PARTITION BY with date between 2 date - sql

Question

I work on Azure SQL Database working with SQL Server
In SQL, I try to have a table by day, but the day is not in the table.
I explain it by the example below:
TABLE STARTER: (Format Date: YYYY-MM-DD)
Date begin
Date End
Category
Value
2021-01-01
2021-01-03
1
0.2
2021-01-02
2021-01-03
1
0.1
2021-01-01
2021-01-02
2
0.3
For the result, I try to have this TABLE RESULT:
Date
Category
Value
2021-01-01
1
0.2
2021-01-01
2
0.3
2021-01-02
1
0.3 (0.2+0.1)
2021-01-02
2
0.3
2021-01-03
1
0.3 (0.2+0.1)
For each day, I want to sum the value if the day is between the beginning and the end of the date. I need to do that for each category.
In terms of SQL code I try to do something like that:
SELECT SUM(CAST(value as float)) OVER (PARTITION BY Date begin, Category) as value,
Date begin,
Category,
Value
FROM TABLE STARTER
This code calculates only the value that has the same Date begin but don't consider all date between Date begin and Date End.
So in my code, it doesn't calculate the sum of the value for the 02-01-2021 of Category 1 because it doesn't write explicitly. (between 01-01-2021 and 03-01-2021)
Is it possible to do that in SQL?
Thanks so much for your help!

Answer 1

You can use a recursive CTE to expand the date ranges into the list of separate days. Then, it's matter of joining and aggregating.
For example:
with
r as (
select category,
min(date_begin) as date_begin, max(date_end) as date_end
from starter
group by category
),
d as (
select category, date_begin as d from r
union all
select d.category, dateadd(day, 1, d.d)
from d
join r on r.category = d.category
where d.d < r.date_end
)
select d.d, d.category, sum(s.value) as value
from d
join starter s on s.category = d.category
and d.d between s.date_begin and s.date_end
group by d.category, d.d;
Result:
d category value
----------- --------- -----
2021-01-01 1 0.20
2021-01-01 2 0.30
2021-01-02 1 0.30
2021-01-02 2 0.30
2021-01-03 1 0.30
See running example at db<>fiddle.
Note: Starting in SQL Server 2022 it seems there is/will be a new GENERATE_SERIES() function that will make this query much shorter.