Split date range in weeks in oracle SQL - sql

Question

I need to split the following data in Oracle SQL:
WITH sample_data AS
(SELECT DATE '2020-12-16' Start_Date, DATE '2021-01-07' End_Date FROM DUAL)
in week ranges for every working week(from Monday to Friday) of this given period. The final result it should look like this:
NEW_STARTDATE NEW_END_DATE
2020-12-16 2020-12-18
2020-12-21 2020-12-25
2020-12-28 2021-01-01
2021-01-04 2021-01-07
So in this example, the first row starts with the initial start date (2020-12-16) which is on Wednesday and continues with the new end date(2020-12-18) which is the next Friday, and so on with ranges of working weeks until the actual end date of this period.

Answer 1

You can use:
WITH sample_data ( start_date, end_date ) AS (
SELECT DATE '2020-12-16', DATE '2021-01-07' FROM DUAL
),
weeks ( start_date, week_start, week_end, end_date ) AS (
SELECT start_date,
TRUNC( start_date, 'IW' ),
LEAST( TRUNC( start_date, 'IW' ) + INTERVAL '4' DAY, end_date ),
end_date
FROM sample_data
UNION ALL
SELECT start_date,
week_start + INTERVAL '7' DAY,
LEAST( week_end + INTERVAL '7' DAY, end_date ),
end_date
FROM weeks
WHERE week_start + INTERVAL '7' DAY <= end_date
)
SELECT GREATEST( week_start, start_date ) AS new_start_date,
week_end AS new_end_date
FROM weeks
WHERE GREATEST( week_start, start_date ) <= week_end;
Which outputs (where the NLS_DATE_FORMAT is YYYY-MM-DD (DY)):
NEW_START_DATE
NEW_END_DATE
2020-12-16 (WED)
2020-12-18 (FRI)
2020-12-21 (MON)
2020-12-25 (FRI)
2020-12-28 (MON)
2021-01-01 (FRI)
2021-01-04 (MON)
2021-01-07 (THU)
db<>fiddle here

Answer 2

Here is one way - compute the Monday dates for the ISO week of the input dates (start_date and end_date), while also keeping track of which is the first and which is the last such Monday in the same hierarchical (connect by) query. Then produce the requested output; for the first week, check that the start_date is not a Saturday or a Sunday (if it is, that "first week" should not produce a row in the output); that is done in the where clause. This is illustrated in the sample dates I used for testing - the input "start date" is a Saturday, so the first "new start date" is the following Monday.
with
sample_data (start_date, end_date) as (
select date '2020-12-12', date '2021-01-02' from dual
)
, mondays (dt, rn, mx) as (
select trunc(start_date, 'iw') + 7 * (level - 1), level, max(level) over ()
from sample_data
connect by level <= 1 + (trunc(end_date, 'iw') - trunc(start_date, 'iw'))/7
)
select case rn when 1 then greatest(start_date, dt)
else dt end as new_start_date,
case rn when mx then least(dt + 4, end_date)
else dt + 4 end as new_end_date
from sample_data cross join mondays
where rn >= 2 or start_date <= dt + 4
;
NEW_START_DATE NEW_END_DATE
--------------- ---------------
MON 14-DEC-2020 FRI 18-DEC-2020
MON 21-DEC-2020 FRI 25-DEC-2020
MON 28-DEC-2020 FRI 01-JAN-2021