Not allowing Multiple CASE Subquery Results when using IN - sql-server

Question

I'm working with the following query:
SELECT *
FROM TableA a
WHERE a.FieldA IN (
CASE
--select subquery returns a single value
WHEN a.FieldB = 'Value1'
THEN (select b.ID from TableB b where b.FK_Field = '123')
--select subquery returns multiple values
WHEN a.FieldB = 'Value2'
THEN (select c.ID from TableC c where c.FK_Field = '123')
END
)
The first case select statement returns only a single b.ID. If I just have that statement, my code works.
The second case statement, however, returns multiple c.IDs. When I add that check, I get the following error:
Subquery returned more than 1 value. This is not permitted when the
subquery follows =, !=, <, <= , >, >= or when the subquery is used as
an expression.
If I would have WHERE a.FieldA =, then I understand that the subquery can only return 1 value. I however have WHERE a.FieldA IN, so why is it complaining if there are multiple values returned?
How can I implement this kind of check?

Answer 1

As #marc_s explained in a comment:
CASE in T-SQL is an expression (like a+b) which returns a single,
atomic value - you cannot use it to selectively run SQL snippets that
return entire result sets
In order to resolve this error, I removed the CASE statement and instead used a bunch of AND and OR statements to accomplish the same kind of check.
SELECT *
FROM TableA a
WHERE
(a.FieldB = 'Value1'
AND a.FieldA IN (select b.ID from TableB b where b.FK_Field = '123'))
OR (a.FieldB = 'Value2'
AND a.FieldA IN (select c.ID from TableC c where c.FK_Field = '123'))
This code is a bit messier than a CASE statement, but it works.

Answer 2

Lots of ways of doing this, here is one way using union all and a correlated EXISTS statement
;WITH cte AS (
SELECT 'Value1' as FieldB, b.Id
FROM
TableB
WHERE
b.FK_FieldId = '123'
UNION ALL
SELECT 'Value1' as FieldB, c.Id
FROM
TableB
WHERE
c.FK_FieldId = '123'
)
SELECT *
FROM
TableA a
WHERE
EXISTS (SELECT 1
FROM
cte c
WHERE
a.FieldB = c.FieldB
AND a.FieldA = c.Id)
The problem with the way you have written it is that you are getting a non-scalar value (meaning more than 1 row) where sql is expecting a scalar value. In the case expression only scalar values can be used in the THEN part as well as some rules in WHEN as well. To solve you need to break apart your case expression to multiple where statements and/or use some other technique such as the one above.
Or you could write your case expression like this:
SELECT *
FROM
TableA a
WHERE
(CASE
WHEN a.FieldB = 'Value1' AND a.FieldA IN (select b.ID from TableB b where b.FK_Field = '123') THEN 1
WHEN a.FieldB = 'Value2' AND a.FieldA IN (select c.ID from TableC c where c.FK_Field = '123') THEN 1
ELSE 0
END) = 1

Answer 3

Don't use case in predicates if it is not necessary - using case make your argument non-SARG (you qry will not use index).
SELECT *
FROM TableA a
WHERE EXISTS(
SELECT NULL
FROM TableB b
WHERE a.FieldB = 'Value1'
AND b.FK_Field = '123'
AND a.FieldA = b.ID))
OR EXISTS(
SELECT NULL
FROM TableC c
WHERE a.FieldB = 'Value2'
AND c.FK_Field = '123'
AND a.FieldB = c.ID))
Use indexes. And try to make your qry readable.
Or if you would like use semi-join:
SELECT *
FROM TableA a
WHERE a.FieldA IN (
SELECT b.ID
FROM TableB b
WHERE a.FieldB = 'Value1'
AND b.FK_Field = '123'))
OR a.FieldB IN (
SELECT c.ID
FROM TableC c
WHERE a.FieldB = 'Value2'
AND c.FK_Field = '123'))
Both of this solutions are with SARG.