Question

从一列唯一（非重复）值中选择一对先前未选择的不同随机值的最有效方法是什么？

我目前的方法是（将每对值保持在一个单独的“mytable_associations”表中）：

SELECT * FROM
(
 SELECT id,count(*) AS associations_count FROM mytable 
 INNER JOIN mytable_associations 
 WHERE (myvalue=myvalue1 OR myvalue=myvalue2) 
 GROUP BY myvalue 
 HAVING associations_count<(SELECT count(*) FROM mytable)-1
 ORDER BY rand() limit 1
) mytable1 
LEFT JOIN 
(SELECT myvalue AS myvalue2 FROM mytable) mytable2
ON mytable1.myvalue1<>mytable2.myvalue2
WHERE
(
 SELECT myvalue1 FROM mytable_associations 
 WHERE
 myvalue1=mytable1.myvalue1 AND myvalue2=mytable2.myvalue2
 OR
 myvalue1=mytable2.myvalue2 AND myvalue2=mytable1.myvalue1
) IS NULL;

（然后当然用这个新关联更新mytable_associations）

正如您所看到的，可以从一些优化中获益匪浅。

（抱歉代码中的缩进很差，我真的不知道如何缩进mysql命令。）

你能帮助我吗？

（P.S。这是我在这里发布的第一个问题：当然我做了很多错事，我明白随之而来的火焰，但请不要太难我;）

Answer 1

任何涉及order by rand()的解决方案效率都会很低。有关替代方案，请参阅：

我对Selecting random rows with MySQL
Jan Kneschke on Order by RAND()

要排除您已选择过的数字，请按以下步骤操作（这是伪代码）：

$c1 = SELECT COUNT(DISTINCT myvalue) FROM mytable
$c2 = SELECT COUNT(*) FROM mytable_associations

$offset = ROUND( RAND() * ($c1 * ($c1-1) - $c2) )

SELECT v.* FROM (
  SELECT LEAST(m1.myvalue,my2.myvalue) AS myvalue1,
    GREATEST(m1.myvalue,my2.myvalue) AS myvalue2
  FROM (SELECT DISTINCT myvalue FROM mytable) AS m1 
  INNER JOIN (SELECT DISTINCT myvalue FROM mytable) AS m2
    ON m1.myvalue <> m2.myvalue
) AS v
LEFT OUTER JOIN mytable_associations AS a USING (myvalue1,myvalue2)
WHERE a.myvalue1 IS NULL
LIMIT 1 OFFSET $offset

确保myvalue1＆lt; myvalue2，并在mytable_associations中按顺序存储它们，您可以简化连接。

mysql优化：从一列唯一值中选择一个先前未选择的随机不同值对

1 个答案: