从开始日期和结束日期的表中加入事件

Question

我有一个表来记录员工是否已上班和下班，但将它们作为单独的行进行跟踪。

输入：

+------------+------------+---------------+
| EmployeeID | Event_Date | Event_Type    |
+------------+------------+---------------+
| 2450770    | 2020/01/02 | 'Clocked Out' | -- Doesn't have a clocked in time within desired time range
| 2195326    | 2020/01/06 | 'Clocked In'  |
| 2195326    | 2020/01/10 | 'Clocked Out' |
| 800455     | 2020/01/15 | 'Clocked In'  |
| 2450770    | 2020/01/15 | 'Clocked In'  | -- No clock out time yet
| 800455     | 2020/01/22 | 'Clocked Out' | 
| 2195326    | 2020/01/23 | 'Clocked In'  | 
| 2331340    | 2020/01/25 | 'Clocked In'  |
| 2195326    | 2020/01/27 | 'Clocked Out' |
| 2331340    | 2020/02/01 | 'Clocked Out' |
| 2515957    | 2020/02/05 | 'Clocked In'  |
|            |            |               | -- etc

所需输出：

+------------+------------+-------------+
| EmployeeID | Clocked_In | Clocked_Out |
+------------+------------+-------------+
| 2195326    | 2020/01/06 | 2020/01/10  |
| 800455     | 2020/01/15 | 2020/01/22  |
| 2450770    | 2020/01/15 | NULL        |
| 2195326    | 2020/01/23 | 2020/01/27  |
| 2331340    | 2020/01/25 | 2020/02/01  |
+------------+------------+-------------+

这个问题似乎很简单，但我还不太清楚：我如何将那些Clocked In和Clocked Out事件放在一起，这样我就可以看到有人在一月份进场，并带有相应的进场日期（ （如果存在）没有日期限制？并发症包括没有匹配上班日期和下班日期的人员，在不同月份内上班和下班的人员，一个月内多次上班/下班的人员。据我所知，没有人连续排两次（或连续排两次）的情况。

Answer 1

类似的操作会获取数据的行号，并按EmployeeID进行分区，并按Event_Date进行排序。一旦有了，就加入员工编号和下一行编号。

通过这种方式可以处理大部分（即使不是全部）并发症，包括输入日期大于输出日期。

DROP TABLE IF EXISTS #temp
CREATE TABLE #temp (EmployeeID VARCHAR(256), Event_Date DATE, Event_Type varchar(256))

INSERT INTO #temp
VALUES 
( '2450770', '2020/01/02', 'Clocked Out'),
( '2195326', '2020/01/06', 'Clocked In'),
( '2195326', '2020/01/10', 'Clocked Out'),
( '800455', '2020/01/15', 'Clocked In'), 
( '2450770', '2020/01/15', 'Clocked In'), 
( '800455', '2020/01/22', 'Clocked Out'),
( '2195326', '2020/01/23', 'Clocked In'), 
( '2331340', '2020/01/25', 'Clocked In'), 
( '2195326', '2020/01/27', 'Clocked Out'),
( '2331340', '2020/02/01', 'Clocked Out'),
( '2515957', '2020/02/05', 'Clocked In')

DROP TABLE IF EXISTS #output
SELECT 
    EmployeeID,
    Event_Date,
    Event_Type,
    ROW_NUMBER() OVER(PARTITION BY EmployeeID ORDER BY Event_Date) AS RN
INTO #output
FROM #temp

SELECT 
    I.EmployeeID, 
    I.Event_Date AS [Clocked In], 
    O.Event_Date AS [Clocked Out]
FROM #output I 
LEFT JOIN #output O ON O.EmployeeID + O.RN = I.EmployeeID + (I.RN + 1) AND O.Event_Type = 'Clocked Out'
WHERE I.Event_Type = 'Clocked In'