在合并过程中将选择用作源并插入到多个表中

Question

I can do plenty with the basics of SQL, but I'm struggling with this more complex task. I have three tables set up like this.

Person

• identifier
• attribute
• type

Employee

• identifier
• hobby_id

Hobby

• hobby_id
• attribute

I need to select rows from Person of specific types that exist in the employee table. Then, if the employee table has a hobby_id (not null) then I need to update the Hobby row with the attribute from the corresponding person row. If the employee row does NOT have a hobby_id, I need to create a new one with the person attribute and set the employee Hobby_id to the new id.

I've tried a lot of approaches. I started with an IF/ELSE syntax that was getting really messy so I tried a CASE statement, but it appears I can't do inserts or updates in those. I'm currently trying to use a MERGE because it looked robust enough to work. My current 'iteration' that I'm working with is:

USE MY_DB

SET @nextHobbyId = (SELECT MAX( HOBBY_ID ) FROM HOBBY)

MERGE DBO.HOBBY AS TARGET

USING (
    SELECT ATTRIBUTE, HOBBY_ID
    FROM AA.PERSONe 
        INNER JOIN DBO.EMPLOYEE v 
        ON e.IDENTIFIER= v.IDENTIFIER
) AS SOURCE (hobbyId, ATTRIBUTE)

ON (TARGET.HOBBY_ID = SOURCE.HOBBY_ID) 

WHEN MATCHED 
THEN UPDATE SET 
    TARGET.ATTRIBUTE= SOURCE.ATTRIBUTE

WHEN NOT MATCHED BY TARGET 
THEN INSERT (HOBBY_ID, ATTRIBUTE)
VALUES (@nextHobbyId + 1, ATTRIBUTE);

You can see I don't even attempt to update the employee table with the new hobby_id in this iteration because I currently get an error that "ATTRIBUTE is an invalid column name" on the very last line.

Things to note, hobby ID does not have an identity but the code in our app that used to be the only thing to touch the table used something like :

SELECT NEXT VALUE FOR...

when doing inserts so I try to mimic that by selecting max value + 1. While my code will likely never run at the same time as the original code, I'm worried about a possible race condition.

How can I make this work? I'm not attached to a merge if its not the way to go.

Answer 1

If I understood correctly, it can be done fairly simple:

;with CTE as (
    select * from  Person where identifier is in (select identifier from Employees
)

update Hobby set ... --some updates
where hobby_id in (select hobby_id from CTE where hobby_id is not null)

insert into Hobby
... -- here, through join most probably you can perform insert or any other updating operation