Oracle 递归查询详解

查看原文：http://www.ibloger.net/article/256.html

创建表和主外键

CREATE TABLE SC_DISTRICT
(
  ID         NUMBER(10)                  NOT NULL,
  PARENT_ID  NUMBER(10),
  NAME       VARCHAR2(255 BYTE)          NOT NULL
);

ALTER TABLE SC_DISTRICT ADD (
  CONSTRAINT SC_DISTRICT_PK
 PRIMARY KEY
 (ID));

ALTER TABLE SC_DISTRICT ADD (
  CONSTRAINT SC_DISTRICT_R01 
 FOREIGN KEY (PARENT_ID) 
 REFERENCES SC_DISTRICT (ID)
);

插入数据

insert into A_DISTRICT (id, parent_id, name) values (1, null, '河南省');
insert into A_DISTRICT (id, parent_id, name) values (2, null, '山东省');
insert into A_DISTRICT (id, parent_id, name) values (3, 1, '郑州市');
insert into A_DISTRICT (id, parent_id, name) values (4, 1, '开封市');
insert into A_DISTRICT (id, parent_id, name) values (5, 1, '商丘市');
insert into A_DISTRICT (id, parent_id, name) values (6, 3, '中原区');
insert into A_DISTRICT (id, parent_id, name) values (7, 3, '金水区');
insert into A_DISTRICT (id, parent_id, name) values (8, 3, '二七区');
insert into A_DISTRICT (id, parent_id, name) values (9, 3, '城管回族区');
insert into A_DISTRICT (id, parent_id, name) values (10, 3, '上街区');
insert into A_DISTRICT (id, parent_id, name) values (11, 3, '惠济区');
insert into A_DISTRICT (id, parent_id, name) values (12, 4, '开封县');
insert into A_DISTRICT (id, parent_id, name) values (13, 4, '杞县');
insert into A_DISTRICT (id, parent_id, name) values (14, 4, '兰考县');
insert into A_DISTRICT (id, parent_id, name) values (15, 4, '通许县');
insert into A_DISTRICT (id, parent_id, name) values (16, 13, '城关镇');
insert into A_DISTRICT (id, parent_id, name) values (17, 13, '五里河镇');
insert into A_DISTRICT (id, parent_id, name) values (18, 13, '邢口镇');
insert into A_DISTRICT (id, parent_id, name) values (19, 13, '柿园乡');
insert into A_DISTRICT (id, parent_id, name) values (20, 13, '城郊乡');

生成数据如下

递归语法和主要使用函数

      select * from 表 start with 条件入口 connect by prior id =  parent_id（向下）

      select * from 表 start with 条件入口 connect by prior parent_id = id（向上）

如果不写prior，表明前序遍历 未指明遍历方向，所以不进行递归，习惯上会在语句后面加上order by排序和group by分组

      connect_by_root(列名)  查询根节点

      connect_by_isleaf   查询是否叶子节点，0是，1不是

      sys_connect_by_path  查询递归路径

实战

查询根节点

select id, parent_id, name from a_district start with parent_id is null connect by id = parent_id order by id

递归查询 杞县 的所有上级节点（包含当前节点）

select id, parent_id, name from a_district start with name='杞县' connect by prior parent_id = id order by id

递归查询 杞县 的所有上级节点（包含当前节点和根节点）

select id, parent_id, name,connect_by_root(id) city_id, connect_by_root(name) city_name from a_district 
       start with name='杞县' connect by prior parent_id = id order by id

递归查询 杞县 的所有子城镇乡（包含当前节点）

select id, parent_id, name from a_district start with name='杞县' connect by prior id = parent_id order by id

递归查询 开封市 的所有城镇乡（包含当前节点和根节点）

select id, parent_id, name, connect_by_root(id) city_id, connect_by_root(name) city_name from a_district 
       start with  name='开封市' connect by prior id = parent_id order by id

上行递归查询 城郊乡 深度、是否包含叶子节点

select id, name, parent_id, level, connect_by_isleaf isleaf from a_district start with name = '城郊乡' connect by prior parent_id = id order by id

下行递归查询 河南省 深度、是否包含叶子节点

select id, name, parent_id, level, connect_by_isleaf isleaf from a_district start with name = '河南省' connect by prior id = parent_id order by id

上行递归查询 杞县 路径

select id, name, parent_id, substr(sys_connect_by_path(name,'->'),3) name_path from a_district 
       start with name='杞县' connect by prior parent_id = id order by id

下行递归查询 河南省 路径

select id, name, parent_id, substr(sys_connect_by_path(name,'->'),3) name_path from a_district 
       start with name='河南省' connect by prior id = parent_id order by id

综合使用

select id, name, parent_id, level, connect_by_isleaf isleaf, substr(sys_connect_by_path(name,'->'),3) name_path from a_district
        start with name='河南省' connect by prior id = parent_id order by id

SYS_CONNECT_BY_PATH函数

        在Oracle中，SYS_CONNECT_BY_PATH函数主要作用是可以把一个父节点下的所有子节点通过某个字符进行区分，然后连接在一个列中显示。

        sys_connect_by_path(字段名, 2个字段之间的连接符号)，注意这里的连接符号不要使用逗号，oracle会报错，如果一定要用，可以使用replace替换一下，REPLACE(字段名,原字符,',')。

        还有，这个函数使用之前必须先建立一个树，否则无用。