计算用户唯一的开始结束时间集合

我需要以下情况的帮助。计算用户唯一的开始结束时间集合

此表在SQL Server 2005上。如果您觉得2008年的某些功能会有帮助，我们也可以访问SQL Server 2008。尽管如此，我可以做到这一点。 :)

我有一个系统100多个用户的登录时间。他们可以同时登录到不同的频道。

我需要有端到端登录注销时候，不论通道

的

我猜，这可以通过一个存储过程来完成。但如果它没有达到性能，我可以接受任何其他选择。

以下AgentResults2是我的基础表。

AgentOutcome表是我想要的样子。

基表

FUser_id   FLoginDt     FLogoutDt 
30001    2012-02-17 12:32:15.473 2012-02-17 14:15:36.547 
30001    2012-02-17 12:49:48.177 2012-02-17 14:10:01.097 
30001    2012-02-17 14:30:49.293 2012-02-17 15:41:02.387

该用户登录到一个信道以12.32呆在那里直到14:15此期间，他也登录到另一个channelat 12点49分，在那里呆直到14内的一个例子： 10。并于14:30至15:41进行了新的登录。我想从这里得到两行是

FUser_id   FLoginDt     FLogoutDt 
30001    2012-02-17 12:32:15.473 2012-02-17 14:15:36.547 
30001    2012-02-17 14:30:49.293 2012-02-17 15:41:02.387

，当你在用户39395

SET DATEFORMAT DMY 

IF OBJECT_ID('#AgentResults2', 'U') IS NOT NULL DROP TABLE #AgentResults2; 

CREATE TABLE #AgentResults2 (FUser_id varchar(48), FLoginDt datetime, FLogoutDt DateTime) 

Insert Into #AgentResults2 (FUser_id , FLoginDt, FLogoutDt) 
Select '30001','17/02/2012 8:09:23.117 AM' ,'17/02/2012 8:39:28.527 AM' UNION ALL 
Select '30001','17/02/2012 8:50:26.087 AM','17/02/2012 9:31:32.040 AM' UNION ALL 
Select '30001','17/02/2012 10:28:42.430 AM','17/02/2012 10:54:13.880 AM' UNION ALL 
Select '30001','17/02/2012 10:59:20.567 AM','17/02/2012 12:00:36.030 PM' UNION ALL 
Select '30001','17/02/2012 12:32:15.473 PM','17/02/2012 2:15:36.547 PM' UNION ALL 
Select '30001','17/02/2012 12:49:48.177 PM','17/02/2012 2:10:01.097 PM' UNION ALL 
Select '30001','17/02/2012 2:30:49.293 PM','17/02/2012 3:41:02.387 PM' UNION ALL 
Select '30001','17/02/2012 3:44:05.800 PM','17/02/2012 4:01:33.613 PM' UNION ALL 
Select '39300','17/02/2012 8:06:31.250 AM','17/02/2012 3:51:31.930 PM' UNION ALL 
Select '39300','17/02/2012 10:15:08.923 AM','17/02/2012 10:21:29.833 AM' UNION ALL 
Select '39363','17/02/2012 9:58:44.287 AM','17/02/2012 11:20:08.950 AM' UNION ALL 
Select '39363','17/02/2012 11:20:29.203 AM','17/02/2012 1:27:36.717 PM' UNION ALL 
Select '39363','17/02/2012 1:27:36.717 PM','17/02/2012 3:05:12.383 PM' UNION ALL 
Select '39363','17/02/2012 3:05:31.527 PM','17/02/2012 4:57:13.733 PM' UNION ALL 
Select '39363','17/02/2012 4:57:44.640 PM','17/02/2012 5:59:20.273 PM' UNION ALL 
Select '39395','17/02/2012 9:37:14.353 AM','17/02/2012 10:15:52.397 AM' UNION ALL 
Select '39395','17/02/2012 10:15:28.427 AM','17/02/2012 10:58:47.080 AM' UNION ALL 
Select '39395','17/02/2012 10:57:03.590 AM','17/02/2012 11:53:47.933 AM' UNION ALL 
Select '39395','17/02/2012 11:51:31.567 AM','17/02/2012 2:26:27.640 PM' UNION ALL 
Select '39395','17/02/2012 2:31:30.247 PM','17/02/2012 4:04:44.217 PM' UNION ALL 
Select '39395','17/02/2012 4:10:49.013 PM','17/02/2012 4:11:26.983 PM' UNION ALL 
Select '39395','17/02/2012 4:17:16.813 PM','17/02/2012 5:55:47.187 PM' UNION ALL 
Select '39395','17/02/2012 4:55:25.900 PM','17/02/2012 5:26:07.310 PM' ; 

Select * from #AgentResults2; 

