What's wrong with my group_concat SQL query? - mysql

Question

I have three tables:
CREATE TABLE `b10g_entries` (
`id` bigint(20) unsigned NOT NULL AUTO_INCREMENT,
`permalink` text NOT NULL,
`title` varchar(300) NOT NULL,
`fullcontent` text NOT NULL,
`introcontent` text NOT NULL,
`dateadded` datetime NOT NULL,
`lastedited` datetime NOT NULL,
`author` varchar(40) NOT NULL,
`comments` int(11) NOT NULL DEFAULT '0',
`published` tinyint(1) unsigned NOT NULL DEFAULT '0',
PRIMARY KEY (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=299 DEFAULT CHARSET=utf8
CREATE TABLE `b10g_tag_map` (
`id` bigint(20) NOT NULL AUTO_INCREMENT,
`tag_id` bigint(20) unsigned DEFAULT NULL,
`entry_id` bigint(20) unsigned DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=8 DEFAULT CHARSET=utf8
CREATE TABLE `b10g_tags` (
`id` bigint(20) NOT NULL AUTO_INCREMENT,
`name` text NOT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=5 DEFAULT CHARSET=utf8
And i'm trying to get first 25 blog entries with their tags (that's why I use many-to-many relationship) using this query:
SELECT b10g_entries.*, GROUP_CONCAT( b10g_tags.name SEPARATOR ', ')
AS tags FROM b10g_entries
LEFT JOIN b10g_tag_map ON b10g_entries.id = b10g_tag_map.entry_id
LEFT JOIN b10g_tags ON b10g_tag_map.tag_id = b10g_tags.id LIMIT 0, 25;
But I only get one record back. What's wrong with this query?

Answer 1

Add a GROUP BY clause.
Now, you're getting a list of ALL tags found anywhere in the set. Instead, you only want the ones within the group (by entry).
SELECT b10g_entries.*, GROUP_CONCAT( b10g_tags.name SEPARATOR ', ')
AS tags FROM b10g_entries
LEFT JOIN b10g_tag_map ON b10g_entries.id = b10g_tag_map.entry_id
LEFT JOIN b10g_tags ON b10g_tag_map.tag_id = b10g_tags.id
GROUP BY b10g_entries.id

Answer 2

You have a GROUP_CONCAT() aggregate function, but have not used a GROUP BY clause, so your result will be one row.
Note that in MySQL it is permissible to use a GROUP BY with only one column specified while many more appear in the SELECT list, but that is not portable to other RDBMS. So instead, I have joined b10g_entries in a second time to connect all the other columns from that table, while only using the id in the GROUP BY.
SELECT
b10g_entries_all.*,
GROUP_CONCAT( b10g_tags.name SEPARATOR ', ') AS tags
FROM
/* Main table, used gor GROUP BY aggregate */
b10g_entries
/* self join to pull in other columns without needing to put them in GROUP BY */
JOIN b10g_entries b10g_entries_all ON b10g_entries.id = b10g_entries_all.id
LEFT JOIN b10g_tag_map ON b10g_entries.id = b10g_tag_map.entry_id
LEFT JOIN b10g_tags ON b10g_tag_map.tag_id = b10g_tags.id
/* group on the entry id */
GROUP BY b10g_entries.id
LIMIT 0, 25;