How do you handle complex data structures (OOP) in a mySQL DB?

Answer 1

Members:

id    (autoinc)
name
password
theme_id

Posts:

id    (autoinc)
member_id
title
date

Tags:

id    (autoinc)
name

Tag_Relations:

tag_id
post_id

Posts is your "array" of posts, with the member_id column linking each post to its user. Tags is your "array" of tags, Tag Relations links each tag to one or more posts.

Here is an example of how you could get all posts & tags for a user with one query:

SELECT Members.name, Posts.title, Tag_Relations.item_id, Tags.name 
FROM Members LEFT 
JOIN Posts ON Members.id = Posts.member_id 
LEFT JOIN Tag_Relations ON Tag_Relations.post_id = Posts.id 
LEFT JOIN Tags ON Tags.id = Tag_Relations.tag_id 
WHERE Members.id = 2779;

+----------+-----------------------------------+------------+---------+
| name     | title                             | item_recid | name    |
+----------+-----------------------------------+------------+---------+
| Mike     | One Post's Title                  |        973 | Houses! | 
| Mike     | One Post's Title                  |        973 | Cars    | 
| Mike     | One Post's Title                  |        973 | Hats    | 
| Mike     | Another Post's Title              |        973 | Cars    | 
| Mike     | Yet another post                  |        975 | Homes   | 
| Mike     | Guess what?!                      |        976 | Houses! | 
| Mike     | Another one :)                    |        977 | Noses   | 
| Mike     | Another one :)                    |        977 | Mouth   | 
| Mike     | Another one :)                    |        977 | Head    | 
| Mike     | Another one :)                    |        977 | Knees   | 
+----------+-----------------------------------+------------+---------+

Answer 2

Learn about normal forms (several good tutorials online including this one). Database engines are extremely efficient in doing JOIN operations between flat tables that have been indexed appropriately.

The basic idea is that you identify the entities in your database (e.g. the users/posts/themes you mentioned), and the relations between them (one-to-one, one-to-many, or many-to-many). This allows you to split your data up into flat tables which can be efficiently reassembled.