IF OBJECT_ID('#AgentOutcome', 'U') IS NOT NULL DROP TABLE #AgentOutcome; 

CREATE TABLE #AgentOutcome (FUser_id varchar(48), FLoginDt datetime, FLogoutDt DateTime) 

Insert Into #AgentOutcome (FUser_id , FLoginDt, FLogoutDt) 

Select '30001','17/02/2012 8:09:23.117 AM','17/02/2012 8:39:28.527 AM' UNION ALL 
Select '30001','17/02/2012 8:50:26.087 AM','17/02/2012 9:31:32.040 AM' UNION ALL 
Select '30001','17/02/2012 10:28:42.430 AM','17/02/2012 10:54:13.880 AM' UNION ALL 
Select '30001','17/02/2012 10:59:20.567 AM','17/02/2012 12:00:36.030 PM' UNION ALL 
Select '30001','17/02/2012 12:32:15.473 PM','17/02/2012 2:15:36.547 PM' UNION ALL 
Select '30001','17/02/2012 2:30:49.293 PM','17/02/2012 3:41:02.387 PM' UNION ALL 
Select '30001','17/02/2012 3:44:05.800 PM','17/02/2012 4:01:33.613 PM' UNION ALL 
Select '39300','17/02/2012 8:06:31.250 AM','17/02/2012 3:51:31.930 PM' UNION ALL 
Select '39363','17/02/2012 9:58:44.287 AM','17/02/2012 11:20:08.950 AM' UNION ALL 
Select '39363','17/02/2012 11:20:29.203 AM','17/02/2012 3:05:12.383 PM' UNION ALL 
Select '39363','17/02/2012 3:05:31.527 PM','17/02/2012 4:57:13.733 PM' UNION ALL 
Select '39363','17/02/2012 4:57:44.640 PM','17/02/2012 5:59:20.273 PM' UNION ALL 
Select '39395','17/02/2012 9:37:14.353 AM','17/02/2012 2:26:27.640 PM' UNION ALL 
Select '39395','17/02/2012 2:31:30.247 PM','17/02/2012 4:04:44.217 PM' UNION ALL 
Select '39395','17/02/2012 4:10:49.013 PM','17/02/2012 4:11:26.983 PM' UNION ALL 
Select '39395','17/02/2012 4:17:16.813 PM','17/02/2012 5:55:47.187 PM' ; 

Select * from #AgentOutcome; 

DROP TABLE #AgentResults2; 
DROP TABLE #AgentOutcome;

我已经检查了其他相关的主题，他们接近，但更具体的聚合就变得复杂并满足min开始& mx结束时间。

我正在使用的代码是

;with Ranges as ( 
    select FUser_id,FloginDt,FlogoutDt 
    from #AgentResults2 
    union all 
     select r.FUser_id,r.FloginDt,t.FLogoutdt 
    from 
     Ranges r 
      inner join 
     #AgentResults2 t 
      on 
       r.FUser_id = t.FUser_Id and DATEDIFF(ms,r.Flogoutdt,t.FloginDt) = 0 
), 
ExtendedRanges as ( 
select FUser_id,MIN(FloginDt) as FloginDt,Flogoutdt 
from Ranges 
group by FUser_ID,Flogoutdt 
) 
select FUser_Id,FloginDt,MAX(Flogoutdt) 
from ExtendedRanges 
group by FUser_id,FloginDt 
order by 1 
    OPTION (MAXRECURSION 0)

来源

2012-02-21 CeeVei

也许搜索SQL重叠范围，例如[这个问题]（http://stackoverflow.com/questions/4345910/merge-rows-based-on-date-in-sql-server） – 2012-02-21 08:14:34

谢谢@Damien_The_Unbeliever指向我的另一个问题。看完这个问题后，我很高兴 - 认为就是这样。但是在做出相应的代码之后，我没有看到预期的结果。另一个问题有一个诀窍，就是将第二天结合在同一行中，但是，如果同一毫秒包含在同一行中，则我的结果是两倍1）。 2）如果不到endtime还包括做一行。这并不完全符合我的情况。我附上了我在下一条评论中使用的代码。 – CeeVei 2012-02-21 13:05:16

