T-SQL - 两个日期的两个日期时间之间的每小时秒数

【问题标题】：T-SQL - Seconds per hour between two datetimes across 2 datesT-SQL - 两个日期的两个日期时间之间的每小时秒数
【发布时间】：2018-09-11 13:38:11
【问题描述】：

我有以下用户登录和注销时间数据：

 | UserID | StartDate         |EndDate              |
 | 1033 | 06/24/2018 00:11:51 | 06/24/2018 01:03:38 |
 | 1033 | 06/24/2018 02:12:38 | 06/24/2018 02:15:51 |
 | 1033 | 06/24/2018 02:28:08 | 06/24/2018 02:36:31 |
 | 1033 | 06/24/2018 03:07:13 | 06/24/2018 06:02:05 |
 | 1033 | 06/24/2018 07:33:39 | 06/24/2018 07:33:40 |
 | 1033 | 06/24/2018 08:19:19 | 06/24/2018 12:20:03 |
 | 1033 | 06/24/2018 12:26:55 | 06/24/2018 13:30:17 |
 | 1033 | 06/24/2018 14:07:42 | 06/24/2018 14:53:03 |
 | 1033 | 06/24/2018 15:15:20 | 06/24/2018 15:33:01 |
 | 1033 | 06/24/2018 16:42:00 | 06/24/2018 16:58:13 |
 | 1033 | 06/24/2018 17:35:04 | 06/24/2018 17:49:01 |
 | 1033 | 06/24/2018 18:49:26 | 06/24/2018 19:26:18 |
 | 1033 | 06/24/2018 20:06:46 | 06/24/2018 21:00:07 |
 | 1033 | 06/24/2018 22:35:51 | 06/24/2018 22:43:57 |
 | 1033 | 06/24/2018 23:00:52 | 06/25/2018 01:24:53 |
 | 1033 | 06/25/2018 02:01:58 | 06/25/2018 02:03:47 |
 | 1033 | 06/25/2018 03:01:57 | 06/25/2018 03:45:59 |
 | 1033 | 06/25/2018 04:24:16 | 06/25/2018 04:43:52 |
 | 1033 | 06/25/2018 05:16:15 | 06/25/2018 07:39:28 |
 | 1033 | 06/25/2018 08:49:23 | 06/25/2018 09:12:06 |
 | 1033 | 06/25/2018 13:38:20 | 06/25/2018 15:16:25 |
 | 1033 | 06/25/2018 15:16:28 | 06/25/2018 16:54:34 |
 | 1033 | 06/25/2018 17:35:24 | 06/25/2018 18:25:38 |
 | 1033 | 06/25/2018 18:58:41 | 06/25/2018 19:20:56 |
 | 1033 | 06/25/2018 19:46:27 | 06/25/2018 19:47:33 |
 | 1033 | 06/25/2018 20:14:08 | 06/25/2018 20:40:20 |
 | 1033 | 06/25/2018 21:11:01 | 06/26/2018 00:36:56 |
 | 1033 | 06/26/2018 00:50:43 | 06/26/2018 09:43:53 |
 | 1033 | 06/26/2018 10:32:58 | 06/26/2018 10:33:38 |
 | 1033 | 06/26/2018 11:01:29 | 06/26/2018 11:41:24 |
 | 1033 | 06/26/2018 13:56:29 | 06/26/2018 14:52:08 |
 | 1033 | 06/26/2018 15:40:07 | 06/26/2018 16:38:18 |
 | 1033 | 06/26/2018 16:56:33 | 06/26/2018 17:19:14 |
 | 1033 | 06/26/2018 18:37:33 | 06/26/2018 19:10:44 |
 | 1033 | 06/26/2018 19:34:44 | 06/26/2018 21:30:06 |
 | 1033 | 06/26/2018 21:43:55 | 06/26/2018 21:47:51 |
 | 1033 | 06/26/2018 23:03:17 | 06/27/2018 04:26:50 |
 | 1033 | 06/27/2018 07:41:10 | 06/27/2018 07:41:11 |
 | 1033 | 06/27/2018 07:41:23 | 06/27/2018 09:56:05 |
 | 1033 | 06/27/2018 11:31:27 | 06/27/2018 12:05:42 |
 | 1033 | 06/27/2018 12:48:28 | 06/27/2018 12:49:12 |
 | 1033 | 06/27/2018 13:43:48 | 06/27/2018 14:13:04 |
 | 1033 | 06/27/2018 15:13:32 | 06/27/2018 15:46:44 |
 | 1033 | 06/27/2018 17:09:44 | 06/27/2018 17:16:15 |
 | 1033 | 06/27/2018 18:01:28 | 06/27/2018 18:35:04 |
