创建新的索引时清理数据库

Question

我正在创建两个字段的新唯一索引：

ALTER TABLE `tools`
ADD UNIQUE INDEX `name_url` (`name`, `url`);

问题是db已经有500000个工具，其他表引用它。删除双打的最佳策略是什么？

Answer 1

要更新引用此表的另一个表，请执行：

UPDATE otherTable AS o
JOIN tools AS t1 on o.tools_id = t1.id
JOIN (SELECT name, url, MIN(id) AS minid
      FROM tools
      GROUP BY name, url) AS t2 
ON t1.name = t2.name AND t1.url = t2.url AND t1.id != t2.minid
SET o.tools_id = t2.minid

这会将所有外键设置为指向每个(name, url)对的最小ID。您必须为外键指向tools的每个其他表执行此操作。

然后删除所有额外的行：

DELETE t1.* FROM tools AS t1
JOIN (SELECT name, url, MIN(id) AS minid
      FROM tools
      GROUP BY name, url) AS t2 
ON t1.name = t2.name AND t1.url = t2.url AND t1.id != t2.minid

完成所有这些操作后，您可以添加唯一索引。