MySQL stores JSON format data

Mysql5.7 New features after version ,Mysql Provides a native Json type ,Json Values will no longer be stored as strings , Instead, it uses a method that allows fast reading of text elements （document elements） Internal binary （internal binary） Format . stay Json When the column is inserted or updated, it will be automatically verified Json Text , Text that fails validation will produce an error message

JSON The data type is recommended for static data storage that is not updated frequently

Create table t_user

CREATE TABLE `t_user_tag` (
  `id` int NOT NULL AUTO_INCREMENT,
  `tag_name` varchar(255) COLLATE utf8mb4_general_ci DEFAULT NULL COMMENT ' Tag name ',
  PRIMARY KEY (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_general_ci;

The login method field uses json Format , It is divided into phone,wechat,qq,email,zhifubao wait

insert data ：

insert into t_user values (1,'tom', 25, '{"email": "[email protected]", "phone": "13200001111", "wechat": "147258369"}');
insert into t_user values (2,'jack', 30, '{"phone": "13500001111"}');
insert into t_user values (3,'lily', 18, '{"qq": "147258369", "phone": "13600001111"}');
insert into t_user values (4,'lily', 45, '{"wechat":"1884875663"}');

Inquire about

user name , cell-phone number , WeChat ID

select name,
(JSON_EXTRACT(login_info, '$.phone')) phone,
JSON_UNQUOTE(JSON_EXTRACT(login_info, '$.wechat')) wechat
from t_user;

It can be seen that

JSON_UNQUOTE The function is to Remove json Quotes for Strings , Convert the value to string type

JSON_EXTRACT The function is to extract json value

The concise writing is equivalent to the above

select name,
login_info ->> '$.phone' phone,
login_info ->> '$.wechat' wechat
from t_user;

->> expression Equate to JSON_UNQUOTE(JSON_EXTRACT(login_info, ‘$.wechat’))

--  Use json The fields in are used as query criteria 
select name,
login_info ->> '$.phone' phone,
login_info ->> '$.wechat' wechat
from t_user
where login_info ->> '$.phone' = '13200001111';

json data Add index

to login_info Add an index to the mobile number in the field

--  to login_info This json Medium phone Add index 
alter table t_user add COLUMN phone varchar(11) as (login_info ->> '$.phone');
alter table t_user add UNIQUE INDEX idx_uq_phone(phone);

Above SQL First, create a virtual column phone, This column is made up of functions login_info->>“$.phone” Calculated . Then create a unique index on this virtual column idx_uq_phone. At this time, through the virtual column phone The query , You can see that the optimizer will use the newly created idx_uq_phone Indexes

--  Look at the index 
EXPLAIN
select *
from t_user
where phone = '13200001111';

Let's look at the table structure , I found that the index was increased

Use scenarios

Some businesses need user portraits （ That is to label users ）, Then according to the user's label , Through data mining technology , Make corresponding product recommendations . such as ：

• In the e-commerce industry , According to the user's dressing preferences , Recommend corresponding products ;

• In the music industry , According to the music style users like and the singers they often listen to , Recommend the corresponding song ;

• In the financial industry , According to the user's risk preference and investment experience , Recommend corresponding financial products .

Here , I strongly recommend that you use JSON Type stores user portrait information in the database , And combine JSON The characteristics of array type and multivalued index are used for efficient query .

Create a user portrait definition table ：

CREATE TABLE `t_tag` (
  `id` int NOT NULL AUTO_INCREMENT,
  `tag_name` varchar(255) COLLATE utf8mb4_general_ci DEFAULT NULL COMMENT ' Tag name ',
  PRIMARY KEY (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_general_ci;

insert into t_tag values (null, '70 after ');
insert into t_tag values (null, '80 after ');
insert into t_tag values (null, '90 after ');
insert into t_tag values (null, '00 after ');
insert into t_tag values (null, '10 after ');
insert into t_tag values (null, ' Love sports ');
insert into t_tag values (null, ' Love listening to music ');
insert into t_tag values (null, ' I love watching movies ');
insert into t_tag values (null, ' Highly educated ');
insert into t_tag values (null, ' Small endowment ');
insert into t_tag values (null, ' A car ');
insert into t_tag values (null, ' There are children ');
insert into t_tag values (null, ' I like online shopping ');
insert into t_tag values (null, ' Like some takeout ');
insert into t_tag values (null, ' lolita ');

Create user label middle table

CREATE TABLE `t_user_tag` (
  `user_id` int NOT NULL COMMENT ' user id',
  `tag_id` json NOT NULL COMMENT ' User label id',
  PRIMARY KEY (`user_id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_general_ci;

insert data , Use an array to store

insert into t_user_tag values (1,'[2,4,6]');
insert into t_user_tag values (2,'[1,3,7]');
insert into t_user_tag values (3,'[8,10,12]');

MySQL 8.0.17 Version starting support Multi-Valued Indexes, Used in JSON Create index on array , And through the function member of、json_contains、json_overlaps To quickly retrieve index data . So you can watch UserTag To create a Multi-Valued Indexes：

ALTER TABLE t_user_tag
ADD INDEX idx_user_tags ((cast((tag_id->"$") as unsigned array)));

Inquire about those who love watching movies

select * from t_user_tag
where 8 MEMBER OF(tag_id -> '$');

I love watching movies , With children

select * from t_user_tag
where JSON_CONTAINS(tag_id -> '$', '[8,10]');

• Use JSON data type , Recommend to use MySQL 8.0.17 Version above , Better performance , It also supports Multi-Valued Indexes;

• JSON The advantage of data types is that there is no need to define columns in advance , The data itself is well descriptive ;

• Don't use data with obvious relationship JSON Storage , Such as user balance 、 User name 、 User ID card, etc , These are the data that every user must include ;

• JSON The data type is recommended for static data storage that is not updated frequently .