6

我的表是多天和多频道的预定电视节目列表。

SELECT * FROM [Scheduled_Programmes]

Channel   Date          Time     Title
1         2012-09-19    06:00    Family Guy
2         2012-09-19    06:01    CSI Miami
3         2012-09-19    06:20    News
1         2012-09-19    06:30    Heroes
2         2012-09-19    07:01    Spiderman
3         2012-09-19    06:40    Batman
1         2012-09-19    07:30    Micky Mouse
2         2012-09-19    07:31    CSI New York
3         2012-09-19    07:10    Friends
1         2012-09-19    07:55    The Wire
2         2012-09-19    08:00    Dodgeball
3         2012-09-19    07:35    Gossip Girl

我正在尝试创建的结果集是 What's on Now 和 What's on Next。

Let's assume the current datetime is (D/M/Y HH:MM) 19/09/2012 07:15

所以像：

          Channel 1     Channel 2       Channel 3
NOW       Heroes        Spiderman       Friends  
NEXT      Micky Mous    CSI New York    Gossip Girl

我一直在绞尽脑汁寻找最好的方法来做到这一点，而不必为每个频道硬编码单独的查询。我想我现在已经有了过度思考的阶段，所以如果有人能指出我正确的方向，那就太好了。

谢谢

PS：如果它有所作为，我在 Microsoft SQL Server 2012 上

4

5 回答 5

3

这看起来真的像是你会让你的 GIU 格式化和旋转的东西，但这是我的尝试。

SELECT * FROM (
    SELECT * FROM (
        SELECT X.Status, X.Channel, X.Title FROM (
        SELECT 'NOW' as Status, Channel, Title, RANK() OVER (PARTITION BY Channel ORDER BY Time DESC) RANKED FROM Sceduled_Programs SP
        WHERE DateTime <= '7:15') X
        WHERE X.RANKED = 1
    ) A
    UNION ALL
        SELECT * FROM (
        SELECT Y.Status, Y.Channel, Y.Title FROM (
        SELECT 'NEXT' as Status, Channel, Title, RANK() OVER (PARTITION BY Channel ORDER BY Time ASC) RANKED FROM Sceduled_Programs SP
        WHERE DateTime > '7:15') Y
        WHERE Y.RANKED = 1
    ) B
) DataToPivot
PIVOT (MAX(Title) FOR Channel IN ([1], [2], [3])) AS PivotTable

编辑：我在这里只使用时间，但只是添加日期。您应该真正考虑组合日期和时间列。

Edit2：要添加日期，只需将时间替换为此。甚至应该在日期范围内工作。

WHERE CAST(Date AS DATETIME) + CAST(Time AS DATETIME) > '19/09/2012 07:15'
于 2012-09-18T12:32:46.330 回答
2

这是SQLFiddle 示例。如果程序在前一天开始并在当前结束或在当前开始并在第二天结束（对于 NOW 和 NEXT 结果），此查询也很有效。

对于当前日期，只需替换cast('09/19/2012 07:15' as datetime)为CURRENT_TIMESTAMP：

with T as 
(select channel, title, 
       (date+cast(Time as datetime )) as D_Time 
       from Scheduled_Programmes
)

select nn,
   max(CASE when channel=1 then Title end) as Chanel1, 
   max(CASE when channel=2 then Title end) as Chanel2,
   max(CASE when channel=3 then Title end) as Chanel3


from
(
select 'NOW' nn,Channel,Title,D_time,
    row_number() over (PARTITION by channel 
                      order by D_time desc) rn 
    from T
    where D_time<=cast('09/19/2012 07:15' as datetime)
union
select 'NEXT'nn,Channel,Title,D_time,
     row_number() over (PARTITION by channel 
                        order by D_time asc) rn 
     from T
     where D_time>cast('09/19/2012 07:15' as datetime)
) ATable where rn=1
group by nn
order by nn desc;
于 2012-09-18T13:16:55.780 回答
0 
select distinct Channel
, (
select top 2 Title from 
        (
         select Title
         from [Scheduled_Programmes] s
         where s.Channel = o.Channel
         and Date >='2012-09-19' and Time >='07:15' 
         group by Channel
         order by Date, Time desc
        ) x
) as Prog
from [Scheduled_Programmes] o

我没有尝试过（由于此时 SQL Server 不可用），但这应该可以。

于 2012-09-18T12:09:46.203 回答
0

要获取数据和交叉表，请使用以下内容。您可以在第一次选择中重命名通道列，但您必须在数据透视列表中指定有效通道。

Nb 可以通过将日期和时间字段组合到单个日期时间字段来简化代码。

declare @current_date datetime = '20120919 07:15'

select *
from (
  select whenon, a1.channel, title from (
  select a.channel, MAX(CAST(Date AS DATETIME) + CAST(Time AS DATETIME)) as DateTime, 'now' as whenon
  from Scheduled_Programmes a
  join (
    select channel, MIN(CAST(Date AS DATETIME) + CAST(Time AS DATETIME)) as NextTime
    from Scheduled_Programmes 
    where CAST(Date AS DATETIME) + CAST(Time AS DATETIME) > @current_date
    group by channel
  ) b on a.channel = b.channel and CAST(Date AS DATETIME) + CAST(Time AS DATETIME) < NextTime
  group by a.channel
  union
  select channel, MIN(CAST(Date AS DATETIME) + CAST(Time AS DATETIME)), 'next' from Scheduled_Programmes where CAST(Date AS DATETIME) + CAST(Time AS DATETIME) > @current_date
  group by channel

  ) a1
  join Scheduled_Programmes b1 on a1.channel = b1.channel and a1.DateTime = CAST(Date AS DATETIME) + CAST(Time AS DATETIME)
) sub
pivot (max(title) for channel in ([1], [2], [3])) pivottable
order by whenon desc
于 2012-09-18T12:24:13.670 回答
0

另一个可能的选项 - 计算时间差并获取最小的 +ve 值（下一个）和最小的 -ve 值（现在） - 假设 DATE 和 TIME 类型字段，还返回自开始/直到开始的分钟数。

;with T(Channel, Date, Time, Title, delta) as (
select 
    *,
    datediff(minute, '19 sep 2012 07:15:00', dateadd(day, datediff(day, 0, date), cast(time as datetime2)))
from 
    Scheduled_Programmes
)
select 
case when T.delta >= 0 then 'Next' else 'Now' end,
T.* 
from T
inner join (
    select
        channel,
        min(case when delta > 0 then delta end) as starts_in,
        max(case when delta < 0 then delta end) as started
    from t
    group by channel
) T2
    on T.Channel = T2.Channel and (t.delta = T2.starts_in or t.delta=T2.started)
order by Channel, delta
于 2012-09-18T13:53:24.717 回答