如何 select 与固定时间范围 (B) 重叠的时间范围 (A)，其中 A 的开始时间或结束时间超出 SQL 服务器中的 B 范围

Question

This problem corresponds to a real-world scenario wherein an employee is given a shift schedule (ex. 8am-5pm). In this shift, an employee is entitled to have one or more fixed breaks (ex. 10am-10:30am & 3pm-30:30pm). If an employee clocks out before 10am or after 10:30 am (ex. 9:45am-10:20am or 10:20am-10:40am), he/she will get a penalty deduction, called mid-breaks , for clocking out for breaks that are not in schedule.

Scheduled Fixed Breaks:

Start Time     End Time
------------------------
10:00 am       10:30 am
03:00 pm       03:30 pm

Time Logs:

Time          Type
---------------------
08:00 am      IN
10:00 am      OUT
10:40 am      IN
12:00 pm      OUT (lunch break)
01:00 pm      IN  (lunch break)
02:45 pm      OUT
03:30 pm      IN
04:15 PM      OUT
04:25 PM      IN
05:00 pm      OUT

Expected Output

Actual Break Start    Actual Break End    Mid-break (penalty)    Fixed Break violated
---------------------------------------------------------------------------------
10:00 am              10:40 am            10 minutes             10:00 am - 10:30 am
02:45 pm              03:30 pm            15 minutes             03:00 pm - 03:30 pm
04:15 pm              04:25 pm            10 minutes             03:00 pm - 03:30 pm

1. The 1st mid-break is 10 minutes because the 10:40 am clock out is beyond the 10:00 am - 10:30 am break.
2. The 2nd mid-break is 15 minutes because the employee clocked out 15 minutes before the scheduled 10:00 am - 10:30 am break.
3. The last mid-break penalty took 10 minutes because the employee clocked out in neither of the scheduled breaks.

Note: To make things simpler, AM time logs will be validated only against the AM breaks and PM time logs will be validate only against PM breaks. So the 2:45 breaks will not violate the break schedule for 10am.

I was only able to pair INs and OUTs (as shown in the example below) but don't know how query the mid-breaks as explained above.

Sample query to pair INs and OUTs:

SELECT
    CASE WHEN t.LogType = 'CheckOut' THEN NULL ELSE t.CheckTime END [Time IN],
    CASE WHEN t.LogType = 'CheckOut' THEN t.CheckTime ELSE x.CheckTime END [Time Out]
FROM TimeLogs t
OUTER APPLY
(
    SELECT TOP 1 Id, CheckTime, LogType
    FROM TimeLogs
    WHERE t.CheckTime < CheckTime
    AND t.LogType <> LogType
    AND t.EmployeeId = EmployeeId
    ORDER BY CAST(CheckTime AS TIME)
) x
WHERE t.EmployeeId = @EmployeeId
    AND (t.LogType = 'CheckIn' )
ORDER BY CAST(t.CheckTime AS TIME)

Time In     Time Out  
---------------------- 
08:00 am    10:00 am
10:40 am    12:00 pm
01:00 pm    02:45 pm
03:30 pm    04:15 pm
04:25 pm    05:00 pm

The only remaining problem for me now is how to query the expected output (see above expected output table above) to calculate the mid-breaks. Any help from anyone is highly appreciated. Thanks!

Answer 1

So, here a solution I found out working. However, it might require additional fixes if you encounter - as mentioned before - incorrect logging like sequential logging of one type etc. ...

According to your example, the lunch break was accepted even if it was not included in the breaks table - I added it... you might hard-code this if required.

DECLARE @tBreaks TABLE(
ID int,
StartTime time,
EndTime time
)

INSERT INTO @tBreaks VALUES(1,'10:00:00','10:30:00'),(2,'15:00:00','15:30:00'), (3,'12:00:00','13:00:00')

DECLARE @tTimeLog TABLE(
lTime TIME,
lType NVARCHAR(5)
)

INSERT INTO @tTimeLog VALUES
('08:00:00', 'IN')
,('10:00:00', 'OUT')
,('10:40:00', 'IN')
,('12:00:00', 'OUT')
,('13:00:00', 'IN')
,('14:45:00', 'OUT')
,('15:30:00', 'IN')
,('16:15:00', 'OUT')
,('16:25:00', 'IN')
,('17:00:00', 'OUT')

;WITH cteBreaks AS(
SELECT *, ROW_NUMBER() OVER (ORDER BY StartTime DESC) rn
  FROM @tBreaks
),
cteIn AS(
SELECT lTime AS TIn, LEAD(lTime) OVER (ORDER BY lTime) NextTIn
  FROM @tTimeLog
  WHERE lType = 'IN'
),
cteOut AS(
SELECT lTime AS TOut
  FROM @tTimeLog
  WHERE lType = 'OUT'
),
cteLogTimes AS(
SELECT i.Tin, o.TOut, i.NextTIn
  FROM cteIn i
  LEFT JOIN cteOut o ON o.TOut > i.TIn AND o.TOut <= ISNULL(i.NextTIn, o.TOut)
)
SELECT CONVERT(varchar(15), l.TOut, 100) AS ActualBreakStart
       ,CONVERT(varchar(15), l.NextTIn, 100) AS ActualBreakEnd
       ,CASE WHEN l.TOut < t2.StartTime THEN DATEDIFF(MINUTE, l.TOut, t2.StartTime) ELSE 0 END +
           CASE WHEN l.TOut <= t2.StartTime AND l.NextTIn > t2.EndTime THEN DATEDIFF(MINUTE, t2.EndTime, l.NextTIn) ELSE 0 END +
           CASE WHEN l.TOut > t2.EndTime AND t2.rn = 1 THEN DATEDIFF(MINUTE, l.TOut, l.NextTIn) ELSE 0 END AS Penalty
       ,CONCAT(CONVERT(varchar(15), t2.StartTime), ' - ', CONVERT(varchar(15), t2.EndTime)) AS FixedBreakViolated
  FROM cteLogTimes l
  LEFT JOIN cteBreaks t ON l.TOut >= t.StartTime AND l.NextTIn <= t.EndTime
  LEFT JOIN cteBreaks t2 ON (l.TOut BETWEEN t2.StartTime AND t2.EndTime) OR (l.NextTIn BETWEEN t2.StartTime AND t2.EndTime) OR (l.TOut > t2.EndTime AND t2.rn = 1) OR (l.TOut < t2.StartTime AND l.NextTIn > t2.EndTime)
  WHERE l.NextTIn IS NOT NULL
    AND t.ID IS NULL

The result looks like this: