recreate historical ProducerID - tsql

Question

In a system that handles milk quality data, there is a table for lab results. Here is an example of what the data looks like:
This is well and good, except that the ProducerID can change over time. For instance, if a milk producer changes processors, a new ProducerID gets assigned to that milk producer. Unfortunately, this ID change phenomenon is not within control of the company that's using this system.
To track ProducerID changes this system has a table that records that information. Here is an example which corresponds with the lab data above:
The problem is, when a ProducerID change occurs, the ProducerID in the lab results table is updated to whatever the new ProducerID is on ALL the rows. I am tasked with adding another column to the lab results table, ProducerID_Corrected, and populating that using the data in the ProducerID changes table to reflect what the correct ProducerID was for each lab result.
However, I am unsure how to go about populating the ProducerID_Corrected column.
Here is what the lab results should look like:
Here is the code to recreate the 2 tables shown above:
select
OldProducerID,
NewProducerID,
EffectiveDate
into #ProducerIDChanges
from
(
values
(1451, 1539, '3/7/2017'),
(1539, 1650, '3/12/2017')
) d (OldProducerID, NewProducerID, EffectiveDate);
select * from #ProducerIDChanges;
select
ProducerID,
TestDate,
TestType,
TestValue,
ProducerID_Corrected = cast(null as int)
into #LabResults
from
(
values
(1650, '3/1/2017', 'butterfat', 4.7),
(1650, '3/6/2017', 'butterfat', 4.1),
(1650, '3/7/2017', 'butterfat', 3.9),
(1650, '3/8/2017', 'butterfat', 4.0),
(1650, '3/10/2017', 'butterfat', 4.5),
(1650, '3/12/2017', 'butterfat', 4.6)
) d (ProducerID, TestDate, TestType, TestValue);
select * from #LabResults;

Answer 1

Here is the solution used for this problem. It yields a #LabResults table that matches the screen shot of expected results in the original question.
--add and populate a column that can be used for joining to #LabResults
alter table #ProducerIDChanges add CurrentProducerID int;
go
while exists (select top 1 1
from #ProducerIDChanges a
inner join #ProducerIDChanges b
on isnull(a.CurrentProducerID, a.NewProducerID) = b.OldProducerID
and cast(a.EffectiveDate as date) < cast(b.EffectiveDate as date))
begin
update a
set a.CurrentProducerID = b.NewProducerID
from #ProducerIDChanges a
inner join #ProducerIDChanges b
on isnull(a.CurrentProducerID, a.NewProducerID) = b.OldProducerID
and cast(a.EffectiveDate as date) < cast(b.EffectiveDate as date)
end;
update #ProducerIDChanges
set CurrentProducerID = NewProducerID
where CurrentProducerID is null;
--create a temp table that holds effective and termination dates for each ProducerID
select
CurrentProducerID,
FormerProducerID,
EffectiveDate = isnull(effective, '1/1/1900'),
TerminationDate = isnull(termination, '1/1/3000')
into #ProducerIDsWithDates
from
(
select CurrentProducerID, FormerProducerID = OldProducerID, DateType = 'termination', DateValue = cast(dateadd(d, -1, EffectiveDate) as date) from #ProducerIDChanges
union all
select CurrentProducerID, FormerProducerID = NewProducerID, DateType = 'effective', DateValue = EffectiveDate from #ProducerIDChanges
) d
pivot (min(DateValue) for DateType in (effective, termination)) p;
--update ProducerID_Corrected column in #LabResults table
update l
set l.ProducerID_Corrected = d.FormerProducerID
from #LabResults l
inner join #ProducerIDsWithDates d
on l.ProducerID = d.CurrentProducerID
and l.TestDate between d.EffectiveDate and d.TerminationDate;