Question

I have two tables: User and User2Friend. I need to get user’s friends. Query should handle two columns as FriendId: FriendId and UserId because we don’t know who added whom. In result both are friends.

The query:

SELECT DISTINCT
    `Id`
    , `UserName`
FROM `User`
WHERE `Id` IN
    (SELECT
        IF(`UserId` = 25, `FriendId`, `UserId`) -- This important part - friends are always mutual
    FROM `User2Friend`
    WHERE `UserId` = 25
    OR `FriendId` = 25)
LIMIT 10

The schema:

CREATE TABLE `User` (
  `Id` int(10) unsigned NOT NULL AUTO_INCREMENT,
  `UserName` varchar(16) DEFAULT NULL,
  PRIMARY KEY (`Id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;
CREATE TABLE `User2Friend` (
  `Id` int(11) unsigned NOT NULL AUTO_INCREMENT,
  `UserId` int(10) unsigned NOT NULL,
  `FriendId` int(10) unsigned NOT NULL,
  PRIMARY KEY (`Id`),
  UNIQUE KEY `UserIdFriendId` (`UserId`, `FriendId`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;

If user A added user B as friend, user B appears in user A friend list, after that user B must have user A in friend list too. I don’t really like subquery approach because I cannot use LIMIT in sub-select.

I’m expecting few millions of users, so performance is important.

Do you think it would be better to split this query to two separate ones?

Or maybe I need to have tricky JOIN?

Answer 1

Okay! I have split it to two queries…