SQL Server 2008 exec stored procedure on 3rd strike - sql-server

Question

I've been asked to create a process that triggers a stored procedure when an employee hits their 3rd strike. The strikes relate to absence, so if an employee is off 3 times in a 3 month period it hits the trigger.
But... this only applies to single instances of absence, so if a person is off; for example on the 11/01/2016, 12/01/2016 & 13/01/2016 then this is one instance. Meaning I can't do a count on the number of days off sick.
Data I have available and is a fixed process I can't update:
Date | EmpID | EmpName
01/01/2016 | JS01 | John Spartan
02/01/2016 | JS01 | John Spartan
03/01/2016 | JS01 | John Spartan
08/01/2016 | JS01 | John Spartan
19/02/2016 | JS01 | John Spartan
12/02/2016 | JS01 | John Spartan
Based on the above there are more than 2 instances. So this would trigger the procedure
IF EXISTS (<Query Here>)
BEGIN
EXEC usp_ThreeStrikes
END
Is there a way to do this in T-SQL?

Answer 1

If you can't add columns to help query with the task (eg. InstanceID query would group by to find out number of instances), I think best solution would be to create aggregate CLR function for the task.
https://msdn.microsoft.com/en-us/library/ms131056.aspx

Answer 2

You can try the below approach:
Add an additional column to your table to differentiate if the record should be considered for next strike or not (means after 1 instance, it should not be considered the 2nd time)
Create a SQL Update Trigger to call the procedure, based on the below condition:
Get the records whose column is considered for next strike (same column what you have created in first step)
For those particular records check the count if its greater or equal to 3 and call the stored procedure
For those particular records, update the additional column (created in step 1) to not consider it for the subsequent strikes
Hope this helps.

Answer 3

Here you have a query that lists the empid's that were absent three or more times per quarter. You can modify this query in your trigger to only select in the empids/quarters that are present in the inserted table in your trigger.
PS: I've added some random absences to show that the query only selectes when the number of absences is three or more.
CREATE TABLE #absences(dt DATE,empid NVARCHAR(128),empname NVARCHAR(128));
INSERT INTO #absences(dt,empid,empname)VALUES
('20151212','JS02','John Spartan2'),
('20151213','JS02','John Spartan2'),
('20151010','JS01','John Spartan'),
('20151011','JS01','John Spartan'),
('20151217','JS02','John Spartan2'),
('20151219','JS02','John Spartan2'),
('20160101','JS01','John Spartan'),
('20160102','JS01','John Spartan'),
('20160103','JS01','John Spartan'),
('20160108','JS01','John Spartan'),
('20160201','JS02','John Spartan2'),
('20160203','JS02','John Spartan2'),
('20160219','JS01','John Spartan'),
('20160212','JS01','John Spartan');
SELECT
empid,
[quarter]=DATEADD(QUARTER,DATEDIFF(QUARTER,0,o.dt),0)
FROM
#absences AS o
WHERE
NOT EXISTS (
SELECT 1
FROM #absences AS i
WHERE i.empid=o.empid AND
DATEDIFF(QUARTER,0,i.dt)=DATEDIFF(QUARTER,0,o.dt) AND
i.dt=DATEADD(DAY,-1,o.dt)
)
GROUP BY
empid,
DATEDIFF(QUARTER,0,o.dt)
HAVING
COUNT(*)>=3;
DROP TABLE #absences;
Result:
+-------+-------------------------+
| empid | quarter |
+-------+-------------------------+
| JS02 | 2015-10-01 00:00:00.000 |
| JS01 | 2016-01-01 00:00:00.000 |
+-------+-------------------------+