SUM(DISTINCT) Based on Other Columns

Question

I currently have a table that looks something like this:

+------+-------+------------+------------+
| id   | rate  | first_name | last_name  |
+------+-----         


        

Answer 1

You can do this by making the values you are summing distinct. This is possible but is very very ugly.

First, you can turn a string into a number by taking a hash. The SQL below does an MD5 hash of the first and last name, which returns 32 hexadecimal digits. SUBSTRING takes the first 8 of these, and CONV turns that into a 10 digit number (it's theoretically possible this won't be unique):

CONV(SUBSTRING(MD5(CONCAT(first_name,last_name)), 1, 8), 16, 10)

Then you divide that by a very big number and add it to the rate. You'll end up with a rate like 8.0000019351087950. You have to use FORMAT to avoid MySQL truncating the decimal places. This rate will now be unique for each first name and last name.

FORMAT(rate + CONV(SUBSTRING(MD5(CONCAT(first_name,last_name)), 1, 8), 16, 10)/1000000000000000, 16)

And then if you do the SUM DISTINCT over that it will only count the 8 once. Then you need to FLOOR the result to get rid of the extra decimal places:

FLOOR(SUM(DISTINCT FORMAT(rate + CONV(SUBSTRING(MD5(CONCAT(first_name,last_name)), 1, 8), 16, 10)/1000000000000000, 16)))

I found this approach while doing a much more complicated query which joined and grouped several tables. I'm still not sure if I'll use it as it is pretty horrible, but it does work. It's also 6 years too late to be of any use to the person who answered the question.