MySQL 学习笔记 4：视图

图源：ubiq.co

简单的说，视图就是“固化的SQL查询”。

这里看一个简单示例，我们有一个表，保存学生信息：

mysql> select * from student limit 10;
+-----+---------------+-----------+----------+
| id  | average_score | level     | name     |
+-----+---------------+-----------+----------+
| 573 |            27 | FRESH_MAN | Tom      |
| 574 |            76 | JUNIOR    | Icexmoon |
| 575 |             2 | JUNIOR    | BrusLee  |
| 576 |            56 | SOPHOMORE | Harry    |
| 577 |            44 | JUNIOR    | JackChen |
| 578 |            96 | JUNIOR    | Jimmy    |
| 579 |            73 | JUNIOR    | LiLei    |
| 580 |             4 | SENIOR    | XiaoMing |
| 581 |            21 | SOPHOMORE | Adam     |
| 582 |            40 | JUNIOR    | Alex     |
+-----+---------------+-----------+----------+

level字段表示年级，如果我们需要查询大一的学生：

mysql> select * from student where level='FRESH_MAN';
+-----+---------------+-----------+--------+
| id  | average_score | level     | name   |
+-----+---------------+-----------+--------+
| 573 |            27 | FRESH_MAN | Tom    |
| 586 |            61 | FRESH_MAN | Blake  |
| 591 |             6 | FRESH_MAN | Bruce  |
| 599 |            90 | FRESH_MAN | Xiaoli |
+-----+---------------+-----------+--------+

假设我们还需要统计大一学生的总平均分：

mysql> select sum(average_score)/count(id) as sum_average_score, sum(average_score) as total_average_score from student where level='FRESH_MAN';
+-------------------+---------------------+
| sum_average_score | total_average_score |
+-------------------+---------------------+
|           46.0000 |                 184 |
+-------------------+---------------------+

如果要频繁编写类似的查询语句，查询某某年级的学生的总平均分就很麻烦。

我们可以换个思路，能否创建一个“固化的查询结果”，内容是每个年级的统计结果，这样我们的查询语句就很简单了，只要针对这个统计结果按照年级查询即可。

首先我们看怎么分组统计不同年级的均分：

mysql> select `level`,count(id) as num,sum(average_score) as total_score, sum(average_score)/count(id) as sum_average_score
    -> from student
    -> group by `level`;
+-----------+-----+-------------+-------------------+
| level     | num | total_score | sum_average_score |
+-----------+-----+-------------+-------------------+
| FRESH_MAN |   4 |         184 |           46.0000 |
| JUNIOR    |  11 |         526 |           47.8182 |
| SOPHOMORE |   6 |         328 |           54.6667 |
| SENIOR    |   6 |         190 |           31.6667 |
+-----------+-----+-------------+-------------------+

创建视图

用 SQLyog 创建视图会出现类似下面的模版 SQL：

CREATE
    /*[ALGORITHM = {UNDEFINED | MERGE | TEMPTABLE}]
    [DEFINER = { user | CURRENT_USER }]
    [SQL SECURITY { DEFINER | INVOKER }]*/
    VIEW `jpa`.`student_stat_view` 
    AS
(SELECT * FROM ...);

student_stat_view是视图名称，视图名称必须在库中唯一（包括表名）。

我们只需要在AS之后添加SELECT语句即可完成视图创建语句：

CREATE
    VIEW `jpa`.`student_stat_view` 
    AS
SELECT `level`,COUNT(id) AS num,SUM(average_score) AS total_score, SUM(average_score)/COUNT(id) AS sum_average_score
FROM student
GROUP BY `level`;

查询视图

视图创建好后可以在 SQLyog 中看到，也可以使用以下命令查询：

show table status where comment='view';

要查看视图创建时的 SQL 语句：

show create view student_stat_view;

