It appears that MySQL doesn't have array variables. What should I use instead?
There seem to be two alternatives suggested: A set-type scalar and temporary tables. The question I linked to suggests the former. But is it good practice to use these instead of array variables? Alternatively, if I go with sets, what would be the set-based idiom equivalent to foreach?
Well, I've been using temporary tables instead of array variables. Not the greatest solution, but it works.
Note that you don't need to formally define their fields, just create them using a SELECT:
DROP TEMPORARY TABLE IF EXISTS my_temp_table;
CREATE TEMPORARY TABLE my_temp_table
SELECT first_name FROM people WHERE last_name = 'Smith';
(See also Create temporary table from select statement without using Create Table.)
You can achieve this in MySQL using WHILE loop:
SET #myArrayOfValue = '2,5,2,23,6,';
WHILE (LOCATE(',', #myArrayOfValue) > 0)
DO
SET #value = ELT(1, #myArrayOfValue);
SET #myArrayOfValue= SUBSTRING(#myArrayOfValue, LOCATE(',',#myArrayOfValue) + 1);
INSERT INTO `EXEMPLE` VALUES(#value, 'hello');
END WHILE;
EDIT:
Alternatively you can do it using UNION ALL:
INSERT INTO `EXEMPLE`
(
`value`, `message`
)
(
SELECT 2 AS `value`, 'hello' AS `message`
UNION ALL
SELECT 5 AS `value`, 'hello' AS `message`
UNION ALL
SELECT 2 AS `value`, 'hello' AS `message`
UNION ALL
...
);
Try using FIND_IN_SET() function of MySql
e.g.
SET #c = 'xxx,yyy,zzz';
SELECT * from countries
WHERE FIND_IN_SET(countryname,#c);
Note: You don't have to SET variable in StoredProcedure if you are passing parameter with CSV values.
Nowadays using a JSON array would be an obvious answer.
Since this is an old but still relevant question I produced a short example.
JSON functions are available since mySQL 5.7.x / MariaDB 10.2.3
I prefer this solution over ELT() because it's really more like an array and this 'array' can be reused in the code.
But be careful: It (JSON) is certainly much slower than using a temporary table. Its just more handy. imo.
Here is how to use a JSON array:
SET #myjson = '["gmail.com","mail.ru","arcor.de","gmx.de","t-online.de",
"web.de","googlemail.com","freenet.de","yahoo.de","gmx.net",
"me.com","bluewin.ch","hotmail.com","hotmail.de","live.de",
"icloud.com","hotmail.co.uk","yahoo.co.jp","yandex.ru"]';
SELECT JSON_LENGTH(#myjson);
-- result: 19
SELECT JSON_VALUE(#myjson, '$[0]');
-- result: gmail.com
And here a little example to show how it works in a function/procedure:
DELIMITER //
CREATE OR REPLACE FUNCTION example() RETURNS varchar(1000) DETERMINISTIC
BEGIN
DECLARE _result varchar(1000) DEFAULT '';
DECLARE _counter INT DEFAULT 0;
DECLARE _value varchar(50);
SET #myjson = '["gmail.com","mail.ru","arcor.de","gmx.de","t-online.de",
"web.de","googlemail.com","freenet.de","yahoo.de","gmx.net",
"me.com","bluewin.ch","hotmail.com","hotmail.de","live.de",
"icloud.com","hotmail.co.uk","yahoo.co.jp","yandex.ru"]';
WHILE _counter < JSON_LENGTH(#myjson) DO
-- do whatever, e.g. add-up strings...
SET _result = CONCAT(_result, _counter, '-', JSON_VALUE(#myjson, CONCAT('$[',_counter,']')), '#');
SET _counter = _counter + 1;
END WHILE;
RETURN _result;
END //
DELIMITER ;
SELECT example();
Dont know about the arrays, but there is a way to store comma-separated lists in normal VARCHAR column.
And when you need to find something in that list you can use the FIND_IN_SET() function.
I know that this is a bit of a late response, but I recently had to solve a similar problem and thought that this may be useful to others.
