SQLite - performing calculation on preceding rows - sqlite

Question

The problem is this. If I have a quantity in any location, I want to perform the calculation below to each member of that job_no.
The idea is that if there's a quantity in loc3, the same quantity was previously in loc1 and loc2.
So, how do I get 10 in loc1 and loc2 may be another way to put it..?
select s.job_no, s.part, s.location, s.qty,
coalesce(ptime.setup_time, '-') as setup_time,
coalesce(ptime.cycle_time, '-') as cycle_time,
ci.rate
from stock as s join part_timings as pt
on pt.part = s.part
join locations as l on s.location = l.location
left join part_timings as ptime on s.part = ptime.part
and ptime.location = s.location
join costs_internal as ci
group by s.part, s.location
order by s.part, l.stage
job_no | part | location | qty | setup_time | cycle_time | rate | total
123 p1 loc1 0 60 30 0.5 ?
123 p1 loc2 0 30 15 0.5 ?
123 p1 loc3 10 60 15 0.5 ?
123 p1 loc4 0 60 15 0.5 ?
123 p1 loc5 0 60 15 0.5 ?
123 p1 loc6 0 60 15 0.5 ?
123 p1 loc7 20 60 15 0.5 ?
calculation to get total:
coalesce(round((pt.cycle_time * s.qty * ci.rate) +
(pt.setup_time * ci.rate), 2), '-')
EDIT:
I've added loc4 to loc7.
loc3 would need to have the calculation applied to loc1 and loc2 (qty 10).
loc7 would need to have the calculation applied to all locations that are before it (qty 20).
Maybe I'm not explaining it perfectly, struggle to get my intentions across sometimes with SQL!

Answer 1

Using a simplified version of your data...
select * from stock;
job_no qty location
---------- ---------- ----------
123 0 loc1
123 0 loc2
123 10 loc3
123 0 loc4
456 0 loc1
456 20 loc2
You can use a sub-select to get the quantity for each job and join with it to get the stock for each job.
select stock.*, stocked.qty
from stock
join (select * from stock s where s.qty != 0) as stocked
on stock.job_no = stocked.job_no;
job_no qty location qty
---------- ---------- ---------- ----------
123 0 loc1 10
123 0 loc2 10
123 0 loc4 10
123 10 loc3 10
456 0 loc1 20
456 20 loc2 20
stocked has the row for each job which is currently stocked.
Note that unless you've made a restriction, there may be more than one stocked row for a job.
loc7 would need to have the calculation applied to all locations that are before it (qty 20).
With this data...
sqlite> select * from stock order by job_no, location;
job_no qty location
---------- ---------- ----------
123 0 loc1
123 0 loc2
123 10 loc3
123 0 loc4
123 0 loc5
123 0 loc6
123 20 loc7
456 0 loc1
456 20 loc2
To accomplish this, instead of joining on the subselect do it on a per column basis else we'll get multiple values stocked locations. (There's probably also a way to do it with a join)
In order to make sure we select only previous locations (or our own) it's necessary to check that stock.location <= stocked.location. In order to ensure we get the closest one, order them by location and select only the first one.
select stock.*, (
select stocked.qty
from stock stocked
where stock.job_no = stocked.job_no
and qty != 0
and stock.location <= stocked.location
order by stocked.location asc
limit 1
) as stocked_qty
from stock
order by job_no, location;
job_no qty location stocked_qty
---------- ---------- ---------- -----------
123 0 loc1 10
123 0 loc2 10
123 10 loc3 10
123 0 loc4 20
123 0 loc5 20
123 0 loc6 20
123 20 loc7 20
456 0 loc1 20
456 20 loc2 20
This may be inefficient as a column subselect. It's important that job_no, qty, and location are all indexed.