每个 Int 扩展每一行的 SQL 查询。在 COL_A 和 COL_B 之间的范围内

Posted 2023-04-12

【中文标题】每个 Int 扩展每一行的 SQL 查询。在 COL_A 和 COL_B 之间的范围内

【英文标题】：SQL Query Expanding Each Row Every Int. In Range Between COL_A and COL_B

【发布时间】：2018-12-31 21:31:09

【问题描述】：

我的 SQL Server 有一个表。两列都是整数，但以 YYYYMM 的形式表示日期。我想查询这个表并返回第三列，对于每一行，在两列的范围内，每一年/月都包含一个 YYYYMM 形式的整数。

这是表格：

+------------+------------+
| beg_YYYYMM | end_YYYYMM |
+------------+------------+
|     201802 |     201805 |
|     201711 |     201801 |
+------------+------------+

期望的输出：

+------------+------------+----------------+
| beg_YYYYMM | end_YYYYMM | month_in_range |
+------------+------------+----------------+
|     201802 |     201805 |         201802 |
|     201802 |     201805 |         201803 |
|     201802 |     201805 |         201804 |
|     201802 |     201805 |         201805 |
|     201711 |     201801 |         201711 |
|     201711 |     201801 |         201712 |
|     201711 |     201801 |         201801 |
+------------+------------+----------------+

【问题讨论】：

我的建议是，出于此查询的目的，将所有数据转换为 date 数据类型 - 这样您就可以利用可用的 datetime 函数。然后您可以将结果转换回unusual int 格式。

【参考方案1】：

使用递归 CTE：

with cte as (
      select beg_YYYYMM, end_YYYYMM,
             convert(date, convert(varchar(255), beg_YYYYMM) + '01') as dte,
             convert(date, convert(varchar(255), end_YYYYMM) + '01') as end_dte
      from t
      union all
      select beg_YYYYMM, end_YYYYMM,
             dateadd(month, 1, dte),
             end_dte
      from cte
      where dte < end_dte
     )
select beg_yyyymm, end_yyyymm, 
       year(dte) * 100 + month(dte) as yyyymm
  from cte
 order by dte;

Here 是一个 dbfiddle。

【讨论】：

比我的优雅多了。我羞愧地低下了头。

【参考方案2】：

我真的不认为我会实际部署它，但我想知道它会是什么样子。如果你创建一个 TVF 做丑陋的工作，那还不错。我使用 Aaron Bertrand 的 DateDim code 进行了快速修改，以获取传入的两个日期之间的月初日期。

CREATE OR ALTER FUNCTION dbo.tvf_MonthRange (@beg_YYYYMM int, @end_YYYYMM int)  
RETURNS @Results TABLE   
 (month_in_range int)
AS  
BEGIN  

--Have to convert ints to dates
    DECLARE @BegDate DATE; 
    SET @BegDate =  DATEFROMPARTS(CAST(SUBSTRING(CAST(@beg_YYYYMM AS varchar(6)),1,4) AS INT), CAST(SUBSTRING(CAST(@beg_YYYYMM AS varchar(6)),5,2) AS INT), 1);

--This needs to be the second day of the month for the code below to work.
    DECLARE @EndDate DATE;
    SET @EndDate =  DATEFROMPARTS(CAST(SUBSTRING(CAST(@end_YYYYMM AS varchar(6)),1,4) AS INT), CAST(SUBSTRING(CAST(@end_YYYYMM AS varchar(6)),5,2) AS INT), 2);

    INSERT INTO @Results
    SELECT (DATEPART(YEAR, d) *100) + DATEPART(MONTH, d)
    FROM
        (
          SELECT d = DATEADD(DAY, rn - 1, @BegDate)
          FROM 
          (
            SELECT TOP (DATEDIFF(DAY, @BegDate, @EndDate)) 
              rn = ROW_NUMBER() OVER (ORDER BY s1.[object_id])
            FROM sys.all_objects AS s1
            CROSS JOIN sys.all_objects AS s2
            -- on my system this would support > 5 million days
            ORDER BY s1.[object_id]
          ) AS x
        ) AS y
    WHERE DATEPART(DAY, d) = 1;

    RETURN;
END 

那你就可以这样称呼了。

DECLARE @Months TABLE (beg_YYYYMM int, end_YYYYMM int)
INSERT INTO @MONTHS SELECT 201802, 201805 
INSERT INTO @MONTHS SELECT 201711, 201801 

SELECT *
FROM @Months m
CROSS APPLY dbo.tvf_MonthRange(m.beg_YYYYMM, m.end_YYYYMM) mr ;

星期一不是个好主意吗？

【参考方案3】：

使用计数表的解决方案在较大的数据集上可能会更快。

with dates as(
select 
    201501 as beg_YYYYMM
    ,201504 as end_YYYYMM
union all
select '201711', '201801'
union all
select '201807', '201812'

),

--Tally table
ctedaterange AS (
SELECT top 15
    rn = Row_number() OVER(ORDER BY (SELECT NULL)) -1
FROM sys.objects a  
)

SELECT
    dates.*
    ,months_in_range = convert(varchar(6), Dateadd(mm, rn, cast(cast(dates.beg_YYYYMM as varchar(8)) +'01' as date)), 112)

FROM dates
    cross join ctedaterange

WHERE  
    rn <= Datediff(mm, cast(cast(dates.beg_YYYYMM as varchar(8)) +'01' as date), cast(cast(dates.end_YYYYMM as varchar(8)) +'01' as date))

ORDER BY 
    beg_YYYYMM
    ,Dateadd(mm, rn, cast(cast(dates.beg_YYYYMM as varchar(8)) +'01' as date))

Here's the db<>fiddle