Question

我有一个表格，其中包含有关客户，客户（姓名，地址）的数据，其中包含“John Doe”，“Some Street 123”等行。对于表中的每一行，我想在Person（id，name）表中插入一行，并在Address（id，person_id，address）表中插入一行。

我可以通过为Customer中的每一行运行两个insert语句来实现此目的：

insert into Person(name) values (@name);
insert into Address(person_id, address) values (scope_identity(), @address);

但这效率低下。我想批量进行插入，有点像这样：

-- This works, the problem is with the Address table...
insert into Person(name)
select name from Customer

-- This looks good but does not work because name is not unique.
insert into Address(person_id, address)
select p.person_id, c.address
from Customer c join Person p on c.name = p.name

Answer 1

没有办法解释，因为你丢失了第一次插入的每一行的scope_identity（）值。

解决方法可能是将Customer主键字段添加到Person表中，然后使用以下字段连接第二个插入字段：

在插入之前在Person上创建customerID字段

alter table Person add customerID int null;

然后批量插入：

-- inserting customerID
insert into Person(name, customerID)
select name, customerID from Customer

-- joining on customerID.
insert into Address(person_id, address)
select p.person_id, c.address
  from Customer c 
  join Person p on c.customerID = p.customerID

之后，您可以从Person表中删除customerID字段：

alter table Person drop column customerID

Answer 2

将其留给像我这样找到这篇文章的Google旅行者。

我找到了这个解决方案，它似乎很好用，并且不需要任何时髦的模式更改： https://dba.stackexchange.com/questions/160210/splitting-data-into-two-tables-in-one-go

他们使用MERGE语句对第一个表（正在生成将在其他地方使用的标识的表）执行初始插入。之所以使用MERGE语句，是因为它允许您使用OUTPUT语句，您可以使用该语句输出新的标识值以及源表中的标识值（相对于在标准OUTPUT上使用INSERT语句，不允许您输出源表标识）。您可以将此输出数据插入映射表，然后使用该映射表执行第二次插入。

这是该解决方案的示例代码：

------------------------------------------------------------------------------

------------------------------------------------------------------------------
-- Set up sample schema and data
------------------------------------------------------------------------------
--Source Data
IF OBJECT_ID('dbo.tmp1') IS NOT NULL DROP TABLE dbo.tmp1 --SELECT * FROM dbo.tmp1
CREATE TABLE dbo.tmp1 (tmp1ID INT IDENTITY(1,1), Col1 CHAR(1) NOT NULL, Col2 CHAR(1) NOT NULL, Col3 CHAR(1) NOT NULL, Col4 CHAR(1) NOT NULL, Col5 CHAR(1) NOT NULL, Col6 CHAR(1) NOT NULL)

INSERT INTO dbo.tmp1 (Col1, Col2, Col3, Col4, Col5, Col6)
SELECT x.c1, x.c2, x.c3, x.c4, x.c5, x.c6
FROM (VALUES ('A','B','C','D','E','F'),
             ('G','H','I','J','K','L'),
             ('M','N','O','P','Q','R')
) x(c1,c2,c3,c4,c5,c6)

IF OBJECT_ID('dbo.tmp3') IS NOT NULL DROP TABLE dbo.tmp3 --SELECT * FROM dbo.tmp3
IF OBJECT_ID('dbo.tmp2') IS NOT NULL DROP TABLE dbo.tmp2 --SELECT * FROM dbo.tmp2

--Taget tables to split into
CREATE TABLE dbo.tmp2 (
      tmp2ID INT IDENTITY(1,1) NOT NULL CONSTRAINT PK_tmp2 PRIMARY KEY CLUSTERED (tmp2ID ASC)
    , Col1 CHAR(1) NOT NULL
    , Col2 CHAR(1) NOT NULL
    , Col3 CHAR(1) NOT NULL
)

CREATE TABLE dbo.tmp3 (
      tmp2ID INT NOT NULL
    , Col4 CHAR(1) NOT NULL
    , Col5 CHAR(1) NOT NULL
    , Col6 CHAR(1) NOT NULL
    , CONSTRAINT FK_tmp3_tmp2ID FOREIGN KEY(tmp2ID) REFERENCES dbo.tmp2 (tmp2ID)
)
------------------------------------------------------------------------------

------------------------------------------------------------------------------
-- Split data into two tables
------------------------------------------------------------------------------
DECLARE @Mapping TABLE (tmp1ID INT NOT NULL, tmp2ID INT NOT NULL);

--Use merge statment to output the source data PK as well as the newly inserted identity to generate a mapping table
MERGE INTO dbo.tmp2 AS tgt
USING dbo.tmp1 AS src ON (1=0)
WHEN NOT MATCHED THEN
    INSERT (    Col1,     Col2,     Col3)
    VALUES (src.Col1, src.Col2, src.Col3)
OUTPUT src.tmp1ID, Inserted.tmp2ID INTO @Mapping (tmp1ID, tmp2ID);

--Use the mapping table to insert the split data into the second table
INSERT INTO dbo.tmp3 (tmp2ID, Col4, Col5, Col6)
SELECT t2.tmp2ID, t1.Col4, t1.Col5, t1.Col6
FROM dbo.tmp2 t2
    JOIN @Mapping m ON m.tmp2ID = t2.tmp2ID
    JOIN dbo.tmp1 t1 ON t1.tmp1ID = m.tmp1ID

SELECT tmp2ID, Col1, Col2, Col3 FROM dbo.tmp2
SELECT tmp2ID, Col4, Col5, Col6 FROM dbo.tmp3
------------------------------------------------------------------------------

------------------------------------------------------------------------------
-- Clean up
------------------------------------------------------------------------------
DROP TABLE dbo.tmp1
DROP TABLE dbo.tmp3
DROP TABLE dbo.tmp2
------------------------------------------------------------------------------

------------------------------------------------------------------------------
GO

Answer 3

最好在两个表中创建一些独特类型的字段与它们相关。另外你想要加入，因为你没有条件的唯一字段

从属插入语句

3 个答案: