Consecutive date query... Please help

Question

Dear All,
I have this table "att_emp_leaves" that records employees leaves information in the following columns:
Employee_ID number(8)
From_Date date
To_Date
Now I have to trace records that have consectutive date records, like leave taken again when the previous leaves ends up. The following records will help you understand this.
Employee_ID , From_Date , To_Date
2711, 19-DEC-07, 07-JAN-08
2711, 08-JAN-08 , 10-JAN-08
or
3134 , 12-DEC-07 , 03-JAN-08
3134 , 04-JAN-08 , 04-JAN-08
In first case employee applied for leaves from 19-dec to 7-jan and then the very next date again from 8-jan till 10-jan.
In second record set again employee having id 3134 applies leaves just after his previous leaves were ended up. To date of the first record is 03-JAN while 04-Jan just comes after 3-jan i.e. 4-jan.
There are other records too in this table that might not have consetuive entries of dates and they are to be ignored, records like
14210 , 01-JAN-08, 02-JAN-08
14210 , 09-JAN-08, 10-JAN-08
This record is to be ignored.
Kindly help me write such a query. I will be very greatful to you.
Regards, Imran Baig

Answer 1

Here's a version that will return 1 record per employee_id per sequence of multiple leaves. I added some additional test data to to show that it will handle leave sequences greate than 2, as well as multiple leave sequences. Employee_id 2711 has single sequence of 3 consecutive leaves while employee_id 3134 has two sequences one of length 2 and another 1 year later of length 3.
with att_emp_leaves as (select 2711 employee_id, to_date('19-DEC-07','dd-mon-yy') from_date, to_date('07-JAN-08','dd-mon-yy') to_date from dual
  union all select 2711, to_date('08-JAN-08','dd-mon-yy'), to_date('10-JAN-08','dd-mon-yy') from dual
  union all select 2711, to_date('11-JAN-08','dd-mon-yy'), to_date('13-JAN-08','dd-mon-yy') from dual
  union all select 3134, to_date('12-DEC-07','dd-mon-yy'), to_date('03-JAN-08','dd-mon-yy') from dual
  union all select 3134, to_date('04-JAN-08','dd-mon-yy'), to_date('04-JAN-08','dd-mon-yy') from dual
  union all select 3134, to_date('12-DEC-08','dd-mon-yy'), to_date('03-JAN-09','dd-mon-yy') from dual
  union all select 3134, to_date('04-JAN-09','dd-mon-yy'), to_date('04-JAN-09','dd-mon-yy') from dual
  union all select 3134, to_date('05-JAN-09','dd-mon-yy'), to_date('07-JAN-09','dd-mon-yy') from dual
  union all select 14210,to_date('01-JAN-08','dd-mon-yy'), to_date('02-JAN-08','dd-mon-yy') from dual
  union all select 14210,to_date('09-JAN-08','dd-mon-yy'), to_date('10-JAN-08','dd-mon-yy') from dual
), t1 as (
  select d.*,
      connect_by_root from_date original_from_date,
      sys_connect_by_path(d.to_date,'->') end_dates,
      level leave_cnt
  from att_emp_leaves d
  connect by prior employee_id = employee_id and prior d.to_date = from_date-1
  start with (employee_id, from_date) in
    (select employee_id, from_date
     from att_emp_leaves d1
     where not exists
       (select 1 from att_emp_leaves d2
        where d2.employee_id=d1.employee_id and d2.to_date = d1.from_date-1))
), t2 as (
  select t1.*, rank() over (partition by employee_id, original_from_date order by leave_cnt desc) rn
  from t1
select t2.*, t2.to_date-original_from_date leave_duration from t2
where rn=1 and leave_cnt > 1
order by employee_id, original_from_date, to_date
EMPLOYEE_ID FROM_DATE   TO_DATE     ORIGINAL_FROM_DATE  END_DATES                                LEAVE_CNT  RN  LEAVE_DURATION
2711        11-JAN-2008 13-JAN-2008 19-DEC-2007         ->07-JAN-2008->10-JAN-2008->13-JAN-2008  3          1   25            
3134        04-JAN-2008 04-JAN-2008 12-DEC-2007         ->03-JAN-2008->04-JAN-2008               2          1   23            
3134        05-JAN-2009 07-JAN-2009 12-DEC-2008         ->03-JAN-2009->04-JAN-2009->07-JAN-2009  3          1   26            
3 rows selected