mysql层次结构自联接,检索所有子类别

时间:2015-08-12 12:33:14

标签: mysql sql recursion

我的表categories包含以下列:

category_id
category_name
parent_id

我需要获取给定主要类别的所有级别的所有子类别的列表,因此,例如,如果我给出一些lvl 3类别的id,我将返回所有lvl 4,5,6 ...类别的列表是那一类3类儿童。

不需要保留任何层次结构,只需要一个简单的列表。

我首先考虑过使用几个连接和子查询来做这件事但是我认为类别会更加深入,因此这不是一种方法。

由于我刚刚开始使用SQL,我仍然不知道如何编写递归查询,因此这将是一个很好的帮助和学习材料。

1 个答案:

答案 0 :(得分:0)

首先阅读底部的Please Note。好的,你回来了。

为类似递归的层次结构检索创建存储过程。

请注意,您不是按级别想要它,但可以很容易地完成。

架构:

create table category
(   category_id int not null auto_increment primary key,
    category_name varchar(40) not null,
    parent_id int null,  -- index on this column not a shabby idea
    unique key (category_name)
);

insert category(category_name,parent_id) values ('car',null),('food',null); -- 1,2
insert category(category_name,parent_id) values ('ford',1),('chevy',1),('fruit',2); -- 3,4,5
insert category(category_name,parent_id) values ('economy',3),('escort',6),('exhaust',7); -- 6,7,8
insert category(category_name,parent_id) values ('chassis',7),('loud',8),('banana',5); -- 9,10,11
-- ok granted I could have explicity inserted category_id to make it more obvious

创建存储过程:

-- drop procedure showHierarchyBelow;
delimiter $$
create procedure showHierarchyBelow
(
catname varchar(40)
)
BEGIN
    -- deleteMe parameter means i am anywhere in hierarchy of role
    -- and i want me and all my offspring deleted (no orphaning of children or theirs)
    declare bDoneYet boolean default false;
    declare working_on int;
    declare theCount int;
    declare findFirst int;

    select ifnull(category_id,0) into findFirst from category where category_name=catname;

    CREATE TABLE xx_RecursishHelper_xx
    (   -- it's recurshish, not recursive
        category_id int not null,
        processed int not null
    );
    if isnull(findFirst) then
        set findFirst=0;
    end if;
    insert into xx_RecursishHelper_xx (category_id,processed) select findFirst,0;
    if (findFirst=0) then
        set bDoneYet=true;
    else
        set bDoneYet=false;
    end if;

    while (!bDoneYet) do
        -- I am not proud of this next line, but oh well
        select count(*) into theCount from xx_RecursishHelper_xx where processed=0;

        if (theCount=0) then 
            -- found em all
            set bDoneYet=true;
        else
            -- one not processed yet, insert its children for processing
            SELECT category_id INTO working_on FROM xx_RecursishHelper_xx where processed=0 limit 1;
            insert into xx_RecursishHelper_xx (category_id,processed)
            select category_id,0 from category
            where parent_id=working_on;

            -- mark the one we "processed for children" as processed
            update xx_RecursishHelper_xx set processed=1 where category_id=working_on;
        end if;
    end while;

    delete from xx_RecursishHelper_xx where category_id=findFirst;

    select x.category_id,c.category_name
    from xx_RecursishHelper_xx x
    join category c
    on c.category_id=x.category_id;

    drop table xx_RecursishHelper_xx;
END
$$

测试存储过程:

call showHierarchyBelow('food');
+-------------+---------------+
| category_id | category_name |
+-------------+---------------+
|           5 | fruit         |
|          11 | banana        |
+-------------+---------------+

call showHierarchyBelow('car');
+-------------+---------------+
| category_id | category_name |
+-------------+---------------+
|           3 | ford          |
|           4 | chevy         |
|           6 | economy       |
|           7 | escort        |
|           8 | exhaust       |
|           9 | chassis       |
|          10 | loud          |
+-------------+---------------+

call showHierarchyBelow('ford');
+-------------+---------------+
| category_id | category_name |
+-------------+---------------+
|           6 | economy       |
|           7 | escort        |
|           8 | exhaust       |
|           9 | chassis       |
|          10 | loud          |
+-------------+---------------+

call showHierarchyBelow('xxx');
-- no rows

注意我几个月前根据您的需要修改了我的Answer

请注意

以上仅用于说明目的。在现实世界中,我永远不会在存储过程中创建表。 DDL开销很重要。相反,我会使用预先存在的非临时表和会话概念。并为会话完成清理行。所以不要把上面的东西当作一个稻草人,等着你让它变得更加高效。问这是否令人困惑。