Struggling with a dynamic pivot on multiple columns with one being concatenated - sql-server

Question

I have a table with data like so:
Employee PRDate Type Code Amount Subject Eligible
1234 1/1/2015 D 1 100.00 100.00 0.00
1234 1/1/2015 D 2 200.00 0.00 0.00
5678 1/1/2015 D 1 500.00 40.00 500.00
1234 1/1/2015 E 1 300.00 30.00 300.00
5678 1/1/2015 E 1 700.00 700.00 500.00
1234 1/1/2015 E 2 400.00 200.00 0.00
1234 1/8/2015 L 55 40.00 40.00 40.00
And I need for the data to be displayed like this:
Employee PRDate D1Amt D1Subj D1Elig D2Amt D2Subj D2Elig E1Amt E1Subj E1Elig E2Amt E2Subj E2Elig L55Amt L55Subj L55Elig
1234 1/1/2015 100.00 100.00 0.00 200.00 0.00 0.00 300.00 30.00 300.00 400.00 200.00 0.00 40.00 40.00 40.00
4678 1/1/2015 500.00 40.00 500.00 700.00 700.00 500.00
I can pivot on one column but when I try combining the Type and Code columns to get the one I get conversion errors (Type is a varchar and code is a tinyint). I'm not sure how to get to the desired results other than dynamic pivot. Can the desired results be achieved?
I've gotten this far but I can't figure out how to combine the type, code and each money columns (amount, subject and eligible) to get the data under the correct columns.
IF EXISTS (
SELECT *
FROM sys.tables
WHERE name LIKE '#temp285865%')
DROP TABLE #temp285865;
Create table dbo.#temp285865
(
EDLCodetemp varchar(10)
);
INSERT INTO #temp285865
(
[EDLCodetemp]
)
SELECT DISTINCT EDLCode
FROM #results
ORDER BY EDLCode;
-- Building a comma separated list of EDLCodes in #edltemp
DECLARE #cols varchar(1000);
SELECT #cols = COALESCE(#cols + ',[' + [EDLCodetemp] + ']', '[' + [EDLCodetemp] + ']')
FROM #temp285865;
-- Building the query appending columns
DECLARE #query varchar(4000);
SET #query =
'SELECT [CoName],
[PRCo],
[PRGroup],
[PREndDate],
[PaySeq],
[EDLType],
[Hours],
[SubjectAmt],
[EligibleAmt],
[PaidMth],
[LastName],
[FirstName],
[UseOver],
[OverAmt],
[Amount],
[PRGRDescrip],
[LimitPeriod],
[LimitMth],
[PREHEmployee],
[SortName],
[PaybackAmt],
[PaybackOverAmt],
[PaybackOverYN],
[PRDTEmployee],
[TrueEarns], '
+ #cols + ' FROM
(
SELECT [CoName],
[PRCo],
[PRGroup],
[PREndDate],
[PaySeq],
[EDLType],
[Hours],
[SubjectAmt],
[EligibleAmt],
[PaidMth],
[LastName],
[FirstName],
[PRDLDescrip],
[PRECDescrip],
[UseOver],
[OverAmt],
[Amount],
[PRGRDescrip],
[LimitPeriod],
[LimitMth],
[PREHEmployee],
[SortName],
[PaybackAmt],
[PaybackOverAmt],
[PaybackOverYN],
[PRDTEmployee],
[TrueEarns],
[EDLCode]
FROM #results
) p
PIVOT (
MAX(EDLCode)
FOR [EDLCode] IN (' + #cols + ')
)
as pvt';
EXEC(#query);
DROP TABLE #temp285865;

Answer 1

The following PIVOT with dynamic SQL would give you the result you want, based on the input data you provided (I changed the PRDate in the last row though).
The first statement builds an intermediate table #bt with the column names you want and the associated value. Then the column names are built in #cols for the dynamic SQL statement. Finally the intermediate table #bt is pivoted with a dynamic SQL statement using the #cols to pivot.
SET NOCOUNT ON;
CREATE TABLE #t(
Employee INT,
PRDate DATETIME,
Type CHAR(1),
Code TINYINT,
Amount DECIMAL(28,2),
Subject DECIMAL(28,2),
Eligible DECIMAL(28,2)
);
INSERT INTO #t(Employee,PRDate,Type,Code,Amount,Subject,Eligible)VALUES(1234,'2015-01-01','D',1,100.00,100.00,0.00);
INSERT INTO #t(Employee,PRDate,Type,Code,Amount,Subject,Eligible)VALUES(1234,'2015-01-01','D',2,200.00,0.00,0.00);
INSERT INTO #t(Employee,PRDate,Type,Code,Amount,Subject,Eligible)VALUES(5678,'2015-01-01','D',1,500.00,40.00,500.00);
INSERT INTO #t(Employee,PRDate,Type,Code,Amount,Subject,Eligible)VALUES(1234,'2015-01-01','E',1,300.00,30.00,300.00);
INSERT INTO #t(Employee,PRDate,Type,Code,Amount,Subject,Eligible)VALUES(5678,'2015-01-01','E',1,700.00,700.00,500.00);
INSERT INTO #t(Employee,PRDate,Type,Code,Amount,Subject,Eligible)VALUES(1234,'2015-01-01','E',2,400.00,200.00,0.00);
INSERT INTO #t(Employee,PRDate,Type,Code,Amount,Subject,Eligible)VALUES(1234,'2015-01-01','L',55,40.00,40.00,40.00);
SELECT
Employee,
PRDate,
Type+CAST(Code AS VARCHAR(3))+ca.name AS colname,
ca.val
INTO
#bt
FROM
#t
CROSS APPLY(
SELECT Amount AS val,'Amt' AS name
UNION ALL
SELECT Subject AS val,'Subj' AS name
UNION ALL
SELECT Eligible AS val,'Elig' AS name
) AS ca;
/* If you need to SUM for all dates, instead use this statement to create #bt
SELECT
Employee,
Type+CAST(Code AS VARCHAR(3))+ca.name AS colname,
ca.val
INTO
#bt
FROM
(
SELECT
Employee,
Type,
Code,
SUM(Amount) AS Amount,
SUM(Subject) AS Subject,
SUM(Eligible) AS Eligible
FROM
#t
GROUP BY
Employee,
Type,
Code
) AS t
CROSS APPLY(
SELECT Amount AS val,'Amt' AS name
UNION ALL
SELECT Subject AS val,'Subj' AS name
UNION ALL
SELECT Eligible AS val,'Elig' AS name
) AS ca;
*/
DECLARE #cols VARCHAR(8000);
SET #cols=STUFF(
(SELECT DISTINCT
',['+colname+']'
FROM
#bt
FOR XML PATH('')),
1,
1,
''
);
DECLARE #sql VARCHAR(MAX);
SET #sql='
SELECT
*
FROM
#bt
PIVOT(
MAX(val)
FOR colname IN ('+#cols+')
) AS piv
';
EXEC (#sql);
DROP TABLE #bt;
DROP TABLE #t;
The result is the following:
Employee PRDate D1Amt D1Elig D1Subj D2Amt D2Elig D2Subj E1Amt E1Elig E1Subj E2Amt E2Elig E2Subj L55Amt L55Elig L55Subj
1234 2015-01-01 100.00 0.00 100.00 200.00 0.00 0.00 300.00 300.00 30.00 400.00 0.00 200.00 40.00 40.00 40.00
5678 2015-01-01 500.00 500.00 40.00 NULL NULL NULL 700.00 500.00 700.00 NULL NULL NULL NULL NULL NULL