消除并减少重叠日期范围

Question

我有一组日期范围,包括部分和完全重叠的日期,如下所示:

UserID  StartDate   EndDate 
======  ==========  ==========
1       2011-01-01  2011-01-02  <- A
1       2011-01-01  2011-01-10  <- A
1       2011-01-08  2011-02-15  <- A
1       2011-02-20  2011-03-10  <- B
2       2011-01-01  2011-01-20  <- C
2       2011-01-15  2011-01-25  <- C

使用T-SQL,我想为每个用户创建一组新的数据,消除重叠数据,扩展范围并在需要时删除冗余数据,结果如下所示:

UserID  StartDate   EndDate 
======  ==========  ==========
1       2011-01-01  2011-02-15 ('A', three rows combined, extending the range)
1       2011-02-20  2011-03-10 ('B', no change, no overlaps here)
2       2011-01-01  2011-01-25 ('C', two rows combined)

如果需要游标很好,但如果没有它就可以做得更好.

Answer 1

对于SQL Server 2005+

-- sample table with data
declare @t table(UserID int, StartDate datetime, EndDate datetime)
insert @t select
1, '20110101', '20110102' union all select
1, '20110101', '20110110' union all select
1, '20110108', '20110215' union all select
1, '20110220', '20110310' union all select
2, '20110101', '20110120' union all select
2, '20110115', '20110125'

-- your query starts below

select UserID, Min(NewStartDate) StartDate, MAX(enddate) EndDate
from
(
    select *,
        NewStartDate = t.startdate+v.number,
        NewStartDateGroup =
            dateadd(d,
                    1- DENSE_RANK() over (partition by UserID order by t.startdate+v.number),
                    t.startdate+v.number)
    from @t t
    inner join master..spt_values v
      on v.type='P' and v.number <= DATEDIFF(d, startdate, EndDate)
) X
group by UserID, NewStartDateGroup
order by UserID, StartDate

笔记:

1. 替换@t为您的表名