Need to find out gap in data

Question

Hi All
I have the following records:
--Drop Table
drop table agreement;
drop table GRP_INFO;
--Create table
create table agreement
Agreement_Id Number(5),
Coverage_Effective_Date Date,
COVERAGE_termination_date date
Create Table GRP_INFO
Agreement_Id Number(5),
Grp_Id Number(5),
Effective_Date Date,
TERMINATION_DATE Date
--Insertion
Insert Into Agreement
Select 100,'01JAN-2013','31-DEC-2013'
From Dual;
Insert Into Agreement
Select 200,'01JAN-2013','31-DEC-2013'
From Dual;
Insert Into Agreement
Select 300,'01JAN-2013','31-DEC-2013'
From Dual;
Insert Into Agreement
Select 400,'01JAN-2013','31-DEC-2013'
From Dual;
Insert Into Grp_Info
Select 100,1,'01-JAN-2013','31-MAR-2013'
From Dual
UNION ALL
Select 100,2,'01-APR-2013','02-APR-2013'
From Dual
UNION ALL
Select 100,3,'03-APR-2013','15-APR-2013'
From Dual
UNION ALL
Select 100,4,'03-APR-2013','15-APR-2013'
From Dual
UNION ALL
Select 100,5,'01-JUN-2013','31-DEC-2013'
From Dual
Union All
Select 200,6,'01-JAN-2013','02-APR-2013'
From Dual
Union All
Select 200,7,'03-APR-2013','15-APR-2013'
From Dual
Union All
Select 200,8,'03-APR-2013','15-APR-2013'
From Dual
Union All
Select 200,9,'01-JUN-2013','30-NOV-2013'
From Dual
Union All
Select 300,10,'01-JAN-2013','15-APR-2013'
From Dual
Union All
Select 300,11,'16-APR-2013','31-DEC-2013'
From Dual
Union All
Select 400,12,'02-JAN-2013','31-DEC-2013'
From Dual;
COMMIT;
--Query on agreement table
Select * from Agreement;
--Query Result
agreement_id                    coverage_effective_date                     coverage_termination_date
100                                         01-JAN-13                                                   31-DEC-13
200                                         01-JAN-13                                                   31-DEC-13
300                                         01-JAN-13                                                   31-DEC-13
400                                         01-JAN-13                                                   31-DEC-13
--Query on grp_info table
agreement_id                  grp_id       effective_date          termination_date
100                                              1                     01-JAN-13                       31-MAR-13
100                                              2                     01-APR-13                       02-APR-13
100                                              3                     03-APR-13                       15-APR-13
100                                              4                     03-APR-13                       15-APR-13
100                                              5                     01-JUN-13                        31-DEC-13
200                                              6                     01-JAN-13                        02-APR-13
200                                              7                     03-APR-13                        15-APR-13
200                                              8                     03-APR-13                        15-APR-13
200                                              9                     01-JUN-13                         30-NOV-13
300                                              10                   01-JAN-13                         15-APR-13
300                                              11                   16-APR-13                         31-DEC-13
400                                              12                   02-JAN-13                          31-DEC-13
--Result
agreement_id
100
200
400
--Logic for the above result
Each agreement_id have multiple grp_id and all grp_id or atleast one should cover all period agreement_id for example:
- Agreement_id 100 has coverage effective date is 01-Jan-2013 and coverage_termination_date is 31-Dec-2013 and if you look into all record against agreement_id 100 then you can find that the period from 16-Apr-2013 till 31-May-2013 are missing so I need this agreement_id.
- Agreement_id 200 has coverage effective date is 01-Jan-2013 and coverage_termination_date is 31-Dec-2013 and if you look into all record against agreement_id 200 then you can find that the period from 01-DEC-2013 till 31-DEC-2013 are missing so I need this agreement_id.
- Agreement_id 300 has coverage effective date is 01-Jan-2013 and coverage_termination_date is 31-Dec-2013 and if you look into all record against agreement_id 300 then you can find that no period is missing so I don't need this agreement_id.
- Agreement_id 400 has coverage effective date is 01-Jan-2013 and coverage_termination_date is 31-Dec-2013 and if you look into all record against agreement_id 300 then you can find that the period from 01-JAN-2013 till 01-JAN-2013 is missing so I need this agreement_id.
Please let me know if you have any questions related to my scenario and I really appreciate if someone provide me solution for this issue.
Regards
Shumail

Answer 1

Hi,
Here's one way:
WITH got_gap  AS
    SELECT agreement_id, effective_date, termination_date
    ,      CASE
               WHEN  effective_date >
                     1 + MAX (termination_date) OVER ( PARTITION BY  agreement_id
                                                       ORDER BY      effective_date
                                                       ROWS BETWEEN  UNBOUNDED PRECEDING
                                                       AND           1         PRECEDING
               THEN  1
           END  AS gap
    FROM    grp_info
SELECT    g.agreement_id
FROM      got_gap g
JOIN      agreement a  ON  a.agreement_id = g.agreement_id
GROUP BY  g.agreement_id
HAVING    COUNT (g.gap)             > 0
OR        MIN (g.effective_date)    > MIN (a.coverage_effective_date)
OR        MAX (g.termination_date)  < MAX (a.coverage_termination_date)
ORDER BY  g.agreement_id
Output:
AGREEMENT_ID
         100
         200
         400
This makes no assumptions about effective_date and termination_date, except that effective_date <= termination_date on each row.  It's okay if different rows for the same agreement_id overlap, or if one encompasses another.
Don't try to insert VARCHAR2 values (such as '01-JAN-2013') into DATE columns.  Use TO_DATE, or DATE literals.