Pivot sum in PostgreSQL - postgresql

Question

I'm using PosgreSQL 4.2. And I have the following code:
SELECT ID, Num1, Num2 FROM Tab1
WHERE ID IN(1,2,3,4.....50);
Returns results as:
ID | Num1 | Num2
----+--------------------
1 | 100 | 0
2 | 50 | 1
3 | 30 | 2
4 | 110 | 3
5 | 33 | 4
6 | 46 | 5
7 | 36 | 6
8 | 19 | 7
9 | 20 | 8
10 | 31 | 9
11 | 68 | 10
12 | 123 | 11
13 | 588 | 0
14 | 231 | 1
15 | 136 | 2
I want to Pivot sum to return result with pairs of number in IN clause, and result return like this:
| ID | Meaning
-------------------------
Num1 | 150 | 1+2(num1)
Num2 | 1 | 1+2(num2)
Num1 | 140 | 3+4(num1)
Num2 | 5 | 3+4(num2)
Num1 | 79 | 5+6(num1)
Num2 | 9 | 5+6(num2)
.........................
How can I do that?

Answer 1

PostgreSQL 4.2. No you must have another application ans confusing your application versions:
In 1996, the project was renamed to PostgreSQL to reflect its support
for SQL. The online presence at the website PostgreSQL.org began on
October 22, 1996.[26] The first PostgreSQL release formed version 6.0
on January 29, 1997.
Wikipedia. You might want to run the query: select version();
However, any supported version is sufficient for this request. Actually the SQL necessary to supply the necessary summations is simple once you understand LEAD (LAG) functions. The LEAD function permits access to the following row. Using that then the query:
select id, nid, n1s, n2s
from ( select id
, coalesce(lead(id) nid
, num1 + coalesce(lead(num1) over (order by id),0) n1s
, num2 + coalesce(lead(num2) over (order by id),0) n2s
, row_number() over() rn
from tab1
) i
where rn%2 = 1;
Provides all the necessary data for the presentation layer to format the the result as desired. However, that's not the result requested, that requires some SQL gymnastics.
We begin the gymnastics show by wrapping the above into an CTE adding a little setup for the show to follow. The main event then breaks the results into 2 sets in order to add the syntactical sugar tags before bringing then back together. So for the big show:
with joiner(id,nid,n1s,n2s,rn) as
( select *
from ( select id
, coalesce(lead(id) over (order by id),0)
, num1 + coalesce(lead(num1) over (order by id),0)
, num2 + coalesce(lead(num2) over (order by id),0)
, row_number() over() rn
from tab1
) i
where rn%2 = 1
)
select "Name","Sum","Meaning"
from (select 'Num1' as "Name"
, n1s as "Sum"
, concat(id::text, case when nid = 0
then null
else '+' || nid::text
end
) || ' (num1)'as "Meaning"
, rn
from joiner
union all
select 'Num2'
, n2s
, concat(id::text, case when nid = 0
then null
else '+' || nid::text
end
) || ' (num2)'
, rn
from joiner
) j
order by rn, "Name"
See fiddle. Note: I used "Sum" instead of ID for the column title, as it indicates the id not at all. That is just in the "Meaning" column.