Hierarchical query using oracle SQL - sql

Question

We have following table to query data from,
EMPLOYEE_ABSENCE_TAB
emp_id
absence_id
from_date
to_date
absence_continuation
100
1
01/01/2022
03/01/2022
100
2
01/02/2022
05/02/2022
1
200
3
01/01/2022
07/01/2022
200
4
10/01/2022
14/01/2022
3
200
5
16/01/2022
20/01/2022
4
300
6
01/01/2022
14/01/2022
We need to connect data hierarchically based on the following logic.
Output should be emp_id, absence_days, from_date and to_date.
absence_days : number of absence days = sum(to_date - from_date)
from_date : first from_date where absence_continuation is NULL
to_date : last to_date connect by prior absence_id = absence_continuation
Expected Outcome
emp_id
absence_days
from_date
to_date
100
8
01/01/2022
05/02/2022
200
17
01/01/2022
20/01/2022
300
14
01/01/2022
14/01/2022
I tried hierarchical but could able to come up with working solution. How can this be achieved using hierarchical query in Oracle SQL?

Answer 1

You don't need to consider the hierarchy in the data (assuming that there are no overlaps); you just need to aggregate:
SELECT emp_id,
SUM(to_date - from_date + 1) AS absence_days,
MIN(from_date) AS from_date,
MAX(to_date) AS to_date
FROM EMPLOYEE_ABSENCE_TAB
GROUP BY emp_id
Which, for the sample data:
CREATE TABLE EMPLOYEE_ABSENCE_TAB
(emp_id, absence_id, from_date, to_date, absence_continuation) AS
SELECT 100, 1, DATE '2022-01-01', DATE '2022-01-03', NULL FROM DUAL UNION ALL
SELECT 100, 2, DATE '2022-02-01', DATE '2022-02-05', 1 FROM DUAL UNION ALL
SELECT 200, 3, DATE '2022-01-01', DATE '2022-01-07', NULL FROM DUAL UNION ALL
SELECT 200, 4, DATE '2022-01-10', DATE '2022-01-14', 3 FROM DUAL UNION ALL
SELECT 200, 5, DATE '2022-01-16', DATE '2022-01-20', 4 FROM DUAL UNION ALL
SELECT 300, 6, DATE '2022-01-01', DATE '2022-01-14', NULL FROM DUAL;
Outputs:
EMP_ID
ABSENCE_DAYS
FROM_DATE
TO_DATE
100
8
2022-01-01 00:00:00
2022-02-05 00:00:00
200
17
2022-01-01 00:00:00
2022-01-20 00:00:00
300
14
2022-01-01 00:00:00
2022-01-14 00:00:00
If you did want to consider the hierarchy and get the duration of each absence (and not the total absences per employee) then you can use:
SELECT emp_id,
absence_id,
SUM(to_date - from_date + 1) AS absence_days,
MIN(from_date) AS from_date,
MAX(to_date) AS to_date
FROM (
SELECT emp_id,
CONNECT_BY_ROOT absence_id AS absence_id,
to_date - from_date + 1 AS absence_days,
from_date,
to_date
FROM EMPLOYEE_ABSENCE_TAB
START WITH absence_continuation IS NULL
CONNECT BY PRIOR absence_id = absence_continuation
)
GROUP BY
emp_id,
absence_id
Which outputs the same (with an additional absence_id column for the id of the first part of the absence) for your sample data as you only have a single absence for each employee; if you had more absences then it would aggregate each absence separately.
db<>fiddle here