通过DATEADD分组日期和DATEDIFF

Question

我被时期试图组一串日期，即​​“月”

首先我声明表：

--inserts 36 dates every 10 days 
create table sales (sales_date datetime, sales_amount decimal(12,2)) 
insert into sales (sales_date,sales_amount) 
select '20140101',1000 union all 
select '20140110',1000 union all 
select '20140120',1000 union all 
select '20140130',1000 union all 
... 

然后我通过查询分组期间条目（尺寸=精度*月）

DECLARE @precision int = 1 --number of months 

SELECT 
dateadd(month, (datediff(month, 0, sales_date)/@precision) * @precision, 0) as FromDate, 
dateadd(month, @precision + (datediff(month, 0, sales_date)/@precision) * @precision, 0) as ToDate, 
count(*) as number 
FROM sales 
GROUP BY 
dateadd(month, (datediff(month, 0, sales_date)/@precision) * @precision, 0), 
dateadd(month, @precision + (datediff(month, 0, sales_date)/@precision) * @precision, 0) 

结果：

Test 1 : precision of 1 month 
From : 01/01/2014, To : 01/02/2014, number : 4 
From : 01/02/2014, To : 01/03/2014, number : 2 
From : 01/03/2014, To : 01/04/2014, number : 3 
From : 01/05/2014, To : 01/06/2014, number : 4 
From : 01/06/2014, To : 01/07/2014, number : 3 
From : 01/07/2014, To : 01/08/2014, number : 3 
From : 01/08/2014, To : 01/09/2014, number : 3 
From : 01/09/2014, To : 01/10/2014, number : 3 
From : 01/10/2014, To : 01/11/2014, number : 3 
From : 01/11/2014, To : 01/12/2014, number : 3 
From : 01/12/2014, To : 01/01/2015, number : 2 

Test 2 : precision of 2 months 
From : 01/01/2014, To : 01/03/2014, number : 6 
From : 01/03/2014, To : 01/05/2014, number : 6 
From : 01/05/2014, To : 01/07/2014, number : 7 
From : 01/07/2014, To : 01/09/2014, number : 6 
From : 01/09/2014, To : 01/11/2014, number : 6 
From : 01/11/2014, To : 01/01/2015, number : 5 

Test 3 : precision of 3 months 
From : 01/01/2014, To : 01/04/2014, number : 9 
From : 01/04/2014, To : 01/07/2014, number : 10 
From : 01/07/2014, To : 01/10/2014, number : 9 
From : 01/10/2014, To : 01/01/2015, number : 8 

Test 4 : precision of 4 months 
From : 01/01/2014, To : 01/05/2014, number : 12 
From : 01/05/2014, To : 01/09/2014, number : 13 
From : 01/09/2014, To : 01/01/2015, number : 11 

一切工作正常，直到测试5：

Test 5 : precision of 5 months 
From : 01/10/2013, To : 01/03/2014, number : 6 
From : 01/03/2014, To : 01/08/2014, number : 16 
From : 01/08/2014, To : 01/01/2015, number : 14 

第一行很讨厌，我需要它在2014年1月1日像以前的精度

有什么办法来克服这个问题开始？

Answer 1

你并不需要通过精密乘以精度为FROM列

这里分裂，再次发生了什么

datediff(month, 0, sales_date)/5 -> 273.6 as it is integer division, it is converting to 273 

进一步273*5 = 1365乘以（SO 3月是失去了应有的到整数除法）

所以这一块应该为你

DECLARE @precision int = 5 --number of months 

SELECT 
dateadd(month, (datediff(month, 0, sales_date)) , 0) as [From], 
dateadd(month, @precision + (datediff(month, 0, sales_date)) , 0) as [To], 
count(*) as number 
FROM sales 
GROUP BY 
dateadd(month, (datediff(month, 0, sales_date)) , 0), 
dateadd(month, @precision + (datediff(month, 0, sales_date)), 0) 

工作