Question

我有一个旧表，其中有很多列，我想将其拆分成3个表，并具有多对多关系。

旧表没有标识列。

旧桌子：

CustomerNumber
FirstName
LastName
Address
Postal
City
....

新表格

客户：

Id
Customernumber
Firstname
Lastname
... ect

地址：

Id
Address
Postal
City
... ect

CustomerAddress

Id
CustomerId
AddressId

现在如何使用SQL将旧表散布到新表中？

我尝试过使用MERGE，但那一次最多只能处理一张桌子。一种选择是使用CURSOR，但我读到使用它或迭代是一个坏主意，但目前这是我为此找到的唯一解决方案。

declare 
@CustomerId bigint,
@CustomerNumber float, 
@Status int,
@Address varchar(50),
@RoadNumber int,
@LastEdited datetime,

@AddressId bigint

declare my_cursor cursor
local static read_only forward_only
for
    select CustomerNumber, Address, Housenumber, Status, Date
    from [db1].dbo.OldCustomer k
    where 
        FIRMANR in (1, 40, 60, 80, 90, 120, 180, 400)

    open my_cursor
    fetch next from my_cursor into @CustomerNumber, @Address, @RoadNumber, @Status, @LastEdited
while @@FETCH_STATUS = 0
begin
    --if the customer already exists we get the identity
    if exists (select Id from [db2].dbo.Customers where CustomerNumber = @CustomerNumber)
        select @CustomerId = Id from [db2].dbo.Customers where CustomerNumber = @CustomerNumber

    --if the customer does not exit we need to insert and retrieve the new Identity value
    else
        begin
            -- insert the customer
            insert into [db2].dbo.Customers (CustomerNumber, [Status], LastEdited) values (@CustomerNumber, @Status, @LastEdited)
            set @CustomerId = SCOPE_IDENTITY()
        end

    -- get address if it already exists
    if exists (select Id from [db2].dbo.Addresses where Road = @Address and Roadnumber = @RoadNumber)
        select @AddressId = Id from [db2].dbo.Addresses where Road = @Address and Roadnumber = @RoadNumber
    else
        begin
            -- insert new addresses
            insert into [db2].dbo.Addresses (Road,Roadnumber) values (@Address, @RoadNumber)
            set @AddressId = SCOPE_IDENTITY()
        end

    -- insert customer => address reference if it does not exist
    if not exists (select Id from [db2].dbo.CustomerAddress where CustomerId = @CustomerId and AddressId = @AddressId)
        -- insert customer => address reference
        insert into [db2].dbo.CustomerAddress(CustomerId,AddressId) values (@CustomerId, @AddressId)


    fetch next from my_cursor into @CustomerNumber, @Address, @RoadNumber, @Status, @LastEdited
end

close my_cursor
deallocate my_cursor

Answer 1

这可能是最简单的：

为老客户添加一个名为AddressId的列
使用唯一的ID填充它（您可以重复使用Customer ID，或者如果其GUID使用NewID（））
为地址和客户地址创建新表
将部分旧客户数据插入每个新表中
来自客户的与地址有关的列

--populate new Address table
INSERT INTO Address(id,col1,col2...)
SELECT AddressID, col1, col2... FROM Customer

--populate new CustomerAddress table
INSERT INTO CustomerAddress(CustomerId,AddressId)
SELECT Id, AddressID FROM Customer

使用SQLS，您可以将其作为一个简单的脚本执行此操作，如果需要的话，也可以使用事务进行。不需要存储过程，游标，合并等。

不要给CustomerAddress自己的ID列； CustomerAddress的主键是CustomerId和AddressId的组合；制作复合PK，而不是单独的

Tbh，我可能没有CustomerAddress表，而在customer中只有BillingAddressId，WorkAddressId，HomeAddressId，ShippingAddressId列，但是如何管理此内容取决于您；如果您有很多地址并且类型各异，那么可以肯定地进行M：M分解，但是如果实际上您的客户最多只能有3个地址，等等，我会坚持使用一个命名列来说明该地址是什么在客户中

如果要使用CustomerAddress表，请考虑添加一列来声明地址的类型/原因

MS SQL将旧表拆分为多对多关系

1 个答案: