那些年我们写过的T-SQL（下篇）

INSERT VALUES

标准方式：INSERT INTO dbo.Orders(orderid, orderdate, empid, custid) VALUES(10001, '20160207', 3, 'A')

INSERT SELECT

使用便捷：INSERT INTO dbo.Orders(orderid, orderdate, empid, custid) SELECT 10001, '20120207', 2, 'B' UNION ALL SELECT 10002, '20120207', 3, 'C'

INSERT EXEC

INSERT INTO dbo.Orders(orderid, orderdate, empid, custid)     EXEC Sales.GetOrder @country = 'China'

SELECT INTO

SELECT courtry, region, city INTO dbo.locations FROM Sales.Customers EXCEPT SELECT courtry, region, city FROM HR.Employees

BULK INSERT

类似SSIS的导入功能 BULK INSERT dbo.Orders FROM 'C;orders.txt' WITH (DATAFILETYPE = 'char', FIELDTERMINATOR = ',', ROWTERMINATOR = 'n' )

创建标识列

CREATE TABLE dbo.T1 (keycol INT NOT NULL IDENTITY(1, 1) CONSTRAINT PK_T1 PRIMARY KEY)

插入数据（有标识列的情况下）

注意忽略标识列，INSERT INTO dbo.T1(datacol) VALUES('AAAAA')

获取当前标识号

SELECT SCOPE_IDENTITY(), @@identity, IDENT_CURRENT('Sales.Orders') 第一列获取当前作用域下的标识号，第二列获取会话生成的最后一个标识号（无论作用域），最后一个获取全局的标记号，与会话无关

显示插入标识

SET IDENTITY_INSERT表名 ON/OFF

创建序列

CREATE SEQUENCE dbo.SeqOrderIDS AS INT MIN VALUE 1 CYCLE;

使用序列

SELECT NEXT VALUE FOR dbo.SeqOrderIDS

一般删除

DELETE FROM dbo.Orders WHERE orderdate < '20160207'

TRANCATE语句

TRANCATE TABLE dbo.T1，测试时经常使用，更高效，其实相当于先删除表再创建表，而不是delete那样按条删除。DELETE是用完全日志模式，TRANCATE使用最小日志模式

基于联接的DELETE

场景：从Orders表中删除所有与Customers表中美国客户相关的行 标准方式：DELETE FROM dbo.Orders WHERE EXISTS ( SELECT * FROM dbo.Customer AS WHERE Orders.Custid = C.Custid AND C.Country = 'USA') 联接DELETE：DELETE FROM O FROM dbo.Orders AS O JOIN dbo.Customers AS C ON O.custid = C.custid WHERE C.country = 'China'

一般更新

UPDATE dbo.OrderDetails SET discount = discount + 0.05 WHERE productid = 50

基于联接的UPDATE

场景：对客户1的所有订单明细增加5%折扣 标准方式：UPDATE dbo.OrderDetails SET discount += 0.05 WHERE EXISTS( SELECT * FROM dbo.Orders AS O WHERE O.orderid = OrderDetails.orderid AND O.custid = 1) 联接UPDATE：UPDATE OD SET discount += 0.05 FROM dbo.OrderDetails AS OD JOIN dbo.Orders AS O ON OD.orderid = O.orderid WHERE O.custid =1

赋值UPDATE

DECLARE @nextval AS INT; UPDATE dbo.Sequences SET @nextval = val+= 1 WHERE id = 'SEQ1'

合并数据MERGE

MERGE INTO dbo.Customers AS TGT USING dbo.CustomerStage AS SRC ON TGT.custid = SRC.custid WHEN MATCHED THEN UPDATE SET TGT.company = SRC.companyname, TGT.phone = SRC.phone, TGT.address = SRC.address WHEN NOT MATCHED THEN INSERT (custid, companyname, phone, address) VALUES (SRC.custid, SRC.companyname, SRC.phone, SRC.address)

通过表表达式修改数据

WITH C AS( SELECT custid, OD.orderid, productid, discount, discount + 0.05 AS newdiscount FROM dbo.OrderDetails AS OD JOIN dbo.Orders AS O ON OD.orderid = O.orderid WHERE O.custid = 1 ) UPDATE C SET discount = newdiscount 补充一点是，内部的查询操作支持TOP关键字哦

OUTPUT字句

场景：从Orders表中删除所有与Customers表中美国客户相关的行 标准方式：DELETE FROM dbo.Orders WHERE EXISTS ( SELECT * FROM dbo.Customer AS WHERE Orders.Custid = C.Custid AND C.Country = 'USA') 联接DELETE：DELETE FROM O FROM dbo.Orders AS O JOIN dbo.Customers AS C ON O.custid = C.custid WHERE C.country = 'China'

INSERT OUTPUT

INSERT INTO dbo.T1(datacol) OUTPUT inserted.keycol, inserted.datacol SELECT lastname FROM HR.Employees WHERE country = 'China' 其中datacol是需要返回的属性

DELETE OUTPUT

DELETE FROM dbo.Orders OUTPUT deleted.orderid, deleted.orderid, deleted.empid, deleted.custid WHERE orderdate < '20160101'

UPDATE OUTPUT

UPDATE dbo.OrderDetails SET discount += 0.05 OUTPUT inserted.productid, deleted.discount AS olddiscount, inserted.discount AS newdiscount WHERE productid = 51

能否授予请求排它锁

No

No

No

No

能否授予请求共享锁

No

Yes

No

Yes

能否授予意向排它锁

No

No

Yes

Yes

能否授予意向排它锁

No

Yes

Yes

Yes

查询会话相关锁信息

SELECT request_session_id AS spid, resource_type AS restype, resource_database_id AS dbid,     DB_NAME(resource_database_id) AS dbname, resource_description AS res, resource_associated_entity_id AS resid,     request_mode AS mode, request_status AS status FROM sys.dm_tran_locks 其中spid表示进程ID，restype表示锁定的资源类型(KEY, PAGE, Database, object)，mode表示锁模式，status表示是否授予了锁

查询连接相关信息

SELECT session_id AS spid, connect_time, last_read, last_write, most_recent_sql_handle     FROM sys.dm_exec_connections WHERE session_id IN (60, 61) 其中connect_time表示连接时间，write&read_time表示读写时间，most_recent_sql_handle表示该连接中最近的批处理语句的句柄。接下来可以通过一个简单的APPLY表运算符获取相应的SQL语句 SELECT session_id AS spid, text     FROM sys.dm_exec_connections CROSS APPLY sys.dm_exec_sql_text(most_recent_sql_handle) AS st     WHERE session_id IN (60, 61)

查询会话相关信息

SELECT session_id as spid, login_time, host_name, program_name ,login_name, nt_user_name, last_request_start_time, last_request_end_time FROM sys.dm_exec_sessions WHERE session_id in (60, 61) 其中包括会话的登陆时间、主机名、程序名、登录名、WindowsNT用户名，最后的请求开始和结束时间等信息

查询请求相关信息

SELECT session_id AS spid, blocking_session_id, command, sql_handle, database_id, wait_type, wait_time, wait_resource FROM sys.dm_exec_requests WHERE blocking_session_id > 0 其中包括阻塞该会话的某个会话ID、阻塞的毫秒数等，可以通过blocking_session_id > 0判断是否为阻塞会话

处理阻塞

可以通过kill <spid>方式关闭会话，此外还可以设置会话中锁的时间，包括0立即超时，-1无超时（默认值），和n>0超时毫秒数

READ UNCOMMITTED

Yes

Yes

Yes

Yes

No

No

READ COMMITTED

No

Yes

Yes

Yes

No

No

READ COMMITTED SNAPSHOT

No

Yes

Yes

Yes

No

Yes

REPEATABLE READ

No

No

No

Yes

No

No

SERIALIZABLE

No

No

No

No

No

No

SNAPSHOT

No

No

No

No

Yes

Yes

NOLOCK（不加锁）

此选项被选中时，SQL Server 在读取或修改数据时不加任何锁。 在这种情况下，用户有可能读取到未完成事务（Uncommited Transaction）或回滚(Roll Back)中的数据, 即所谓的"脏数据"，等于 READ UNCOMMITTED事务隔离级别