Background
Consider the table below called 'mytable':
The problem was to keep only latest 3 records and delete any older records whose systemid=1 (there could be many other records in the table with other systemid values)
It would be good if you could do this simply using the statement
DELETE FROM mytable WHERE id IN (SELECT id FROM `mytable` WHERE systemid=1 ORDER BY id DESC LIMIT 3)
However this is not yet supported in MySQL and if you try this then you will get an error like
...doesn't yet support 'LIMIT & IN/ALL/SOME subquery'
So a workaround is needed whereby an array of values is passed to the IN selector using variable. However, as variables need to be single values, I would need to simulate an array. The trick is to create the array as a comma separated list of values (string) and assign this to the variable as follows
SET #myvar = (SELECT GROUP_CONCAT(id SEPARATOR ',') AS myval FROM (SELECT * FROM `mytable` WHERE systemid=1 ORDER BY id DESC LIMIT 3 ) A GROUP BY A.systemid);
The result stored in #myvar is
5,6,7
Next, the FIND_IN_SET selector is used to select from the simulated array
SELECT * FROM mytable WHERE FIND_IN_SET(id,#myvar);
The combined final result is as follows:
SET #myvar = (SELECT GROUP_CONCAT(id SEPARATOR ',') AS myval FROM (SELECT * FROM `mytable` WHERE systemid=1 ORDER BY id DESC LIMIT 3 ) A GROUP BY A.systemid);
DELETE FROM mytable WHERE FIND_IN_SET(id,#myvar);
I am aware that this is a very specific case. However it can be modified to suit just about any other case where a variable needs to store an array of values.
I hope that this helps.
DELIMITER $$
CREATE DEFINER=`mysqldb`#`%` PROCEDURE `abc`()
BEGIN
BEGIN
set #value :='11,2,3,1,';
WHILE (LOCATE(',', #value) > 0) DO
SET #V_DESIGNATION = SUBSTRING(#value,1, LOCATE(',',#value)-1);
SET #value = SUBSTRING(#value, LOCATE(',',#value) + 1);
select #V_DESIGNATION;
END WHILE;
END;
END$$
DELIMITER ;
Maybe create a temporary memory table with columns (key, value) if you want associative arrays. Having a memory table is the closest thing to having arrays in mysql
Here’s how I did it.
First, I created a function that checks whether a Long/Integer/whatever value is in a list of values separated by commas:
CREATE DEFINER = 'root'#'localhost' FUNCTION `is_id_in_ids`(
`strIDs` VARCHAR(255),
`_id` BIGINT
)
RETURNS BIT(1)
NOT DETERMINISTIC
CONTAINS SQL
SQL SECURITY DEFINER
COMMENT ''
BEGIN
DECLARE strLen INT DEFAULT 0;
DECLARE subStrLen INT DEFAULT 0;
DECLARE subs VARCHAR(255);
IF strIDs IS NULL THEN
SET strIDs = '';
END IF;
do_this:
LOOP
SET strLen = LENGTH(strIDs);
SET subs = SUBSTRING_INDEX(strIDs, ',', 1);
if ( CAST(subs AS UNSIGNED) = _id ) THEN
-- founded
return(1);
END IF;
SET subStrLen = LENGTH(SUBSTRING_INDEX(strIDs, ',', 1));
SET strIDs = MID(strIDs, subStrLen+2, strLen);
IF strIDs = NULL or trim(strIds) = '' THEN
LEAVE do_this;
END IF;
END LOOP do_this;
-- not founded
return(0);
END;
So now you can search for an ID in a comma-separated list of IDs, like this:
select `is_id_in_ids`('1001,1002,1003',1002);
And you can use this function inside a WHERE clause, like this:
SELECT * FROM table1 WHERE `is_id_in_ids`('1001,1002,1003',table1_id);
This was the only way I found to pass an "array" parameter to a PROCEDURE.
I'm surprised none of the answers mention ELT/FIELD.
ELT/FIELD works very similar to an array especially if you have static data.
FIND_IN_SET also works similar but doesn't have a built in complementary
function but it's easy enough to write one.
mysql> select elt(2,'AA','BB','CC');
+-----------------------+
| elt(2,'AA','BB','CC') |
+-----------------------+
| BB |
+-----------------------+
1 row in set (0.00 sec)
mysql> select field('BB','AA','BB','CC');
+----------------------------+
| field('BB','AA','BB','CC') |
+----------------------------+
| 2 |
+----------------------------+
1 row in set (0.00 sec)
mysql> select find_in_set('BB','AA,BB,CC');
+------------------------------+
| find_in_set('BB','AA,BB,CC') |
+------------------------------+
| 2 |
+------------------------------+
1 row in set (0.00 sec)
mysql> SELECT SUBSTRING_INDEX(SUBSTRING_INDEX('AA,BB,CC',',',2),',',-1);
+-----------------------------------------------------------+
| SUBSTRING_INDEX(SUBSTRING_INDEX('AA,BB,CC',',',2),',',-1) |
+-----------------------------------------------------------+
| BB |
+-----------------------------------------------------------+
1 row in set (0.01 sec)
Is an array variable really necessary?
I ask because I originally landed here wanting to add an array as a MySQL table variable. I was relatively new to database design and trying to think of how I'd do it in a typical programming language fashion.
But databases are different. I thought I wanted an array as a variable, but it turns out that's just not a common MySQL database practice.
Standard Practice
The alternative solution to arrays is to add an additional table, and then reference your original table with a foreign key.
As an example, let's imagine an application that keeps track of all the items every person in a household wants to buy at the store.
The commands for creating the table I originally envisioned would have looked something like this:
#doesn't work
CREATE TABLE Person(
name VARCHAR(50) PRIMARY KEY
buy_list ARRAY
);
I think I envisioned buy_list to be a comma-separated string of items or something like that.
But MySQL doesn't have an array type field, so I really needed something like this:
CREATE TABLE Person(
name VARCHAR(50) PRIMARY KEY
);
CREATE TABLE BuyList(
person VARCHAR(50),
item VARCHAR(50),
PRIMARY KEY (person, item),
CONSTRAINT fk_person FOREIGN KEY (person) REFERENCES Person(name)
);
Here we define a constraint named fk_person. It says that the 'person' field in BuyList is a foreign key. In other words, it's a primary key in another table, specifically the 'name' field in the Person table, which is what REFERENCES denotes.
We also defined the combination of person and item to be the primary key, but technically that's not necessary.
Finally, if you want to get all the items on a person's list, you can run this query:
SELECT item FROM BuyList WHERE person='John';
This gives you all the items on John's list. No arrays necessary!
This is my solution to use a variable containing a list of elements.
You can use it in simple queries (no need to use store procedures or create tables).
I found somewhere else on the site the trick to use the JSON_TABLE function (it works in mysql 8, I dunno of it works in other versions).
set #x = '1,2,3,4' ;
select c.NAME
from colors c
where
c.COD in (
select *
from json_table(
concat('[',#x,']'),
'$[*]' columns (id int path '$') ) t ) ;
Also, you may need to manage the case of one or more variables set to empty_string.
In this case I added another trick (the query does not return error even if x, y, or both x and y are empty strings):
set #x = '' ;
set #y = 'yellow' ;
select c.NAME
from colors
where
if(#y = '', 1 = 1, c.NAME = #y)
and if(#x = '', 1, c.COD) in (
select *
from json_table(
concat('[',if(#x = '', 1, #x),']'),
'$[*]' columns (id int path '$') ) t) ;
This works fine for list of values:
SET #myArrayOfValue = '2,5,2,23,6,';
WHILE (LOCATE(',', #myArrayOfValue) > 0)
DO
SET #value = ELT(1, #myArrayOfValue);
SET #STR = SUBSTRING(#myArrayOfValue, 1, LOCATE(',',#myArrayOfValue)-1);
SET #myArrayOfValue = SUBSTRING(#myArrayOfValue, LOCATE(',', #myArrayOfValue) + 1);
INSERT INTO `Demo` VALUES(#STR, 'hello');
END WHILE;
Both versions using sets didn't work for me (tested with MySQL 5.5). The function ELT() returns the whole set. Considering the WHILE statement is only avaible in PROCEDURE context i added it to my solution:
DROP PROCEDURE IF EXISTS __main__;
DELIMITER $
CREATE PROCEDURE __main__()
BEGIN
SET #myArrayOfValue = '2,5,2,23,6,';
WHILE (LOCATE(',', #myArrayOfValue) > 0)
DO
SET #value = LEFT(#myArrayOfValue, LOCATE(',',#myArrayOfValue) - 1);
SET #myArrayOfValue = SUBSTRING(#myArrayOfValue, LOCATE(',',#myArrayOfValue) + 1);
END WHILE;
END;
$
DELIMITER ;
CALL __main__;
To be honest, i don't think this is a good practice. Even if its realy necessary, this is barely readable and quite slow.
Isn't the point of arrays to be efficient? If you're just iterating through values, I think a cursor on a temporary (or permanent) table makes more sense than seeking commas, no? Also cleaner. Lookup "mysql DECLARE CURSOR".
For random access a temporary table with numerically indexed primary key. Unfortunately the fastest access you'll get is a hash table, not true random access.
Another way to see the same problem.
Hope helpfull
DELIMITER $$
CREATE PROCEDURE ARR(v_value VARCHAR(100))
BEGIN
DECLARE v_tam VARCHAR(100);
DECLARE v_pos VARCHAR(100);
CREATE TEMPORARY TABLE IF NOT EXISTS split (split VARCHAR(50));
SET v_tam = (SELECT (LENGTH(v_value) - LENGTH(REPLACE(v_value,',',''))));
SET v_pos = 1;
WHILE (v_tam >= v_pos)
DO
INSERT INTO split
SELECT SUBSTRING_INDEX(SUBSTRING_INDEX(v_value,',',v_pos),',', -1);
SET v_pos = v_pos + 1;
END WHILE;
SELECT * FROM split;
DROP TEMPORARY TABLE split;
END$$
CALL ARR('1006212,1006404,1003404,1006505,444,');
If we have one table like that
mysql> select * from user_mail;
+------------+-------+
| email | user |
+------------+-------+-
| email1#gmail | 1 |
| email2#gmail | 2 |
+------------+-------+--------+------------+
and the array table:
mysql> select * from user_mail_array;
+------------+-------+-------------+
| email | user | preferences |
+------------+-------+-------------+
| email1#gmail | 1 | 1 |
| email1#gmail | 1 | 2 |
| email1#gmail | 1 | 3 |
| email1#gmail | 1 | 4 |
| email2#gmail | 2 | 5 |
| email2#gmail | 2 | 6 |
We can select the rows of the second table as one array with CONCAT function:
mysql> SELECT t1.*, GROUP_CONCAT(t2.preferences) AS preferences
FROM user_mail t1,user_mail_array t2
where t1.email=t2.email and t1.user=t2.user
GROUP BY t1.email,t1.user;
+------------+-------+--------+------------+-------------+
| email | user | preferences |
+------------+-------+--------+------------+-------------+
|email1#gmail | 1 | 1,3,2,4 |
|email2#gmail | 2 | 5,6 |
+------------+-------+--------+------------+-------------+
In MYSQL version after 5.7.x, you can use JSON type to store an array. You can get value of an array by a key via MYSQL.
Inspired by the function ELT(index number, string1, string2, string3,…),I think the following example works as an array example:
set #i := 1;
while #i <= 3
do
insert into table(val) values (ELT(#i ,'val1','val2','val3'...));
set #i = #i + 1;
end while;
Hope it help.
Here is an example for MySQL for looping through a comma delimited string.
