GROUP_CONCAT in sub-query based on specified values - mysql

Question

User Table:
ID InstructionSets
1 123,124
Instruction Set Table:
ID Name
123 Learning SQL
124 Learning More SQL
Desired Query Result:
UserID SetID SetNames
1 123,124 Learning SQL,Learning More SQL
Current SQL:
SELECT U1.ID AS UserID, U1.InstructionSets AS SetID, (
SELECT GROUP_CONCAT(Name ORDER BY FIELD(I1.ID, U1.InstructionSets))
FROM Instructions I1
WHERE I1.ID IN (U1.InstructionSets)
) AS SetName
FROM Users U1
WHERE `ID` = 1
RESULT
UserID SetID SetNames
1 123,124 Learning SQL
As expected, if I remove the WHERE clause in the sub-query, all of the SetNames appear; but if I specify the required IDs, I only get the name associated with the first ID. Obviously, I also need to fetch the SetNames in the same order as the IDs. Hence ORDER BY in GROUP_CONCAT.
Also:
Is there a better approach (other than storing the user instruction set assignments in a separate table — overkill for this application)? Couldn't see how to use JOIN in this
situation.
Is there a better title for this question?
Thanks.

Answer 1

Instead of IN use LIKE operator like this:
SELECT U1.ID AS UserID, U1.InstructionSets AS SetID, (
SELECT GROUP_CONCAT(Name ORDER BY (I1.ID))
FROM Instructions I1
WHERE CONCAT(',', U1.InstructionSets, ',') LIKE concat('%,', I1.ID, ',%')
) AS SetName
FROM Users U1
WHERE `ID` = 1
See the demo.
Results:
| UserID | SetID | SetName |
| ------ | ------- | ------------------------------ |
| 1 | 123,124 | Learning SQL,Learning More SQL |

Answer 2

We can use FIND_IN_SET(). In this context, using FIELD() function doesn't make sense.
We can also use FIND_IN_SET() in the WHERE clause. (Function returns 0 when the string isn't found in the string list.)
e.g.
SELECT u.id AS userid
, u.instructionsets AS setid
, ( SELECT GROUP_CONCAT(i.name ORDER BY FIND_IN_SET(i.id, u.instructionsets))
FROM `Instructions` i
WHERE FIND_IN_SET(i.id, u.instructionsets))
) AS setname
FROM `Users` u
WHERE u.id = 1
Storing comma separated lists is an anti-pattern; a separate table isn't overkill.
Assuming id is unique in Users table, we could do a join operation with a GROUP BY
SELECT u.id AS userid
, MIN(u.instructionsets) AS setid
, GROUP_CONCAT(i.name ORDER BY FIND_IN_SET(i.id, u.instructionsets))) AS setname
FROM `Users` u
LEFT
JOIN `Instructions` i
ON FIND_IN_SET(i.id, u.instructionsets)
WHERE u.id = 1
GROUP BY u.id