Calculate Employee Timesheet Data - sql

Question

Suppose you have a timeclock table with only a timestamp entry for each timeclock event:
Userid CheckTime
312 2018-05-08 05:52:00
312 2018-05-08 18:06:00
312 2018-05-10 05:55:00
312 2018-05-10 18:00:00
312 2018-05-11 05:58:00
312 2018-05-11 18:00:00
312 2018-05-12 05:35:00
312 2018-05-12 18:00:00
How can I tally the events in SQL Server to show like this?
Day Date In Out Reg OT
Tuesday 5/8/2018 5:52AM 6:06PM 12.00 0.00
Thursday 5/10/2018 5:55AM 6:00PM 12.00 0.00
Friday 5/11/2018 5:58AM 6:00PM 12.00 0.00
Saturday 5/12/2018 5:35AM 6:00PM 12.00 0.42
Plus, we have overnight people who start their shift in one day and carry over to another day.
I'm not sure how to calculate this since it is row based vs column based.
I've tried this...but it doesn't work correctly...
;WITH emp
AS (SELECT [UserID],
[CheckTime],
CAST([CheckTime] AS DATE) AS [Day],
Row_Number()
OVER( PARTITION BY [UserID], CAST([CheckTime] AS DATE)
ORDER BY [UserID], [CheckTime]) AS [RowNumber]
FROM [dbo].[Clock_Data] WHERE CHECKTIME
BETWEEN '2018-05-06' AND '2018-05-13')
SELECT
t1.[UserID],
E.[Last Name]AS [EMPID],
MIN(t1.[CheckTime]) AS [time_in],
MAX(t2.[CheckTime]) AS [time_out],
CAST((SUM(ISNULL(DATEDIFF(ss, t1.[CheckTime],
t2.[CheckTime]) , 0)) / 3600)-1 AS VARCHAR(10)) + '.' +
FROM emp AS t1
LEFT JOIN emp AS t2
ON ( t1.[UserID] = t2.[UserID]
AND t1.[Day] = t2.[Day]
AND t1.[RowNumber] = ( t2.[RowNumber] - 1 )
AND t2.[RowNumber] % 2 = 0
)
INNER JOIN Employees as E on t1.Userid = E.[ID Number]
GROUP BY t1.[UserID], E.[Last Name]
ORDER BY t1.[UserID]

Answer 1

As mentioned in the comments above there are a LOT of complexities in this type of query. Missed/duplicate punches. Daylight saving time. Holidays, weekends. The types of things that might count as O/T for whatever rules you need. But for a nice clean set of data like you have you can do this fairly easily. This is by no means a complete solution because you have a LOT of things to iron out for details. But this should serve as a decent starting point.
declare #Something table
(
Userid int
, CheckTime datetime
)
insert #Something values
(312, '2018-05-08 05:52:00')
, (312, '2018-05-08 18:06:00')
, (312, '2018-05-10 05:55:00')
, (312, '2018-05-10 18:00:00')
, (312, '2018-05-11 05:58:00')
, (312, '2018-05-11 18:00:00')
, (312, '2018-05-12 05:35:00')
, (312, '2018-05-12 18:00:00');
with OrderedResults as
(
select *
, RowNum = ROW_NUMBER() over(partition by Userid order by CheckTime)
from #Something
)
, InPunch as
(
select *
, GroupNum = Row_Number () over(partition by Userid order by RowNum)
from OrderedResults
where RowNum % 2 = 1
)
, OutPunch as
(
select *
, GroupNum = Row_Number () over(partition by Userid order by RowNum)
from OrderedResults
where RowNum % 2 = 0
)
select ip.Userid
, PunchDate = convert(date, ip.CheckTime)
, CheckIn = ip.CheckTime
, CheckOut = op.CheckTime
from InPunch ip
join OutPunch op on op.GroupNum = ip.GroupNum