Question

Say there is a table which stores a hierarchical structure like this:

item_id | hierarchical_id 
--------+-----------------
    1   | ;1;
    2   | ;1;2;
    3   | ;1;2;3;
    4   | ;1;2;4;
    5   | ;1;2;4;5;

The hierarchy stored here is 1 as root, 2 is a child of 1, 3 and 4 are children of 2 and 5 is the child of 4.

The query

SELECT
  -- the substr is used to remove the first and last semicolumns
  regexp_split_to_table(substr(hierarchical_id, 2, length(hierarchical_id) - 2)
                        , E';'
  ) as parent_id,
  item_id,
  hierarchical_id
FROM 
  table

returns

parent_id | item_id | hierarchical_id
----------+---------+-----------------
       1  |    1    | ;1;
       1  |    2    | ;1;2;
       2  |    2    | ;1;2;
       1  |    3    | ;1;2;3;
       3  |    3    | ;1;2;3;
       1  |    4    | ;1;2;3;
       2  |    4    | ;1;2;4;
       4  |    4    | ;1;2;4;
       1  |    5    | ;1;2;4;5;
       2  |    5    | ;1;2;4;5;
       4  |    5    | ;1;2;4;5;
       5  |    5    | ;1;2;4;5;

How can I modify the query to get a 4th column like this:

parent_id | item_id | hierarchical_id | distance
----------+---------+-----------------+---------
       1  |    1    | ;1;             | 0
       1  |    2    | ;1;2;           | 1
       2  |    2    | ;1;2;           | 0
       1  |    3    | ;1;2;3;         | 2 
       2  |    3    | ;1;2;3;         | 1
       3  |    3    | ;1;2;3;         | 0
       1  |    4    | ;1;2;4;         | 2
       2  |    4    | ;1;2;4;         | 1
       4  |    4    | ;1;2;4;         | 0
       1  |    5    | ;1;2;4;5;       | 3
       2  |    5    | ;1;2;4;5;       | 2
       4  |    5    | ;1;2;4;5;       | 1
       5  |    5    | ;1;2;4;5;       | 0

The meaning of distance is the distance between the item_id and the parent_id on the current row. Eg: the distance between a node and itself is 0, and the distance between a node and its parent is 1, the distance between a node and its parent parent is 2 etc. It does not have to start at 0.

row_number would work fine if I could get it to restart at 0 for each group of equal item_ids, since the ids in hierarchical_id are ordered.

Any suggestions?

Answer 1

Window functions give you lots of control; see 4.2.8. Window Function Calls.

The key thing you need is:

row_number() OVER (PARTITON BY item_id ORDER BY hierarchical_id)

Given data:

create table t ( item_id integer, hierarchical_id text );
insert into t (item_id, hierarchical_id) values
(1,';1;'),
(2,';1;2;'),
(3,';1;2;3;'),
(4,';1;2;4;'),
(5,';1;2;4;5;');

the query:

WITH x AS (
  SELECT regexp_split_to_table(substr(hierarchical_id, 2, length(hierarchical_id) - 2), E';') as parent_id,
    item_id,
    hierarchical_id
  FROM t
)
SELECT 
  *,
  row_number() OVER (PARTITION BY item_id ORDER BY parent_id DESC) - 1 AS distance
FROM x
ORDER BY item_id, parent_id;

produces:

 parent_id | item_id | hierarchical_id | distance 
-----------+---------+-----------------+----------
 1         |       1 | ;1;             |        0
 1         |       2 | ;1;2;           |        1
 2         |       2 | ;1;2;           |        0
 1         |       3 | ;1;2;3;         |        2
 2         |       3 | ;1;2;3;         |        1
 3         |       3 | ;1;2;3;         |        0
 1         |       4 | ;1;2;4;         |        2
 2         |       4 | ;1;2;4;         |        1
 4         |       4 | ;1;2;4;         |        0
 1         |       5 | ;1;2;4;5;       |        3
 2         |       5 | ;1;2;4;5;       |        2
 4         |       5 | ;1;2;4;5;       |        1
 5         |       5 | ;1;2;4;5;       |        0

which looks roughly right, but since your expected output doesn’t seem to match the output of the query you provided when I run it (Pg 9.1) it’s hard to know for sure.