随机且唯一的Mysql选择查询

Question

My Mysql database is filled with about 90 image links. However the ID of each row is not really chronological. For example initially it goes from 6-20 then theres a gap where I deleted some. Then it goes from 80 - 120 then another gap ect.

It is a hot or not website. I want the user to be able to go through all 90 images without having to view the same one twice. However at the moment every so often this happens. The same image comes up twice or even three times.

Here is my select query right now:

$sql = mysql_query("SELECT * FROM `table` WHERE id >= (SELECT FLOOR( MAX(id) * RAND()) FROM `content`)  ORDER BY rand() LIMIT 1");

Does anyone know any better solutions to this problem?

Answer 1

You could store the id's of the 'things' that the user has seen, and exclude them in all the next pages (either by a join or a not in). In that way you can always get a random post and not be stuck with the order that the rows have in the database (and what if the user returns later on?).

Answer 2

如何从数据库中获取所有索引，对其进行混排，创建列表，然后将它们从一页传递到另一页，显示最上面的索引并将其从列表中删除，该怎么办？

Answer 3

The problem that you are facing is one with rand() itself. Each time you call it, it produces a different order. Random is random, so a given row could occur twice in a row. What you are really looking for is a random permutation.

One approach is random with a seed followed by limit with offset:

select *
from table t
order by rand(<seed value here>)
limit <offset>, 1;

The two things you need are a seed value. You can do this by just keeping an enumerated value for each user, or using the user id, or something that is an integer and a constant for each user.

The other thing you need is the offset. This is important, because when you get the second item, you need to be sure that the ordering is the same as for the first item.

Another approach is to fetch all the ids in random order the first time and to stash them in the application. You can do this using group_concat() :

select group_concat(id order by rand())
from table t;

Then go through this list instead of returning to the database.