Question

Assume I have a table that can have 0 or more tags:

TABLE: foo
- foo_id (PK)

TABLE: tag
- tag_id (PK)
- name

TABLE: foo_tag
- foo_tag_id (PK)
- foo_id (FK)
- tag_id (FK)

Questions:

1. The above is the most common and simplest way to implement a tagging system that I have seen. I do realize that there are potential problems with tagging systems in terms of scalability. Am I going to have that here?
2. Is there a way to do something like this in PHP or SQL: insert a new tag “bar”. If “bar” does not exist, add it to the tag table and return the last inserted id. If “bar” does exist, do not add it and return the tag_id for it.
3. Is there a “proper” DB term for such a table describe in #2?

Answer 1

Please see http://dev.mysql.com/doc/refman/4.1/en/insert-on-duplicate.html for a workaround to get the last ID on update:

INSERT INTO table (a,b,c) VALUES (1,2,3)
  ON DUPLICATE KEY UPDATE id=LAST_INSERT_ID(id), c=3;

This will update the last insert id to the value of the updated value. In your case:

INSERT INTO tag (name) VALUES ('bar')
  ON DUPLICATE KEY UPDATE tag_id=LAST_INSERT_ID(tag_id);

tag.name needs to be a unique key.