DECLARE v_delimited_string_access_index INT;
DECLARE v_delimited_string_access_value VARCHAR(255);
DECLARE v_can_still_find_values_in_delimited_string BOOLEAN;
SET v_can_still_find_values_in_delimited_string = true;
SET v_delimited_string_access_index = 0;
WHILE (v_can_still_find_values_in_delimited_string) DO
SET v_delimited_string_access_value = get_from_delimiter_split_string(in_array, ',', v_delimited_string_access_index); -- get value from string
SET v_delimited_string_access_index = v_delimited_string_access_index + 1;
IF (v_delimited_string_access_value = '') THEN
SET v_can_still_find_values_in_delimited_string = false; -- no value at this index, stop looping
ELSE
-- DO WHAT YOU WANT WITH v_delimited_string_access_value HERE
END IF;
END WHILE;
this uses the get_from_delimiter_split_string function defined here: https://stackoverflow.com/a/59666211/3068233
I Think I can improve on this answer. Try this:
The parameter 'Pranks' is a CSV. ie. '1,2,3,4.....etc'
CREATE PROCEDURE AddRanks(
IN Pranks TEXT
)
BEGIN
DECLARE VCounter INTEGER;
DECLARE VStringToAdd VARCHAR(50);
SET VCounter = 0;
START TRANSACTION;
REPEAT
SET VStringToAdd = (SELECT TRIM(SUBSTRING_INDEX(Pranks, ',', 1)));
SET Pranks = (SELECT RIGHT(Pranks, TRIM(LENGTH(Pranks) - LENGTH(SUBSTRING_INDEX(Pranks, ',', 1))-1)));
INSERT INTO tbl_rank_names(rank)
VALUES(VStringToAdd);
SET VCounter = VCounter + 1;
UNTIL (Pranks = '')
END REPEAT;
SELECT VCounter AS 'Records added';
COMMIT;
END;
This method makes the searched string of CSV values progressively shorter with each iteration of the loop, which I believe would be better for optimization.
I would try something like this for multiple collections. I'm a MySQL beginner. Sorry about the function names, couldn't decide on what names would be best.
delimiter //
drop procedure init_
//
create procedure init_()
begin
CREATE TEMPORARY TABLE if not exists
val_store(
realm varchar(30)
, id varchar(30)
, val varchar(255)
, primary key ( realm , id )
);
end;
//
drop function if exists get_
//
create function get_( p_realm varchar(30) , p_id varchar(30) )
returns varchar(255)
reads sql data
begin
declare ret_val varchar(255);
declare continue handler for 1146 set ret_val = null;
select val into ret_val from val_store where id = p_id;
return ret_val;
end;
//
drop procedure if exists set_
//
create procedure set_( p_realm varchar(30) , p_id varchar(30) , p_val varchar(255) )
begin
call init_();
insert into val_store (realm,id,val) values (p_realm , p_id , p_val) on duplicate key update val = p_val;
end;
//
drop procedure if exists remove_
//
create procedure remove_( p_realm varchar(30) , p_id varchar(30) )
begin
call init_();
delete from val_store where realm = p_realm and id = p_id;
end;
//
drop procedure if exists erase_
//
create procedure erase_( p_realm varchar(30) )
begin
call init_();
delete from val_store where realm = p_realm;
end;
//
call set_('my_array_table_name','my_key','my_value');
select get_('my_array_table_name','my_key');
Rather than Saving data as a array or in one row only you should be making diffrent rows for every value received. This will make it much simpler to understand rather than putting all together.
Have you tried using PHP's serialize()?
That allows you to store the contents of a variable's array in a string PHP understands and is safe for the database (assuming you've escaped it first).
$array = array(
1 => 'some data',
2 => 'some more'
);
//Assuming you're already connected to the database
$sql = sprintf("INSERT INTO `yourTable` (`rowID`, `rowContent`) VALUES (NULL, '%s')"
, serialize(mysql_real_escape_string($array, $dbConnection)));
mysql_query($sql, $dbConnection) or die(mysql_error());
You can also do the exact same without a numbered array
$array2 = array(
'something' => 'something else'
);
or
$array3 = array(
'somethingNew'
);
MySQL runs pretty much all string comparisons under the default collation... except the REPLACE command. I have a case-insensitive collation and need to run a case-insensitive REPLACE. Is there any way to force REPLACE to use the current collation rather than always doing case-sensitive comparisons? I'm willing to upgrade my MySQL (currently running 5.1) to get added functionality...
mysql> charset utf8 collation utf8_unicode_ci;
Charset changed
mysql> select 'abc' like '%B%';
+------------------+
| 'abc' like '%B%' |
+------------------+
| 1 |
+------------------+
mysql> select replace('aAbBcC', 'a', 'f');
+-----------------------------+
| replace('aAbBcC', 'a', 'f') |
+-----------------------------+
| fAbBcC | <--- *NOT* 'ffbBcC'
+-----------------------------+
If replace(lower()) doesn't work, you'll need to create another function.