 | 1033 | 06/27/2018 19:21:18 | 06/27/2018 19:33:47 |
 | 1033 | 06/27/2018 20:01:51 | 06/27/2018 20:04:29 |
 | 1033 | 06/27/2018 20:45:42 | 06/27/2018 22:13:48 |
 | 1033 | 06/27/2018 23:14:33 | 06/27/2018 23:28:31 |
 | 1033 | 06/27/2018 23:57:57 | 06/28/2018 04:16:47 |
 | 1033 | 06/28/2018 04:48:50 | 06/28/2018 04:50:12 |
 | 1033 | 06/28/2018 06:00:36 | 06/28/2018 08:14:20 |
 | 1033 | 06/28/2018 08:53:19 | 06/28/2018 09:09:52 |
 | 1033 | 06/28/2018 09:28:04 | 06/28/2018 10:07:02 |
 | 1033 | 06/28/2018 10:30:47 | 06/28/2018 11:07:06 |
 | 1033 | 06/28/2018 12:23:48 | 06/28/2018 12:26:52 |
 | 1033 | 06/28/2018 13:12:23 | 06/28/2018 13:24:10 |
 | 1033 | 06/28/2018 13:50:18 | 06/28/2018 13:59:04 |
 | 1033 | 06/28/2018 14:21:08 | 06/28/2018 14:56:30 |
 | 1033 | 06/28/2018 15:20:02 | 06/28/2018 15:46:18 |
 | 1033 | 06/28/2018 16:44:35 | 06/28/2018 17:09:43 |
 | 1033 | 06/28/2018 17:26:54 | 06/28/2018 17:35:20 |
 | 1033 | 06/28/2018 18:20:17 | 06/28/2018 18:42:42 |
 | 1033 | 06/28/2018 18:50:23 | 06/28/2018 19:07:15 |
 | 1033 | 06/28/2018 19:12:00 | 06/28/2018 20:06:46 |
 | 1033 | 06/28/2018 20:15:26 | 06/28/2018 20:46:14 |
 | 1033 | 06/28/2018 21:12:03 | 06/28/2018 21:12:29 |
 | 1033 | 06/28/2018 21:23:12 | 06/28/2018 21:27:14 |
 | 1033 | 06/28/2018 22:04:46 | 06/28/2018 22:17:00 |
 | 1033 | 06/28/2018 22:58:18 | 06/29/2018 01:21:10 |
 | 1033 | 06/29/2018 02:05:34 | 06/29/2018 02:10:05 |
 | 1033 | 06/29/2018 02:15:52 | 06/29/2018 07:07:20 |
 | 1033 | 06/29/2018 07:46:33 | 06/29/2018 08:06:29 |
 | 1033 | 06/29/2018 08:50:00 | 06/29/2018 08:54:24 |
 | 1033 | 06/29/2018 10:16:49 | 06/29/2018 12:47:49 |
 | 1033 | 06/29/2018 14:11:53 | 06/29/2018 15:02:08 |
 | 1033 | 06/29/2018 15:35:25 | 06/29/2018 16:46:58 |
 | 1033 | 06/29/2018 16:49:12 | 06/29/2018 16:55:00 |
 | 1033 | 06/29/2018 17:23:20 | 06/29/2018 17:50:58 |
 | 1033 | 06/29/2018 19:26:29 | 06/29/2018 19:40:41 |
 | 1033 | 06/29/2018 21:28:30 | 06/29/2018 22:00:50 |
 | 1033 | 06/29/2018 22:40:32 | 06/29/2018 22:41:46 |
 | 1033 | 06/29/2018 23:20:08 | 06/30/2018 01:24:54 |
 | 1033 | 06/30/2018 01:39:53 | 06/30/2018 06:21:02 |
 | 1033 | 06/30/2018 09:17:11 | 06/30/2018 09:17:12 |
 | 1033 | 06/30/2018 09:17:20 | 06/30/2018 09:45:50 |
 | 1033 | 06/30/2018 10:52:58 | 06/30/2018 11:46:47 |
 | 1033 | 06/30/2018 12:28:47 | 06/30/2018 14:05:42 |
 | 1033 | 06/30/2018 15:30:42 | 06/30/2018 15:37:32 |
 | 1033 | 06/30/2018 16:28:27 | 06/30/2018 16:39:04 |
 | 1033 | 06/30/2018 17:13:20 | 06/30/2018 18:09:44 |
 | 1033 | 06/30/2018 19:30:26 | 06/30/2018 20:25:35 |
 | 1033 | 06/30/2018 21:30:45 | 06/30/2018 22:25:15 |
 | 1033 | 06/30/2018 23:27:07 | 06/30/2018 23:27:35 |
 | 1033 | 06/30/2018 23:48:45 | 06/30/2018 23:50:08 |

