TSQL Question About Update Column with Hierarchy in Value - tsql

Question

I have this table with two columns, Hierarchy (Alphabet order) and Access as below.
N Hierarchy Access
1 A Y
2 A >B N
3 A >B >C NULL
4 A >B >C >D NULL
5 A >B >C >D >E NULL
6 A >B >C >D >E >F NULL
7 A >B >C >D >E >F >G Y
8 A >B >C >D >E >F >G >J NULL
I need to update Access column with this logic. If the Access value is null, update the Access for this row with Access from next higher hierarchy where Access is not null.
For example,
Row 7's Access is null, the query will update Row 7's Access to Y. Because hierarchy G is Y.
Row 6 will be N, because hierarchy E, D and C is null, and B is N (not null).
Row 5 will be N, because hierarchy D and C is null, and B is N (not null).
Row 4 will be N, because hierarchy C is null, and B is N (not null).
So the desired output would look like this
N Hierarchy Access
1 A Y
2 A >B N
3 A >B >C N
4 A >B >C >D N
5 A >B >C >D >E N
6 A >B >C >D >E >F N
7 A >B >C >D >E >F >G Y
8 A >B >C >D >E >F >G >J Y
How can I achieve this? Thank you.

Answer 1

Honestly, I'm going to firstly repeat my comment:
I really suggest fixing your design. Storing delimited data like this is not a good idea. SQL Server has a built in hierachyid data type, and if you don't want to use that, you are far better off using a primary and foreign key relationship.
This isn't pretty to achieve with your denormalised data. You have to use LIKE expressions and then get rid of the unwanted JOINed rows with a "TOP 1 in each group", by doing something like this:
CREATE TABLE dbo.DelimitedHierarchy (N int NOT NULL,
Hierarchy varchar(200) NOT NULL,
Access char(1) NULL);
GO
INSERT INTO dbo.DelimitedHierarchy (N, Hierarchy, Access)
VALUES (1,'A','Y'),
(2,'A >B','N'),
(3,'A >B >C',NULL),
(4,'A >B >C >D',NULL),
(5,'A >B >C >D >E',NULL),
(6,'A >B >C >D >E >F',NULL),
(7,'A >B >C >D >E >F >G','Y'),
(8,'A >B >C >D >E >F >G >J',NULL);
GO
WITH rCTE AS(
SELECT DH.N,
DH.Hierarchy,
DH.Access,
1 AS Level
FROM dbo.DelimitedHierarchy DH
WHERE DH.Hierarchy NOT LIKE '%>%'
UNION ALL
SELECT DH.N,
DH.Hierarchy,
ISNULL(DH.Access,r.Access),
r.Level + 1
FROM dbo.DelimitedHierarchy DH
JOIN rCTE r ON DH.Hierarchy LIKE r.Hierarchy + ' >%')
SELECT TOP (1) WITH TIES
*
FROM rCTE r
ORDER BY ROW_NUMBER() OVER (PARTITION BY Hierarchy ORDER BY Level DESC);
GO
DROP TABLE dbo.DelimitedHierarchy;
If you had a properly normalised data set, not using hierarchyid then the query would look like this:
CREATE TABLE dbo.NormalisedHierarchy (N int NOT NULL,
value char(1) NOT NULL,
parent char(1) NULL,
Access char(1) NULL);
GO
INSERT INTO dbo.NormalisedHierarchy (N, Value, Parent, Access)
VALUES (1,'A',NULL,'Y'),
(2,'B','A','N'),
(3,'C','B',NULL),
(4,'D','C',NULL),
(5,'E','D',NULL),
(6,'F','E',NULL),
(7,'G','F','Y'),
(8,'J','G',NULL);
GO
WITH rCTE AS(
SELECT NH.N,
NH.[value],
NH.Access,
1 AS Level
FROM dbo.NormalisedHierarchy NH
WHERE NH.parent IS NULL
UNION ALL
SELECT NH.N,
NH.[value],
ISNULL(NH.Access,r.Access),
r.Level + 1
FROM dbo.NormalisedHierarchy NH
JOIN rCTE r ON NH.parent = r.[value])
SELECT *
FROM rCTE r
ORDER BY [Level] ASC;
GO
DROP TABLE dbo.NormalisedHierarchy;
Alternatively, implement a hierarchyid. I don't include a solution for that, as I have never worked with them, and have used self referencing normalised tables.