How to use GROUP BY in MySQL subquery - mysql

Question

I'm using phpMyAdmin for submitting queries. When using GROUP BY in subquery the whole application just hangs without errors until I restart the browser.
I have three tables: files stores information about uploaded files, file_category defines the available categories for files and file_category_r stores relations between files and categories.
I want to count how many files each category has, but some files can have multiple entries in the files table, so I need to group them by files.filename.
I tried two different approaches, both resulting in a hang:
SELECT
fc.*,
(SELECT COUNT(*) FROM file_category_r
WHERE file_category_r.category_id = fc.id
AND file_category_r.file_id IN
(SELECT f2.id FROM
(SELECT * FROM files f3 GROUP BY f3.filename) f2
WHERE f2.mandant_id = 1)
) as file_count
FROM file_category fc ORDER BY name ASC
or
SELECT
fc.*,
(SELECT COUNT(*) FROM file_category_r
WHERE file_category_r.category_id = fc.id
AND file_category_r.file_id IN
(SELECT id FROM files WHERE mandant_id = 1 GROUP BY filename)
) as file_count
FROM file_category fc ORDER BY name ASC
I don't see a problem with my queries, running the subquery alone works ok. Even removing the GROUP BY return the result, but the result is wrong, because it's counting duplicate values.
Here is the table schema:
CREATE TABLE IF NOT EXISTS `files` (
`id` bigint(20) unsigned NOT NULL,
`project_id` bigint(20) unsigned DEFAULT NULL,
`customer_id` bigint(20) unsigned DEFAULT NULL,
`opportunity_id` int(11) DEFAULT NULL,
`task_id` bigint(20) unsigned DEFAULT NULL,
`calendar_event_id` bigint(20) unsigned DEFAULT NULL,
`mandant_id` tinyint(4) DEFAULT NULL,
`time` timestamp NULL DEFAULT CURRENT_TIMESTAMP,
`size` float NOT NULL,
`mime_type` varchar(100) NOT NULL,
`filename` text NOT NULL,
`file` longblob NOT NULL,
`folder_id` int(11) DEFAULT NULL,
`user_id` int(11) DEFAULT NULL,
`is_public` tinyint(1) unsigned NOT NULL DEFAULT '0',
`description` text,
`file_link` varchar(500) DEFAULT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8 AUTO_INCREMENT=104832 ;
CREATE TABLE IF NOT EXISTS `file_category` (
`id` int(11) NOT NULL,
`name` varchar(200) NOT NULL,
`parent` int(11) DEFAULT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8 AUTO_INCREMENT=445 ;
CREATE TABLE IF NOT EXISTS `file_category_r` (
`id` bigint(20) unsigned NOT NULL,
`file_id` bigint(20) unsigned NOT NULL,
`category_id` int(11) NOT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8 AUTO_INCREMENT=300346 ;
What am I doing wrong? The tables are quite big, is it possible the request is too heavy? I'm out of ideas, please help! Thanks!

Answer 1

select fc.name, count(*)
from file_category fc
inner join file_category_r fcr on fc.id = fcr.category_id
group by fc.name
Not quite sure about that "some files can have multiple entries in the files table, so I need to group them by files.filename", though.
Maybe you need something like
select fc.name, count(distinct f.filename)
from file_category fc
inner join file_category_r fcr on fc.id = fcr.category_id
inner join files f on fcr.file_id = f.id
group by fc.name

Answer 2

Often, the use of in can result in inefficient query plans. You can try exists instead:
SELECT fc.*,
(SELECT COUNT(*)
FROM file_category_r fcr
WHERE fcr.category_id = fc.id AND
exists (select 1 from files f where f.mandant_id = 1 and fcr.file_id = f.id)
) as file_count
FROM file_category fc
ORDER BY name ASC;
Now, you should add indexes. Start with file_category_r(category_id, file_id) and files(id, mandant_id).

Answer 3

I use heidisql, not phpmyadmin and your query works fine here. Maybe phpmyadmin has got problems to parse your query.
edit: also, there is a limit for query length. if your "in"-statement is to long, mysql will return an error that phpmyadmin should return.
but if phpmyadmin hangs, i'd try to execute your query my mysqlc or another mysql client like heidisql.