My 2 cents.
Since many people have migrated from MySQL to MariaDB, those people will have available a new function called REGEXP_REPLACE. Use it as you would a normal replace, but the pattern is a regular expression.
This is a working example:
UPDATE `myTable`
SET `myField` = REGEXP_REPLACE(`myField`, '(?i)my insensitive string', 'new string')
WHERE `myField` REGEXP '(?i)my insensitive string'
The option (?i) makes all the subsequent matches case insensitive (if put at the beginning of the pattern like I have then it all is insensitive).
See here for more information: https://mariadb.com/kb/en/mariadb/pcre/
Edit: as of MySQL 8.0 you can now use the regexp_replace function too, see documentation: https://dev.mysql.com/doc/refman/8.0/en/regexp.html
Alternative function for one spoken by fvox.
DELIMITER |
CREATE FUNCTION case_insensitive_replace ( REPLACE_WHERE text, REPLACE_THIS text, REPLACE_WITH text )
RETURNS text
DETERMINISTIC
BEGIN
DECLARE last_occurency int DEFAULT '1';
IF LCASE(REPLACE_THIS) = LCASE(REPLACE_WITH) OR LENGTH(REPLACE_THIS) < 1 THEN
RETURN REPLACE_WHERE;
END IF;
WHILE Locate( LCASE(REPLACE_THIS), LCASE(REPLACE_WHERE), last_occurency ) > 0 DO
BEGIN
SET last_occurency = Locate(LCASE(REPLACE_THIS), LCASE(REPLACE_WHERE));
SET REPLACE_WHERE = Insert( REPLACE_WHERE, last_occurency, LENGTH(REPLACE_THIS), REPLACE_WITH);
SET last_occurency = last_occurency + LENGTH(REPLACE_WITH);
END;
END WHILE;
RETURN REPLACE_WHERE;
END;
|
DELIMITER ;
Small test:
SET #str = BINARY 'New York';
SELECT case_insensitive_replace(#str, 'y', 'K');
Answers: New Kork
This modification of Luist's answer allows one to replace the needle with a differently cased version of the needle (two lines change).
DELIMITER |
CREATE FUNCTION case_insensitive_replace ( REPLACE_WHERE text, REPLACE_THIS text, REPLACE_WITH text )
RETURNS text
DETERMINISTIC
BEGIN
DECLARE last_occurency int DEFAULT '1';
IF LENGTH(REPLACE_THIS) < 1 THEN
RETURN REPLACE_WHERE;
END IF;
WHILE Locate( LCASE(REPLACE_THIS), LCASE(REPLACE_WHERE), last_occurency ) > 0 DO
BEGIN
SET last_occurency = Locate(LCASE(REPLACE_THIS), LCASE(REPLACE_WHERE), last_occurency);
SET REPLACE_WHERE = Insert( REPLACE_WHERE, last_occurency, LENGTH(REPLACE_THIS), REPLACE_WITH);
SET last_occurency = last_occurency + LENGTH(REPLACE_WITH);
END;
END WHILE;
RETURN REPLACE_WHERE;
END;
|
DELIMITER ;
I went with http://pento.net/2009/02/15/case-insensitive-replace-for-mysql/ (in fvox's answer) which performs the case insensitive search with case sensitive replacement and without changing the case of what should be unaffected characters elsewhere in the searched string.
N.B. the comment further down that same page stating that CHAR(255) should be changed to VARCHAR(255) - this seemed to be required for me as well.
In the previous answers, and the pento.net link, the arguments to LOCATE() are lower-cased.
This is a waste of resources, as LOCATE is case-insensitive by default:
mysql> select locate('el', 'HELLo');
+-----------------------+
| locate('el', 'HELLo') |
+-----------------------+
| 2 |
+-----------------------+
You can replace
WHILE Locate( LCASE(REPLACE_THIS), LCASE(REPLACE_WHERE), last_occurency ) > 0 DO
with
WHILE Locate(REPLACE_THIS, REPLACE_WHERE, last_occurency ) > 0 DO
etc.
