AFD
AFD

Reputation: 1419

SQL split values to multiple rows

I have table :

id | name    
1  | a,b,c    
2  | b

i want output like this :

id | name    
1  | a    
1  | b    
1  | c    
2  | b

Upvotes: 137

Views: 311444

Answers (13)

Kavita Padave
Kavita Padave

Reputation: 29

The simplest and most correct way to do this in MySQL is to use a Recursive CTE.

WITH RECURSIVE SplitValues AS (
    SELECT
        id,
        SUBSTRING_INDEX(name, ',', 1) AS split_value,
        IF(LOCATE(',', name) > 0, SUBSTRING(name, LOCATE(',', name) + 1), NULL) AS remaining_values
    FROM tmpNumber
    UNION ALL
    SELECT
        id,
        SUBSTRING_INDEX(remaining_values, ',', 1) AS split_value,
        IF(LOCATE(',', remaining_values) > 0, SUBSTRING(remaining_values, LOCATE(',', remaining_values) + 1), NULL)
    FROM
        SplitValues
    WHERE
        remaining_values IS NOT NULL
)
SELECT
    id,
    split_value
FROM
    SplitValues order by id;

Output Window

Upvotes: 1

Mamed Shahmaliyev
Mamed Shahmaliyev

Reputation: 343

Here is another trick. The number 20 is the maximum number of values in comma separated list.

We use single query, no procedures.

If tbl has more rows than the maximum number of values in single comma separated list you may remove 'inner join tbl a inner join tbl c' part from query. I added this because there are only 2 rows.

CREATE TABLE tbl(id int NOT NULL,name varchar(50),PRIMARY KEY (`id`));
insert into tbl values(1, 'a,b,c'), (2, 'd');
select id ,SUBSTRING_INDEX(SUBSTRING_INDEX(name, ',', k.n), ',', -1) as name
from tbl
INNER JOIN (
    SELECT * 
      FROM (
         SELECT @n:=@n+1 AS n
           FROM tbl inner join tbl a inner join tbl c
           INNER JOIN (SELECT @n:=0) AS _a
      ) AS _a WHERE _a.n <= 20
)AS k ON k.n <= LENGTH(name) - LENGTH(replace(name, ',','')) + 1
order by id

This is a trick to extract nth value in comma separated list:

SUBSTRING_INDEX(SUBSTRING_INDEX(name, ',', k.n), ',', -1)

Upvotes: 2

Leon Straathof
Leon Straathof

Reputation: 31

Because you have to keep adding "select number union all" in the example above which can be a issue if you need a large number of splits.

    select
  tablename.id,
  SUBSTRING_INDEX(SUBSTRING_INDEX(tablename.name, ',', numbers.n), ',', -1) name
from
  (select 1 n union all
   select 2 union all select 3 union all
   select 4 union all select 5) numbers INNER JOIN tablename
  on CHAR_LENGTH(tablename.name)
     -CHAR_LENGTH(REPLACE(tablename.name, ',', ''))>=numbers.n-1
order by
  id, n

I decided a better way was this which only adds a number row for each digit. Example below is good for 1-1000 add another row makes it good for 1-10000 and so on.

    select
  tablename.id,
  SUBSTRING_INDEX(SUBSTRING_INDEX(tablename.name, ',', numbers.n), ',', -1) name
from(SELECT @row := @row + 1 AS n FROM 
(select 0 union all select 1 union all select 2 union all select 3 union all select 4 union all select 5 union all select 6 union all select 7 union all select 8 union all select 9) as t,
(select 0 union all select 1 union all select 2 union all select 3 union all select 4 union all select 5 union all select 6 union all select 7 union all select 8 union all select 9) as t2,
(select 0 union all select 1 union all select 2 union all select 3 union all select 4 union all select 5 union all select 6 union all select 7 union all select 8 union all select 9) as t3,  
(SELECT @row:=0) as numbers)as numbers INNER JOIN tablename
  on CHAR_LENGTH(tablename.name)
     -CHAR_LENGTH(REPLACE(tablename.name, ',', ''))>=numbers.n-1
order by
  id, n