我需要计算每个唯一日期每小时的用户登录时间（即开始日期到结束日期）。

我找到了讨论这个问题的 post，这是我目前的查询，但它没有考虑跨 2 个日期（6/24 到 6/25）的第 1 行的小时数

WITH Numbers (Number) AS
(   SELECT ROW_NUMBER() OVER(ORDER BY N1.N) - 1
    FROM (VALUES (1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N1(N)
    CROSS JOIN (VALUES (1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N2 (N)
    CROSS JOIN (VALUES (1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N3 (N)
    CROSS JOIN (VALUES (1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N4 (N)
), SampleData (userid, StartDate, EndDate) AS
(   SELECT  userid, CONVERT(DATETIME2, StartDate), CONVERT(DATETIME2, EndDate)
    FROM (VALUES 
            (1033, '06/24/2018 23:00:52', '06/25/2018 01:24:53'),
            (1033, '06/25/2018 02:01:58', '06/25/2018 02:03:47'),
            (1033, '06/25/2018 03:01:57', '06/25/2018 03:45:59')
        ) d (userid, StartDate, EndDate)
)
SELECT  d.userid,
        [Date] = CONVERT(DATE, d.StartDate),
        [Hour] = CONVERT(TIME(0), DATEADD(HOUR, DATEPART(HOUR, d.StartDate) + n.Number, 0)),
        Seconds_in = CASE  
                        -- SPECIAL CASE: START HOUR = END HOUR
                        WHEN DATEPART(HOUR, d.StartDate) = DATEPART(HOUR, d.EndDate)
                            AND DATEDIFF(DAY, d.StartDate, d.EndDate) = 0
                            THEN DATEDIFF(second, d.StartDate, d.EndDate)

                        WHEN CONVERT(DATE, d.StartDate)<CONVERT(DATE, d.EndDate) then ?????

                        -- START HOUR
                        WHEN n.Number = 0
                            THEN 3600 - DATEPART(second, d.StartDate)

                        -- END HOUR
                        WHEN n.Number = DATEDIFF(HOUR, d.StartDate, d.EndDate) 
                            THEN DATEPART(second, d.EndDate)

                        -- FULL HOURS IN BETWEEN START AND END
                        ELSE 3600

                    END
FROM    SampleData d
        INNER JOIN Numbers n 
            ON n.Number <= DATEDIFF(HOUR, d.StartDate, d.EndDate)
ORDER BY d.userid,[Date],n.Number;

