I've been working on something similar and after a few tries and fails came up with this:
Example: STRING-TO-TEST-ON = 'ab,cd,ef,gh'
I wanted to extract everything after the last occurrence of "," (comma) from the string... resulting in "gh".
My query is:
SELECT SUBSTR('ab,cd,ef,gh' FROM (LENGTH('ab,cd,ef,gh') - (LOCATE(",",REVERSE('ab,cd,ef,gh'))-1)+1)) AS `wantedString`
Now let me try and explain what I did ...
I had to find the position of the last "," from the string and to calculate the wantedString length, using LOCATE(",",REVERSE('ab,cd,ef,gh'))-1
by reversing the initial string I actually had to find the first occurrence of the "," in the string ... which wasn't hard to do ... and then -1 to actually find the string length without the ",".
calculate the position of my wantedString by subtracting the string length I've calculated at 1st step from the initial string length:
LENGTH('ab,cd,ef,gh') - (LOCATE(",",REVERSE('ab,cd,ef,gh'))-1)+1
I have (+1) because I actually need the string position after the last "," .. and not containing the ",". Hope it makes sense.
I haven't tested the query on large strings so I do not know how slow it is. So if someone actually tests it on a large string I would very happy to know the results.