Upvotes: 3

Trace Ashley
Trace Ashley

Reputation: 37

SELECT id, unnest(string_to_array(name, ',')) AS names
FROM datatable

Hope this helps :D

Upvotes: -5

qupc
qupc

Reputation: 1

Best Practice. Result:

SELECT
SUBSTRING_INDEX(SUBSTRING_INDEX('ab,bc,cd',',',help_id+1),',',-1) AS oid
FROM
(
SELECT @xi:=@xi+1 as help_id from 
(SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5) xc1,
(SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5) xc2,
(SELECT @xi:=-1) xc0
) a
WHERE 
help_id < LENGTH('ab,bc,cd')-LENGTH(REPLACE('ab,bc,cd',',',''))+1

First, create a numbers table:

SELECT @xi:=@xi+1 as help_id from 
(SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5) xc1,
(SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5) xc2,
(SELECT @xi:=-1) xc0;
| help_id  |
| --- |
| 0   |
| 1   |
| 2   |
| 3   |
| ...   |
| 24   |

Second, just split the str:

SELECT SUBSTRING_INDEX(SUBSTRING_INDEX('ab,bc,cd',',',help_id+1),',',-1) AS oid
FROM
numbers_table
WHERE
help_id < LENGTH('ab,bc,cd')-LENGTH(REPLACE('ab,bc,cd',',',''))+1
| oid  |
| --- |
| ab   |
| bc   |
| cd   |

Upvotes: 0

Paul Spiegel
Paul Spiegel

Reputation: 31832

If the name column were a JSON array (like '["a","b","c"]'), then you could extract/unpack it with JSON_TABLE() (available since MySQL 8.0.4):

select t.id, j.name
from mytable t
join json_table(
  t.name,
  '$[*]' columns (name varchar(50) path '$')
) j;

Result:

| id  | name |
| --- | ---- |
| 1   | a    |
| 1   | b    |
| 1   | c    |
| 2   | b    |

View on DB Fiddle

If you store the values in a simple CSV format, then you would first need to convert it to JSON:

select t.id, j.name
from mytable t
join json_table(
  replace(json_array(t.name), ',', '","'),
  '$[*]' columns (name varchar(50) path '$')
) j

Result:

| id  | name |
| --- | ---- |
| 1   | a    |
| 1   | b    |
| 1   | c    |
| 2   | b    |

View on DB Fiddle

Upvotes: 37

Harry Marx
Harry Marx

Reputation: 159

Here is my attempt: The first select presents the csv field to the split. Using recursive CTE, we can create a list of numbers that are limited to the number of terms in the csv field. The number of terms is just the difference in the length of the csv field and itself with all the delimiters removed. Then joining with this numbers, substring_index extracts that term.

with recursive
    T as ( select 'a,b,c,d,e,f' as items),
    N as ( select 1 as n union select n + 1 from N, T
        where n <= length(items) - length(replace(items, ',', '')))
    select distinct substring_index(substring_index(items, ',', n), ',', -1)
group_name from N, T

Upvotes: 10

user9526573
user9526573

Reputation: 51

The original question was for MySQL and SQL in general. The example below is for the new versions of MySQL. Unfortunately, a generic query that would work on any SQL server is not possible. Some servers do no support CTE, others do not have substring_index, yet others have built-in functions for splitting a string into multiple rows.

--- the answer follows ---

Recursive queries are convenient when the server does not provide built-in functionality. They can also be the bottleneck.

