What query would I use to obtain sibling records when using closure tables?

问题

If I have the following schema & data and am employing the closure table pattern:

+----+----------+------------+--------+ | id | ancestor | descendant | length | +----+----------+------------+--------+ | 1 | 2 | 2 | 0 | | 2 | 2 | 12 | 1 | | 3 | 2 | 13 | 1 | | 4 | 2 | 14 | 1 | | 5 | 2 | 15 | 1 | | 10 | 12 | 12 | 0 | | 11 | 13 | 13 | 0 | | 12 | 14 | 14 | 0 | | 13 | 15 | 15 | 0 | | 9 | 17 | 20 | 1 | | 8 | 17 | 19 | 1 | | 7 | 17 | 18 | 1 | | 6 | 17 | 17 | 0 | | 14 | 18 | 18 | 0 | | 15 | 19 | 19 | 0 | | 16 | 20 | 20 | 0 | +----+----------+------------+--------+

What would my join query back to my main table look like to obtain all the sibling rows of row id 2?

+----+----------+------------+--------+ | id | ancestor | descendant | length | +----+----------+------------+--------+ | 3 | 2 | 13 | 1 | | 4 | 2 | 14 | 1 | | 5 | 2 | 15 | 1 | +----+----------+------------+--------+

回答1:

The siblings of a given node would have the same ancestor. However, this would include "1" as well as your list:

select t.*
from table t 
where t.ancestor = (select ancestor from table t2 where t.id = 2);

In your table, I am not sure what it means for ancestor to be the same as descendant. But, I think the following is the query you want:

select t.*
from table t 
where t.ancestor = (select ancestor from table t2 where t2.id = 2) and
      t.ancestor <> t.descendant and
      t.id <> 2;

EDIT:

You can do this as an explicit join like this:

select t.*
from table t join
     table t2
     on t.ancestor = t2.ancestor and
        t2.id = 2 a
where t.id <> 2 and
      t.ancestor <> t.descendant;

Note: I also added the condition t.id <> 2 so "2" is not considered a sibling of itself.

来源：https://stackoverflow.com/questions/22123187/what-query-would-i-use-to-obtain-sibling-records-when-using-closure-tables

标签

mysql

sql

transitive-closure-table