我有一个表,其中包含有关用户登录的信息。我想对 last 个重复记录进行分组。例如:
+---+------------+-------------+-------------+------------------+
| | ip | platform | browser | date |
+---+------------+-------------+-------------+------------------+
| 1 | 127.0.0.1 | Windows | Chrome | 2018-01-01 00:00 |
| 2 | 127.0.0.1 | Windows | Chrome | 2018-01-02 00:00 |
| 3 | 10.0.0.1 | Linux | Firefox | 2018-01-03 00:00 |
| 4 | 127.0.0.1 | Windows | Chrome | 2018-01-04 00:00 |
+---+------------+-------------+-------------+------------------+
将输出:
+-----+------------+-------------+-------------+-------------+
| | ip | platform | browser | num_records |
+-----+------------+-------------+-------------+-------------+
| 1-2 | 127.0.0.1 | Windows | Chrome | 2 |
| 3 | 10.0.0.1 | Linux | Firefox | 1 |
| 4 | 127.0.0.1 | Windows | Chrome | 1 |
+-----+------------+-------------+-------------+-------------+
(为简单起见,我发出了日期,应该有id之类的日期范围)
请注意,id 1,2,4
是相同的,但是1,2
和4
由于时间轴而被分开分组(还有另一条记录将它们分开)。
要查找重复项,我应该考虑以下几列:ip, platform, browser
。如果某些内容与这些列不同,那么它就不是重复的内容。
我可以做到:
SELECT ip, platform, browser, COUNT(1) AS num_records
FROM users_logins
WHERE user_id = 1
GROUP BY ip, platform, browser
但这将对所有记录进行分组,而无需考虑时间轴。
答案 0 :(得分:1)
这是一个孤岛问题。在MySQL 8+中,您可以使用行号的不同之处:
select ip, platform, browser,
count(*) as numrecords,
min(id), max(id),
min(date), max(date)
from (select t.*,
row_number() over (order by date) as seqnum,
row_number() over (partition by ip, platform, browser order by date) as seqnum_2
from t
) t
group by ip, platform, browser, (seqnum - seqnum_2)
order by min(date) desc;