How can I use a Table-Valued Parameter to insert multiple rows and then return their IDs?

Question

In my application, I have a large of number (100+) of rows that I need to insert into the database. Once they get inserted into the database, I need to insert their children, which have a foreign key reference back to the children.

I'm wondering if there's a way to write a stored procedure that can insert all of those rows and return their IDs back to my application?

Answer 1

You have tagged your question with table-value-parameters - you can pass one these to the stored procedure for inserting to the database.

You can use the OUTPUT clause with the INSERTED logical table to get the new ID values and return those from the stored procedure.

Answer 2

Here's an example of using the output clause, like Oded suggested. First, some setup code:

if exists (select * from tempdb.sys.tables where name like '#tmp%')
    drop table #tmp
create table #tmp (id int identity, FirstName varchar(50), LastName varchar(50))

if exists (select * from sys.procedures where name = 'TestProcedure')
    drop procedure TestProcedure
if exists (select * from sys.types where name = 'TestTableType')
    drop type TestTableType

create type TestTableType as table (FirstName varchar(50), LastName varchar(50))
go

Now we can create the stored procedure:

create procedure dbo.TestProcedure
    @List TestTableType readonly
as
insert  #tmp
        (FirstName, LastName)
output  inserted.*
select  FirstName
,       LastName
from    @List l
go

Note the output clause, which tells SQL Server to return the inserted rows back to the client, including the value of the identity column. Now that the code is all set up, we can test it:

declare @List TestTableType
insert @List values ('Rick','Cain'),
    ('Herman', 'Gingrich'),
    ('Newt', 'Paul'),
    ('Ron', 'Perry')

exec  dbo.TestProcedure @List

select * from #tmp

You'll see that the values returned by the procedure match the values in #tmp exactly.

Answer 3

I would do something similar to what Andomar suggests, but using the "INSERT-EXEC" technique described here :

http://www.sommarskog.se/tableparam.html

In the link, search for : INSERT-EXEC

Answer 4

I'm still researching how to do this, but it seems to me if you want to insert a parent table and a child table, it could be done with one sql call by passing both tables in as table value parameters. If you have some artificial identity (relationship) which currently links the two tables like a row number, but is not necesarily stored in either DB table, then when the parent is inserted and gets its identity value, it could be joined to the child table using the unique relationship and the child's foreign keys could be updated. The child could then be inserted.

The output is interesting, but I don't actually need the output, it simply has to be used within the sproc, so i have to do more research to see if the output can be used by, for example, outputing it to a temp table.

Answering my own question the syntax for Output is:

[ OUTPUT <dml_select_list> INTO { @table_variable | output_table } [ ( column_list ) ] ]

so it can be used to create a temp table that is joined with a child.