The following query was written and tested on MySQL version 8.0.16. It will not work on version 5.7-. The old versions do not support Common Table Expression (CTE) and thus recursive queries.

with recursive
  input as (
        select 1 as id, 'a,b,c' as names
      union
        select 2, 'b'
    ),
  recurs as (
        select id, 1 as pos, names as remain, substring_index( names, ',', 1 ) as name
          from input
      union all
        select id, pos + 1, substring( remain, char_length( name ) + 2 ),
            substring_index( substring( remain, char_length( name ) + 2 ), ',', 1 )
          from recurs
          where char_length( remain ) > char_length( name )
    )
select id, name
  from recurs
  order by id, pos;

Upvotes: 1

Here is my solution

-- Create the maximum number of words we want to pick (indexes in n)
with recursive n(i) as (
    select
        1 i
    union all
    select i+1 from n where i < 1000
)
select distinct
    s.id,
    s.oaddress,
    -- n.i,
    -- use the index to pick the nth word, the last words will always repeat. Remove the duplicates with distinct
    if(instr(reverse(trim(substring_index(s.oaddress,' ',n.i))),' ') > 0,
        reverse(substr(reverse(trim(substring_index(s.oaddress,' ',n.i))),1,
            instr(reverse(trim(substring_index(s.oaddress,' ',n.i))),' '))),
        trim(substring_index(s.oaddress,' ',n.i))) oth
from 
    app_schools s,
    n

Upvotes: 0

Andrey
Andrey

Reputation: 1623

My variant: stored procedure that takes table name, field names and delimiter as arguments. Inspired by post http://www.marcogoncalves.com/2011/03/mysql-split-column-string-into-rows/

delimiter $$

DROP PROCEDURE IF EXISTS split_value_into_multiple_rows $$
CREATE PROCEDURE split_value_into_multiple_rows(tablename VARCHAR(20),
    id_column VARCHAR(20), value_column VARCHAR(20), delim CHAR(1))
  BEGIN
    DECLARE id INT DEFAULT 0;
    DECLARE value VARCHAR(255);
    DECLARE occurrences INT DEFAULT 0;
    DECLARE i INT DEFAULT 0;
    DECLARE splitted_value VARCHAR(255);
    DECLARE done INT DEFAULT 0;
    DECLARE cur CURSOR FOR SELECT tmp_table1.id, tmp_table1.value FROM 
        tmp_table1 WHERE tmp_table1.value IS NOT NULL AND tmp_table1.value != '';
    DECLARE CONTINUE HANDLER FOR NOT FOUND SET done = 1;

    SET @expr = CONCAT('CREATE TEMPORARY TABLE tmp_table1 (id INT NOT NULL, value VARCHAR(255)) ENGINE=Memory SELECT ',
        id_column,' id, ', value_column,' value FROM ',tablename);
    PREPARE stmt FROM @expr;
    EXECUTE stmt;
    DEALLOCATE PREPARE stmt;

    DROP TEMPORARY TABLE IF EXISTS tmp_table2;
    CREATE TEMPORARY TABLE tmp_table2 (id INT NOT NULL, value VARCHAR(255) NOT NULL) ENGINE=Memory;

    OPEN cur;
      read_loop: LOOP
        FETCH cur INTO id, value;
        IF done THEN
          LEAVE read_loop;
        END IF;

        SET occurrences = (SELECT CHAR_LENGTH(value) -
                           CHAR_LENGTH(REPLACE(value, delim, '')) + 1);
        SET i=1;
        WHILE i <= occurrences DO
          SET splitted_value = (SELECT TRIM(SUBSTRING_INDEX(
              SUBSTRING_INDEX(value, delim, i), delim, -1)));
          INSERT INTO tmp_table2 VALUES (id, splitted_value);
          SET i = i + 1;
        END WHILE;
      END LOOP;

      SELECT * FROM tmp_table2;
    CLOSE cur;
    DROP TEMPORARY TABLE tmp_table1;
  END; $$

delimiter ;

Usage example (normalization):

CALL split_value_into_multiple_rows('my_contacts', 'contact_id', 'interests', ',');

CREATE TABLE interests (
  interest_id INT NOT NULL AUTO_INCREMENT PRIMARY KEY,
  interest VARCHAR(30) NOT NULL
) SELECT DISTINCT value interest FROM tmp_table2;

