INSERT SELECT and IGNORE_DUP_KEY, how can I retrieve discarded rows? - sql

Question

Using MS-SQL, I have an INSERT ... SELECT statement that populates a table that has a unique key with IGNORE_DUP_KEY = ON.
Is there an easy way I could get the rows that were discarded because they were duplicated? (Preferably after the statement has completed)

Answer 1

The OUTPUT clause is the key here.
create table dbo.IDKsource (
SNumber int not null,
SText varchar(100) not null
) ;
go
insert into dbo.IDKsource
values ( 1, 'aaaaa' ), ( 2, 'bbbbb' ), ( 1, 'cccccc' ), ( 3, 'dddddd' ) ;
go
create table dbo.IDKOntarget (
SNumber int not null unique with ( ignore_dup_key = on ),
SText varchar(100) not null
) ;
go
-- The following lines must all be in one batch!
declare #RecordsWereInserted table (
SNumber int not null ,
SText varchar(100) not null
) ;
insert into dbo.IDKOntarget ( SNumber , SText )
output inserted.* into #RecordsWereInserted
select SNumber , SText
from dbo.IDKSource ;
select SNumber , SText
from dbo.IDKsource
except
select SNumber , SText
from #RecordsWereInserted ;
If you want to keep the inserted data for longer than one batch (or your version of SQL Server does not support table variables) then replace my table variable RecordsWereInserted with an actual table.
Note: My first approach was to use the INSERT statement directly with the EXCEPT but SQL Server will not allow DML statements with EXCEPT, INTERSECT or UNION.