Merge two colums, multiple rows ignoring duplicates - MySQL - mysql

Question

This is, maybe, a stupid question. Not for me: honestly, I'm not so skilled with MySql queries so I'm looking for a little help.
I have a table:
id | type_a | type_b |
__________________________________________________
1 | *color_1*color_2*color_3*| *color_1* |
2 | *color_3* | *color_3*color_2*|
3 | *color_2*color_3* | *color_4* |
4 | *color_1*color_3*color_4*| |
5 | *color_4* | *color_5* |
__________________________________________________
I would like to move "type_b" column content in "type_a" column ignoring duplicate fields (delimited by * and *, ex: color_1. This kind of storage is builded by a Joomla component).
I would like to have this final result:
id | type_a | type_b |
_________________________________________
1 | *color_1*color_2*color_3*| |
2 | *color_3*color_2* | |
3 | *color_2*color_3*color_4*| |
4 | *color_1*color_3*color_4*| |
5 | *color_4*color_5* | |
_________________________________________
What's the best way to accomplish something similar?
Thanks to all!

Answer 1

You could do it with this statement (no, it's not looking nice), assuming the name of your table is example:
UPDATE
example e1
SET
e1.type_a = (
SELECT
CONCAT('*', GROUP_CONCAT(DISTINCT n1.value ORDER BY n1.value SEPARATOR '*'), '*') as type_a
FROM (
SELECT
id,
CASE
WHEN SUBSTRING_INDEX(SUBSTRING_INDEX(TRIM(BOTH '*' FROM e.type_a), '*', n.n), '*', -1) = '' THEN NULL
ELSE SUBSTRING_INDEX(SUBSTRING_INDEX(TRIM(BOTH '*' FROM e.type_a), '*', n.n), '*', -1)
END value
FROM example e CROSS JOIN (
SELECT
a.N + b.N * 10 + 1 AS n
FROM
(SELECT 0 AS N 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) a
,(SELECT 0 AS N 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) b
ORDER BY n
) n
WHERE
n.n <= 1 + (LENGTH(e.type_a) - LENGTH(REPLACE(e.type_a, '*', '')))
UNION
SELECT
id,
CASE
WHEN SUBSTRING_INDEX(SUBSTRING_INDEX(TRIM(BOTH '*' FROM e.type_b), '*', n.n), '*', -1) = '' THEN NULL
ELSE SUBSTRING_INDEX(SUBSTRING_INDEX(TRIM(BOTH '*' FROM e.type_b), '*', n.n), '*', -1)
END value
FROM example e CROSS JOIN (
SELECT
a.N + b.N * 10 + 1 AS n
FROM
(SELECT 0 AS N 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) a
,(SELECT 0 AS N 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) b
ORDER BY n
) n
WHERE
n.n <= 1 + (LENGTH(e.type_b) - LENGTH(REPLACE(e.type_b, '*', '')))
) n1
WHERE
n1.id = e1.id
GROUP BY
id
),
e1.type_b = ''
;
Demo of the SELECT statement
Explanation
Basically I adapted the method of peterm to get the split done. I had to remove the outer * first by TRIM.
To allow the empty string as column value, I've added the CASE construct, to eliminate such values. If your column has NULL values instead, you could substitute the CASE by
SUBSTRING_INDEX(SUBSTRING_INDEX(TRIM(BOTH '*' FROM e.type_a), '*', n.n), '*', -1)
and
SUBSTRING_INDEX(SUBSTRING_INDEX(TRIM(BOTH '*' FROM e.type_a), '*', n.n), '*', -1)
The UNION (without the ALL keyword) of this construct will give us the list of distinct color values and with GROUP BY id and GROUP_CONCAT we'll get the * separated value list. Last we add a leading and a trailing * to match your requirements.
For the update you've got to modificate the select, so that it returns just one column with one row (with the where clause).
Note
As stated by peterm this will allow up to 100 values in your value list. I don't believe you will need more, but if you will, then you've got to adapt the generating of the numbers up to your needs.