数据库树型结构中获取目录串的高性能方法，sys_connect_by_path与自定义函数性能对比

背景：

假设有如下文件夹表T_FOLDER，其DDL语句如下，根据文件夹主键ID获取此文件夹所在的目录串（格式为/id1/id2）？

create table t_folder

(

id number(15) primary key,

name varchar2(32),

parent_id number(15)

);

注：parent_id表示文件夹的父文件夹，parent_id为0表示根文件夹，根文件夹不在表T_FOLDER中存储

数据准备（500万）：

insert into t_folder(id, name, parent_id) values(1, 'name_1', 0);

insert into t_folder(id, name, parent_id) values(2, 'name_2', 0);

insert into t_folder(id, name, parent_id) values(3, 'name_3', 0);

insert into t_folder(id, name, parent_id) values(4, 'name_4', 2);

insert into t_folder(id, name, parent_id) values(5, 'name_5', 2);

insert into t_folder(id, name, parent_id) values(6, 'name_6', 3);

insert into t_folder(id, name, parent_id) values(7, 'name_7', 4);

commit;

declare

begin

for i in 8 .. 5000000 loop

insert into t_folder (id, name, parent_id) values (i, 'name_' || i, 4);

if i mod 8 = 0 then

commit;

end if;

end loop;

end;

/

方法一：（208.422(s)）

select sys_connect_by_path(id, '/')

from t_folder

where id=7

start with parent_id = 0

connect by prior id = parent_id;

方法二：（0.047(s)）

create or replace function f_folder_dir(v_id number) return varchar2 is

v_parent_id number;

v_folder_dir varchar2(4000) := '/' || v_id;

v_tmp_id number := v_id;

begin

loop

select parent_id into v_parent_id

from t_folder

where id = v_tmp_id;

exit when v_parent_id = 0;

v_tmp_id := v_parent_id;

v_folder_dir := '/' || v_parent_id || v_folder_dir;

end loop;

return v_folder_dir;

end;

/

-- 采用f_folder_dir自定义函数, 产生目录串，耗时: 0.047(s)

select f_folder_dir(7) from dual;