理想的结果如下：

 | UserID | Date      | Hour     |  Seconds_in |
 | 1033   | 6/24/2018 | 23:00:00 |  3548       |
 | 1033   | 6/25/2018 | 0:00:00  |  3600       |
 | 1033   | 6/25/2018 | 1:00:00  |  1493       |
 | 1033   | 6/25/2018 | 2:00:00  |  109        |
 | 1033   | 6/25/2018 | 3:00:00  |  2642       |

【问题讨论】：

    标签： sql-server tsql date datetime


    【解决方案1】：

    与之前略有不同的策略

    我已经创建了一个 Temp 表 - 请参阅 [SQLFiddle] 以获得完整的解决方案1

    我创建了一个 Grp 和 StartDate_Plus 列，如下所示

    ;WITH Numbers (Number) AS
(   SELECT ROW_NUMBER() OVER(ORDER BY N1.N) - 1
    FROM (VALUES (1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N1(N)
    CROSS JOIN (VALUES (1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N2 (N)
    CROSS JOIN (VALUES (1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N3 (N)
    CROSS JOIN (VALUES (1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N4 (N)
)
    SELECT
         D.UserId
        ,D.StartDate
        ,D.EndDate
        ,StartDate_Plus = DATEADD(HOUR, N.Number, D.StartDate) --Only for Ordering of ResultSet below
        ,N.Number
        ,Grp = MAX(N.Number)OVER(PARTITION BY UserId, StartDate)
    FROM
        dbo.T1  D
    INNER JOIN Numbers N ON N.Number <= DATEDIFF(HOUR, D.StartDate, D.EndDate)

    这样输出

    UserId  StartDate                   EndDate                     StartDate_Plus              Number  Grp
1033    2018-06-24 00:11:51.0000000 2018-06-24 01:03:38.0000000 2018-06-24 00:11:51.0000000 0   1
1033    2018-06-24 00:11:51.0000000 2018-06-24 01:03:38.0000000 2018-06-24 01:11:51.0000000 1   1
1033    2018-06-24 02:12:38.0000000 2018-06-24 02:15:51.0000000 2018-06-24 02:12:38.0000000 0   0
1033    2018-06-24 02:28:08.0000000 2018-06-24 02:36:31.0000000 2018-06-24 02:28:08.0000000 0   0
1033    2018-06-24 03:07:13.0000000 2018-06-24 06:02:05.0000000 2018-06-24 03:07:13.0000000 0   3
1033    2018-06-24 03:07:13.0000000 2018-06-24 06:02:05.0000000 2018-06-24 04:07:13.0000000 1   3
1033    2018-06-24 03:07:13.0000000 2018-06-24 06:02:05.0000000 2018-06-24 05:07:13.0000000 2   3
1033    2018-06-24 03:07:13.0000000 2018-06-24 06:02:05.0000000 2018-06-24 06:07:13.0000000 3   3

