How to get all in and out time for an particular employee? - sql

Question

My table is as below:
id time_stamp Access Type
1001 2017-09-05 09:35:00 IN
1002 2017-09-05 11:00:00 IN
1001 2017-09-05 12:00:00 OUT
1002 2017-09-05 12:25:00 OUT
1001 2017-09-05 13:00:00 IN
1002 2017-09-05 14:00:00 IN
1001 2017-09-05 17:00:00 OUT
1002 2017-09-05 18:00:00 OUT
I have tried this query below:
SELECT ROW_NUMBER() OVER (
ORDER BY A.emp_reader_id ASC
) AS SNo
,B.emp_code
,B.emp_name
,CASE
WHEN F.event_entry_name = 'IN'
THEN A.DT
END AS in_time
,CASE
WHEN F.event_entry_name = 'OUT'
THEN A.DT
END AS out_time
,cast(left(CONVERT(TIME, a.DT), 5) AS VARCHAR) AS 'time'
,isnull(B.areaname, 'OAE6080036073000006') AS areaname
,C.dept_name
,b.emp_reader_id
,isnull(c.dept_name, '') AS group_name
,CONVERT(CHAR(11), '2017/12/30', 103) AS StartDate
,CONVERT(CHAR(11), '2018/01/11', 103) AS ToDate
,0 AS emp_card_no
FROM dbo.trnevents AS A
LEFT OUTER JOIN dbo.employee AS B ON A.emp_reader_id = B.emp_reader_id
LEFT OUTER JOIN dbo.departments AS C ON B.dept_id = C.dept_id
LEFT OUTER JOIN dbo.DevicePersonnelarea AS E ON A.POINTID = E.areaid
LEFT OUTER JOIN dbo.Event_entry AS F ON A.EVENTID = F.event_entry_id
ORDER BY A.emp_reader_id ASC
It works but it takes like below. Sometime have same in event and out event :
SNo emp_code emp_name in_time out_time time areaname dept_name emp_reader_id group_name StartDate ToDate emp_card_no
1 102 Ihsan Titi NULL 2017-12-30 12:16:26.000 12:16 Dubai Sales 102 Sales 2017/12/30 2018/01/11 0
2 102 Ihsan Titi NULL 2017-12-30 12:16:27.000 12:16 Dubai Sales 102 Sales 2017/12/30 2018/01/11 0
3 102 Ihsan Titi 2017-12-30 12:44:26.000 NULL 12:44 Dubai Sales 102 Sales 2017/12/30 2018/01/11 0
4 102 Ihsan Titi 2017-12-30 16:27:48.000 NULL 16:27 Dubai Sales 102 Sales 2017/12/30 2018/01/11 0
Expected output:
SNo emp_code emp_name in_time out_time time areaname dept_name emp_reader_id group_name StartDate ToDate emp_card_no
1 102 Ihsan Titi 2017-12-30 12:16:26.000 2017-12-30 12:44:26.000 12:16 Dubai Sales 102 Sales 2017/12/30 2018/01/11 0
2 102 Ihsan Titi 2017-12-30 12:50:26.000 2017-12-30 16:27:48.000 12:16 Dubai Sales 102 Sales 2017/12/30 2018/01/11 0
kindly help i stuck here to get like this..

Answer 1

you can use this :
select A_In.emp_reader_id as empId,A_In.Belongs_to,A_In.DeviceSerialNumber,
DT as EntryTime,
(
select min(DT) as OutTime
from trnevents A_Out
where EVENTID like 'IN'
and A_Out.emp_reader_id = A_In.emp_reader_id
and A_Out.DT > A_In.DT and DATEDIFF(day,A_In.Dt,A_Out.DT)=0
) as ExitTime from trnevents A_In where EVENTID like 'OUT'
from trnevents A_In

Answer 2

The way I've approached it below is to say that if an event is the same type as the event before it then treat it as a "rogue".
Rogues always sit on their own, never paired with any other event.
All other events get paired such that IN is the first item and OUT is the second item.
Then I can group everything up to reduce pairs down to single rows.
WITH
rogue_check
AS
(
SELECT
CASE WHEN LAG(F.event_entry_name) OVER (PARTITION BY A.emp_reader_number ORDER BY A.DT) = F.event_entry_name THEN 1 ELSE 0 END AS is_rogue,
*
FROM
trnevents AS A
LEFT JOIN
EVent_entry AS F
ON F.event_entry_id = A.event_id
),
sorted AS
(
SELECT
ROW_NUMBER() OVER ( ORDER BY DT) AS event_sequence_id,
ROW_NUMBER() OVER (PARTITION BY emp_reader_number, is_rogue ORDER BY DT) AS employee_checked_event_sequence_id,
*
FROM
rogue_check
)
SELECT
MIN(event_sequence_id) AS unique_id,
emp_reader_number,
MAX(CASE WHEN event_entry_name = 'IN' THEN DT END) AS time_in,
MAX(CASE WHEN event_entry_name = 'OUT' THEN DT END) AS time_out
FROM
sorted
GROUP BY
emp_reader_number,
is_rogue,
employee_checked_event_sequence_id - CASE WHEN is_rogue = 1 OR event_entry_name = 'IN' THEN 0 ELSE 1 END
ORDER BY
emp_reader_number,
unique_id
;
Example Schema:
CREATE TABLE trnevents (
emp_reader_number INT,
DT DATETIME,
event_id INT
);
CREATE TABLE Event_entry (
event_entry_id INT,
event_entry_name NVARCHAR(32)
);
Example Data:
INSERT INTO Event_entry VALUES (0, N'IN'), (1, N'OUT');
INSERT INTO trnevents VALUES
(1, '2017-01-01 08:00', 0),
(1, '2017-01-01 08:01', 0),
(1, '2017-01-01 12:00', 1),
(1, '2017-01-01 13:00', 0),
(1, '2017-01-01 17:00', 1),
(1, '2017-01-01 17:01', 1)
;
Example Results:
unique_id emp_reader_number time_in time_out
1 1 01/01/2017 08:00:00 01/01/2017 12:00:00
2 1 01/01/2017 08:01:00 null
4 1 01/01/2017 13:00:00 01/01/2017 17:00:00
6 1 null 01/01/2017 17:01:00
The GROUP BY turned out a bit more fiddly than I anticipated on the train and so may cause an expensive SORT in the execution plan for large data sets. I'll also think about an alternative shortly.
Here is a demo with some simple dummy data demonstrating that it works for those cases at least. (Feel free to update it with other cases if they demonstrate any problems)
http://dbfiddle.uk/?rdbms=sqlserver_2017&fiddle=d06680d8ed374666760cdc67182aaacb

Answer 3

You can use a PIVOT
select id, [in], out
from
( select
id, time_stamp, accessType,
(ROW_NUMBER() over (partition by id order by time_stamp) -1 )/ 2 rn
from yourtable ) src
pivot
(min(time_stamp) for accessType in ([in],[out])) p
This assumes that each "in" is followed by an "out" and uses row_number to group those pairs of times.