Sql 按特殊条件筛选日期范围_Sql_Sql Server_Tsql_Intervals_Date Range

Sql 按特殊条件筛选日期范围

sql sql-server tsql

Sql 按特殊条件筛选日期范围,sql,sql-server,tsql,intervals,date-range,Sql,Sql Server,Tsql,Intervals,Date Range,我有一个表，其中包含按StartDate排序的以下数据： | Id | StartDate | EndDate | |----|---------------------|---------------------| | 1 | 2017-03-18 00:00:00 | 2017-03-18 02:00:00 | | 2 | 2017-03-18 01:30:00 | 2017-03-18 04:00:00 |

我有一个表，其中包含按StartDate排序的以下数据：

    | Id | StartDate           | EndDate             |
    |----|---------------------|---------------------|
    | 1  | 2017-03-18 00:00:00 | 2017-03-18 02:00:00 |
    | 2  | 2017-03-18 01:30:00 | 2017-03-18 04:00:00 |
    | 3  | 2017-03-18 01:45:00 | 2017-03-18 03:00:00 |
    | 4  | 2017-03-18 02:10:00 | 2017-03-18 02:30:00 |
    | 5  | 2017-03-18 02:35:00 | 2017-03-18 02:50:00 |
    | 6  | 2017-03-18 03:20:00 | 2017-03-18 03:50:00 |
    | 7  | 2017-03-18 05:00:00 | 2017-03-18 05:30:00 |
    | 8  | 2017-03-18 05:10:00 | 2017-03-18 07:00:00 |
    | 9  | 2017-03-18 05:50:00 | 2017-03-18 08:00:00 |

过滤逻辑：在第一段日期之后，我们发现另一段日期与之不相交。然后，针对所发现的周期重复该逻辑

过滤后应保留：

    | Id | StartDate           | EndDate             |
    |----|---------------------|---------------------|
    | 1  | 2017-03-18 00:00:00 | 2017-03-18 02:00:00 |
    | 4  | 2017-03-18 02:10:00 | 2017-03-18 02:30:00 |
    | 5  | 2017-03-18 02:35:00 | 2017-03-18 02:50:00 |
    | 6  | 2017-03-18 03:20:00 | 2017-03-18 03:50:00 |
    | 7  | 2017-03-18 05:00:00 | 2017-03-18 05:30:00 |
    | 9  | 2017-03-18 05:50:00 | 2017-03-18 08:00:00 |

是否可以在不使用光标的情况下实现此功能

为方便使用记录：

CREATE TABLE #Dates (Id INT, StartDate DATETIME, EndDate DATETIME);
INSERT INTO #Dates SELECT 1, '2017-03-18 00:00:00', '2017-03-18 02:00:00';
INSERT INTO #Dates SELECT 2, '2017-03-18 01:30:00', '2017-03-18 04:00:00';
INSERT INTO #Dates SELECT 3, '2017-03-18 01:45:00', '2017-03-18 03:00:00';
INSERT INTO #Dates SELECT 4, '2017-03-18 02:10:00', '2017-03-18 02:30:00';
INSERT INTO #Dates SELECT 5, '2017-03-18 02:35:00', '2017-03-18 02:50:00';
INSERT INTO #Dates SELECT 6, '2017-03-18 03:20:00', '2017-03-18 03:50:00';
INSERT INTO #Dates SELECT 7, '2017-03-18 05:00:00', '2017-03-18 05:30:00';
INSERT INTO #Dates SELECT 8, '2017-03-18 05:10:00', '2017-03-18 07:00:00';
INSERT INTO #Dates SELECT 9, '2017-03-18 05:50:00', '2017-03-18 08:00:00';
DROP TABLE #Dates;

是的，您可以在没有光标的情况下执行此操作。以下是使用exists的版本，假设在开始日期没有完全相同的版本：

select min(id), min(startdate), max(startdate)
from (select t.*, sum(flag) over (order by startdate) as grp
      from (select t.*,
                   (case when not exists (select 1
                                          from t t2
                                          where t2.startdate < t.startdate and
                                                t2.enddate >= t.startdate
                                         )
                         then 1 else 0
                    end) as flag
            from t
           ) t
      ) t
group by grp;

最里面的子查询计算一个标志以确定行是否重叠。中间层累积此标志以标识每个组，外部层进行聚合。

从SQL Server 2012开始，您可以使用和函数来实现此目的：

CREATE TABLE #Dates (Id INT, StartDate DATETIME, EndDate DATETIME);
INSERT INTO #Dates SELECT 1, '2017-03-18 00:00:00', '2017-03-18 02:00:00';
INSERT INTO #Dates SELECT 2, '2017-03-18 01:30:00', '2017-03-18 04:00:00';
INSERT INTO #Dates SELECT 3, '2017-03-18 01:45:00', '2017-03-18 03:00:00';
INSERT INTO #Dates SELECT 4, '2017-03-18 02:10:00', '2017-03-18 02:30:00';
INSERT INTO #Dates SELECT 5, '2017-03-18 02:35:00', '2017-03-18 02:50:00';
INSERT INTO #Dates SELECT 6, '2017-03-18 03:20:00', '2017-03-18 03:50:00';
INSERT INTO #Dates SELECT 7, '2017-03-18 05:00:00', '2017-03-18 05:30:00';
INSERT INTO #Dates SELECT 8, '2017-03-18 05:10:00', '2017-03-18 07:00:00';
INSERT INTO #Dates SELECT 9, '2017-03-18 05:50:00', '2017-03-18 08:00:00';


