Returning SQL LIKE & GROUP BY results

Question

I have a pretty simple table in a database with 3 columns called people . Example:

user           id             notes
john           01             has red hair, last logged in 02/04/12
tony           02             has brown hair, last logged in 04/03/12
brian          03             has brown hair, last logged in 03/06/13
amanda         04             has blonde hair, last logged in 05/07/14
…

if I want to group by the notes field, and do a count, the 2nd and 3rd rows show as a count of 1 each because the logged in date is different; what I would like to do is chop off the date and amalgamate and count purely by hair colour, eg if I run a query:

SELECT `notes`, COUNT( `user` ) AS Count 
FROM `people` 
WHERE `notes` LIKE "%hair%" GROUP BY `notes`;

I get a result of:

+-----------------------------------------+-------+
|notes                                    | Count |
+-----------------------------------------+-------+
|has red hair, last logged in 02/04/12    |      1|
|has brown hair, last logged in 04/03/12  |      1|
|has brown hair, last logged in 03/06/13  |      1|
|has blonde hair, last logged in 05/07/14 |      1|
+-----------------------------------------+-------+

would like to achieve a result of:

+-------------------+------+
|notes              |Count | 
+-------------------+------+
|has red hair       |     1|
|has brown hair     |     2|
|has blonde hair    |     1|
+-------------------+------+

Is this at all possible?

Cheers.

Answer 1

You can do it using SUBSTRING_INDEX , like this:

SELECT SUBSTRING_INDEX(`notes`, ',', 1), COUNT( `user` ) AS Count
FROM `people`
WHERE `notes` LIKE "%hair%"
GROUP BY SUBSTRING_INDEX(`notes`, ',', 1)

Of course this is a somewhat dirty workaround. A better approach would be to separate out the hair color, but I realize that this may not always be possible.

Demo