【原创】PostgreSQL 对简单树的遍历

昨天我用MySQL来实现了ORACLE的递归语句CONNECT BY， 看起来稍复杂些。今天来看看POSTGRESQL如何实现ORACLE的CONNECT BY。
还是用昨天同样的表以及数据。POSTGRESQL自诩最像ORACLE的数据库，所以大部分语句也就都可以简单而且变相的实现了。
在这点上可以用他自己带的WITH递归功能，还可以用第三方扩展带来的类似connect by 函数。


先来看第一点，用递归的WITH来展现这棵树的路径。

点击(此处)折叠或打开

1. t_girl=# with recursive tmp_country(id,path) as
   
2. t_girl-# (
   
3. t_girl(# select a.id,'/'||b.name as "path" from country_relation as a inner join country as b on (a.id = b.id) where a.parentid is null
   
4. t_girl(# union all
   
5. t_girl(# select a.id,q.path||'/'||b.name as "path" from country_relation as a inner join tmp_country as q on (q.id = a.parentid)
   
6. t_girl(# inner join country as b on (a.id = b.id)
   
7. t_girl(# )
   
8. t_girl-# select a.path from tmp_country as a;
   
9.                      path
   
10. -----------------------------------------------
    
11.  /Earth
    
12.  /Earth/North America
    
13.  /Earth/South America
    
14.  /Earth/Europe
    
15.  /Earth/Asia
    
16.  /Earth/Africa
    
17.  /Earth/Australia
    
18.  /Earth/North America/Canada
    
19.  /Earth/North America/Central America
    
20.  /Earth/North America/Island Nations
    
21.  /Earth/North America/United States
    
22.  /Earth/North America/United States/Alabama
    
23.  /Earth/North America/United States/Alaska
    
24.  /Earth/North America/United States/Arizona
    
25.  /Earth/North America/United States/Arkansas
    
26.  /Earth/North America/United States/California
    
27. (16 rows)
    
28. 
    
29. 
    
30. Time: 3.260 ms

还可以用tablefunc扩展带来的CONNECT BY函数把这棵树遍历出来。
由于昨天设计的两张表通过ID来关联，这个扩展自带的函数要把名字展现出来比较麻烦，索性这里我就用了一张临时表保存我想要的结果。

点击(此处)折叠或打开

1. t_girl=# CREATE TEMPORARY TABLE tmp_country_relation as SELECT b.id,a.name,b.parentid,''::text as parentname FROM country AS a,country_relation AS b WHERE a.id = b.id;
   
2. SELECT 16
   
3. Time: 11.773 ms
   
4. t_girl=#

这里更新了对应的ID为NAME。

点击(此处)折叠或打开

1. t_girl=# update tmp_country_relation set parentname = a.name from country as a where parentid = a.id;
   
2. UPDATE 15
   
3. Time: 1.829 ms

我用TABLEFUNC扩展带来的CONNECT BY 实现这棵树的遍历。

点击(此处)折叠或打开

1. t_girl=# select path from connectby('tmp_country_relation as a','a.name','a.parentname','Earth',0,'/') as g(id text,parentid text,level int,path text) order by level;
   
2.                      path
   
3. ----------------------------------------------
   
4.  Earth
   
5.  Earth/Australia
   
6.  Earth/North America
   
7.  Earth/Africa
   
8.  Earth/South America
   
9.  Earth/Europe
   
10.  Earth/Asia
    
11.  Earth/North America/Island Nations
    
12.  Earth/North America/Canada
    
13.  Earth/North America/Central America
    
14.  Earth/North America/United States
    
15.  Earth/North America/United States/California
    
16.  Earth/North America/United States/Arkansas
    
17.  Earth/North America/United States/Alabama
    
18.  Earth/North America/United States/Alaska
    
19.  Earth/North America/United States/Arizona
    
20. (16 rows)
    
21. 
    
22. 
    
23. Time: 5.974 ms
    
24. t_girl=#