CREATE TABLE contact_interest (
  contact_id INT NOT NULL,
  interest_id INT NOT NULL,
  CONSTRAINT fk_contact_interest_my_contacts_contact_id FOREIGN KEY (contact_id) REFERENCES my_contacts (contact_id),
  CONSTRAINT fk_contact_interest_interests_interest_id FOREIGN KEY (interest_id) REFERENCES interests (interest_id)
) SELECT my_contacts.contact_id, interests.interest_id
    FROM my_contacts, tmp_table2, interests
    WHERE my_contacts.contact_id = tmp_table2.id AND interests.interest = tmp_table2.value;

Upvotes: 6

Imanez
Imanez

Reputation: 514

CREATE PROCEDURE `getVal`()
BEGIN
        declare r_len integer;
        declare r_id integer;
        declare r_val varchar(20);
        declare i integer;
        DECLARE found_row int(10);
        DECLARE row CURSOR FOR select length(replace(val,"|","")),id,val from split;
        create table x(id int,name varchar(20));
      open row;
            select FOUND_ROWS() into found_row ;
            read_loop: LOOP
                IF found_row = 0 THEN
                         LEAVE read_loop;
                END IF;
            set i = 1;  
            FETCH row INTO r_len,r_id,r_val;
            label1: LOOP        
                IF i <= r_len THEN
                  insert into x values( r_id,SUBSTRING(replace(r_val,"|",""),i,1));
                  SET i = i + 1;
                  ITERATE label1;
                END IF;
                LEAVE label1;
            END LOOP label1;
            set found_row = found_row - 1;
            END LOOP;
        close row;
        select * from x;
        drop table x;
END

Upvotes: 1

fthiella
fthiella

Reputation: 49089

If you can create a numbers table, that contains numbers from 1 to the maximum fields to split, you could use a solution like this:

select
  tablename.id,
  SUBSTRING_INDEX(SUBSTRING_INDEX(tablename.name, ',', numbers.n), ',', -1) name
from
  numbers inner join tablename
  on CHAR_LENGTH(tablename.name)
     -CHAR_LENGTH(REPLACE(tablename.name, ',', ''))>=numbers.n-1
order by
  id, n

Please see fiddle here.

If you cannot create a table, then a solution can be this:

select
  tablename.id,
  SUBSTRING_INDEX(SUBSTRING_INDEX(tablename.name, ',', numbers.n), ',', -1) name
from
  (select 1 n union all
   select 2 union all select 3 union all
   select 4 union all select 5) numbers INNER JOIN tablename
  on CHAR_LENGTH(tablename.name)
     -CHAR_LENGTH(REPLACE(tablename.name, ',', ''))>=numbers.n-1
order by
  id, n

an example fiddle is here.

Upvotes: 184

Prahalad Gaggar
Prahalad Gaggar

Reputation: 11609

I have take the reference from here with changed column name.

DELIMITER $$

CREATE FUNCTION strSplit(x VARCHAR(65000), delim VARCHAR(12), pos INTEGER) 
RETURNS VARCHAR(65000)
BEGIN
  DECLARE output VARCHAR(65000);
  SET output = REPLACE(SUBSTRING(SUBSTRING_INDEX(x, delim, pos)
                 , LENGTH(SUBSTRING_INDEX(x, delim, pos - 1)) + 1)
                 , delim
                 , '');
  IF output = '' THEN SET output = null; END IF;
  RETURN output;
END $$


CREATE PROCEDURE BadTableToGoodTable()
BEGIN
  DECLARE i INTEGER;

  SET i = 1;
  REPEAT
    INSERT INTO GoodTable (id, name)
      SELECT id, strSplit(name, ',', i) FROM BadTable
      WHERE strSplit(name, ',', i) IS NOT NULL;
    SET i = i + 1;
    UNTIL ROW_COUNT() = 0
  END REPEAT;
END $$

DELIMITER ;

Upvotes: 9

Related Questions