SQL返回重复的行

Question

I've been developing an extremely basic feed system within my website which (is supposed to) display all the posts where:

1. The posts are not from the logged in user
2. They are all from friends
3. And are ordered by post_id (to get the most recent ones first)

One problem I am having with my SQL though, is that when querying the database tables to grab the rows, it will return one post, but it'll be posted by all users, instead of just the one person that created it.

For example, if user 1 posted the status "Hi" , and there are 5 users in the database; it'll grab that post, then print it to the page 5 times, along with each individual user as the poster. (See below for image example)

SQL :

SELECT posts.*, users.*, friends.* 
FROM posts, users, friends
WHERE posts.poster <> ? 
AND users.user_id = posts.poster
AND friends.areFriends = 1 AND (friends.userA = ? AND friends.userB = posts.poster) 
OR friends.areFriends = 1 AND (friends.userA = posts.poster AND friends.userB = ?)
ORDER BY posts.post_id desc LIMIT 20

For reference:

posts.poster is the person who created the post
users.user_id is the id of the same person friends.areFriends 0 = not a friend, 1 = friend friends.userA / friends.userB is the logged in user or the friend that that user is friends with (if that makes sense)

I have tried every possible combination of things that I know to fix the problem, however I'm still new to "advanced" SQL and can't seem to find anything

I've attempted to explain my problem as best I can, so hopefully someone can help me. I've been trying for the last couple of days to fix it but to no avail,

Cheers.

Answer 1

It should be because of your where conditions. Change it like this:

WHERE posts.poster <> ? 
AND users.user_id = posts.poster
AND friends.areFriends = 1 
AND ((friends.userA = ? AND friends.userB = posts.poster) OR (friends.userA = posts.poster AND friends.userB = ?))

As you wrote the conditions, it will return the rows that fulfill all other conditions OR friends.areFriends = 1 AND (friends.userA = posts.poster AND friends.userB = ?) . So even the rows that don't fulfill users.user_id = posts.poster will be returned. But as I wrote the conditions, the OR is enclosed in a parenthesis.

Answer 2

A classic way of making inner joins is more readable and helps you to understand why the code does not work. Also, it is kind of code convention for SQL, so it also helps other people (like me) to understand you.

I have re-wrote your code to use INNER JOIN . Try this out.

SELECT @userId := ?;

SELECT posts.*, users.*, friends.* 
FROM friends
JOIN users ON users.user_id = 
    CASE friends.userA
    WHEN @userId
    THEN friends.userB
    ELSE friends.userA END
JOIN posts ON posts.poster = users.user_id
WHERE friends.userA = @userId OR friends.userB = @userId
ORDER BY posts.post_id DESC LIMIT 20;

Make sure that there are no dupes in friends table. Following query will help you with this:

SELECT User1, User2, COUNT(*) as DupeCount
FROM 
(
    SELECT
        CASE WHEN friends.userA < friends.userB THEN friends.userA ELSE friends.userB END AS User1,
        CASE WHEN friends.userA < friends.userB THEN friends.userB ELSE friends.userA END AS User2
    FROM friends
) AS F
GROUP BY User1, User2
HAVING COUNT(*) > 1

Answer 3

I assume the query must return posts of the friends of the user ? . First find friends, the their posts, then their other data.

SELECT  pp.*, uu.* 
FROM     (select distinct case when ? = friends.userB then friends.userA else friends.userB end as user_id
         from friends 
         where ? in (friends.userA, friends.userB)
         and friends.areFriends = 1 
         ) f
join posts pp on pp.poster = f.user_id
join users uu on uu.user_id = pp.poster