Question

I have a rather large mysql table that has the following structure:

fieldid | recordid | content
----------------------------
 10     |   01     | Google
 10     |   02     | Yahoo
 10     |   03     | Facebook
 20     |   01     | google.com
 20     |   02     | yahoo.com
 20     |   03     | facebook.com
 30     |   01     | Big search engine
 30     |   02     | Other search engine
 30     |   03     | Largest Social Network
 40     |   01     | Search engine
 40     |   02     | Search engine
 40     |   03     | Social Network

Where fieldid is the id for the field (in this case ’10 is the ‘site name’, ’20’ is the ‘site url’, ’30’ is the ‘site description’ and ’40’ is the ‘category’).
Also, recordid is the id for all the fields in a given website (in this case 01 is Google, 02 is Yahoo and 03 is Facebook).

I would like to combine them into the equivalent of a result set (associative array) filtered by a particular ‘category’ such as:

site            |     url            |     description
----------------|------------------------------------------
Google          | google.com         | Big search engine
Yahoo           | yahoo.com          | Other search engine

Where in this case they are filtered by “Search engine”.

I tried

SELECT
recordid,
GROUP_CONCAT( if( fieldID = 10, content, NULL ) ) AS 'site',
GROUP_CONCAT( if( fieldID = 20, content, NULL ) ) AS 'url',
GROUP_CONCAT( if( fieldID = 30, content, NULL ) ) AS 'description',
GROUP_CONCAT( if( fieldID = 40 and content = 'Search engine', content, NULL ) ) AS 'category'
FROM my_table
GROUP BY recordid
HAVING Category IS NOT NULL

Now, although this works, it is very, very slow because there are far more fieldids than the ones I need to process. Does anyone know of a better way to achieve the same result?

Thanks in advance.

Answer 1

I’d use cascading self joins. Something like:

SELECT t.content AS 'Site', t2.content AS 'URL', t3.content AS 'Description', t4.content AS 'Category'
FROM my_table AS t
LEFT JOIN my_table AS t2 ON (t2.recordid = t.recordid AND t2.fieldid = 20)
LEFT JOIN my_table AS t3 ON (t3.recordid = t.recordid AND t3.fieldid = 30)
LEFT JOIN my_table AS t4 ON (t4.recordid = t.recordid AND t4.fieldid = 40 AND t4.content = 'Search engine')
WHERE t.fieldid = 10

Since the WHERE condition is going to significantly reduce the resultset, the query should perform faster, assuming you have appropriate indexes on (fieldid) and (fieldid, recordid).