Generate dates without gap and vlookup 2 tables - sql

Question

I have 2 tables,
Table 1 - has the available principal balance on the transaction dates
=>(Transactions_Date, Principal balance)
Table 2 - has interest rates and the effective dates
=> ( Effective_Date, Interest)
I have to generate a result set, with the below data
Result_Date - List out all the dates from table1(Transaction_Date) by filling all the gaps between the available dates
Principle Balance - vlookup the Result_Dates in Table 1 and fill in the corresponding Principal_Balance until the next table1.transaction_date.
Interest - vlookup Result_date in Table 2.Effective_date and fill in all the Interest for that date range.
How to implement this in SNOWFLAKE?
Note: I can use CTE but can not create table

Answer 1

a couple of simple steps, find the min/max of dates for all tables, find the least/greatest of those values, spread across a generator range, and then use a row_number to feed into a dateadd, then left join this range of dates to the original data, and use nvl to pick between the this value, and the carried lag via the ignore nulls extension.
with table_1(t_date, p_bal) as (
select * from values
('2023-02-10'::date, 10000),
('2023-01-20'::date, 9000),
('2023-01-03'::date, 8000)
), table_2(e_date, interest) as (
select * from values
('2023-02-05'::date, 1),
('2023-01-15'::date, 2),
('2023-01-01'::date, 3)
), range_of_date as (
select dateadd(day, rn, min_date) as date
from (
select
least(min_t_date, min_e_date) as min_date
,greatest(max_t_date, max_e_date) as max_date
from (
select
min(t_date) as min_t_date
,max(t_date) as max_t_date
from table_1
) as t
cross join (
select
min(e_date) as min_e_date
,max(e_date) as max_e_date
from table_2
) as e
) as a
cross join (
select
row_number() over (order by null) - 1 as rn
from table(generator(ROWCOUNT => 1000))
) as b
having date <= max_date
)
select
*
,nvl(t1.p_bal,lag(t1.p_bal)ignore nulls over (order by rd.date)) as filled_p_bal
,nvl(t2.interest,lag(t2.interest)ignore nulls over(order by rd.date)) as filled_interest
from range_of_date as rd
left join table_1 as t1
on rd.date = t1.t_date
left join table_2 as t2
on rd.date = t2.e_date
order by 1;
gives:
DATE
T_DATE
P_BAL
E_DATE
INTEREST
FILLED_P_BAL
FILLED_INTEREST
2023-01-01
2023-01-01
3
3
2023-01-02
3
2023-01-03
2023-01-03
8,000
8,000
3
2023-01-04
8,000
3
2023-01-05
8,000
3
2023-01-06
8,000
3
2023-01-07
8,000
3
2023-01-08
8,000
3
2023-01-09
8,000
3
2023-01-10
8,000
3
2023-01-11
8,000
3
2023-01-12
8,000
3
2023-01-13
8,000
3
2023-01-14
8,000
3
2023-01-15
2023-01-15
2
8,000
2
2023-01-16
8,000
2
2023-01-17
8,000
2
2023-01-18
8,000
2
2023-01-19
8,000
2
2023-01-20
2023-01-20
9,000
9,000
2
2023-01-21
9,000
2
2023-01-22
9,000
2
2023-01-23
9,000
2
2023-01-24
9,000
2
2023-01-25
9,000
2
2023-01-26
9,000
2
2023-01-27
9,000
2
2023-01-28
9,000
2
2023-01-29
9,000
2
2023-01-30
9,000
2
2023-01-31
9,000
2
2023-02-01
9,000
2
2023-02-02
9,000
2
2023-02-03
9,000
2
2023-02-04
9,000
2
2023-02-05
2023-02-05
1
9,000
1
2023-02-06
9,000
1
2023-02-07
9,000
1
2023-02-08
9,000
1
2023-02-09
9,000
1
2023-02-10
2023-02-10
10,000
10,000
1
Take 2:
To make the data only have values in the range of the table 1 but carry interesting values from earlier, that are not aligned with the first balance value, we need the time range to be wider (it can be a more constrained match, but I have opted for simplicity to me).
with table_1(t_date, p_bal) as (
select * from values
('2023-02-10'::date, 10000),
('2023-01-20'::date, 9000),
('2023-01-03'::date, 8000)
), table_2(e_date, interest) as (
select * from values
('2023-02-05'::date, 1),
('2023-01-15'::date, 2),
('2023-01-01'::date, 3)
), range_1 as (
select
min(t_date) as min_t_date
,max(t_date) as max_t_date
from table_1
), range_of_date as (
select dateadd(day, rn, min_date) as date
from (
select
least(min_t_date, (select min(e_date) from table_2)) as min_date
,max_t_date as max_date
from range_1
) as a
cross join (
select
row_number() over (order by null) - 1 as rn
from table(generator(ROWCOUNT => 1000))
) as b
having date <= max_date
)
select
rd.date
,nvl(t1.p_bal,lag(t1.p_bal)ignore nulls over (order by rd.date)) as filled_p_bal
,nvl(t2.interest,lag(t2.interest)ignore nulls over(order by rd.date)) as filled_interest
from range_of_date as rd
left join table_1 as t1
on rd.date = t1.t_date
left join table_2 as t2
on rd.date = t2.e_date
qualify filled_p_bal is not null
order by 1;