    StartDate_plus 列只是将 [Number] of hours 添加到 StartDate 和 Grp 只是为了为同一事件提供一组多行

    添加这个额外的 CTE，您就可以向 Dummy StartDate_ 和 EndDate_ 显示每个小时的事件超过一小时这样的事件

    ;WITH Numbers (Number) AS
(   SELECT ROW_NUMBER() OVER(ORDER BY N1.N) - 1
    FROM (VALUES (1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N1(N)
    CROSS JOIN (VALUES (1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N2 (N)
    CROSS JOIN (VALUES (1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N3 (N)
    CROSS JOIN (VALUES (1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N4 (N)
),cteStartDate_Plus
AS(
    SELECT
         D.UserId
        ,D.StartDate
        ,D.EndDate
        ,StartDate_Plus = DATEADD(HOUR, N.Number, D.StartDate) --Only for Ordering of ResultSet below
        ,N.Number
        ,Grp = MAX(N.Number)OVER(PARTITION BY UserId, StartDate)
    FROM
        dbo.T1  D
    INNER JOIN Numbers N ON N.Number <= DATEDIFF(HOUR, D.StartDate, D.EndDate)
)
    SELECT TOP 100 PERCENT
         UserId, StartDate, EndDate, Grp
        ,CS.StartDate_Plus
        ,CS.Number
        ,[StartDate_] = CASE --apply some rounding to the StartDate if required
                            WHEN Number = 0 THEN CS.StartDate
                            ELSE DATEADD(hour, DATEDIFF(HOUR, 0, CS.StartDate_Plus), 0)
                        END
        ,[EndDate_] =   CASE --apply some rounding to the EndDate if required
                            WHEN CS.Number = CS.Grp THEN CS.EndDate
                            WHEN CS.StartDate_Plus > CS.EndDate THEN CS.EndDate
                            WHEN CS.Number <> CS.Grp AND CS.StartDate_Plus <= CS.EndDate 
                                THEN DATEADD(HOUR, DATEDIFF(HOUR, 0, DATEADD(MINUTE, 30 + DATEPART(MINUTE, DATEADD(MINUTE, 30, CS.StartDate_Plus)),CS.StartDate_Plus)), 0) 
                        END
        ,[Hour] = CONVERT(TIME(0), DATEADD(HOUR, DATEPART(HOUR, CS.StartDate_Plus), 0))
    FROM
        cteStartDate_Plus CS
    ORDER BY
        UserId, CS.StartDate_Plus

    然后将上述查询封装到另一个名为 CteDummyDates 的 CTE 中，以下查询只是为您提供所需的结果

    SELECT
      UserId
     ,[Date] = CONVERT(DATE, DD.StartDate_Plus)
     ,DD.[Hour]
    ,[Seconds_in] = DATEDIFF(SECOND, DD.StartDate_, DD.EndDate_)
FROM
    cteDummyDates DD
ORDER BY
    DD.UserId, DD.StartDate_

    输出

    UserId  Date        Hour        Seconds_in
1033    2018-06-24  00:00:00    2889
1033    2018-06-24  01:00:00    218
1033    2018-06-24  02:00:00    193
1033    2018-06-24  02:00:00    503
1033    2018-06-24  03:00:00    3167
1033    2018-06-24  04:00:00    3600
1033    2018-06-24  05:00:00    3600
1033    2018-06-24  06:00:00    125
1033    2018-06-24  07:00:00    1
1033    2018-06-24  08:00:00    2441
1033    2018-06-24  09:00:00    3600

    【讨论】：

    • 谢谢！毕竟这看起来很有希望！不幸的是，当我将其应用于整个数据集时，我的 seconds_in 超过了 3600。我添加了更多数据来说明这一点。当我将这些值添加到您的查询中时，它会产生 seconds>3600。
    • 再次感谢！这真是太好了！你太棒了！我现在在我的数据集中看到一些数据，其中 StartDates 可能由于某种原因背靠背（IE 用户连续登录两次然后注销）这给了我两次登录时间一次注销时间并使用您的查询，给出我的时间> 3600。我绞尽脑汁想弄清楚如何排除这些异常值。本质上，我需要排除在注销前有 2 次或更多登录时间的数据。在您的 SQL Fiddle 中，如果您将值 (1033, '2018-06-24 00:11:50', '2018-06-24 01:03:38') 添加为您将看到的第一个值。
    • 改变问题的性质或添加额外的变体在 SO 上是不受欢迎的，尤其是当您得到原始问题的答案时。如果此回复回答了您的问题，则接受它作为答案。打开一个新问题，使用其他变体链接回此问题。
    猜你喜欢
    • 1970-01-01
    • 1970-01-01
    • 2012-05-19
    • 2021-03-14
    • 2014-10-12
    • 2015-06-28
    • 1970-01-01
    • 1970-01-01
    • 1970-01-01
    相关资源
    最近更新 更多
    热门标签
    Java Python linux javascript Mysql C# Docker 算法 前端 SpringBoot Redis Vue spring 设计模式 .net core .net kubernetes c++ 数据库 数据结构 大数据 js 机器学习 微服务 Android Go 程序员 面试 JVM ASP.net core 云原生 人工智能 后端 PHP git CSS golang k8s Nginx Django mybatis 深度学习 多线程 React 架构 devops 爬虫 云计算 Spring Boot LeetCode