Question

i have one query that need some changes, and i don’t get any clue to do this :

this is my query :

select * from user_data a
left join user_group b 
on (a.role like b.role)

actually role value in userdata is (varchar)‘staff’

and role value in group is (varchar)‘staff;security;finance’

so i don’t get result what i expected ..

i imagine the query should be similar to this :

select * from user_data a
left join user_group b 
on (b.role like a.role+";%") // using wildcard

and i still don’t know the right query using wildcard to this case

any one can help?

Answer 1

You can use CONCAT:

select * from user_data a
left join user_group b 
on (b.role like CONCAT(a.role,";%")) // using wildcard

Note – does b.role only have to match a.role at the beginning? what if it was security;staff;finance? You could do CONCAT('%',a.role,'%').

You could do CONCAT('%','a.role','%') to handle matching a.role at any position, but only if you can be sure that you won’t have nested roles.
For example: if b.role is staff and a.role is finance;gardenstaff;security, then this row will be returned from the query even though the role is gardenstaff and not staff.

As an alternative, you can use RLIKE instead of LIKE. This is basically a regular-expressions verson of LIKE.

In particular, the regex [[:<:]]staff[[:>:]] will match the whole word staff. The [[:<:]] and [[:>:]] stand for word boundaries, which stop you from matching the staff in gardenstaff.

So, your query could be:

select * from user_data a
left join user_group b 
on (b.role RLIKE CONCAT('[[:<:]]',a.role,'[[:>:]]'))

And this would work for b.role being anywhere in the semicolon-separated a.role.