如果您发布复杂的代码，最好编辑您的问题，以便您可以正确格式化它。 – 2012-02-21 13:12:55

我觉得这已经得到了它：

;With Ranges as (
    select 
     a1.FUser_Id,a1.FLoginDt,a1.FLogoutDt 
    from 
     #AgentResults2 a1 
      left join 
     #AgentResults2 a_nooverlap_early 
      on 
       a1.FUser_id = a_nooverlap_early.FUser_id and 
       a_nooverlap_early.FLoginDt < a1.FLoginDt and 
       a_nooverlap_early.FLogoutDt >= a1.FLoginDt 
    where 
     a_nooverlap_early.FUser_id is null 
    union all 
    select r.FUser_Id,r.FLoginDt,a1.FLogoutDt 
    from 
     Ranges r 
      inner join 
     #AgentResults2 a1 
      on 
       r.FUser_id = a1.FUser_Id and 
       r.FLogoutDt >= a1.FLoginDt and 
       a1.FLogoutDt > r.FLogoutDt 
) 
select FUser_id,FLoginDt,MAX(FLogoutDt) as FLogoutDt from Ranges group by FUser_id,FLoginDt

的Ranges CTE的第一部分，发现有一个FLoginDt是ISN所有时间段” t与另一个范围重叠。 CTE的递归部分然后试图找出与这些时期的FLogoutDt重叠的任何其他时间段，并且采用后来的FLogoutDt。然后最后的选择将最新发现的FLogoutDt用于任何特定的FUser_id,FLoginDt组合 - 这应该是整个存在重叠的时期。

我还重新写你的样本数据，以便SET DATEFORMAT DMY没有要求 - 日期格式我已经展示了始终都会安全/被明确地由SQL Server转换为datetime：

CREATE TABLE #AgentResults2 (FUser_id varchar(48), FLoginDt datetime, FLogoutDt DateTime) 

Insert Into #AgentResults2 (FUser_id , FLoginDt, FLogoutDt) 
select '30001','2012-02-17T08:09:23.117','2012-02-17T08:39:28.527' union all 
select '30001','2012-02-17T08:50:26.087','2012-02-17T09:31:32.040' union all 
select '30001','2012-02-17T10:28:42.430','2012-02-17T10:54:13.880' union all 
select '30001','2012-02-17T10:59:20.567','2012-02-17T12:00:36.030' union all 
select '30001','2012-02-17T12:32:15.473','2012-02-17T14:15:36.547' union all 
select '30001','2012-02-17T12:49:48.177','2012-02-17T14:10:01.097' union all 
select '30001','2012-02-17T14:30:49.293','2012-02-17T15:41:02.387' union all 
select '30001','2012-02-17T15:44:05.800','2012-02-17T16:01:33.613' union all 
select '39300','2012-02-17T08:06:31.250','2012-02-17T15:51:31.930' union all 
select '39300','2012-02-17T10:15:08.923','2012-02-17T10:21:29.833' union all 
select '39363','2012-02-17T09:58:44.287','2012-02-17T11:20:08.950' union all 
select '39363','2012-02-17T11:20:29.203','2012-02-17T13:27:36.717' union all 
select '39363','2012-02-17T13:27:36.717','2012-02-17T15:05:12.383' union all 
select '39363','2012-02-17T15:05:31.527','2012-02-17T16:57:13.733' union all 
select '39363','2012-02-17T16:57:44.640','2012-02-17T17:59:20.273' union all 
select '39395','2012-02-17T09:37:14.353','2012-02-17T10:15:52.397' union all 
select '39395','2012-02-17T10:15:28.427','2012-02-17T10:58:47.080' union all 
select '39395','2012-02-17T10:57:03.590','2012-02-17T11:53:47.933' union all 
select '39395','2012-02-17T11:51:31.567','2012-02-17T14:26:27.640' union all 
select '39395','2012-02-17T14:31:30.247','2012-02-17T16:04:44.217' union all 
select '39395','2012-02-17T16:10:49.013','2012-02-17T16:11:26.983' union all 
select '39395','2012-02-17T16:17:16.813','2012-02-17T17:55:47.187' union all 
select '39395','2012-02-17T16:55:25.900','2012-02-17T17:26:07.310';

来源

2012-02-22 09:04:11

从ExtendedRanges选择FUser_Id，FloginDt，MAX（Flogoutdt）组有没有分析你的脚本尚未彻底，只是测试，它似乎并没有涵盖所有情况为止。 '39300'行应该合并成一个（我认为），但它们在脚本的输出中保持单独的行。我还添加了这一行：'select'30001'，'2012-02-17T08：39：28.527'，'2012-02-17T08：39：33.117'union all'，我预料这应该会延长'30001 “的第一个范围，并没有（脚本似乎忽略了它）。 – 2012-02-22 12:54:11

@AndriyM - 我有一个比较（应该有'> ='而不是'>'）稍微错了，我想，但我无法重现你39300问题 - 我只在输出中得到一行。 – 2012-02-22 13:08:38

现在没什么不对，两个问题都没有了。我必须说，这非常好！我的版本使用了ROW_NUMBER（），最终得到的查询结果更加严重。 – 2012-02-22 13:30:03

计算用户唯一的开始结束时间集合

回答

相关问题