无法在单个查询中执行此操作。即使有，它也可能效率很低。

我们可以通过存储过程和循环来实现。使用添加的索引，它也应该很快。这使用两个表从输入表（A）中选择节点，并将该节点及其子级插入（B）。然后，它将B交换为A，并重复执行直到直到A中不再存在非叶节点为止。可喜的是，循环迭代的数量将与输入节点和最后一个叶节点之间的级别一样多，在大多数情况下，循环迭代次数为可能没有那么深。此存储过程比在代码中进行外部存储过程要快。

仅供参考，我在安装临时表时遇到了困难，如果遇到“错误2”，请删除临时关键字。

delimiter $$
drop procedure if exists GetLeafNodes $$
create procedure GetLeafNodes(nodeid int)
begin
declare N int default 1;

-- create two working sets of IDs, we'll go back and forth between these two sets
drop temporary table if exists A;
drop temporary table if exists B;
create temporary table A(node int, child int);
create temporary table B(node int, child int);

-- insert our single input node into the working set
insert into A values (null, nodeid);

while (N>0) do
  -- keep selecting child nodes for each node we are now tracking
  -- leaf nodes will end up with the child set to null
  insert into B
  select ifnull(A.child,A.node), tree.ID
    from A
    left outer join DATA_TREE as tree on A.child=tree.parent_id;

  -- now swap A and B
  rename table A to temp, B to A, temp to B;

  -- remove non-leaf nodes from table B
  delete from B;

  -- exit when there are no longer any non-leaf nodes in A
  set N=(select count(*) from A where child is not null);
end while;

-- now output our list of leaf nodes
select node from A;

drop temporary table A;
drop temporary table B;
end $$
DELIMITER ;
call GetLeafNodes(4);

我使用以下样本集进行测试：

CREATE TABLE `DATA_TREE` (
  `ID` int(11) NOT NULL,
  `PARENT_ID` int(11) NOT NULL,
  PRIMARY KEY (`ID`),
  UNIQUE KEY `ID_UNIQUE` (`ID`),
  KEY `fk_DATA_TREE_1_idx` (`PARENT_ID`)
) ENGINE=InnoDB
;

insert into DATA_TREE values
(1,0),(2,1),(3,1),(4,1),(5,3),(6,3),(7,4),(8,4),(9,4),(10,6),(11,6),(12,7),(13,9),(14,9),(15,12),(16,12),(17,12),(18,14);