我们可以像查询普通表那样查询视图：

mysql> select * from student_stat_view;
+-----------+-----+-------------+-------------------+
| level     | num | total_score | sum_average_score |
+-----------+-----+-------------+-------------------+
| FRESH_MAN |   4 |         184 |           46.0000 |
| JUNIOR    |  11 |         526 |           47.8182 |
| SOPHOMORE |   6 |         328 |           54.6667 |
| SENIOR    |   6 |         190 |           31.6667 |
+-----------+-----+-------------+-------------------+

这样就很容易查询出我们需要年级的均分情况：

mysql> select * from student_stat_view where level='FRESH_MAN';
+-----------+-----+-------------+-------------------+
| level     | num | total_score | sum_average_score |
+-----------+-----+-------------+-------------------+
| FRESH_MAN |   4 |         184 |           46.0000 |
+-----------+-----+-------------+-------------------+

如果不使用视图，我们可能需要这么查询：

select `level`,count(id) as num,sum(average_score) as total_score, sum(average_score)/count(id) as sum_average_score
from student
group by `level` having `level`='FRESH_MAN'

视图的优缺点

可以从上面这个示例中看出，视图具有以下优点：

• 简化 SQL 查询。
• 让查询更直观。

视图也有缺点，视图本身并不保存数据，视图的数据都来自查询关联的基本表，视图本身只是一条查询 SQL。所以每次查询视图意味着执行了一条视图相关的 SQL，如果创建大量视图，且包含视图关联视图、视图嵌套视图、视图关联表等复杂查询语句，就可能造成性能问题。

简化多表联查

上面的示例演示了如何用视图简化聚合数据查询，视图还常用于简化多表联查。

假设我们有以下表：

mysql> select * from school;
+----+------------------+
| id | name             |
+----+------------------+
|  2 | 布斯巴顿魔法学校 |
|  1 | 霍格沃茨魔法学校 |
+----+------------------+

mysql> select * from student;
+----+----------+-----------+
| id | name     | school_id |
+----+----------+-----------+
|  1 | Harry    |         1 |
|  2 | icexmoon |         1 |
|  3 | JackChen |         2 |
+----+----------+-----------+

mysql> select * from email;
+----+----------+-----------+------------+
| id | account  | domain    | student_id |
+----+----------+-----------+------------+
|  1 | harry    | qq.com    |          1 |
|  2 | harry    | gmail.com |          1 |
|  3 | icexmoon | qq.com    |          2 |
|  4 | 111      | tom.com   |          2 |
|  5 | 123      | gmail.com |          3 |
|  6 | jack     | qq.com    |          3 |
+----+----------+-----------+------------+

假设我们需要查询使用了某种邮箱的学校，可能需要这么查询：

mysql> SELECT DISTINCT sc.id,sc.name FROM
    -> school AS sc
    -> LEFT JOIN student AS s
    -> ON s.school_id=sc.id
    -> LEFT JOIN email AS e
    -> ON e.student_id=s.id
    -> WHERE e.domain='gmail.com';
+----+------------------+
| id | name             |
+----+------------------+
|  1 | 霍格沃茨魔法学校 |
|  2 | 布斯巴顿魔法学校 |
+----+------------------+

不要问为什么会有这么古怪的要求…

可以用视图简化这个过程，我们可以创建一个表示学校和电子邮件的视图：

CREATE
    VIEW `jpa`.`school_email_view` 
    AS
SELECT sc.id AS school_id,sc.name AS school_name,e.id AS email_id, e.account AS email_account, e.domain AS email_domain FROM
school AS sc
LEFT JOIN student AS s
ON s.school_id=sc.id
LEFT JOIN email AS e
ON e.student_id=s.id;

这个视图直接体现了学校和学校学生拥有的电邮地址之间的关系：

mysql> select * from school_email_view;
+-----------+------------------+----------+---------------+--------------+
| school_id | school_name      | email_id | email_account | email_domain |
+-----------+------------------+----------+---------------+--------------+
|         2 | 布斯巴顿魔法学校 |        6 | jack          | qq.com       |
|         2 | 布斯巴顿魔法学校 |        5 | 123           | gmail.com    |
|         1 | 霍格沃茨魔法学校 |        2 | harry         | gmail.com    |
|         1 | 霍格沃茨魔法学校 |        1 | harry         | qq.com       |
|         1 | 霍格沃茨魔法学校 |        4 | 111           | tom.com      |
|         1 | 霍格沃茨魔法学校 |        3 | icexmoon      | qq.com       |
+-----------+------------------+----------+---------------+--------------+

