How do I make a query shorter and neater? - sql

Question

Im trying to make This query more understandable and neater. But im not sure how to?
SELECT a.Patient_id, COUNT (p.Person_id) AS "Number of Operations", SUM (w.Daily_charge * (a.Discharge_date - a.Admission_date) + ot.Theatre_fee + b.Charges + c.Charges ) AS "Total Payment"
FROM person p, admission a, ward w, operation o, operation_type ot, staff b, staff c
WHERE w.Ward_code = a.Ward_code AND p.Person_id = a.Patient_id
AND a.Admission_id = o.Admission_id AND ot.Op_code = o.Actual_op
AND o.Surgeon = b.Person_id AND o.Anaesthetist = c.Person_id
GROUP BY a.Patient_id, p.Person_id
ORDER BY COUNT (p.Person_id) DESC FETCH FIRST 1 ROWS ONLY;

Answer 1

Any decent formatter would do it for you.
Other than that,
JOIN instead of comma-separate tables in the FROM clause
remove p.person_id from group by clause, there's no use of it as it is
equal to a.patient_id which is correctly put into the clause,
not part of the select statement's column list
So:
select a.patient_id,
count (p.person_id) as "number of operations",
sum (w.daily_charge * (a.discharge_date - a.admission_date) +
ot.theatre_fee + b.charges + c.charges
) as "total payment"
from person p join admission a on a.patient_id = p.person_id
join ward w on w.ward_code = a.ward_code
join operation o on o.admission_id = a.admission_id
join operation_type ot on ot.op_code = o.actual_op
join staff b on o.surgeon = b.person_id
join staff c on o.anaesthetist = c.person_id
group by a.patient_id
order by count (p.person_id) desc
fetch first 1 rows only;

Answer 2

Sadly there is no perfect formatter, although as Ed mentioned in the comments, they can be a start if you review the settings carefully. (It's a tradition in the industry that the default formatter settings are always horrible.)
It's also been said (I think by Steven Feuerstein) that you should only set formatting rules that are supported by your formatter, and of course he makes a good point. But taken with the limitations of all formatters, an industry tradition for horrible formatting and the impossibility of consistent rules for formatting SQL anyway, that puts us PL/SQL developers in a difficult position.
I'd say the first principle of computer code layout is to use vertically aligned blocks to indicate dependency levels (similar to the grids used in graphic design). A lot of the choices then become about how to apply that principle.
We then need to separate the code into logical sections, but at the same time not let it sprawl down the page unnecessarily. I think this is difficult for automated formatters as the rules become a bit fuzzy, e.g. for a join with only one condition I keep it on one line, but if there is more than one I start splitting it out onto multiple lines, one per on or and keyword. The same goes for your complex sum() expression - normally I would place it all on one line, but if it aids readability then I split it up.
Finally, opinions vary on where to place commas in stacked lists, of which SQL has a lot. I say they go on the left, to act like bullet points and also make it easier to add items to the ends of lists. Others will disagree.
select ad.patient_id
, count(*) as "Number of Operations"
, sum(
wa.daily_charge * (ad.discharge_date - ad.admission_date)
+ ot.theatre_fee + ss.charges + sa.charges
) as "Total Payment"
from person pr
join admission ad on ad.patient_id = pr.person_id
join ward wa on wa.ward_code = ad.ward_code
join operation op on op.admission_id = ad.admission_id
join operation_type ot on ot.op_code = op.actual_op
join staff ss on ss.person_id = op.surgeon
join staff sa on sa.person_id = op.anaesthetist
group by ad.patient_id
order by count(*) desc
fetch first row only;