Question

Consider the following (1:N) relationship:

[entity: user] <------ rid key ------> [entity: rid].

consider the data in both tables as:

select * from user;
user-id        rid-key
a-basa         a
b-basa         b
a.a-basa       a.a   
a.b-basa       a.b
a.a.a-basa     a.a.a
a.a.b-basa     a.a.b
a.b.a-basa     a.b.a
a.b.b-basa     a.b.b
a.b.b.a-basa   a.b.b.a
a.b.b.b-basa   a.b.b.b



select * from rid;

rid-key    parent-rid    enabled
a            null        true
b            null        true
a.a          a           true 
a.b          a           false
a.a.a        a.a         true
a.b.a        a.b         true
a.b.b        a.b         true
a.b.b.a      a.b.b       true
......
n rows

I need to design a single query (not stored procedure) which will input a user-id, and the following facts are considered:

If an user is given access to a rid, then it can also access the parent rid of the rid given – the rid itself is enabled (enabled = true).

This should continue till we reach the root rid, ie. parent rid property is null.

In above example, the list of accessible rid for the user 'a.b.b.a-basa' will be:

a.b.b.a
a.b.b
a.b

and for a.a.a-basa:

a.a.a
a.a
a

can we get this list using a single query? Any sql vendor is fine.

Answer 1

There are several models for Hierarchical data. Most models (like the Adjacency List) require some sort of recursion for some queries. With your design that uses the Materialized Path model, what you want is possible without a recursive query.

Tested in MySQL (that has no recursive queries), at SQL-fiddle test-mysql. It can be easily converted for other DBMS, if you modify the string concatenation part:

SELECT 
     COUNT(*)-1 AS steps_up,
     rid2.rid_key AS ancestor_rid_key
FROM 
    u2 
  JOIN
    rid 
      ON u2.rid_key = rid.rid_key
      OR u2.rid_key LIKE CONCAT(rid.rid_key, '.%')
  JOIN
    rid AS rid2 
      ON rid.rid_key = rid2.rid_key
      OR rid.rid_key LIKE CONCAT(rid2.rid_key, '.%')
WHERE
    u2.userid = 'basa'
  AND
    u2.rid_key = 'a.b.b.a' 
GROUP BY 
    rid2.rid_key, rid2.enabled 
HAVING 
    COUNT(*) + (rid2.enabled = 'true') 
  = SUM(rid.enabled = 'true') + 1 ;

It uses this view, which is not strictly needed but it shows that the user.user_id is storing data that you already have in the rid_key column.

CREATE VIEW u2 AS
SELECT 
    SUBSTRING_INDEX(user_id, '-', -1) AS userid
  , rid_key
FROM user ;

One more note is that the above query does not use the parent_rid column at all. And that I’m sure it can be further improved.