How to get column wise sum in SQL? - sql

Question

I have a complex query where I am getting the count of various categories in separate columns.
Here's the output of my query:
district | colA | colB | colC
------------------------------------
DistA | 1 | 1 | 3
DistB | 2 | 0 | 2
DistC | 2 | 1 | 0
DistD | 0 | 3 | 4
..
And here's my query:
select
q1."district",
coalesce(max(case q1."type" when 'colA' then q1."type_count" else 0 end), 0) as "colA",
coalesce(max(case q1."type" when 'colB' then q1."type_count" else 0 end), 0) as "colB",
coalesce(max(case q1."type" when 'colC' then q1."type_count" else 0 end), 0) as "colC"
from (
select
d."name" as "district",
t."name" as "type",
count(t.id) as "type_count"
from
main_entity as m
inner join type_entity as t on
m."type_id" = t.id
inner join district as d on
m."district_id" = d.id
where
m."delete_at" is null
group by
d."name",
t.id
) as q1
group by
q1."district"
I want to modify this query so that I can get the sum of each column in the last row, something like this:
district | colA | colB | colC
------------------------------------
DistA | 1 | 1 | 3
DistB | 2 | 0 | 2
DistC | 2 | 1 | 0
DistD | 0 | 3 | 4
..
Total | 5 | 5 | 9
I have tried using group by + rollup with the above query by just adding the following:
...
group by rollup (q1."district")
It adds a row at the bottom but the values are similar to the values of a row before it, and not the sum of all the rows before it, so basically something like this:
district | colA | colB | colC
------------------------------------
DistA | 1 | 1 | 3
..
DistD | 0 | 3 | 4
Total | 0 | 3 | 4
So, how can I get the column-wise some from my query?

Answer 1

Try this:
With temp as
( --your query from above
select
q1."district",
coalesce(max(case q1."type" when 'colA' then q1."type_count" else 0 end), 0) as "colA",
coalesce(max(case q1."type" when 'colB' then q1."type_count" else 0 end), 0) as "colB",
coalesce(max(case q1."type" when 'colC' then q1."type_count" else 0 end), 0) as "colC"
from (
select
d."name" as "district",
t."name" as "type",
count(t.id) as "type_count"
from
main_entity as m
inner join type_entity as t on
m."type_id" = t.id
inner join district as d on
m."district_id" = d.id
where
m."delete_at" is null
group by
d."name",
t.id
) as q1
group by
q1."district"
)
select t.* from temp t
UNION
select sum(t1.colA),sum(t1.colB),sum(t1.colC) from temp t1