CODEWITHSUNDEEP
mysqlsql
Tomas Reimers
Tomas Reimers

Reputation: 3292

MySQL - Relational Databases and Tagging

I have been scratching my head on this one for a while, so I figured I'd ask stack overflow (Note: I am an SQL novice trying to learn more SQL, so please be respectful and explanatory):

I have one sql table that looks like this called "posts":

id | user
--------------------------------
0  | tim
1  | tim
2  | bob

And another called "tags" that stores the tags on the posts (in the "posts" table) in text:

id | postID | tag
--------------------------------
0  | 0      | php
1  | 2      | php
2  | 0      | mysql
3  | 1      | mysql
4  | 1      | sql
5  | 3      | perl

(To clarify, the concept where: id=0 is tagged php,mysql; id=1 is tagged sql,mysql; id=2 is tagged php; id=3 is tagged perl.)

How could I write a WHERE statement to get posts tagged x, but not y (x and y will be defined by php)?

For example, how could I get all posts tagged mysql but not php?

EDIT

Could you also explain how to add multiple tags to search for (for example get all tagged mysql and recursion but not php)

Upvotes: 3

Views: 172

Answers (3)

Tomas
Tomas

Reputation: 59495

select *
from
    (select distinct postID
    from tags
    where tag = "mysql") as t1
left join
    (select distinct postID
     from tags
     where tag = "php") as t2
using (postID)
where t2.postID is NULL

example 2: get all tagged mysql and recursion but not php:

select *
from
    ((select distinct postID
    from tags
    where tag = "mysql") as t1
join
    (select distinct postID
    from tags
    where tag = "recursion") as t3
using (postID))
left join
    (select distinct postID
     from tags
     where tag = "php") as t2
using (postID)
where t2.postID is NULL

Upvotes: 2

Cade Roux
Cade Roux

Reputation: 89661

Naively, simply and portably:

SELECT *
FROM posts
WHERE EXISTS (SELECT * FROM tags WHERE tags.postid = posts.id AND tags.tag = 'x')
AND NOT EXISTS (SELECT * FROM tags WHERE tags.postid = posts.id AND tags.tag = 'y')

Now depending upon the execution plan, you can do other things to optimize this.

Upvotes: 1

Vincent Savard
Vincent Savard

Reputation: 35927

My suggestion with EXISTS :

SELECT DISTINCT postID
FROM tags t1
WHERE EXISTS(SELECT NULL
             FROM tag t2
             WHERE t2.id = t1.id
               AND t2.tag = 'x')
  AND NOT EXISTS(SELECT NULL
                   FROM tag t2
                  WHERE t2.id = t1.id
                    AND t2.tag = 'y')

Upvotes: 1

Related Questions