SELECT T1.Id, T1.StartDate, T1.EndDate FROM (
    SELECT *,
        LAG(EndDate) OVER(ORDER BY ID) as lagdate  ,
        LEAD(StartDate) OVER(ORDER BY ID) as leaddate  
    FROM #Dates) AS T1
WHERE T1.StartDate > T1.lagdate OR T1.lagdate IS NULL 
   OR T1.EndDate < T1.leaddate  OR T1.leaddate IS NULL

这就是结果：

注意：T1.lagdate为NULL用于获取第一行，T1.leaddate为NULL用于获取最后一行

/*这是过滤逻辑的代码：在第一段日期之后，我们找到另一段不相交的日期。然后，针对所发现的周期重复该逻辑*/

WHILE(@@ROWCOUNT > 0)   -- we did something the last insert
BEGIN

    INSERT INTO @Results
    SELECT TOP 1 * 
    FROM #Dates d
    WHERE d.StartDate = 
        (SELECT MIN(StartDate)
        FROM #Dates
END 

SELECT *
FROM @Results
ORDER BY 1

DROP TABLE #Dates;

Id  StartDate               EndDate
1   2017-03-18 00:00:00.000 2017-03-18 02:00:00.000
4   2017-03-18 02:10:00.000 2017-03-18 02:30:00.000
5   2017-03-18 02:35:00.000 2017-03-18 02:50:00.000
6   2017-03-18 03:20:00.000 2017-03-18 03:50:00.000
7   2017-03-18 05:00:00.000 2017-03-18 05:30:00.000
9   2017-03-18 05:50:00.000 2017-03-18 08:00:00.000

找到的解决方案：

您使用的sql server版本是什么？我使用MSSQL server 2014您可以使用LAG函数来实现这一点，更多信息请查看我的答案details@IvanAntonov感谢您输入示例数据-使每个人都能更轻松地使用。抱歉，解决方案无法正常工作。这项任务比看起来要困难得多。您将在1、5、6、7中显示Id为的记录。@IvanAntonov等待您的反馈：非常感谢您的帮助！但是，此解决方案仅适用于当前输入数据，如果您稍微更改输入数据，该解决方案将产生错误的结果：使用此数据：插入日期选择1，'2017-03-18 04:00:00'，'2017-03-18 10:30:00'；插入日期选择2，“2017-03-18 05:00:00”，“2017-03-18 15:00:00”；插入日期选择3，“2017-03-18 07:00:00”，“2017-03-18 20:00:00”；在日期中插入选择4，“2017-03-18 11:00:00”，“2017-03-18 23:00:00”；在日期中插入选择5，‘2017-03-18 12:30:00’、‘2017-03-18 19:00:00’；您可以选择Id为1、5的记录，但按照逻辑，应该获得Id为1、5的记录4@IvanAntonov很高兴你的问题解决了。幸运的是，我描述问题的本质是错误的。目前，不存在的解决方案应该输出第二个表中的记录。但是，这个解决方案不能正常工作，我看不到修复它的可能性：。@IvanAntonov。我认为问题仅仅在于，不存在的国家正在寻找时间框架中的任何重叠。但是，只有在开始重叠时才需要设置标志。干得好。祝你好运

WHILE(@@ROWCOUNT > 0)   -- we did something the last insert
BEGIN

    INSERT INTO @Results
    SELECT TOP 1 * 
    FROM #Dates d
    WHERE d.StartDate = 
        (SELECT MIN(StartDate)
        FROM #Dates
END 

SELECT *
FROM @Results
ORDER BY 1

DROP TABLE #Dates;

Id  StartDate               EndDate
1   2017-03-18 00:00:00.000 2017-03-18 02:00:00.000
4   2017-03-18 02:10:00.000 2017-03-18 02:30:00.000
5   2017-03-18 02:35:00.000 2017-03-18 02:50:00.000
6   2017-03-18 03:20:00.000 2017-03-18 03:50:00.000
7   2017-03-18 05:00:00.000 2017-03-18 05:30:00.000
9   2017-03-18 05:50:00.000 2017-03-18 08:00:00.000

WITH PIP AS (
    SELECT [D1].*, [T].*
    FROM #Dates [D1]
    OUTER APPLY (
        SELECT TOP 1 Id AS NextId FROM #Dates [D2]
        WHERE 
            NOT ([D1].StartDate <= [D2].EndDate AND [D1].EndDate >= [D2].StartDate) AND [D2].Id > [D1].Id
        ORDER BY [D2].StartDate
    ) AS [T]
), POP AS (
    SELECT [T].Id, [T].StartDate, [T].EndDate, [T].NextId
        FROM PIP [T]
    WHERE [T].Id = 1
    UNION ALL
    SELECT [X].Id, [X].StartDate, [X].EndDate, [X].NextId
        FROM PIP [X]
    JOIN POP [H] ON [H].NextId = [X].Id
)
SELECT * FROM POP;