In case of 'special' characters there is unexpected behaviour:
SELECT case_insensitive_replace('A', 'Ã', 'a')
Gives
a
Which is unexpected... since we only want to replace the à not A
What is even more weird:
SELECT LOCATE('Ã', 'A');
gives
0
Which is the correct result... seems to have to do with encoding of the parameters of the stored procedure...
I like to use a search and replace function I created when I need to replace without worrying about the case of the original or search strings. This routine bails out quickly if you pass in an empty/null search string or a null replace string without altering the incoming string. I also added a safe count down just in case somehow the search keep looping. This way we don't get stuck in a loop forever. Alter the starting number if you think it is too low.
delimiter //
DROP FUNCTION IF EXISTS `replace_nocase`//
CREATE FUNCTION `replace_nocase`(raw text, find_str varchar(1000), replace_str varchar(1000)) RETURNS text
CHARACTER SET utf8
DETERMINISTIC
BEGIN
declare ret text;
declare len int;
declare hit int;
declare safe int;
if find_str is null or find_str='' or replace_str is null then
return raw;
end if;
set safe=10000;
set ret=raw;
set len=length(find_str);
set hit=LOCATE(find_str,ret);
while hit>0 and safe>0 do
set ret=concat(substring(ret,1,hit-1),replace_str,substring(ret,hit+len));
set hit=LOCATE(find_str,ret,hit+1);
set safe=safe-1;
end while;
return ret;
END//
This question is a bit old but I ran into the same problem and the answers given didn't allow me to solve it entirely.
I wanted the result to retain the case of the original string.
So I made a small modification to the replace_ci function proposed by fvox :
DELIMITER $$
DROP FUNCTION IF EXISTS `replace_ci`$$
CREATE FUNCTION `replace_ci` (str TEXT, needle CHAR(255), str_rep CHAR(255))
RETURNS TEXT
DETERMINISTIC
BEGIN
DECLARE return_str TEXT DEFAULT '';
DECLARE lower_str TEXT;
DECLARE lower_needle TEXT;
DECLARE tmp_needle TEXT;
DECLARE str_origin_char CHAR(1);
DECLARE str_rep_char CHAR(1);
DECLARE final_str_rep TEXT DEFAULT '';
DECLARE pos INT DEFAULT 1;
DECLARE old_pos INT DEFAULT 1;
DECLARE needle_pos INT DEFAULT 1;
IF needle = '' THEN
RETURN str;
END IF;
SELECT LOWER(str) INTO lower_str;
SELECT LOWER(needle) INTO lower_needle;
SELECT LOCATE(lower_needle, lower_str, pos) INTO pos;
WHILE pos > 0 DO
SELECT substr(str, pos, char_length(needle)) INTO tmp_needle;
SELECT '' INTO final_str_rep;
SELECT 1 INTO needle_pos;
WHILE needle_pos <= char_length(tmp_needle) DO
SELECT substr(tmp_needle, needle_pos, 1) INTO str_origin_char;
SELECT SUBSTR(str_rep, needle_pos, 1) INTO str_rep_char;
SELECT CONCAT(final_str_rep, IF(BINARY str_origin_char = LOWER(str_origin_char), LOWER(str_rep_char), IF(BINARY str_origin_char = UPPER(str_origin_char), UPPER(str_rep_char), str_rep_char))) INTO final_str_rep;
SELECT (needle_pos + 1) INTO needle_pos;
END WHILE;
SELECT CONCAT(return_str, SUBSTR(str, old_pos, pos - old_pos), final_str_rep) INTO return_str;
SELECT pos + CHAR_LENGTH(needle) INTO pos;
SELECT pos INTO old_pos;
SELECT LOCATE(lower_needle, lower_str, pos) INTO pos;
END WHILE;
SELECT CONCAT(return_str, SUBSTR(str, old_pos, CHAR_LENGTH(str))) INTO return_str;
RETURN return_str;
END$$
DELIMITER ;
Example of use :
SELECT replace_ci( 'MySQL', 'm', 'e' ) as replaced;
Will return :
| replaced |
| --- |
| EySQL |