Question

I’m trying to query a hierarchy of data in a single database table from the bottom up (I don’t want to include parents that don’t have a particular type of child due to authorities). The schema and sample data are as follows:

create table Users(
id int,
name varchar(100));

insert into Users values (1, 'Jill');

create table nodes(
    id int,
    name varchar(100),
    parent int,
    nodetype int);

insert into nodes values (1, 'A', 0, 1);
insert into nodes values (2, 'B', 0, 1);
insert into nodes values (3, 'C', 1, 1);
insert into nodes values (4, 'D', 3, 2);
insert into nodes values (5, 'E', 1, 1);
insert into nodes values (6, 'F', 5, 2);
insert into nodes values (7, 'G', 5, 2);

create table nodeAccess(
    userid int,
    nodeid int,
    access int);

insert into nodeAccess values (1, 1, 1);
insert into nodeAccess values (1, 2, 1);
insert into nodeAccess values (1, 3, 1);
insert into nodeAccess values (1, 4, 1);
insert into nodeAccess values (1, 5, 1);
insert into nodeAccess values (1, 6, 0);
insert into nodeAccess values (1, 7, 1);


with Tree(id, name, nodetype, parent)
as
(
    select n.id, n.name, n.nodetype, n.parent
    from nodes as n
    inner join nodeAccess as na on na.nodeid = n.id
    where na.access =1 and na.userid=1 and n.nodetype=2

    union all

    select n.id, n.name, n.nodetype, n.parent
    from nodes as n
    inner join Tree as t on t.parent = n.id
    inner join nodeAccess as na on na.nodeid = n.id
    where na.access =1 and na.userid=1 and n.nodetype=1
)
select * from Tree

Yields:

  id    name    nodetype    parent
   4    D        2             3
   7    G        2             5
   5    E        1             1
   1    A        1             0
   3    C        1             1
   1    A        1             0

How can I not include the duplicates in the result set? The queries against the real tables have many more nodes at the lowest levels and hence many more duplicates of the parent nodes. The solution needs to work with at least SQL Server 2005.

Thanks in advance!

Answer 1

The simplest (not necessarily the most efficient) solution:

...
)
SELECT DISTINCT id,name,nodetype,parent FROM Tree;

This changes the order from your sample output because the DISTINCT operator implements a sort. If there is some intentional ordering there I cannot detect it but you can add an ORDER BY if you know the order you want.