Always have SQL query return atleast one record - sql

Question

SELECT column_a,
CASE WHEN column_a > 10 THEN 0 ELSE 1 END AS column_b
FROM table_a;
This does not always returns record, my requirement is to always have at least 1 row returned.
Following a solution here I tried couple of workaround such as union but none of it works.
SELECT 2 as column_a,
CASE WHEN column_a > 10 THEN 0 ELSE 1 END AS column_b
FROM table_a
UNION
SELECT -1,-1 FROM table_a;
Is there a way I can make this query return a record even when none is found?

Answer 1

Well, query won't return anything if TABLE_A is empty (as there's no WHERE clause). In any other case, it'll return something.
Though, perhaps you meant to union it with a "dummy" select statement that selects from DUAL, not table_a (see line #6). Line #4 is here just to make the first select return no rows.
SQL> WITH table_a (column_a) AS (SELECT 3 FROM DUAL)
2 SELECT 2 AS column_a, CASE WHEN column_a > 10 THEN 0 ELSE 1 END AS column_b
3 FROM table_a
4 WHERE 1 = 2
5 UNION
6 SELECT -1, -1 FROM DUAL; --> DUAL here, not TABLE_A
COLUMN_A COLUMN_B
---------- ----------
-1 -1
SQL>

Answer 2

What is the problem you're trying to solve? Your title says "PL/SQL" which implies that this is happening in a PL/SQL block. In which case the standard answer would be just to catch the no_data_found exception and populate whatever variables you're fetching the data into with some appropriate default values.
BEGIN
SELECT column_a,
CASE WHEN column_a > 10 THEN 0 ELSE 1 END AS column_b
INTO local_variable1, local_variable2
FROM table_a;
EXCEPTION
WHEN no_data_found
THEN
local_variable1 := 1;
local_variable2 := -1;
END;
Likely, it would make sense to encapsulate this in a stored function.
If you really want to ensure that the query returns 1 row if there is no data in table_a, you could do something like this
SELECT column_a,
CASE WHEN column_a > 10 THEN 0 ELSE 1 END AS column_b
FROM table_a
UNION ALL
SELECT 1, -1
FROM dual
WHERE NOT EXISTS( SELECT 1 FROM table_a );

Answer 3

You can use:
SELECT *
FROM (
SELECT column_a,
CASE WHEN column_a > 10 THEN 0 ELSE 1 END AS column_b
FROM table_a
UNION ALL -- Use UNION ALL to keep duplicates from table_a.
SELECT -1,-1 FROM DUAL -- Use the DUAL table which always has exactly 1 row.
)
WHERE column_b >= 0 -- All the rows from table_a, if any exist.
OR ROWNUM = 1 -- Or, just the first row.
;
Which, for the empty table:
CREATE TABLE table_a (column_a INT);
Outputs:
COLUMN_A
COLUMN_B
-1
-1
And if you insert data:
INSERT INTO table_a (column_a)
SELECT 1 FROM DUAL UNION ALL
SELECT 2 FROM DUAL UNION ALL
SELECT 6 FROM DUAL UNION ALL
SELECT 10 FROM DUAL UNION ALL
SELECT 11 FROM DUAL UNION ALL
SELECT 11 FROM DUAL UNION ALL
SELECT 15 FROM DUAL;
Then the query outputs:
COLUMN_A
COLUMN_B
1
1
2
1
6
1
10
1
11
0
11
0
15
0
db<>fiddle here