因为,其之间的关系已经不是一个键可以描述清楚的,所以只有用匿名类,表示组合键。这个例子有点像SelectMany中的ManyToMany的那个。
还有一种composite key的,就是两个表之间是用composite key表示关系的。这种情况很简单,不需像该例中使用匿名类。该例翻译的T-sql为
FROM [dbo].[Orders] AS [t0], [dbo].[Products] AS [t1], [dbo].[Order Details] AS [t2]
WHERE ([t0].[OrderID] = [t2].[OrderID]) AND ([t1].[ProductID] = [t2].[ProductID])
-- Context: SqlProvider(Sql2005) Model: AttributedMetaModel Build: 2.0.20612.0
最后一个例子,没有看出什么好玩的来,不讲了。
写到这里,c#3.0的入门系列已经接近尾声了。我们一起学习了Dlinq的最基本操作。还剩下Union, In, Like还有一些聚合函数等操作。将会在下面几章中介绍。不知道大家对什么还感兴趣的,或者我能够提供帮助的,尽管问。
关于Linq To Sql 中的,Create, update, Delete 操作,以及Store procedure 及UDF等,更像是运用函数,而不是语言。所以,不在C#语言中讲。在考虑是不是开个什么Linq To Sql的深入应用。
写blog是对自己个人知识的总结,也是对自己表达功底的考验。因本人水平有限,错误再所难免,还请大家指出并谅解。