Oracle start with connect by prior... 递归查询

create table create table prior_test ( parentid number(10), subid number(10) ); -- 字段类型最好用 number，而不是 varchar2，因为测试 SQL 需要比较 id -- insert insert into prior_test values ( 1, 2 ); insert into prior_test values ( 1, 3 ); insert into prior_test values ( 2, 4 ); insert into prior_test values ( 2, 5 ); insert into prior_test values ( 3, 6 ); insert into prior_test values ( 3, 7 ); insert into prior_test values ( 5, 8 ); insert into prior_test values ( 5, 9 ); insert into prior_test values ( 7, 10 ); insert into prior_test values ( 7, 11 ); insert into prior_test values ( 10, 12 ); insert into prior_test values ( 10, 13 );

 

-- select
select * from prior_test

 

-- SQL-1
select t.parentid, t.subid, level 
from prior_test t
start with t.subid = 7
connect by subid = prior parentid
order by level desc

 

-- SQL-2
select t.parentid, t.subid, level 
from prior_test t
start with t.subid = 7
connect by prior subid = parentid
order by level desc

SQL 解析：

start with 子句：遍历起始条件

connect by 子句：连接条件

　　关键词 prior，prior 跟父节点列 parentid 放在一起，prior parentid 就是往父节点方向遍历；prior跟子节点列 subid 放在一起，prior subid 则往子节点方向遍历。

　　parentid、subid 两列谁放在 ‘=‘ 前都无所谓，关键是 prior 后面的字段。（比较上面查询语句 SQL-1 与 SQL-2）

order by 子句：排序

 

观察下面 SQL-3 与 SQL-4 分别执行的结果集：

-- SQL-3
select t.parentid, t.subid, level 
from prior_test t
start with t.subid = 7
connect by prior subid = parentid
order by level desc

 

-- SQL-4
select t.parentid, t.subid, level 
from prior_test t
start with t.parentid = 7
connect by prior subid = parentid
order by level desc

结论：start with subid 与 start with parentid 结果集是不同的。

 

加入 where 子句

-- SQL-5
select t.parentid, t.subid, level from bb_test t
where t.parentid > 3
start with t.subid = 12
connect by subid = prior parentid
order by level desc

SQL 执行顺序是：先 start with 子句, 在是 connect by 子句, 最后是 where 子句！

where 只是树结构的修剪，不改变树的层次结构。

Oracle start with connect by prior... 递归查询

标签：分享   span   code   desc   技术分享   creat   with   varchar2   into