HOLDLOCK（保持锁）

此选项被选中时，SQL Server 会将此共享锁保持至整个事务结束，而不会在途中释放，等于SERIALIZABLE事务隔离级别

UPDLOCK（修改锁）

此选项被选中时，SQL Server 在读取数据时使用修改锁来代替共享锁，并将此锁保持至整个事务或命令结束。使用此选项能够保证多个进程能同时读取数据但只有该进程能修改数据。

TABLOCK（表锁）

此选项被选中时，SQL Server 将在整个表上置共享锁直至该命令结束。 这个选项保证其他进程只能读取而不能修改数据。

PAGLOCK（页锁）

当被选中时，SQL Server 使用共享页锁。

TABLOCKX（排它表锁）

强制使用独占表级锁，这个锁在事务期间阻止任何其他事务使用这个表

READPAST

让sql server跳过任何锁定行，执行事务，适用于READ UNCOMMITTED事务隔离级别只跳过RID锁，不跳过页，区域和表锁

ROWLOCK

强制使用行锁

变量

DECLARE @i AS INT; SET @i = 10;

批

表示一个单元分析和执行的命令组，变量存在于批的生命周期中，并且一个批中只能包含一个DDL语句。 USE TSQL2012; GO

语句块和流元素

相对于PL/SQL，T-SQL中语法相对简单，结构完整性要求没有那么高 语句块: BEGIN END 逻辑流：IF BEGIN XXX END ELSE IF BEGIN XXX END ELSE BEGIN XXX END 循环流: WHILE @i < 10 BEGIN XXX END，支持BREAK和CONTINUE

游标

游标使用的步骤：1.基于查询声明游标;2.打开游标;3.从游标记录中提取属性值给变量;4.遍历游标记录并迭代;5.关闭游标;6.释放游标 DECLARE C CURSOR FAST_FORWARD FOR     SELECT custid, ordermonth, qty, FROM Sales.CustOrders ORDER BY custid, ordermonth OPEN C FETCH NEXT FROM C INTO @custid, @ordermonth, @qty SELECT @precustid = @custid, @runqty = 0; WHILE @@FETCH_STATUS = 0 BEGIN IF @custid <> @precustid SELECT @precustid = @custid, @runqty = 0 SET @runqty = @runqty + @qty INSERT INTO @Result VALUES(@custid, @ordermonth, @qty, @runqty) FETCH NEXT FROM C INTO @custid, @ordermonth, @qty END CLOSE C

用户函数、存储过程

前者之前介绍内嵌表值函数时以有例子，这儿只介绍StoreProcedure，场景为获取某客户指定日期内订单并返回记录数 CREATE PROCEDURE Sales.GetCustomerOrders     @custid AS INT, @fromdate AS DATETIME ='19010101', @todate AS DATETIME ='99991231', @numrows AS INT OUTPUT AS SET NOCOUNT NO SELECT orderid, custid, empid, orderdate FROM Sales.Orders WHERE custid = @custid AND orderdate > @fromdate AND orderdate < @todate SET @numrows = @@rowcount GO

触发器

CREATE TRIGGER trg_T1 ON dbo.T1 AFTER INSERT AS INSERT INTO dbo.T1_Audit(keycol, datacol) SELECT keycol, datacol FROM inserted GO

动态SQL

DECLARE @sql AS VARCHAR(100) SET @sql = 'PRINT ''XIONGER''' EXEC(@sql),此外为了防止SQL注入，还可以使用sp_executesql来达到参数化存储过程数据参数的目的。

错误处理

BEGIN TRY END TRY BEGIN CATCH IF XXX ELSE THROW END CATCH

本地临时表

IF OBJECT_ID('tempdbo.dbo.#MYTemp') IS NOT NULL CREATE TABLE #MYTemp(orderyear INT NOT NULL PRIMARY KEY)

全局临时表

将本地临时表中的#换成##即可

表变量

DECLARE @MyOrder TABLE(orderyear INT NOT NULL PRIMARY KEY)

创建架构

SET SCHEMA HR AUTHPRIZATION dbo

常见连接字符串

Data Source=myServerAddress;Initial Catalog=myDataBase;Integrated Security=SSPI;