Reputation: 447
I'm having an issue figuring the mysql to find links that have two specific 'tags' and the same 'hashid' when doing a JOIN of two tables
Assume my tables look like this:
Links
md5 url title numberofsaves
-----------------------------------------
a0a0 google.com foo 200
b1b1 yahoo.com yahoo 100
Tags
md5 tag
---------------
a0a0 awesome
a0a0 useful
a0a0 cool
b1b1 useful
b1b1 boring
I want to return rows that have tags of BOTH 'useful' and 'awesome'
The current (working/ fast) query for finding links by 1 tag:
SELECT links.title, links.numsaves FROM links LEFT JOIN tags ON links.md5=tags.md5 WHERE tags.tag = 'useful' ORDER BY links.numberofsaves DESC LIMIT 20
After reading an article I tried to use the following:
SELECT links.title, links.numsaves FROM links LEFT JOIN tags ON links.md5=tags.md5 GROUP BY tags.md5 HAVING SUM(tags.tag='useful') AND SUM(tags.tag='awesome') ORDER BY links.numberofsaves DESC LIMIT 20
This does work but it is so unbelievably slow as to be unusable.
Anyone know the solution?
Upvotes: 4
Views: 1854
Reputation: 263693
The type of problem is called Relational Division
SELECT a.md5,
a.url,
a.title
FROM Links a
INNER JOIN Tags b
ON a.md5 = b.md5
WHERE b.Tag IN ('awesome', 'useful') -- <<== list of desired tags
GROUP BY a.md5, a.url, a.title
HAVING COUNT(*) = 2 -- <<== number of tags defined
OUTPUT
╔══════╦════════════╦═══════╗
║ MD5 ║ URL ║ TITLE ║
╠══════╬════════════╬═══════╣
║ a0a0 ║ google.com ║ foo ║
╚══════╩════════════╩═══════╝
Upvotes: 10