When I'm doing this query I'm getting NULL values. Why? - oracle

Question

I have a table called fact_trip which has a column as fare_final and I want to find the difference between values.
SQL> desc fact_trip
Name Null? Type
----------------------------------------- -------- ----------------------------
TRIP_UUID NOT NULL VARCHAR2(20)
DATESTR DATE
PRODUCT_TYPE_NAME VARCHAR2(20)
CITY_ID NUMBER
DRIVER_UUID VARCHAR2(50)
IS_COMPLETED VARCHAR2(10)
ETA NUMBER
ATA NUMBER
UFP_FARE NUMBER(4,2)
FARE_FINAL NUMBER(4,2)
So when I do this I'm getting NULL as output:
SQL> select sum(fare_final) from fact_trip where to_char(datestr, 'W')=1 - (select sum(fare_final) from fact_trip where to_char(datestr, 'W')=2);
SUM(FARE_FINAL)
---------------
I even tried doing those select queries individually like this:
SQL> select sum(fare_final) from fact_trip where to_char(datestr, 'W')=1;
SUM(FARE_FINAL)
---------------
1821.6
SQL> select sum(fare_final) from fact_trip where to_char(datestr, 'W')=2;
SUM(FARE_FINAL)
---------------
67
Which is of course fetching result. But then when I run those queries for what I want a difference of them, it's showing NULL. Like I want difference of them (1821.6 - 67).
Can anybody tell what's wrong in it?
Thank You!

Answer 1

SQL is not a worksheet, where you may evaluate any expression. You cannot use an expression as a SQL command: you have to eitherselect an expression from something, or calculate it in PL/SQL block.
You want to calculate the difference, so you assumed your code is: subquery - subquery. But because of described above this is not syntactically correct command, and parser tries to find a parse tree which is syntactically correct. And it actually finds one:
select sum(fare_final)
from fact_trip
where to_char(datestr, 'W') = (
1 - (
select sum(fare_final)
from fact_trip
where to_char(datestr, 'W') = 2
)
)
Of course, such week doesn't exist and you get null as a result.
To turn your code into a correct command, the simplest way is to select the expression from a dual table with appropriate brackets to specify calculation precedence:
select
(select sum(fare_final) from fact_trip where to_char(datestr, 'W')='1')
- (select sum(fare_final) from fact_trip where to_char(datestr, 'W')='2') as res
from dual
But more performant and set-based approach would be:
select
sum(
case to_char(datestr, 'W')
when '1' then 1
when '2' then -1
end * fare_final
) as res
from fact_trip
where to_char(datestr, 'W') in ('1', '2')
Below is the sample code with results of all that was described above:
select * from t
ID | VAL
-: | --:
1 | 3
2 | 6
3 | 9
select sum(val)
from t
where id = 1 - (
select sum(val)
from t
where id = 2
)
| SUM(VAL) |
| -------: |
| null |
select
(
(select sum(val)
from t
where id = 1) - (
select sum(val)
from t
where id = 2
)
) as q
from dual
| Q |
| -: |
| -3 |
select
sum(
case id
when 1 then 1
when 2 then -1
end * val
) as res
from t
where id in (1, 2)
| RES |
| --: |
| -3 |
db<>fiddle here

Answer 2

Should be
SELECT SUM (
CASE
WHEN TO_CHAR (datestr, 'W') = 1 THEN fare_final
WHEN TO_CHAR (datestr, 'W') = 2 THEN -fare_final
END)
FROM fact_trip;
I don't have your table so I'll illustrate it on Scott's EMP, computing the total for analysts and clerks:
SQL> SELECT job, sal FROM emp ORDER BY job;
JOB SAL
--------- ----------
ANALYST 3000 --> total for analysts is 6000
ANALYST 3000
CLERK 1300 --> total for clerks is 4150
CLERK 950
CLERK 800 --> difference: 6000 - 4150 = 1850
CLERK 1100
MANAGER 2850
MANAGER 2975
MANAGER 2450
PRESIDENT 5000
SALESMAN 1500
SALESMAN 1250
SALESMAN 1250
SALESMAN 1600
14 rows selected.
SQL> SELECT SUM (
2 CASE WHEN job = 'ANALYST' THEN sal
3 WHEN job = 'CLERK' THEN -sal
4 END) total
5 FROM emp;
TOTAL
----------
1850
SQL>