在不使用过程的情况下以 'range' 格式序列化日期

Question

我需要在列中查找日期范围并以简洁的格式将它们序列化（start - end 用于范围或 date 用于单个日期范围） .

我的 CTE (readings) returns 数据集类似于：

ID          VALUE   DATE
1234567     A       2012-05-09
1234567     A       2012-05-10
1234567     A       2012-05-11
1234567     A       2012-05-16
1234567     A       2012-05-17
1234567     A       2012-05-20
1234567     B       2012-05-11
1234567     B       2012-05-12
1234567     B       2012-05-13
1234567     B       2012-05-14

我已经能够得到：

ID          VALUE   TOTAL_DAYS  DATES
1234567     A       6           2012-05-09; 2012-05-10; 2012-05-11; 2012-05-16; 2012-05-17; 2012-05-20
1234567     B       4           2012-05-11; 2012-05-12; 2012-05-13; 2012-05-14

使用：

readings AS (
...
)
,
reading_aggr AS (

    SELECT  ID, [VALUE]
            ,count(distinct date) TOTAL_DAYS
            ,STUFF((
                SELECT '; ' + cast(date as varchar)
                FROM    readings r0
                WHERE   id=r0.id
                    AND value=r0.value
                ORDER BY date
                FOR XML PATH(''),TYPE).value('(./text())[1]','VARCHAR(MAX)'
            ),1,2,'') AS DATES
    FROM    readings
    GROUP BY id, [value]
)

SELECT * FROM readings_aggr

我想这样格式化：

ID          VALUE   TOTAL_DAYS  DATES
1234567     A       6           2012-05-09 - 2012-05-11; 2012-05-16 - 2012-05-17; 2012-05-20
1234567     B       4           2012-05-11 - 2012-05-14

如果不使用程序方法，这可能吗？

Answer 1

您可能可以使用 CLR 聚合来执行此操作。

这是一个来自 MSDN 的示例，它将您的数据连接在一起。只需将逗号更改为分号，您就可以使用当前格式进行更清晰的查询。

https://msdn.microsoft.com/en-us/library/ms165055%28v=vs.90%29.aspx

一旦到位，您可以调整累积 and/or 终止方法以尽可能查看数据和输出范围。您可能希望将值累积到类似 SortedList 而不是 StringBuilder 的对象中，然后在 Terminate 方法中进行范围分析。

Answer 2

您可以使用此查询：

 SELECT ID, VALUE, MIN([DATE]) AS startDate, MAX([DATE]) AS endDate
   FROM (
      SELECT ID, VALUE, DATE, 
             DATEDIFF(Day, '1900-01-01' , [DATE])- ROW_NUMBER() OVER( PARTITION BY ID, VALUE ORDER BY [DATE] ) AS DateGroup
      FROM readings ) rGroups
   GROUP BY ID, VALUE, DateGroup

得到一个table包含你数据的所有开始-结束间隔的表达式：

ID      VALUE   startDate   endDate
--------------------------------------
1234567 A       2012-05-09  2012-05-11
1234567 A       2012-05-16  2012-05-17
1234567 A       2012-05-20  2012-05-20
1234567 B       2012-05-11  2012-05-14

然后在 reading_aggr:

中使用上面的查询

;WITH start_end_readings AS (
   SELECT ID, VALUE, MIN([DATE]) AS startDate, MAX([DATE]) AS endDate
   FROM (
      SELECT ID, VALUE, DATE, DATEDIFF(Day, '1900-01-01' , [DATE])- ROW_NUMBER() OVER( PARTITION BY ID, VALUE ORDER BY [DATE] ) AS DateGroup
      FROM readings ) rGroups
   GROUP BY ID, VALUE, DateGroup
), readings_aggr AS (

    SELECT  ID, [VALUE]
            ,count(distinct date) TOTAL_DAYS 
            ,STUFF((
                SELECT '; ' + cast(startDate as varchar) + 
                       CASE WHEN startDate <> endDate THEN ' - ' + cast(endDate as varchar)
                            ELSE ''
                       END
                FROM    start_end_readings r0
                WHERE   r1.id=r0.id AND r1.value=r0.value
                ORDER BY startDate 
                FOR XML PATH(''),TYPE).value('(./text())[1]','VARCHAR(MAX)'
            ),1,2,'') AS DATES
    FROM    readings AS r1
    GROUP BY id, [value]
)
SELECT * FROM readings_aggr

获得想要的结果：

ID      VALUE   TOTAL_DAYS  DATES
===========================================================================
1234567 A       6           2012-05-09 - 2012-05-11; 2012-05-16 - 2012-05-17; 2012-05-20
1234567 B       4           2012-05-11 - 2012-05-14

SQL Fiddle Demo here

Answer 3

你可以这样做：

DECLARE @t TABLE ( ID INT, V CHAR(1), D DATE )

INSERT  INTO @t
VALUES  ( 1234567, 'A', '2012-05-09' ),
        ( 1234567, 'A', '2012-05-10' ),
        ( 1234567, 'A', '2012-05-11' ),
        ( 1234567, 'A', '2012-05-16' ),
        ( 1234567, 'A', '2012-05-17' ),
        ( 1234567, 'A', '2012-05-20' ),
        ( 1234567, 'B', '2012-05-11' ),
        ( 1234567, 'B', '2012-05-12' ),
        ( 1234567, 'B', '2012-05-13' ),
        ( 1234567, 'B', '2012-05-14' );

WITH    cte1
          AS ( SELECT   ID ,
                        V ,
                        CASE WHEN MIN(D) <> MAX(D)
                             THEN CONVERT(NVARCHAR(MAX), MIN(D), 121) + ' - '
                                  + CONVERT(NVARCHAR(MAX), MAX(D), 121)
                             ELSE CONVERT(NVARCHAR(MAX), MIN(D), 121)
                        END AS D ,
                        COUNT(*) AS cn
               FROM     ( SELECT    ID ,
                                    V ,
                                    D ,
                                    DATEADD(dd,
                                            -ROW_NUMBER() OVER ( PARTITION BY V ORDER BY D ),
                                            D) AS rn
                          FROM      @t
                        ) a
               GROUP BY ID ,
                        V ,
                        rn
             ),-- SELECT * FROM cte1,
        cte2
          AS ( SELECT   ID ,
                        V ,
                        SUM(cn) TOTAL_DAYS ,
                        STUFF((SELECT   '; ' + D
                               FROM     cte1 r0
                               WHERE    cte1.id = r0.id
                                        AND cte1.V = r0.V
                        FOR   XML PATH('') ,
                                  TYPE).value('(./text())[1]', 'VARCHAR(MAX)'),
                              1, 2, '') AS DATES
               FROM     cte1
               GROUP BY id ,
                        V
             )
    SELECT  *
    FROM    cte2

输出：

ID      V   TOTAL_DAYS  DATES
1234567 A   6           2012-05-09 - 2012-05-11; 2012-05-16 - 2012-05-17; 2012-05-20
1234567 B   4           2012-05-11 - 2012-05-14

这个想法是先获得岛屿（https://www.simple-talk.com/sql/t-sql-programming/the-sql-of-gaps-and-islands-in-sequences/）然后应用你的东西。我知道@Betsos 超过了我，但这有点不同。但是思路是一样的。

在不使用过程的情况下以 'range' 格式序列化日期

Serialize dates in a 'range' format without using a procedure

sql

sql-server

sql-server-2012