Question

Purpose:
Every half-hour I query database to find a post that havn’t published to publish,every time I query by different category,if no posts in this category,switch to next category.
The if check done by SQL

pseudo-code

SELECT FROM post WHERE post.cat_ID='7' AND post_date > '$last_query_time_cat_ID_7' 
IF NO_POSTS_MATCH SELECT FROM post WHERE post.cat_ID='8' AND post_date>'$last_query_time_cat_ID_8'
...

pseudo explain:select one post of one category by its last query time of this category,if no new post of this category select from another category by its last query time.
My question:
Is there such SQL to do this work?
This can be done by php to check if no posts return,then make another sql query.This could make many queries.
Somebody show me the sql-pseudo-code of this problem?

Answer 1

Could look like this:

SELECT p.post_ID, p.cat_ID, p.post_date
FROM   post p
JOIN   (
    SELECT cat_ID
          ,max(query_time) AS last_query
    FROM   post 
    WHERE  cat_ID >= 7
    GROUP  BY 1
    ORDER  BY 1) l USING (cat_ID)
-- in some RDBMS you need ON l.cat_ID = p.cat_ID in the JOIN clause
WHERE  p.post_date > l.last_query
ORDER  BY p.cat_ID, p.post_date DESC
LIMIT  1;  -- In T-SQL you need to write "SELECT TOP 1" instead

Major points

• First get the timestamp of the last query per category in a subquery. We are only interested in cat_ID >= 7

• Join this to the base table and filter only rows with a later post_date than last_query per cat_ID.

• Sort it by cat_ID – you seem to want smallest cat_ID first. Secondary sort by post_date descending, so the last post comes first.

• If there are none in cat_ID 7, cat_ID 8 will be first etc. Voila.