MySQL field userid comma separated save by userid query

my sql Average level , Just a note . Unable to save is the optimal solution . For reference only .

scene

There is a message table , There is a recipient field in which multiple users are listed as , Separate and save information . I need to be based on userid To search for information . For ease of understanding , Let me reduce the table structure . My table structure is as follows ：

message_id	receiver	sender	message	create_time
1	user1,user2	system	hello	2022-06-30 16:30:12
2	user1,user3	system	world	2022-06-30 16:31:19
3	user3,user4,user5	user1	java	2022-06-30 16:31:22
4	user7,user8	user2	neovim	2022-06-30 16:31:53
5	itkey,lxyoucan	system	itkey	2022-06-30 16:32:17
6	user,user11	good	very	2022-06-30 16:38:18

Table structure

create table t_message
(
    message_id     bigint auto_increment comment ' news ID'
        primary key,
    receiver       varchar(4000)    null comment ' The recipient ',
    sender         varchar(200)     not null comment ' Sender ',
    message        varchar(4000)    not null comment ' The message content ',
    create_time    datetime         null comment ' Creation time '
)
    comment ' Message center ';

data

INSERT INTO t_message (message_id, receiver, sender, message, create_time) VALUES (1, 'user1,user2', 'system', 'hello', '2022-06-30 16:30:12');
INSERT INTO t_message (message_id, receiver, sender, message, create_time) VALUES (2, 'user1,user3', 'system', 'world', '2022-06-30 16:31:19');
INSERT INTO t_message (message_id, receiver, sender, message, create_time) VALUES (3, 'user3,user4,user5', 'user1', 'java', '2022-06-30 16:31:22');
INSERT INTO t_message (message_id, receiver, sender, message, create_time) VALUES (4, 'user7,user8', 'user2', 'neovim', '2022-06-30 16:31:53');
INSERT INTO t_message (message_id, receiver, sender, message, create_time) VALUES (5, 'itkey,lxyoucan', 'system', 'itkey', '2022-06-30 16:32:17');
INSERT INTO t_message (message_id, receiver, sender, message, create_time) VALUES (6, 'user,user11', 'good', 'very', '2022-06-30 16:38:18');

practice

[ False demonstration ]like Fuzzy matching

First of all, my first thought is to use like Statement for fuzzy query . It seems simple and easy to use , But there will be a bug.

Such as query user3：

select * from t_message where receiver like '%user3%';

message_id	receiver	sender	message	create_time
2	user1,user3	system	world	2022-06-30 16:31:19
3	user3,user4,user5	user1	java	2022-06-30 16:31:22

No problem , But if I want to inquire user User data ？

select * from t_message where receiver like '%user%';

message_id	receiver	sender	message	create_time
1	user1,user2	system	hello	2022-06-30 16:30:12
2	user1,user3	system	world	2022-06-30 16:31:19
3	user3,user4,user5	user1	java	2022-06-30 16:31:22
4	user7,user8	user2	neovim	2022-06-30 16:31:53
6	user,user11	good	very	2022-06-30 16:38:18

If you find that the data queried at this time is not prepared ,userid The shorter it is , The less accurate it is .

workable proposition

First of all , Probably not the best solution , For reference only . If you have a better way , Welcome to comment area .

help_topic

Before we start, let's have a brief understanding help_topic, It is mainly used to turn one line into multiple lines .

select substring_index(substring_index('82,83,84,85,86', ',', help_topic_id + 1), ',', -1) as Id
from mysql.help_topic
where help_topic_id < (length('82,83,84,85,86') - length(replace('82,83,84,85,86', ',', '')) + 1);

Execution results ：

Id
82
83
84
85
86

analysis

At first glance, it seems more convoluted , I try to write in detail .

1. substring_index The role of : Get the... On the left side of the target string n The left part of the separator ,n When it is negative, return to the right n The right part of a .

2. help_topic It's the database mysql A table of , This table provides details of the query help topic for a given keyword （ Detailed help information ）
   Table field meaning :

   • help_topic_id： Help topic details correspond to... In the table record ID
   • name： Help topic given keyword name .
   • help_category_id： Help topic categories ID, And help_category In the table help_category_id Field values are equal .
   • description： Help topic details （ Here's what we really want to see with help information , for example ： Tell us how to use the syntax and precautions of XX sentence ）.
   • example： Sample information for help topics （ Tell us an example of how statements are used ）.
   • url： This help topic corresponds to MySQL In the official online manual URL Link address .

ps: This article sql The sentence is actually the same as help_topic The table has nothing to do with , Just borrowing help_topic Tabular help_topic_id The value of is (0,1,2,…), Therefore, a database containing only ID Fields , It can also achieve the effect of turning to multiple lines

where Query criteria

(length('82,83,84,85,86') - length(replace('82,83,84,85,86', ',', '')) + 1)

It's been a long time , In fact, it is for calculation , Number of separated partial messages . That is to say , The number of +1. In fact, the ultimate goal is just to get one 0,1,2,3,4 A list of . There is such a message , It can be used to intercept strings .

You can put this sql Carry out one paragraph at a time , Until you fully understand .

Final realization

SELECT
    message_id,
    substring_index( substring_index( msg.receiver, ',', topic.help_topic_id + 1 ), ',',- 1 ) AS receiver,
    sender,
    message,
    create_time
FROM
    t_message msg
        JOIN mysql.help_topic topic ON topic.help_topic_id < ( length( msg.receiver ) - length( REPLACE ( msg.receiver, ',', '' ) ) + 1 );

The query results are as follows ：

message_id	receiver	sender	message	create_time
1	user1	system	hello	2022-06-30 16:30:12
1	user2	system	hello	2022-06-30 16:30:12
2	user1	system	world	2022-06-30 16:31:19
2	user3	system	world	2022-06-30 16:31:19
3	user3	user1	java	2022-06-30 16:31:22
3	user4	user1	java	2022-06-30 16:31:22
3	user5	user1	java	2022-06-30 16:31:22
4	user7	user2	neovim	2022-06-30 16:31:53
4	user8	user2	neovim	2022-06-30 16:31:53
5	itkey	system	itkey	2022-06-30 16:32:17
5	lxyoucan	system	itkey	2022-06-30 16:32:17
6	user	good	very	2022-06-30 16:38:18
6	user11	good	very	2022-06-30 16:38:18

After getting this watch , The query is relatively simple . I'm more worried about , The performance may not be particularly high .
At this time, we still press user To query , It is written as follows ：

select * from (
                  SELECT message_id,
                         substring_index(substring_index(msg.receiver, ',', topic.help_topic_id + 1), ',',
                                         - 1) AS receiver,
                         sender,
                         message,
                         create_time
                  FROM t_message msg
                           JOIN mysql.help_topic topic ON topic.help_topic_id <
                                                          (length(msg.receiver) - length(REPLACE(msg.receiver, ',', '')) + 1)
              ) as tmsg where tmsg.receiver='user';

message_id	receiver	sender	message	create_time
6	user	good	very	2022-06-30 16:38:18

The result of this query is correct .

Reduce the use of subqueries , It can be written as follows ：

SELECT
    message_id,
    sender,
    message,
    create_time
FROM
    t_message msg
        JOIN mysql.help_topic topic ON topic.help_topic_id < ( length( msg.receiver ) - length( REPLACE ( msg.receiver, ',', '' ) ) + 1 )
where substring_index( substring_index( msg.receiver, ',', topic.help_topic_id + 1 ), ',',- 1 ) = 'user';

Query results ：

message_id	sender	message	create_time
6	good	very	2022-06-30 16:38:18

summary

The accuracy of the final plan should be no problem , But there is still room for optimization in terms of performance .

Reference resources

• https://blog.csdn.net/lshxuexi/article/details/113571795
• https://blog.csdn.net/hfsaini/article/details/86477371
• https://www.jianshu.com/p/2c17b3360059