Dynamic Pivot multiple columns in SQL Server

Question

I have a table like this

Id   Name   FromAddress   ToAddress
1    Joey      ABC          JKL
2    Joey      DEF          MNP
3    Joey      GHI          OQR


        

Answer 1

ok, I created a temp table to do some testing on. The solution requires an unpivot first. I recommend running with/without the extra test data to get a sense of some other behaviors that surround this solution -- the weirdness that comes with the MAX aggregation and lack of new rows that you might have expected when changing the value in 'name'.

GL. Hope it helps.

-------------------------
-- Some test data here
CREATE table #addresses ( Id int, Name varchar(5), FromAddress varchar(5), ToAddress varchar(5))
insert into #addresses(id, Name, FromAddress, ToAddress) Values
(1,'Joey', 'ABC', 'JKL')
, (2,'Joey', 'DEF', 'MNO')
, (3,'Joey', 'GHI', 'PQR')
, (4,'Spike', 'XXX', 'YYY')
, (1,'Spike', 'QQQ', 'RRR')

-------------------------
--  Solution starts here.  create a temp table and unpivot your data into it.
--  Your initial technique of does not work, PIVOT only supports one aggregation

CREATE table #unpvt(RowColCode varchar(20), vals varchar(20))
Insert into #unpvt
SELECT ColCode + '_' + Cast(ID as varchar(2)) as RowColCode, vals
FROM #addresses a
UNPIVOT
    (vals for ColCode in (Name,FromAddress,ToAddress)) c

-------------------------
--  Read the temp table for a column list

declare @ColList nvarchar(max)
set @ColList = STUFF((
    SELECT distinct ',[' + t.RowColCode + ']'
            FROM #unpvt t
            FOR XML PATH(''), TYPE
            ).value('.', 'NVARCHAR(MAX)'),1,1,'')


-------------------------
-- 're pivot' the data using your new column list

declare @qry varchar(max)
set @qry = '

select *
from
    #unpvt
    PIVOT(
    MAX(vals)
    FOR RowColCode in (' +@ColList +  ') 
    ) rslt
'


execute(@qry)  

Answer 2

I believe the problem is the IN () expression in the PIVOTs. The column list explicitly has to be a list of fields names, not a function and not a list of varchar literals or function values. You've got a REPLACE() function in there. The engine expects to be looking for a field named [REPLACE] and then gets confused by the open parentheses that shows up.

This is valid (square brackets for emphasis):

SELECT VendorID, Employee, Orders
FROM 
   (SELECT VendorID, Emp1, Emp2, Emp3, Emp4, Emp5
   FROM pvt) p
UNPIVOT
   (Orders FOR Employee IN 
      ([Emp1], [Emp2], [Emp3], [Emp4], [Emp5])
)AS unpvt;

This is not:

SELECT VendorID, Employee, Orders
FROM 
   (SELECT VendorID, Emp1, Emp2, Emp3, Emp4, Emp5
   FROM pvt) p
UNPIVOT
   (Orders FOR Employee IN 
      ('Emp1', 'Emp2', 'Emp3', 'Emp4', 'Emp5')
)AS unpvt;

And this is not valid:

SELECT VendorID, Employee, Orders
FROM 
   (SELECT VendorID, Emp1, Emp2, Emp3, Emp4, Emp5
   FROM pvt) p
UNPIVOT
   (Orders FOR Employee IN 
      (REPLACE('Emp1','1','A'), REPLACE('Emp2','2','B'))
)AS unpvt;

Replace the execute(@query) with a select @query or print @query to see the query your code generated and troubleshoot the syntax in a query analyzer that way. Then work backwards.

You want to do the REPLACE() at the same level you're building the query. The query that ends up in the @query variable should already have the column names fixed.

Alternately, you could generate @colsFromLabels, @colsToLabels, @colsFrom and @colsTo with the former two have the 'from' and to bits added and the latter two just being column names.

Your desired output is a little gross as far as square bracket escaping, too.