Question

我想根据时间戳和用户ID更新几行所以我的数据看起来像这样

| INT(11) | TIMESTAMP        | VARCHAR(50) |
| user id | timestamp        | like id     |  
| 1       | 10/20/2014 10:30 |             |
| 1       | 10/20/2014 9:45  |             |  
| 1       | 10/20/2014 22:30 |             |
| 2       | 10/20/2014 9:45  |             |  
| 3       | 10/20/2014 10:30 |             |
| 3       | 10/20/2014 9:45  |             |

我想像这样更新具有相同用户ID和时间戳的行，如果在具有相同用户ID的任何其他行中的一小时（同一天）内有另一个条目，则timestamp匹配。

| user id | timestamp        | like id |  
| 1       | 10/20/2014 10:30 | ASDF-1  |
| 1       | 10/20/2014 9:45  | ASDF-1  |
| 1       | 10/20/2014 11:30 | ASDF-1  |
| 1       | 10/20/2014 21:45 | ADGG-2  |   
| 1       | 10/20/2014 22:30 | ADGG-2  |
| 2       | 10/20/2014 9:45  | GFDQ-3  |  
| 3       | 10/20/2014 10:30 | QWER-4  |
| 3       | 10/20/2014 9:45  | QWER-4  |

这可以在MYSQL更新语句中执行吗？我还没有想到办法解决这个问题。

编辑：我还想为每个分组的like_ids生成一个随机的50char字符串

Answer 1

这是一个返回所需内容的SQL查询：

select t.userid, t.timestamp,
       (@rn := if(@u = userid and @ts <= timestamp + interval 1 hour,
                  if(@ts := timestamp, @rn, @rn),
                  if(@u := userid, if(@ts := timestamp, @rn + 1, @rn + 1), @rn + 1)
                 )
       ) as likeid
from table t cross join
     (select @rn := 0, @u := 0, @ts := 0) vars
order by userid, timestamp;

嵌套if的所有奇怪语法都是将变量的赋值放在一个语句中。

然后您可以将其用于更新：

update t join
       (select t.userid, t.timestamp,
               (@rn := if(@u = userid and @ts <= timestamp + interval 1 hour,
                          if(@ts := timestamp, @rn, @rn),
                          if(@u := userid, if(@ts := timestamp, @rn + 1, @rn + 1), @rn + 1)
                         )
               ) as likeid
        from table t cross join
             (select @rn := 0, @u := 0, @ts := 0) vars
        order by userid, timestamp
       ) tnew
       on t.userid = tnew.userid and t.timestamp = tnew.timestamp
    set t.likeid = tnew.likeid;

根据时间戳更新行

1 个答案: