如何在mysql中将单行拆分为多列

Question

Simply Asking, Is there any function available in mysql to split single row elements in to multiple columns ? I have a table row with the fields, user_id, user_name, user_location.

In this a user can add multiple locations. I am imploding the locations and storing it in a table as a single row using php.

When i am showing the user records in a grid view, I am getting problem for pagination as i am showing the records by splitting the user_locations. So I need to split the user_locations ( single row to multiple columns).

Is there any function available in mysql to split and count the records by character ( % ).

For Example the user_location having US%UK%JAPAN%CANADA

How can i split this record in to 4 columns. I need to check for the count values (4) also. thanks in advance.

Answer 1

First normalize the string, removing empty locations and making sure there's a % at the end:

select replace(concat(user_location,'%'),'%%','%') as str
from YourTable where user_id = 1

Then we can count the number of entries with a trick. Replace '%' with '% ', and count the number of spaces added to the string. For example:

select length(replace(str, '%', '% ')) - length(str)
    as LocationCount    
from (
    select replace(concat(user_location,'%'),'%%','%') as str
    from YourTable where user_id = 1
) normalized

Using substring_index, we can add columns for a number of locations:

select length(replace(str, '%', '% ')) - length(str)
    as LocationCount    
, substring_index(substring_index(str,'%',1),'%',-1) as Loc1
, substring_index(substring_index(str,'%',2),'%',-1) as Loc2
, substring_index(substring_index(str,'%',3),'%',-1) as Loc3
from (
    select replace(concat(user_location,'%'),'%%','%') as str
    from YourTable where user_id = 1
) normalized

For your example US%UK%JAPAN%CANADA , this prints:

LocationCount  Loc1    Loc2    Loc3
4              US      UK      JAPAN

So you see it can be done, but parsing strings isn't one of SQL's strengths.

Answer 2

The "right thing" would be splitting the locations off to another table and establish a many-to-many relationship between them.

create table users (
   id int not null auto_increment primary key,
   name varchar(64)
)

create table locations (
   id int not null auto_increment primary key,
   name varchar(64)
)

create table users_locations (
   id int not null auto_increment primary key,
   user_id int not null,
   location_id int not null,
   unique index user_location_unique_together (user_id, location_id)
)

Then, ensure referential integrity either using foreign keys (and InnoDB engine) or triggers.

Answer 3

this should do it

DELIMITER $$

DROP PROCEDURE IF EXISTS `CSV2LST`$$

CREATE DEFINER=`root`@`%` PROCEDURE `CSV2LST`(IN csv_ TEXT)
BEGIN

 SET @s=CONCAT('select \"',REPLACE(csv_,',','\" union select \"'),'\";');
 PREPARE stmt FROM @s;
 EXECUTE stmt;
 DEALLOCATE PREPARE stmt; 
END$$

DELIMITER ;

Answer 4

You should do this in your client application, not on the database.

When you make a SQL query you must statically specify the columns you want to get, that is, you tell the DB the columns you want in your resultset BEFORE executing it. For instance, if you have a datetime stored, you may do something like select month(birthday), select year(birthday) from ... , so in this case we split the column birthday into 2 other columns, but it is specified in the query what columns we will have.

In your case, you would have to get exactly that US%UK%JAPAN%CANADA string from the database, and then you split it later in your software, ie

/* get data from database */
/* ... */
$user_location = ... /* extract the field from the resultset */
$user_locations = explode("%", $user_location);

Answer 5

This is a bad design, If you can change it, store the data in 2 tables:

table users: id, name, surname ...

table users_location: user_id (fk), location

users_location would have a foreign key to users thorugh user_id field