SQL Server 中第一条记录结束日期和下一条记录开始日期之间的重叠

Question

我有以下类型的数据，我需要以下类型的输出。

输入：

id   startdate  enddate
1    21/01/2019 23/01/2019
1    23/01/2019 24/01/2019
1    24/01/2029 27/01/2019
1    29/01/2019 02/02/2019

输出：

id  startdate   enddate
1   21/01/2019  27/01/2019
1   29/01/2019  02/02/2019

我们需要使用匹配第一条记录enddate和第n条记录startdate的逻辑。

Answer 1

这是一个间隙和孤岛问题，您希望将“相邻”日期组合在一起。这是使用窗口函数的一种方法：想法是将当前开始日期与“上一个”行的结束日期进行比较，并使用窗口总和来定义组：

select id, min(startdate) startdate, max(enddate) enddate
from (
    select t.*,
        sum(case when startdate = lag_enddate then 0 else 1 end) over(partition by id order by startdate) grp
    from (
        select t.*,
            lag(enddate) over(partition by id order by startdate) lag_enddate
        from mytable t
    ) t
) t
group by id, grp

Demo on DB Fiddle - 感谢 Sander 首先创建了 DDL 语句：

编号 |开始日期 |结束日期 -: | :--------- | :--------- 1 | 2019-01-21 | 2019-01-27 1 | 2019-01-29 | 2019-02-02

Answer 2

看看

1. NEXT VALUE FOR 方法，适用于 2016 年及以后

2. 使用 CTE 或子查询（在 2008 年有效），您可以在其中使用以前的值作为连接来连接您自己的表。这是我使用的示例脚本，用于显示备份增长

   declare @backupType char(1)
          , @DatabaseName sysname
   
    set @DatabaseName   = db_name() --> Name of current database, null for all databaseson server
    set @backupType     ='D'        /* valid options are:
                                    D = Database 
                                    I = Database Differential 
                                    L = Log 
                                    F = File or Filegroup 
                                    G = File Differential 
                                    P = Partial 
                                    Q = Partial Differential
                                    */
   
   
   
    select  backup_start_date
          , backup_finish_date
          , DurationSec
          , database_name,backup_size
          , PreviouseBackupSize
          , backup_size-PreviouseBackupSize as growth
          ,KbSec= format(KbSec,'N2')
    FROM (
    select backup_start_date 
         , backup_finish_date
         , datediff(second,backup_start_date,b.backup_finish_date) as DurationSec 
         , b.database_name
         , b.backup_size/1024./1024. as backup_size
   
         ,case when datediff(second,backup_start_date,b.backup_finish_date) >0 
            then ( b.backup_size/1024.)/datediff(second,backup_start_date,b.backup_finish_date)
            else 0 end as KbSec
    --   , b.compressed_backup_size
         , (      
            select top (1) p.backup_size/1024./1024. 
              from msdb.dbo.backupset p 
             where p.database_name = b.database_name
               and p.database_backup_lsn< b.database_backup_lsn
                and type=@backupType
             order by p.database_backup_lsn desc
            ) as PreviouseBackupSize
    from msdb.dbo.backupset as b
    where @DatabaseName IS NULL OR database_name =@DatabaseName
    and type=@backupType
    )as A
    order by backup_start_date desc
   

3. 使用“光标本地 fast_forward”逐行遍历数据，并使用临时表存储和比较上一个值

Answer 3

这是一个可以使用公用表表达式的解决方案。

样本数据

create table data
(
  id int,
  startdate date,
  enddate date
);

insert into data (id, startdate, enddate) values
(1, '2019-01-21', '2019-01-23'),
(1, '2019-01-23', '2019-01-24'),
(1, '2019-01-24', '2019-01-27'),
(1, '2019-01-29', '2019-02-02');

解决方案

-- determine start dates
with cte_start as
(
  select s.id,
         s.startdate
  from data s
  where not exists ( select 'x'
                     from data e
                     where e.id = s.id
                       and e.enddate = s.startdate )
),
-- determine date boundaries
cte_startnext as
(
  select s.id,
         s.startdate,
         lead(s.startdate) over (partition by s.id order by s.startdate) as startdate_next
  from cte_start s
)
-- determine periods
select sn.id,
       sn.startdate,
       e.enddate
from cte_startnext sn
cross apply ( select top 1 e.enddate
              from data e
              where e.id = sn.id
                and  e.startdate >= sn.startdate
                and (e.startdate <  sn.startdate_next or sn.startdate_next is null)
              order by e.enddate desc ) e
order by sn.id,
         sn.startdate;

结果

id startdate  enddate
-- ---------- ----------
1  2019-01-21 2019-01-27
1  2019-01-29 2019-02-02

Fiddle 查看解决方案和中间 CTE 结果的建立。