Date Ranges Query

Question

Hi Experts,
A very happy new year to all of you(to some in advance)!!
I have a table where I have the employee id and task assigned with start dates & end dates. Overlapping dates of tasks is a data issue however I know how to spot them & eradicate them. So there will be no overlapping task dates
Employee_id
Task_No
Task_Start_date
Task_End_Date
Sample Data
1     T1     01-Jan-2014     28-Feb-2014
1     T2     01-Mar-2014     31-Dec-2014
2     T1     23-Jan-2014     31-Dec-2014
2     T2     01-Jan-2015     31-Dec-2073 (Means end of time)
3     T3     01-Jan-2014     15-Jul-2014
3     T4     01-Aug-2014     31-Dec-2014
4     T5     01-Jan-2014     31-Dec-2073
I want to devise a query where i provide the end date & it will list out all the employees who are free for even one day starting 01-Jan-2014. So for example If I give 31-Dec-2014, it will list out
EmpId     (First day on which the employee is free)
2               01-Jan-2014
3               16-Jul-2014
If I give end date = end of time(31-Dec-2013), Expected Result -
EmpId     (First day on which the employee is free)
1               01-Jan-2015
2               01-Jan-2014
3               16-Jul-2014
If I give end date = 31 Jan 2014, Expected Result -
EmpId     (First day on which the employee is free)
1               01-Jan-2015
I devised following query, however it does not catch employee id 2. Also it does not provide flexibility to change end date-
select *
  from (select employee_id,
               task_start_date,
               task_end_date,
               lag(task_end_date, 1, task_start_date) over(partition by employee_id order by task_start_date) prev_end_date
          from shop.employee_tasks
         where task_end_date >= trunc(sysdate))
where task_start_date - prev_end_date > 1
Thanks in advance!!
Regards,

Answer 1

This is an example of what I call the "free time" query: you have the dates when you are busy and you want the dates when you are free.
You can find a beautiful solution to the basic problem here: Ask Tom "SQL Query to find gaps in date ranges" (search for Antony Boucher's solution). Please note that this solution works even with overlapping date ranges.
To apply the solution here, first create the test data (please do this yourself in later questions).
create table t(Employee_id, task_no, Task_Start_date, task_end_date)
as select
1, 'T1', to_date('01-jan-2014'), to_date('28-feb-2014') from dual union all select
1, 'T2', to_date('01-Mar-2014'), to_date('31-Dec-2014') from dual union all select
2, 'T1', to_date('23-jan-2014'), to_date('31-dec-2014') from dual union all select
2, 'T2', to_date('01-jan-2015'), to_date('31-dec-2073') from dual union all select
3, 'T3', to_date('01-jan-2014'), to_date('15-jul-2014') from dual union all select
3, 'T4', to_date('01-aug-2014'), to_date('31-dec-2014') from dual union all select
4, 'T5', to_date('01-Jan-2014'), to_date('31-Dec-2073') from dual;
In the query, you have to add records for yesterday and for the "end date" you want. This allows you to find "free time" before and after the date ranges in your table.
Now you partition by Employee_id and order by Task_Start_date. Using the max(task_end_date) analytic function, you get the latest end date up to now. Add 1 to this and you get the first free date (maybe). To make sure that date is free, it has to be before the next start date.
variable end_date varchar2(64)
exec :end_date := '31-Dec-2073';
with boundaries as (
  select trunc(sysdate)-1 task_start_date, trunc(sysdate)-1 task_end_date from dual
  union all
  select to_date(:end_date)+1, to_date(:end_date)+1 from dual
), data as (
  select * from t
where task_end_date >= trunc(sysdate)
  and task_start_date < :end_date
  union all
  select distinct a.employee_id, null, b.task_start_date, b.task_end_date
  from t a, boundaries b
select employee_id, min(free_start) free_start
from (
  select employee_id,
  max(task_end_date) over (partition by employee_id order by task_start_date)+1 free_start,
  lead(task_start_date) over (partition by employee_id order by task_start_date)-1 free_end
  from data
where free_start <= free_end
group by employee_id;
EMPLOYEE_ID
FREE_START
1
01-JAN-15
2
01-JAN-14
3
16-JUL-14