将外键与SQL Server中的详细信息表关联的最佳方法是什么？

Question

I'm trying to aggregate details and creating a header record for them. What is the best way to assign the foreign key back into the details?

The details table will have an ID(identity) and HeaderID (foreign key to headers table). After aggregating the details and inserting into the Header table, how do I associate the Header ID back into the individual details?

I thought about using triggers to update the Details table with the @@Identity from the Header ID. However, since there is no association between the two just yet (that's what I'm trying to build), I'm not sure what to have in the WHERE clause to do the update.

The below query should give you some sample data. If you execute it, where the NULL values are in the #OrderLine table is where I'm having an issue associating the OrderId back.

Create Table #Order
(OrderId int identity primary key,
TypeId char(1),
Total decimal(5,2)
)

Create Table #OrderLine
(OrderLineId int identity,
OrderId int constraint FK_OrderId foreign key (OrderId) references [#Order](OrderId),
LineItem int,
TypeId char(1),
Item varchar(30),
Price decimal (5,2)
)

Insert into #OrderLine (TypeId,Item, Price)
Values('S', 'Tennis Racket', 120)
Insert into #OrderLine (TypeId,Item, Price)
Values('C', 'Red Dress', 80)
Insert into #OrderLine (TypeId,Item, Price)
Values('S', 'Basketball', 30)
Insert into #OrderLine (TypeId,Item, Price)
Values('C', 'Dress Shirt', 60)
Insert into #OrderLine (TypeId,Item, Price)
Values('S', 'Pingpong Balls', 10)
Insert into #OrderLine (TypeId,Item, Price)
Values('S', 'Soccer Ball', 25)
Insert into #OrderLine (TypeId,Item, Price)
Values('C', 'Shorts', 20)

Insert into #Order(TypeId, Total)
Select  #OrderLine.TypeId
        ,SUM(#OrderLine.Price)
From    #OrderLine
Group by #OrderLine.TypeId

Select * From #Order
Select * From #OrderLine

Drop Table #Order
Drop Table #OrderLine

Answer 1

I think I found one way to do. For the LineItem, I used ROW_NUMBER() and PARTITION. For the OrderId, I just updated the details based on the aggregation criteria for where the OrderId is NULL.

If anyone has a better suggestion, please let me know.

Create Table #Order
(OrderId int identity primary key,
TypeId char(1),
Total decimal(5,2)
)

Create Table #OrderLine
(OrderLineId int identity,
OrderId int constraint FK_OrderId foreign key (OrderId) references [#Order](OrderId),
LineItem int,
TypeId char(1),
Item varchar(30),
Price decimal (5,2)
)

Insert into #OrderLine (TypeId,Item, Price)
Values('S', 'Tennis Racket', 120)
Insert into #OrderLine (TypeId,Item, Price)
Values('C', 'Red Dress', 80)
Insert into #OrderLine (TypeId,Item, Price)
Values('S', 'Basketball', 30)
Insert into #OrderLine (TypeId,Item, Price)
Values('C', 'Dress Shirt', 60)
Insert into #OrderLine (TypeId,Item, Price)
Values('S', 'Pingpong Balls', 10)
Insert into #OrderLine (TypeId,Item, Price)
Values('S', 'Soccer Ball', 25)
Insert into #OrderLine (TypeId,Item, Price)
Values('C', 'Shorts', 20)

Insert into #Order(TypeId, Total)
Select  #OrderLine.TypeId
        ,SUM(#OrderLine.Price)
From    #OrderLine
Group by #OrderLine.TypeId

--Update OrderId back to #OrderLine
Update #OrderLine
Set OrderId = #Order.OrderId
From    #Order
Where   #Order.TypeId = #Orderline.TypeId
and     #OrderLine.OrderId is null

--Assign the LineItem per Order
Update #OrderLine
Set LineItem = TT.LineItemId
From    #OrderLine as T
join    (Select ROW_NUMBER() OVER (PARTITION BY OrderId, TypeId ORDER BY OrderLineId) as LineItemId
            ,OrderLineId
            ,OrderId
            ,TypeId
            From #OrderLine) as TT on TT.OrderLineId = T.OrderLineId
Where   T.LineItem is null

Select * From #OrderLine
Select * From #Order

Drop Table #Order
Drop Table #OrderLine