Using XML PATH to manipulate row data when grouping - sql

Question

In a sql query I'm trying to do two things.
I have a table Attendance like this
TABLE
EID | PID | In_Time | Out_Time | Shift
__________________________________________________________
100 | S001 | 2014-05-01 07:10 | 2014-05-01 19:20 | D
100 | S001 | 2014-05-04 07:00 | 2014-05-04 19:00 | D
100 | S001 | 2014-05-04 19:00 | 2014-05-05 07:00 | N
EID -EmployeeID
PID -PointID (Location)
D - Day Shift
N - Night Shift
When I group by all fields except Shift, (When grouping only the DatePart of In_Time fields will be taken) I want to get this
INTERMEDIATE STEP
EID | DAY | Shift |
___________________
100 | 01 | D |
100 | 04 | D/N |
Finally I want to PIVOT this to get following result
EXPECTED FINAL RESULT
EmployeeID | 01 | 02 | 03 | 04 |
__________________________________
100 | D | _ | _ | D/N |
I Use following query in this purpose but I'm getting slightly different result.
SELECT EID AS EmployeeID, [1],[2],[3],[4]
FROM (
SELECT
EID, datepart(dd,in_time) as [DAY],
STUFF((
SELECT '/ ' + Shift
FROM Attendance
WHERE ([in_time] = Results.[in_time] )
FOR XML PATH(''),TYPE).value('(./text())[1]','VARCHAR(MAX)')
,1,2,'') AS Shifts
FROM Attendance Results
WHERE EID = '100' AND PID ='C002'
GROUP BY EID , in_time
) AS SourceTable
PIVOT
(
MAX (Shifts )
FOR [DAY] IN ( [1],[2],[3],[4])
) AS PivotTable
This is the result of the query
EmployeeID | 01 | 02 | 03 | 04 |
________________________________
100 | D | _ | _ | N/N |
So something wrong in my query and could you pleas help me to sort this out? What's I'm missing in this query? Do you know any better way to do this?
EDIT: I just realized that the above code works well as long as the Attendance table contains records relevant to a single employee and a single Point (PID).
If the above table has details of several employees (EIDs) who works at different locations (PIDs) then the out put is wrong. So I see the code is not consistent but my knowledge in sql seems not enough to sort this out with out a help :(

Answer 1

the below answer may help you. The SQL looks lengthy, but the logic is simple, just grouping D and N data and finally joining. may be you can work from this.
DECLARE #Lu_Dt table
(id int identity(1,1),Dt Date)
INSERT INTO #Lu_Dt VALUES('2014-05-01'),
('2014-05-02'),('2014-05-03'),
('2014-05-04'),('2014-05-05')
DECLARE #tab table
(EID int,PID varchar(15),In_Time datetime,Out_Time datetime,[Shift] Char(1))
INSERT INTO #tab Values
(100,'S001','2014-05-01 07:10','2014-05-01 19:20','D'),
(100,'S001','2014-05-04 07:00','2014-05-04 19:00','D'),
(100,'S001','2014-05-04 19:00','2014-05-05 07:00','N')
SELECT * FROM #tab
DECLARE #refTab table
(id int identity(1,1),EID int, Dt Date,[shift] varchar(3))
INSERT INTO #refTab
SELECT Lu.EID,Lu.Dt,coalesce(LU1.[Flag],LU2.[Flag]) [shift]
FROM (SELECT *,100 EID FROM #Lu_Dt) Lu
LEFT JOIN (SELECT D.EID,D.In_Time,'D/N' [Flag]
FROM (SELECT EID,CAST(In_Time AS DATE) In_Time FROM #tab Where Shift = 'D') D
JOIN (SELECT EID,CAST(In_Time AS DATE) In_Time FROM #tab Where Shift = 'N') N
ON D.In_Time = N.In_Time AND D.EID = N.EID) LU1 ON Lu.Dt = LU1.In_Time
LEFT JOIN (SELECT CAST(In_Time AS DATE) In_Time,'D' [Flag] FROM #tab Where Shift = 'D') LU2 ON Lu.Dt = LU2.In_Time
LEFT JOIN (SELECT CAST(In_Time AS DATE) In_Time,'N' [Flag] FROM #tab Where Shift = 'N') LU3 ON Lu.Dt = LU3.In_Time
SELECT * FROM #refTab
SELECT *
FROM (SELECT EID,Dt,[shift] from #refTab) As src
PIVOT
(MAX([shift]) for Dt in
( [2014-05-01],
[2014-05-02],
[2014-05-03],
[2014-05-04],
[2014-05-05])) AS PivotTable;
Result
As there is a lot of hard coding, it cant be used as such.Hope this helped.