检查重复记录
-- 检查重复code1 select count(identity) num, identity from event_log where code='code1' group by identity having count(identity) > 1 order by num desc
删除重复记录
DELETE FROM event_log WHERE `code`='code1' AND identity IN ( SELECT identity from ( SELECT identity FROM event_log WHERE code='code1' GROUP BY identity HAVING count(identity) > 1 ) a ) AND id NOT IN ( SELECT keepId FROM ( SELECT min(id) keepId FROM event_log WHERE code='code1' GROUP BY identity HAVING count(identity) > 1 ) b )
其中 a 和 b 两个中间表的作用是, 避免执行时出现 You can't specify target table 'xxxxx' for update in FROM clause 错误
分组按时间正序取第一条记录, 巧妙地使用了not exists
select d.* from t_charge d where not exists (select 1 from t_charge where user_id = d.user_id and created_at < d.created_at)
按时间倒序则是
select f.* from t_charge f where not exists (select 1 from t_charge where user_id = f.user_id and created_at > f.created_at)