Overlapping date problem, I'm a bit stumped...

Question

Hi,
My first post here, woohooo!!
I'm trying to write some SQL for overlapping dates, hoping someone could point me in the write direction because I'm a bit stuck :(
Heres the scenario....
Tables are simillar to this (although I've changed it for the sake of this example).
TABLE A, TABLE B, TABLE C look like this
- EMPLOYEE_NUMBER VARCHAR2
- START_DATE DATE
- END_DATE DATE
These three tables contain information on an employees working period. The dates in each table can overlap each other.
For example
TABLE A
AAAA, 05-10-1991, 05-10-2020
AAAA, 05-01-2000, 20-05-2003
BBBB, 09-12-1999, 02-07-2009
TABLE B
AAAA, 01-11-2002, 01-09-2003
TABLE C
AAAA, 15-08-2003, 12-05-2004
Heres the tricky bit. There is a hierarchy, any dates in table C will overide any of the dates in table A and B. Table B will override any dates Table A. I need a query to break up the dates to show one consecutive timeline for any given employee. The query would obviously need to use all 3 tables.
Therefore I would need a query to output the following for Employee AAAA.
EMPLOYEE, START_DATE, END_DATE
AAAA, 05-10-1991, 04-01-2000
AAAA, 05-01-2000, 31-10-2002
AAAA, 01-11-2002, 14-08-2003
AAAA, 15-08-2003, 12-05-2004
AAAA, 13-05-2004, 05-10-2020
I've worked out how to write this using PLSQL although its pretty feral code, hoping there might be a cleaner method using SQL. Any ideas anyone?
Thanks!
Edited by: user2322009 on 1/03/2009 21:17

Answer 1

in addition to avinash solution and using the data that is provided by you.
with t as (SELECT 'AAAA' emp_id, TO_DATE('05-10-1991','DD-MM-YYYY') start_dt, TO_DATE('05-10-2020','DD-MM-YYYY') end_dt FROM DUAL UNION ALL
    SELECT 'AAAA', TO_DATE('05-01-2000','DD-MM-YYYY'), TO_DATE('20-05-2003','DD-MM-YYYY') FROM DUAL UNION ALL
    SELECT 'AAAA', TO_DATE('01-11-2002','DD-MM-YYYY'), TO_DATE('01-09-2003','DD-MM-YYYY') FROM DUAL UNION ALL
    SELECT 'AAAA', TO_DATE('15-08-2003','DD-MM-YYYY'), TO_DATE('12-05-2004','DD-MM-YYYY') FROM DUAL )
, t1 as (    select emp_id,max(end_dt) m_dt from t
            group by emp_id)
, t2 as (select t1.emp_id, max(end_dt) m_dt from t,t1
             where t.emp_id = t1.emp_id
             and end_dt <> m_dt
             group by t1.emp_id
SELECT emp_id,start_DT, CASE WHEN start_dt_pr < end_dt THEN start_dt_pr -1 ELSE end_dt END  FROM (
    SELECT emp_id,start_dt,NVL(LEAD(start_dt,1) OVER(ORDER BY start_dt),end_dt)  start_dt_pr, end_dt FROM (
    SELECT emp_id,start_dt,end_dt FROM T ORDER BY start_dt ))
union all
select t1.emp_id,t2.m_dt + 1, t1.m_dt from t1,t2
where t1.emp_id = t2.emp_id;See if this helps.
Avinash solutions was missing one record if i use the data provided by you. That is the last record.
Following was added to get the last record. Experts here will correct me if this is not the right way to do.
, t1 as (    select emp_id,max(end_dt) m_dt from t
            group by emp_id)
, t2 as (select t1.emp_id, max(end_dt) m_dt from t,t1
             where t.emp_id = t1.emp_id
             and end_dt <> m_dt
             group by t1.emp_id
             )Alvinder