Question

I have two tables: a schedule table that contains information about how an employee is scheduled and a numbers table in which each number corresponds to a date.

The tables look like:

[Employee Schedule]

ID          Employee ID Project ID  Day ID
----------- ----------- ----------- -----------
1           64          2           168
2           64          2           169
3           64          2           170
4           64          2           171
5           64          1           169
6           64          1           170
7           64          1           171
8           64          1           172
9           64          2           182
10          64          2           183
11          64          2           184

and

[Day Numbers]

ID          Day
----------- ----------
168         2009-06-18
169         2009-06-19
170         2009-06-20
171         2009-06-21
172         2009-06-22
173         2009-06-23
174         2009-06-24
175         2009-06-25
176         2009-06-26
177         2009-06-27
178         2009-06-28
179         2009-06-29
180         2009-06-30
181         2009-07-01
182         2009-07-02
183         2009-07-03
184         2009-07-04

As you can see, Employee 64 is scheduled on project 1 from 2009-06-19 to 2009-06-22 and project 2 from 2009-06-18 to 2009-06-21 and again from 2009-07-02 to 2009-07-04.

My question is: what algorithm can I use to quickly determine the spans of the employee’s schedule in a fashion such that I can display it as follows?

 Employee ID Project ID Duration
 ----------- ---------- ------------
 64          1          2009-06-19 to 2009-06-22
 64          2          2009-06-18 to 2009-06-21
 64          2          2009-07-02 to 2009-07-04

I can do this on the SQL side or the code side. I have Linq at my disposal if I need it. The table doesn’t need to be compiled by SQL. This will happen dynamically on a website and should be as efficient as possible. I don’t want to have to iterate through each and look for breaks in contiguous days if I don’t have to.

Answer 1

Assuming the Day IDs are always sequential for a partial solution…

select *
  from employee_schedule a                    
 where not exists( select *                          
                     from employee_schedule b        
                    where a.employeeid = b.employeeid
                      and a.projectid  = b.projectid 
                      and (a.dayid - 1) = b.dayid )

lists the start day IDs:

 ID      EMPLOYEEID       PROJECTID           DAYID 
 1              64               2             168 
 5              64               1             169 
 9              64               2             182 



select *
  from employee_schedule a                   
 where not exists( select *                         
                     from employee_schedule b       
                    where a.employeeid = b.employeei
                      and a.projectid  = b.projectid
                      and (a.dayid + 1) = b.dayid )

lists the end day IDs:

  ID      EMPLOYEEID       PROJECTID           DAYID 
  4              64               2             171 
  8              64               1             172 
 11              64               2             184