Question

I have a self reference table to store hierarchical values in order to show up them in a TreeView or so on ,according to James Crowley article (Tree structures in ASP.NET and SQL Server)

Our table would look something like this:

Id  ParentId      Name        Depth    Lineage
1   NULL        Root Node       0       /1/
2   1           Child A         1       /1/2/
3   1           Child B         1       /1/3/
4   1           Child C         1       /1/4/
5   2           Child D         2       /1/2/5/

To get path of a node (for example id=5) he suggest following query against table

SELECT *
FROM dfTree
WHERE (SELECT lineage
       FROM dfTree
       WHERE id = 5) LIKE lineage + '%'

result would be :

Id  ParentId      Name        Depth    Lineage
1   NULL        Root Node       0       /1/
2   1           Child A         1       /1/2/
5   2           Child D         2       /1/2/5/

And Is Acceptable

But how about to have a result set when there are multiple IDs which i want to have their path ? therefore for example in above example instead of Id=5 i would like to pass multiple values something like this :

SELECT *
FROM dfTree
WHERE (SELECT lineage
       FROM dfTree
       WHERE id IN (5,6,8,9)) LIKE lineage + '%'

But the above statement make no sense and it is invalid sql server expression
How could i solve this problem ?
Thanks in advance

Answer 1

This query…

SELECT DISTINCT T2.*
FROM 
    (
        SELECT lineage
        FROM dfTree
        WHERE id IN (4, 5)
    ) T1
    JOIN dfTree T2
    ON
        T1.Lineage LIKE T2.Lineage + '%'

…returns the following result on your test data:

Id   ParentId   Name        Depth   Lineage
1    NULL       Root Node   0       /1/
2    1          Child A     1       /1/2/
4    1          Child C     1       /1/4/
5    2          Child D     2       /1/2/5/

As you can see, all paths are “merged” together – for example, the path component Id=1 belongs to both the path: /1/4/ and the path: /1/2/5/, yet exists in the result set only once.

---

One the other hand, if you do need to distinguish between different paths, you’d need to do something like this:

SELECT T2.*, T1.Id LeafId
FROM 
    (
        SELECT id, lineage
        FROM dfTree
        WHERE id IN (4, 5)
    ) T1
    JOIN dfTree T2
    ON
        T1.Lineage LIKE T2.Lineage + '%'

Result:

Id   ParentId   Name        Depth   Lineage    LeafId
1    NULL       Root Node   0       /1/        4
4    1          Child C     1       /1/4/      4
1    NULL       Root Node   0       /1/        5
2    1          Child A     1       /1/2/      5
5    2          Child D     2       /1/2/5/    5

In this case, each path is identified by its leaf. This assumes there are no diamond-shaped dependencies (i.e. this is a real tree and not just any DAG); if there are, then you’d need to use T1.Lineage instead of T1.Id to identify the path.