利用视图可以很容易实现之前的查询：

mysql> select distinct school_id,school_name from school_email_view where email_domain='gmail.com';
+-----------+------------------+
| school_id | school_name      |
+-----------+------------------+
|         1 | 霍格沃茨魔法学校 |
|         2 | 布斯巴顿魔法学校 |
+-----------+------------------+

这样做还有个好处，我们可以无需理解三张表的联结关系，只需要按照需求查询视图即可。

简化字段计算

视图的另一个常见用途是对字段进行重新计算和处理，比如在上边的email表中，电子邮件帐号和网站是单独存放的字段，如果我们想要获取类似123@gmail.com这样的电邮地址，就需要在查询语句中字符串拼接，或者在代码端处理。

完全可以用视图简化这个过程：

CREATE
    VIEW `jpa`.`email_address_view` 
    AS
select id as email_id,concat(account,'@',domain) as email_address
from email
order by id;

查询视图：

mysql> select * from email_address_view;
+----------+-----------------+
| email_id | email_address   |
+----------+-----------------+
|        1 | harry@qq.com    |
|        2 | harry@gmail.com |
|        3 | icexmoon@qq.com |
|        4 | 111@tom.com     |
|        5 | 123@gmail.com   |
|        6 | jack@qq.com     |
+----------+-----------------+

相比维护一个冗余字段存储完整电邮地址，这样做的好处在于如果有帐号或者domain修改，email_address也会改变，无需我们重新维护。但这样是有代价的，是牺牲性能带来的。所以这是个取舍的问题，如果电子邮件改变的不频繁，并且需要我们进行性能优化，我们就可以考虑使用冗余字段存储，并且为添加和更新语句创建触发器，来解决维护冗余字段的问题。

视图和表联查

视图可以和表或其它视图联结查询：

mysql> select e.*,ev.`email_address`
    -> from email as e
    -> left join email_address_view as ev
    -> on e.id=ev.`email_id`;
+----+----------+-----------+------------+-----------------+
| id | account  | domain    | student_id | email_address   |
+----+----------+-----------+------------+-----------------+
|  1 | harry    | qq.com    |          1 | harry@qq.com    |
|  2 | harry    | gmail.com |          1 | harry@gmail.com |
|  3 | icexmoon | qq.com    |          2 | icexmoon@qq.com |
|  4 | 111      | tom.com   |          2 | 111@tom.com     |
|  5 | 123      | gmail.com |          3 | 123@gmail.com   |
|  6 | jack     | qq.com    |          3 | jack@qq.com     |
+----+----------+-----------+------------+-----------------+

视图的限制

视图也存在一些限制，比如视图不能使用索引。

这很容易理解，视图本身并不存储内容，只是一条查询语句。而索引是创建在列上的，并且会创建一个B+树结构的存储数据。

修改 & 删除

用 SQLyog 修改视图会出现类似下面的模版：

DELIMITER $$

ALTER ALGORITHM=UNDEFINED DEFINER=`root`@`localhost` SQL SECURITY DEFINER VIEW `school_email_view` AS 
select
  `sc`.`id`     AS `school_id`,
  `sc`.`name`   AS `school_name`,
  `e`.`id`      AS `email_id`,
  `e`.`account` AS `email_account`,
  `e`.`domain`  AS `email_domain`
from ((`school` `sc`
    left join `student` `s`
      on ((`s`.`school_id` = `sc`.`id`)))
   left join `email` `e`
     on ((`e`.`student_id` = `s`.`id`)))$$

DELIMITER ;

所以可以使用ALTER ... VIEW ... AS语句修改视图。

此外也可以先删除视图：

drop view school_email_view;

再重新创建。

The End，谢谢阅读。

参考资料

• 《MySQL 必知必会》