How to use OUTPUT clause of SQL Server for update

Question

DECLARE @t2 AS TABLE(id INT)  

INSERT INTO dbo.EntityMaster
        (EntityType)
OUTPUT INSERTED.EntityId INTO @t2
SELECT 'G' FROM #tmp

#tmp is a temporary table that contains data loaded from an xml. I need to generate EntityId for each record contained in #tmp . It can be done by inserting record first into EntityMaster table then insert this entityid back into #tmp for each record.

Instead of inserting record into @t2 , I need to update #tmp for each record.

Any possibility?

Answer 1

Try Something like this, you still have to use the temp table but it's not too bad to read and it gets the job done.

CREATE TABLE #tmp
(
    tmpID INT IDENTITY(1,1) PRIMARY KEY CLUSTERED,
    xmlData VARCHAR(255),
    EntityId INT
)
DECLARE @t2 TABLE
(
    tmpID INT,
    EntityId INT
)

MERGE dbo.EntityMaster AS EM
USING
(
    SELECT tmpID,
        xmlData,
        EntityId
    FROM #tmp
) AS X
    ON EM.EntityId = X.EntityId
WHEN NOT MATCHED THEN
    INSERT (EntityType)
    VALUES (X.xmlData)
OUTPUT X.tmpID, INSERTED.EntityId
INTO @t2 (tmpID, EntityId);

UPDATE T
SET EntityId = T2.EntityId
FROM @t2 T2
INNER JOIN #tmp T
    ON T2.tmpID = T.tmpID

Answer 2

This would be easier to accomplish in the SQL that inserts the XML records into the #tmp table.

Consider the following @recs table, which could be thought of as a set of records generated from XML:

declare @recs table (val varchar(255))
insert into @recs values ('this'), ('is'), ('a'), ('test')

You can easily add an incrementing integer to each record like this:

select row_number() over (order by (select 1)) as id, val from @recs

The result looks like this:

id  val
1   this
2   is
3   a
4   test

Could you use row_number() over (order by (select1)) to generate the ids you need at the same time the records are being inserted into #tmp ?