SQL: Sum timestamp intervals of the same day - sql

Question

I'm setting up a new SQL query to summarize records from a table of employee attendance. These records are downloaded from a fingerprint or RFID sensor and recorded on the same table. I want to get the amount of hours on the workplace.
Everything works fine if the employee comes in and leaves the workplace once at day. The device generates 2 records on the table and that's are not a problem. It's easy identify entrance and exit time.
But I don't realize how to solve if the person comes in, take a break (leaves workplace) and after that he comes in again until the exit time.
Assume that they are always even records (arrival and exit timestamps) on each interval. Also, an employee never checks in one day and leaves on the next day.
I have the following query. Remember: this only gets the minimum timestamp (arrival time) and the max timestamp (leave time).
SELECT Userid, Name, Date, Entrance, Exit, Hours FROM
(SELECT Userid AS user,
CONVERT(VARCHAR, CONVERT(TIME, min(Checktime))) AS Entrance,
CONVERT(VARCHAR, CONVERT(TIME, max(Checktime))) AS Exit,
CONVERT(VARCHAR, CONVERT(TIME, max(Checktime)-min(CheckTime))) AS Hours,
CONVERT(VARCHAR, CONVERT(DATE, CheckTime)) AS Fecha,
COUNT(*) AS Regs,
SUM(edited) AS edited FROM attendance
WHERE CONVERT(DATE, CheckTime) < CONVERT(DATE, GETDATE())
GROUP BY Userid, CONVERT(DATE, CheckTime)) AS Hs
INNER JOIN Userinfo
ON Userinfo.Userid = Hs.user
ORDER BY Date DESC, Name ASC;
For example, if the table has the following records:
id | Logid | Userid | CheckTime | edited
1 | 10 | 1 | 2019-06-18 8:00:00 | 0
2 | 11 | 1 | 2019-06-18 12:00:00 | 0
3 | 12 | 1 | 2019-06-18 15:00:00 | 0
4 | 13 | 1 | 2019-06-18 17:00:00 | 0
5 | 14 | 2 | 2019-06-18 8:00:00 | 0
6 | 15 | 2 | 2019-06-18 17:00:00 | 0
What I get:
Userid | Name | Date | Entrance | Exit | Hours | edited
1 | Gandalf | 2019-06-18 | 8:00:00 | 17:00:00 | 9:00:00 | 0
2 | Frodo | 2019-06-18 | 8:00:00 | 17:00:00 | 9:00:00 | 0
What I need:
Userid | Name | Date | Entrance | Exit | Hours | edited
1 | Gandalf | 2019-06-18 | 8:00:00 | 17:00:00 | 6:00:00 | 0
2 | Frodo | 2019-06-18 | 8:00:00 | 17:00:00 | 9:00:00 | 0
The total time was calculated from (12:00:00 - 8:00:00) + (17:00:00 - 15:00:00).
The columns "Entrance" and "Exit" on this case are not necessary at all.
Have you an idea how I can solve this? Thank you very much!

Answer 1

This assume you have pair enter/exit and handle multiple breaks.
SQL DEMO
with cte as (
SELECT *, ROW_NUMBER() OVER (PARTITION BY [Userid], cast ([CheckTime] as Date)
ORDER BY [CheckTime]) as rn
FROM Table1 t1
)
SELECT c1.[Userid],
cast (c1.[CheckTime] as Date) as the_day,
SUM (DATEDIFF (hh, c1.[CheckTime], c2.[CheckTime])) as total_hours
FROM cte c1
JOIN cte c2
ON c1.rn = c2.rn -1
AND c1.[Userid] = c2.[Userid]
AND c1.rn % 2 = 1
GROUP BY c1.[Userid],
cast (c1.[CheckTime] as Date) ;
OUTPUT
| Userid | the_day | total_hours |
|--------|------------|-------------|
| 1 | 2019-06-18 | 6 |
| 2 | 2019-06-18 | 9 |
NOTE:
General syntax for DATEDIFF:
DATEDIFF(datepart, start_date, end_date)
Just realize the function DATEDIFF is used to calculate the time interval between two date values and return it as an integer.
So if you have 08:00 and 09:30 using hh as datepart you still get 1h. Maybe is better use mi and divide by 60

Answer 2

Perfect! Juan Carlos's solution works great!
I'm posting this because I've edited some of his code to match the original post requirements.
The code is exactly the same. Only I've changed/added a few lines
with cte as (
SELECT *, ROW_NUMBER() OVER (PARTITION BY [Userid], cast ([CheckTime] as Date)
ORDER BY [CheckTime]) as rn
FROM Table1 t1
WHERE CAST(CheckTime AS DATE) = '2019-06-17' -- Filter by specific date
)
SELECT c1.[Userid],
cast (c1.[CheckTime] as Date) as the_day,
-- Return time as HH:MM
CONVERT(VARCHAR, SUM (DATEDIFF (SECOND , c1.[CheckTime], c2.[CheckTime]))/3600) + ':' + right('00' + CONVERT(VARCHAR, CONVERT(FLOAT, (SUM (DATEDIFF (SECOND , c1.[CheckTime], c2.[CheckTime]))/60) - ((SUM (DATEDIFF (SECOND , c1.[CheckTime], c2.[CheckTime]))/3600)*60))),2) as total_time
FROM cte c1
JOIN cte c2
ON c1.rn = c2.rn -1
AND c1.[Userid] = c2.[Userid]
AND c1.rn % 2 = 1
GROUP BY c1.[Userid],
cast (c1.[CheckTime] as Date);
This query returns:
| Userid | the_day | total_time |
|--------|------------|-------------|
| 1 | 2019-06-18 | 6:00 |
| 2 | 2019-06-18 | 9:00 |