SQL Server : recursive update statement - sql

Question

I'm somewhat new to SQL and I am trying to figure out the best way of doing this without hardcoding update statements in SQL Server 2012.
Basically I have a hierarchical table of companies (think of a supply chain) with columns (CompanyID, ParentID, ParentPriceAdj, CompanyPriceAdj). Each company gets assigned a price adjustment by their parent that modifies a list price in the PartMaster table and final price gets calculated by cascading the adjustments from parent to child.
If a parents price adjustment gets updated, I want that to reflect on all of his child companies and so forth
aka:
When updating a CompanyPriceAdj for a given updatedcompanyID, I want to recursively find the child CompanyID's (ParentId = updatedCompanyID) and update their ParentPriceAdj to ParentCompany's (parentPriceAdj * (1 + CompanyPriceAdj))
CompanyId ParentID ParentPriceAdj CompanyPriceAdj
5 6 0.96 .10
6 8 1 .20
7 6 0.96 .15
8 11 1 0
10 6 0.96 0
11 12 1 0
I was thinking of using a stored procedure that updates then repeatedly calls itself for every child that was just updated and then subsequently updates his children.... until the company has no children
I've tried looking around couldn't find any examples like this
This is what I have right now
ALTER PROCEDURE [dbo].[UpdatePricing]
#updatedCompanyID int, #PriceAdj decimal
AS
BEGIN
SET NOCOUNT ON;
WHILE (Select CompanyID From CompanyInfo Where ParentID = #updatedCompanyID) IS NOT NULL
UPDATE CompanyInfo
SET ParentPriceAdj = #PriceAdj * (1+CompanyPriceAdj),
#updatedCompanyId = CompanyID,
#PriceAdj = CompanyPriceAdj
WHERE ParentID = #updatedCompanyID
--- some method to call itself again for each (#updatedCompanyID, #PriceAdj)
END

Answer 1

Recursive CTE can be used to walk hierarchy, something like:
ALTER PROCEDURE [dbo].[UpdatePricing]
(
#companyID int,
#PriceAdj decimal
)
as
begin
set nocount on
update CompanyInfo
set CompanyPriceAdj = #PriceAdj
where CompanyID = #companyID
;with Hierarchy(CompanyID, ParentID, InPriceAdj, OutPriceAdj)
as (
select D.CompanyID, D.ParentID, cast(D.ParentPriceAdj as float),
cast(D.ParentPriceAdj as float) * cast(1 + D.CompanyPriceAdj as float)
from CompanyInfo D
where CompanyID = #companyID
union all
select D.CompanyID, D.ParentID,
H.OutPriceAdj, H.OutPriceAdj * (1 + D.CompanyPriceAdj)
from Hierarchy H
join CompanyInfo D on D.ParentID = H.CompanyID
)
update D
set D.ParentPriceAdj = H.InPriceAdj
from CompanyInfo D
join Hierarchy H on H.CompanyID = D.CompanyID
where
D.CompanyID != #companyID
end

Answer 2

You can use WITH expression in t-sql to get all parent records for given child record. And can update each record in record set accordingly with your logic.
Here are links for WITH expression --
http://msdn.microsoft.com/en-us/library/ms175972.aspx
http://msdn.microsoft.com/en-us/library/ms186243(v=sql.105).aspx

Answer 3

Have you looked at using a CURSOR? It's a bit of overhead, but it may give you what you need. In case you're not familiar, below is a basic outline. It allows you pull each company ID one at a time, make your changes, and move on. You can nest them, include them in loops or include loops in them, whatever you need to go through a list performing actions until there are no actions left needing attention.
declare #updatedCompanyID Varchar(5)
DECLARE D_cursor CURSOR FOR Select updatedCompanyID from
OPEN D_cursor
FETCH NEXT FROM D_cursor into #updatedCompanyID
WHILE ##FETCH_STATUS =0
BEGIN
--Run Your Code
WHERE parentID = #updatedCompanyID
FETCH NEXT FROM D_cursor into #updatedCompanyID
END
Close D_cursor
DEALLOCATE D_cursor
Hope this is helpful. I apologize if I've misunderstood.