Question

I have tables containing the following fields among others in a application for a transportation company…

jempe_users

+------------------------------+
| Field                        |
+------------------------------+
| user_id                      |
| user_user_group_id           |
| user_type                    | 
| user_real_name               |
| user_drv_code                |
+------------------------------+

jempe_user_types:

+---------------------+
| Field               |
+---------------------+
| user_type_id        |
| user_type_name      |
+---------------------+

employee_clock_events:

+-----------------------+
| Field                 |
+-----------------------+
| ece_id                |
| ece_employee_id       |
| ece_type              |
| ece_datetime          |
| ece_active            |
+-----------------------+

What I need to do is suggest three drivers that are currently working and not on break that can be assigned to pick up a patient.

The jempe_users tables contains employees as well as some clients of the transportation company. The user types table has a record where the user_type_name = driver and the user_type field in the jempe_users table references the jempe_user_types table.

The employee_clock_events table is used to clock employees in and out during their shift. The ece_type field is enum(‘shift_start’, ‘shift_brake’, ‘shift_resume’, ‘shift_stop’). The ece_datetime field stores a timesatmp of when the shift event took place.

So what I need to do is find driver user types in user_group 16, 17, or 18 that are currently working today and who’s most recent shift ece_type is either ‘shift_start’ or ‘shift_resume’.

I’ve tried dozens of variations of the following but can’t seem to select employees WHO ARE CURRENTLY working. I seem to have no problems selected employees that have at least worked at some point today.

Tried variations of this…

SELECT user_id, driver_name, driver_code
FROM
(
    SELECT
        u.user_id AS user_id,
        u.user_real_name AS driver_name,
        u.user_drv_code AS driver_code,
        e.ece_type AS event_type,
        e.ece_datetime AS event_datetime
    FROM employee_clock_events AS e
    INNER JOIN jempe_users AS u 
        ON e.ece_employee_id = u.user_id
    LEFT JOIN jempe_user_types AS ut
        ON u.user_type = ut.user_type_id
    WHERE
        e.ece_shift_date = '2011-09-13'
        AND ut.user_type_name = 'driver'
        AND u.user_user_group_id IN (16, 17, 18)
        AND NOT ISNULL(u.user_drv_code)
        AND u.user_active = 1
) AS TEMPORARY
GROUP BY user_id
HAVING MAX(event_datetime)

And this…

SELECT u.user_id AS user_id, u.user_real_name AS driver_name, u.user_drv_code AS driver_code, MAX(e.ece_datetime) AS current_event_datetime
FROM employee_clock_events AS e
LEFT JOIN jempe_users AS u 
    ON e.ece_employee_id = u.user_id
LEFT JOIN jempe_user_types AS ut
    ON u.user_type = ut.user_type_id
WHERE
    (e.ece_type = 'shift_start' OR e.ece_type = 'shift_resume')
    AND e.ece_shift_date = '2011-09-13'
    AND ut.user_type_name = 'driver'
    AND u.user_user_group_id IN (16, 17, 18)
    AND NOT ISNULL(u.user_drv_code)
    AND u.user_active = 1
GROUP BY e.ece_employee_id

Where am I going wrong? Thanks for the help.

Answer 1

Your problem is determining what the most recent clock event is per employee. That part of the problem can be solved as follows:

 CREATE VIEW MRClockByEmployee (ece_employee_ID, ece_most_recent_datetime) AS
    AS SELECT ece_employee_ID, MAX(ece_datetime) 
    FROM employee_clock_events
    GROUP BY ece_employee_id;

 CREATE VIEW MRClockInfoByEmployee 
     (ece_id, ece_employee_ID, ece_type, ece_datetime, ece_active) AS
 SELECT CE.* FROM employee_clock_events CE JOIN MRClockByEmployee MR
    ON CE.ece_employee_ID = MR.ece_employee_id AND
       CE.ece_datetime = MR.ece_most_recent_datetime

 SELECT ece_employee_id FROM MRClockInfoByEmployee
    WHERE type IN ('shift_start', 'shift_resume')

You can then JOIN the contents of that second view back to your other tables to get the other information and take the TOP 3 records for your three possible drivers.

Note that you could do all that in a single, more complex, query but I think the concept of “employee current status” is important and reusable enough that it’s worthwhile encapsulating the logic in a view.

The final solution (where you JOIN the tables together) will not use a GROUP BY. That is because GROUP BY must be used relatively early in the solution algorithm in order to create a list of employee ids and most recent datetime values; that list is then joined back to the employee table to get the a list of most recent records from that table.