SQL - Is it possible to make custom column by row values? - mysql

Question

Does anyone know if it's possible to turn something like the table below.. While still being able to use a SELECT query.
+----+-----------+---------------+------------+
| id | listingId | value | identifier |
+----+-----------+---------------+------------+
| 1 | 1a | Alaskan Husky | race |
| 2 | 1a | High | activity |
| 3 | 1a | White | colour |
| 4 | 1b | Akita | race |
| 5 | 1b | Medium | activty |
| 6 | 1b | Grey | colour |
+----+-----------+---------------+------------+
To something like this, while still being able to use a select query..
+----+-----------+---------------+---------+--------+
| id | listingId | race | activty | colour |
+----+-----------+---------------+---------+--------+
| 1 | 1a | Alaskan Husky | High | White |
| 2 | 1b | Akita | Medium | Grey |
+----+-----------+---------------+---------+--------+
I also want to be able to search this new "table". Let's say, the user has picked a filter with something like this:
Race: Alaskan Husky
Activity: Medium, High
It should then return the listingId of 1a.
The reason being for this, is that I can't do a proper SELECT query, when all the values are placed in different rows..
As you can see, the listingId should be the variable that groups it all together, and makes the value in the identifier column, a new column.
The reason why I don't just use the second table as default, is because each listing can have different filters and filter groups. And I need to be able to select specific listings that meet a user specified filter.
Thanks.

Answer 1

You can use conditional aggregation :
select min(id) id, listingId,
max(case when [identifier] = 'race' then [value] end) race,
max(case when [identifier] = 'activity' then [value] end) activity,
max(case when [identifier] = 'colour' then [value] end) colour
from table t
group by listingId;

Answer 2

This might do the trick, just a few self joins. I think there a few ways you could accomplish this but for me this is pretty easy to read and maintain.
CREATE TABLE [dbo].[the_table](
[id] [int] NULL,
[listingId] [varchar](50) NULL,
[value] [varchar](50) NULL,
[identifier] [varchar](50) NULL
) ON [PRIMARY]
GO
INSERT INTO the_table (id, listingId, value, identifier)
VALUES (1, '1a', 'alaskan huskey', 'race'),
(2, '1a', 'high', 'activity'),
(3, '1a', 'white', 'colour'),
(4, '1b', 'akita', 'race'),
(5, '1b', 'medium', 'activity'),
(6, '1b', 'grey', 'colour')
SELECT * FROM
(SELECT a.id, a.listingID, a.value AS race, b.value as activity, c.value as colour
FROM the_table a
INNER JOIN the_table b
ON a.listingId = b.listingId
INNER JOIN the_table c
ON a.listingId = c.listingId
WHERE a.identifier = 'race'
AND b.identifier = 'activity'
AND c.identifier = 'colour') AS t
WHERE t.colour = 'white'
Output:

Answer 3

You can achieve it using Pivot
DECLARE #MyTable TABLE (Id INT, listingId VARCHAR(20),Value VARCHAR(20), identifier varchar(20))
INSERT INTO #MyTable VALUES
(1,'1a','Alaskan Husky' ,'race'),
(2,'1a','High' ,'activity'),
(3,'1a','White' ,'colour'),
(4,'1b','Akita' ,'race'),
(5,'1b','Medium','activity'),
(6,'1b','Grey' ,'colour')
SELECT *
FROM
(
SELECT listingId, identifier,Value FROM #MyTable)t
PIVOT(MIN(Value)
FOR identifier
IN (race,activity,colour)
)p