T-SQL Different Comparison in WHERE Clause Using CASE - tsql

Question

I wanted to do different comparison expression in where clause, but it keep prompt me invalid syntax. The reason i want to use this because i wanted to enforce the evaluation according to the Type. If i use ((t1.Amount <= #valueB) OR (t1.Type IN (1) AND t1.Amount > 0 AND #valueB < 0)) , it seems to be not expression i trying to impose into my query. If first evaluation success, i would not want the following to execute, however OR in sql is not short-circuit operator.
DECLARE #name AS VARCHAR(200) = 'super'
DECLARE #valueA AS INT = 1 -- Can be 1 or 2
DECLARE #valueB AS DECIMAL = 0.5
SELECT * FROM TABLE_1 t1 WITH(NOLOCK)
WHERE t1.Name = #name
AND
CASE WHEN t1.Type = 1 THEN (t1.Amount > 0 AND #valueB < 0)
ELSE (t1.Amount <= #valueB)
END

Answer 1

You can't use it that way.
TRY THIS
SELECT * FROM TABLE_1 t1 WITH(NOLOCK)
WHERE t1.Name = #name AND
((t1.Type = 1 AND t1.Amount > 0 AND #valueB < 0)
OR (t1.Amount <= #valueB))
If could elaborate further what you are trying to achieve , I'll be able to help.

Answer 2

Zohar Peled's comment rightly points out that CASE is an expression in T-SQL, it is not a flow-of-control statement.
What you are trying to do is:
IF <expression_A> = <value_A>
THEN <apply_condition_B>
ELSE <apply_condition_C>
To achieve the following in SQL, you'd need to write it out using Boolean logic using the AND and OR operators as:
(
( <expression_A> = <value_A> AND <apply_condition_B> )
OR ( <expression_A> is not <value_A> AND <apply_condition_C> )
)
The first AND that comes before <apply_condition_B> acts like the THEN keyword, while the combination of the OR and the second AND act like the ELSE keyword.
Taking your query into consideration, we get the following mapping:
<expression_A> = t1.Type
<value_A> = 1
<apply_condition_B> = (t1.Amount > 0 AND #valueB < 0)
<apply_condition_C> = (t1.Amount <= #valueB)
Putting that into query form would give us:
WHERE t1.Name = #name
AND
(
( t1.Type = 1 AND (t1.Amount > 0 AND #valueB < 0) )
OR ( t1.Type != 1 AND (t1.Amount <= #valueB) )
)
-- Note - this is not fully complete, as it assumes t1.Type is not NULL
The above is a simple rewrite of your intended logic. However, do note that SQL operates on three-valued logic which means that t1.Type can be 1, it can be a value other than 1, or it can be NULL. The above code fragment handles only two of these 3 possibilities. To correctly handle t1.Type being NULL you can use the COALESCE() function to return a specified value (zero, below) in case it is NULL:
WHERE t1.Name = #name
AND
(
( t1.Type = 1 AND (t1.Amount > 0 AND #valueB < 0) )
OR ( COALESCE(t1.Type, 0) != 1 AND (t1.Amount <= #valueB) )
)
Tip: Before using this sort of construct directly in your actual query, I'd suggest you play around with it using simpler sample data to understand how it works.

Answer 3

Try this
DECLARE #name AS VARCHAR(200) = 'super'
DECLARE #valueA AS INT = 1 -- Can be 1 or 2
DECLARE #valueB AS DECIMAL = 0.5
SELECT *
FROM TABLE_1 t1 WITH (NOLOCK)
WHERE t1.NAME = #name
AND (t1.Type = 1 AND (t1.Amount > 0 AND #valueB < 0)
OR t1.Type <> 1 AND (t1.Amount <= #valueB))
GO
OR Use Dynamic SQL if you feel like it
DECLARE #name AS VARCHAR(200) = 'super'
DECLARE #valueA AS INT = 1 -- Can be 1 or 2
DECLARE #valueB AS DECIMAL = 0.5
DECLARE #SQL VARCHAR(MAX)
DECLARE #Type INT
SELECT TOP 1 #Type = t1.Type
FROM TABLE_1 t1 WITH (NOLOCK)
WHERE t1.NAME = #name
SET #SQL = 'SELECT *
FROM TABLE_1 t1 WITH (NOLOCK)
WHERE t1.NAME = #name'
SET #SQL = CASE
WHEN #Type = 1
THEN #SQL + CHAR(13)+CHAR(10) + ' AND (t1.Amount > 0 AND #valueB < 0)'
ELSE #SQL + CHAR(13)+CHAR(10) + ' AND (t1.Amount <= #valueB)'
END
--PRINT #SQL
EXEC (#SQL)
GO