Calculation to be made while updating a field - mysql

Question

Origin Dest Date Amount 50% Due
92509 0021 2013-07-30 00:00:00.000 5.37 0.00
92509 0021 2013-07-30 00:00:00.000 5.37 0.00
92509 0021 2013-07-30 00:00:00.000 5.37 0.00
92509 0021 2013-07-31 00:00:00.000 5.37 2.69
92509 0021 2013-07-31 00:00:00.000 5.37 2.69
92509 0021 2013-07-31 00:00:00.000 5.37 2.69
92509 0021 2013-08-01 00:00:00.000 5.37 2.69
92509 0021 2013-08-01 00:00:00.000 5.37 2.69
42101 0029 2013-03-06 00:00:00.000 6.06 0.00
42101 0029 2013-03-06 00:00:00.000 6.06 0.00
42101 0029 2013-03-07 00:00:00.000 6.06 3.03
42101 0029 2013-03-07 00:00:00.000 6.06 3.03
42101 0030 2013-03-06 00:00:00.000 6.06 0.00
42101 0030 2013-03-06 00:00:00.000 6.06 0.00
42101 0030 2013-03-07 00:00:00.000 6.06 3.03
42101 0030 2013-03-07 00:00:00.000 6.06 3.03
So I have a table something similar to what i shown above. Right now, the 50% Due field is empty. I need to fill that field with values as shown above.
The 50% due field should populate values that are half of what is present in the Amount field. But, it should fill zero for the initial date (2013-07-30 00:00:00.000) and for the consecutive days it should fill half of what is present in the Amount field.
I have a lot of rows like these that needs to get updated. Also there are rows with different Origin and Destination.
I am dealing with some freight parcels. The data describes the parcels that were sent to the same destination from same origin on consecutive days. The parcels that were sent on consecutive days could have been sent together on the initial date itself. So I am trying to generate a claim for those parcels that were sent on consecutive days to the same destination from the same origin. And the 50% Due would be the claim!
I am fairly new to SQL! This seems to be very complicated for me. Please help.

Answer 1

If you want to update all origin/dest combinations that had an entry on the prior day then you can do this:
update t
set [50% due]=coalesce(cast(p.amount/2.0 as smallmoney),0.00)
from [table] t
left join [table] p
on t.origin=p.origin and t.dest=p.dest
and dateadd(D,-1,cast(t.[date] as date))=cast(p.[date] as date)
or use a cte and LAG in SQL 2012 and later versions:
;with previous as
(
select origin,dest,[date]
,LAG(cast([date] as date),1,cast([date] as date)) OVER (PARTITION BY origin,dest ORDER BY cast([date] as date)) as previous
from (select distinct origin, dest, [date]
from [table]) a
)
update t
set [50% Due]=case when dateadd(D,-1,cast(t.[date] as date))<>cte.previous then 0.00 else cast(t.[amount]/2.0 as smallmoney) end
from [table] t
join previous cte
on cte.origin=t.origin and cte.dest=t.dest and cte.[date]=t.[date]
If you want to update all your records based on the earliest day that the origin/dest combination occurred, then this will work in SQL Server:
;with earliest as
(
select origin,dest,min(cast([date] as date)) earliest
from [table]
group by origin,dest
)
update t
set [50% Due]=case when cast(t.[date] as date)=cte.earliest then 0.00 else cast(t.[amount]/2.0 as smallmoney) end
from [table] t
join earliest cte
on cte.origin=t.origin and cte.dest=t.dest
If you want to update all your records only based on the earliest day in the table and you don't care about the orgin/dest combination then you don't need the cte to group the earliest dates and you can simply do the compare in the update.
UPDATE t
set [50% Due]=case when cast(t.[date] as date)=(select min(cast(t.date as date))) then 0.00 else cast(t.[amount]/2.0 as smallmoney) end