Join Nearest Date - mysql

Question

I have read through a ton of responses on here but nothing is working quite as well as I would like. I currently have a working query that includes 2 sub queries, the problem is that it takes about 10 seconds to execute. I was wondering if there is any way to make this go quicker, maybe with a join. I just can't seem to get my head out of the box it is in. Please let me know your thoughts.
Here is the working query:
Select concat(a.emp_firstname, ' ', a.emp_lastname) as names
, if(if (a.emp_gender = 1, 'Male', a.emp_gender)=2, 'Female',
if (a.emp_gender = 1, 'Male', a.emp_gender)) as emp_gender
, c.name
, a.emp_work_telephone
, a.emp_hm_telephone, a.emp_work_email
, a.custom7, a.employee_id
, a.city_code, a.provin_code, d.name as status,
(SELECT cast(concat(DATE_FORMAT(e.app_datetime, '%H:%i'), ' ', e.app_facility) as char(100))
FROM li_appointments.li_appointments as e where e.terp_id = a.employee_id
and e.app_datetime <= str_to_date('06/26/13 at 3:20 PM', '%m/%d/%Y at %h:%i %p')
and date(e.app_datetime) = date(str_to_date('06/26/13 at 3:20 PM', '%m/%d/%Y at %h:%i %p'))
order by e.app_datetime desc limit 1) as prevapp,
(SELECT cast(concat(DATE_FORMAT(e.app_datetime, '%H:%i'), ' ', e.app_facility) as char(100))
FROM li_appointments.li_appointments as e
where e.terp_id = a.employee_id
and e.app_datetime > str_to_date('06/26/13 at 3:20 PM', '%m/%d/%Y at %h:%i %p')
and date(e.app_datetime) = date(str_to_date('06/26/13 at 3:20 PM', '%m/%d/%Y at %h:%i %p'))
order by e.app_datetime desc limit 1) as nextapp
from hs_hr_employee as a
Join hs_hr_emp_skill as b on a.emp_number = b.emp_number
Join ohrm_skill as c on b.skill_id = c.id
Join orangehrm_li.ohrm_employment_status as d on a.emp_status = d.id
where c.name like '%Arabic%'
and d.name = 'Active' order by rand();
EXPLAIN results:
+----+--------------------+-------+--------+---------------------+------------+---------+---------------------------+-------+----------------------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+--------------------+-------+--------+---------------------+------------+---------+---------------------------+-------+----------------------------------------------+
| 1 | PRIMARY | d | ALL | PRIMARY | | | | 10 | Using where; Using temporary; Using filesort |
| 1 | PRIMARY | a | ref | PRIMARY,emp_status | emp_status | 5 | orangehrm_li.d.id | 48 | Using where |
| 1 | PRIMARY | b | ref | emp_number,skill_id | emp_number | 4 | orangehrm_li.a.emp_number | 1 | |
| 1 | PRIMARY | c | eq_ref | PRIMARY | PRIMARY | 4 | orangehrm_li.b.skill_id | 1 | Using where |
| 3 | DEPENDENT SUBQUERY | e | ALL | | | | | 28165 | Using where; Using filesort |
| 2 | DEPENDENT SUBQUERY | e | ALL | | | | | 28165 | Using where; Using filesort |
+----+--------------------+-------+--------+---------------------+------------+---------+---------------------------+-------+----------------------------------------------+

Answer 1

Your tables appear small enough to do as I have done here. First, the inner-most query starts with all employees and did TWO left-joins immediately to the appointment table... By getting a MAX() of the appointments less than the date in question gets the "Previous Appointment", and getting the MIN() appointment AFTER the date in question gets the "Next Appointment". So now, for a single person, I have both their ID and possible previous and next appointments based on their specific times.
Now, I take that result and re-join first to the appointment tables (left joined again) but this time based on same person (Terp_ID) AND their respective Previous and Next appointments date/time. This would only be a problem if you had multiple entries with the exact same date/time for a single person, and that would just result in multiple records.
So now, I have each person with the specifics of previous and next appointments available.
The rest is simple joining to the other tables to get only employee status of "Active", and the skill set of "Arabic" criteria (which I have at their respective JOIN criteria), otherwise you could just move these to a WHERE clause.
As for the "Date/Time" basis of the query, I used #variable once so it could be used against both left-joins to appointments. Finally, I grabbed the respective fields you wanted. This SHOULD work, yet without your data, might need some tweaking.
SELECT
EmpPrevNext.Employee_ID,
EmpPrevNext.PrevApnt,
EmpPrevNext.NextApnt,
concat(Emp2.emp_firstname, ' ', Emp2.emp_lastname) as names,
if ( Emp2.emp_gender = 1, 'Male', 'Female' ) as emp_gender,
Emp2.emp_work_telephone,
Emp2.emp_hm_telephone,
Emp2.emp_work_email,
Emp2.custom7,
Emp2.city_code,
Emp2.provin_code,
cast( concat( DATE_FORMAT(PriorApp2.app_datetime, '%H:%i'), ' ', PriorApp2.app_facility) as char(100))
as PriorAppointment,
cast( concat( DATE_FORMAT(NextApp2.app_datetime, '%H:%i'), ' ', NextApp2.app_facility) as char(100))
as NextAppointment,
EStat.`name` as EmployeeStatus,
Skill.`name` as SkillName
FROM
( SELECT
Emp.Employee_ID,
MAX( PriorApp.app_DateTime ) as PrevApnt,
MIN( NextApp.app_DateTime ) as NextApnt
from
( select #DateBasis := '06/26/13 at 3:20 PM' ) sqlvars,
hs_hr_employee as Emp
LEFT JOIN li_appointments.li_appointments as PriorApp
ON Emp.Employee_ID = NextApp.Terp_ID
AND PriorApp.app_DateTime <= #DateBasis
LEFT JOIN li_appointments.li_appointments as NextApp
ON Emp.Employee_ID = NextApp.Terp_ID
AND NextApp.app_DateTime > #DateBasis
group by
Emp.Employee_ID ) EmpPrevNext
LEFT JOIN li_appointments.li_appointments as PriorApp2
ON EmpPrevNext.Employee_ID = PriorApp2.Terp_ID
AND EmpPrevNext.PrevApnt = PriorApp2.app_DateTime
LEFT JOIN li_appointments.li_appointments as NextApp2
ON EmpPrevNext.Employee_ID = NextApp2.Terp_ID
AND EmpPrevNext.NextApnt = NextApp2.app_DateTime
JOIN hs_hr_employee as Emp2
ON EmpPrevNext.Employee_ID = Emp2.Employee_ID
JOIN orangehrm_li.ohrm_employment_status as EStat
ON Emp2.Emp_Status = EStat.ID
AND EStat.`name` = 'Active'
JOIN hs_hr_emp_skill as EmpSkill
ON Emp2.emp_number = EmpSkill.emp_number
JOIN ohrm_skill as Skill
on EmpSkill.skill_id = Skill.id
AND Skill.`name` like '%Arabic%'
order by
rand();
Make sure your appointment table has an index on (Terp_ID, app_datetime )