Question

我有一张看起来像这样的表

ID  start_dt    end_dt
--------------------------
1   1951-12-05  1951-12-21
2   1951-12-19  1951-12-31
3   1957-12-05  1957-12-19
4   1995-12-06  1995-12-20
5   1996-06-24  1996-07-08
6   1997-05-12  1997-05-26
7   1997-10-07  1997-10-21
8   1997-12-25  1998-01-08
9   1998-01-19  1998-02-02
10  1998-08-05  1998-08-19

我想知道start_dt和end_dt之间每个日期的包含次数。

从我的例子中，结果集应该看起来像这样

date         count
------------------
1951-12-05   1
1951-12-06   1
...
1951-12-19   2
1951-12-20   2
1951-12-21   2
...
1998-08-19   1

最好的方法是什么？

编辑：为了澄清，我需要在日期范围内（start_dt和end_dt之间）至少出现一次的每个日期，以便在我的结果集中获得一行，并且我希望此日期适合的范围数量在它旁边

希望这会有所帮助

Answer 1

当您需要将2个值（范围）转换为一系列行时，您可以使用数字表格（如果您不熟悉这个想法，请参阅Aaron Bertrand的The SQL Server Numbers Table文章）。

我使用的是更短更简单的数据，但您应该明白这一点。

declare @dates table (id int not null, start_dt date not null, end_dt date not null)
insert @dates values (1, '20160601', '20160603'),
                     (2, '20160603', '20160605'),
                     (3, '20160610', '20160612')

;with cte as (
    select
        row_number() over (order by so1.object_id) - 1 as n
    from
        sys.objects so1
            cross join sys.objects so2
)
select  
    dateadd(d, c.n, d.start_dt) as [date],
    count(*)
from
    @dates d
        join cte c on dateadd(d, c.n, d.start_dt) <= d.end_dt
group by 
    dateadd(d, c.n, d.start_dt)
order by 
    dateadd(d, c.n, d.start_dt)

Answer 2

您可以使用CTE

WITH CTE AS(SELECT start_dt AS dates FROM Table
            UNION ALL
            SELECT end_dt AS dates FROM Table)
SELECT CAST(dates as DATE) as Date, COUNT(dates) AS Count
FROM CTE c
GROUP BY c.dates
order by Count desc

如果您的列属于DATETIME数据类型，或许您需要更广泛的内容。这种方式将GROUP BY整天：

WITH CTE AS(SELECT CAST(start_dt AS DATE) AS dates FROM Table
            UNION ALL
            SELECT CAST(end_dt AS DATE) AS dates FROM Table)
SELECT Dates as Date, COUNT(Dates) AS Count
FROM CTE c
GROUP BY c.dates
order by Count desc

Answer 3

如果在sys.objects和start_dt之间不超过几天（<80左右，具体取决于end_dt表），您可以使用此方法（受到启发）里斯＆＃39;。）

DECLARE @dates TABLE (id int not null, start_dt date not null, end_dt date not null)
INSERT @dates VALUES
    (1, '1951-12-05', '1951-12-21'),
    (2, '1951-12-19', '1951-12-31'),
    (3, '1957-12-05', '1957-12-19'),
    (4, '1995-12-06', '1995-12-20'),
    (5, '1996-06-24', '1996-07-08'),
    (6, '1997-05-12', '1997-05-26'),
    (7, '1997-10-07', '1997-10-21'),
    (8, '1997-12-25', '1998-01-08'),
    (9, '1998-01-19', '1998-02-02'),
    (10, '1998-08-05', '1998-08-19');

WITH RawData AS (
SELECT  
    DATEADD(d, n.n, d.start_dt) AS [date]
FROM @dates d
INNER JOIN (
    SELECT ROW_NUMBER() OVER (ORDER BY object_id) - 1 AS n FROM sys.objects
    ) n ON DATEADD(d, n.n, d.start_dt) <= d.end_dt
)
SELECT [date], COUNT(*) [count]
FROM RawData
GROUP BY [date]
ORDER BY [date]

我认为即使有1000个日期范围，这也可能需要很长时间。也许您正在使用包含更多字段的表，甚至缺少某些索引？

计算两个日期列之间包含日期的次数

3 个答案: