MySQL8.0新特性(四)通用表表达式(CTE)

一、概述

MySQL 8支持公用表表达式，包括非递归和递归两种。

公用表表达式允许使用命名的临时结果集，这是通过允许在SELECT语句和某些其他语句前面使用WITH子句来实现的。

不能在同一查询中两次引用派生表（子查询），因为那样的话，查询会根据派生表的引用次数计算两次或多次，这会引发严重的性能问题。使用CTE后，子查询只会计算一次。

二、非递归CTE

2.1 派生表（子查询）

select … from (subquery) as derived, table_name …;

2.2 CTE

with derived as (
    subquery
)
select … from derived, table_name …;

CTE可能在SELECT/UPDATE/DELETE之前，包括with derived as (subquery)的子查询，例如：

with derived as (
    subquery
)
delete from table_name 
where table_name.col_name in (
    select col_name from derived
);

2.3 CTE可以引用其他CTE

with derived_one as (
    subquery
),
derived_two as (
    select … from derived_one
)
select … from derived_one, derived_two …;

三、递归CTE

3.1 语法解释

递归CTE是一种特殊的CTE，其子查询会引用自己的名字。WITH子句必须以WITH RECURSIVE开头。递归CTE子查询包括两部分：seed查询和recursive查询，由UNION[ALL]或UNION DISTINCT分隔。

• seed select被执行一次以创建初始数据子集；
• recursive select被重复执行以返回数据的子集，直到获得完整的结果集。

当迭代不会生成任何新行时，递归会停止。这对挖掘层次结构（父/子或部分/子部分）非常有用。

with recursive {temp_table_name}(n) as (
    select … from table_name  /* "seed select" */
    union all 
    select … from {temp_table_name}, table_name  /* "recursive select" */
)
select … from derived …;

假设要打印从1到5的所有数字：

mysql> with recursive derived(n) as (
    -> select 1  /* "seed select" */
    -> union all 
    -> select n + 1 from derived where n < 5  /* "recursive select" */
    -> )
    -> select * from derived;
+------+
| n    |
+------+
|    1 |
|    2 |
|    3 |
|    4 |
|    5 |
+------+
5 rows in set (0.00 sec)

在每次迭代中，SELECT都会生成一个带有新值的行，比前一行的值n多1。第一次迭代在初始行集合（1）上运行并生成值为1 + 1 = 2的行；第二次迭代对第一次迭代的行集合（2）进行操作并生成值为2 + 1 = 3的行。以此类推，一直持续到n不再小于5时，递归结束。

3.2 案例

假设要执行分层数据遍历，以便为每个员工生成一个组织结构图（即从CEO到每个员工的路径），也可以使用递归CTE。

3.2.1 创建带有manager_id的测试表

mysql> create table employees.employees_mgr ( 
    -> id int primary key not null,
    -> name varchar(100) not null,
    -> manager_id int null,
    -> index (manager_id),
    -> foreign key (manager_id) references employees_mgr (id)
    -> );
Query OK, 0 rows affected (0.03 sec)

3.2.2 插入示例数据

mysql> insert into employees.employees_mgr values 
    -> (333, "Yasmina", null), /* Yasmina is the CEO (manager_id is null) */
    -> (198, "John", 333), /* John has id 198 and reports to 333 (Yasmina) */
    -> (692, "Tarek", 333),
    -> (29, "Pedro ", 198),
    -> (4610, "Sarah", 29),
    -> (72, "Pierre", 29),
    -> (123, "Adil", 692);
Query OK, 7 rows affected (0.01 sec)
Records: 7  Duplicates: 0  Warnings: 0

mysql> select id, name, manager_id from employees.employees_mgr;
+------+---------+------------+
| id   | name    | manager_id |
+------+---------+------------+
|   29 | Pedro   |        198 |
|   72 | Pierre  |         29 |
|  123 | Adil    |        692 |
|  198 | John    |        333 |
|  333 | Yasmina |       NULL |
|  692 | Tarek   |        333 |
| 4610 | Sarah   |         29 |
+------+---------+------------+
7 rows in set (0.00 sec)

3.2.3 执行递归CTE

mysql> with recursive employee_paths (id, name, path) as (
    -> select id, name, cast(id as char(200))
    -> from employees.employees_mgr
    -> where manager_id is null
    -> union all
    -> select e.id, e.name, concat(ep.path, '->', e.id)
    -> from employee_paths as ep
    -> join employees.employees_mgr as e
    -> on ep.id = e.manager_id
    -> )
    -> select * from employee_paths;
+------+---------+--------------------+
| id   | name    | path               |
+------+---------+--------------------+
|  333 | Yasmina | 333                |
|  198 | John    | 333->198           |
|  692 | Tarek   | 333->692           |
|   29 | Pedro   | 333->198->29       |
|  123 | Adil    | 333->692->123      |
|   72 | Pierre  | 333->198->29->72   |
| 4610 | Sarah   | 333->198->29->4610 |
+------+---------+--------------------+
7 rows in set (0.00 sec)

with recursive employee_paths (id, name, path) as是CTE的名称，列是(id, name, path)。

select id, name, cast(id as char(200)) from employees.employees_mgr where manager_id is null是查询CEO的seed查询（没有在CEO之上的管理者）。

select e.id, e.name, concat(ep.path, '->', e.id) from employee_paths as ep join employees.employees_mgr as e on ep.id = e.manager_id是递归查询。

递归查询生成的每一行，会查找直接向前一行生成的员工做汇报的所有员工。对于每个员工，该行的信息包括员工ID、姓名和员工管理链，该链是在最后添加了员工ID的管理链。