Question

The SELECT statement below is from a stored procedure I am using to retrieve data which is subsequently sent to JSON:

SELECT *
FROM (
    SELECT CAST(DateTimeUTC as SmallDateTime) as [DateTime], DataValue, VariableID
    FROM DataValues
    WHERE SiteID = @siteID and VariableID BETWEEN 9 and 10 and DateTimeUTC >= DATEADD (day, @pastDays, getdate())
     ) TableDate
PIVOT (SUM(DataValue) FOR VariableID IN ([9],[10])) PivotTable ORDER BY [DateTime]

What I would like to accomplish is to modify the procedure to accept a column range to pivot. In the above example I am only retrieving values for two variables. What if I want to retrieve VariableID 1 through 10, or 1 through 50? There has to be a way to retrieve 1 through 10 in a way other than:

 VariableID IN ([1],[2],[3],[4],[5],[6],[7],[8],[9],[10])

The Using PIVOT and UNPIVOT on MSDN doesn’t mention a way to specify a range.

I realize that it is possible to use dynamic SQL in a stored procedure, but given my own knowledge of SQL and the knowledge of those who will have to maintain this long term I am hesitant to introduce additional complexity by using dynamic SQL. There is a conceptually similar question on here, see TSQL Pivot Long List of Columns, and it was answered with a dynamic SQL solution.

Answer 1

Unfortunately, the PIVOT function does not have the ability to generate the list of columns, without using dynamic sql.

So your SQL code will be similar to this:

DECLARE @cols AS NVARCHAR(MAX),
    @query  AS NVARCHAR(MAX)

select @cols = STUFF((SELECT distinct ',' + QUOTENAME(VariableID) 
                    from DataValues
            FOR XML PATH(''), TYPE
            ).value('.', 'NVARCHAR(MAX)') 
        ,1,1,'')

set @query = 'SELECT [DateTime], ' + @cols + ' from 
             (
                SELECT CAST(DateTimeUTC as SmallDateTime) as [DateTime], 
                    DataValue, 
                    VariableID
                FROM DataValues
                WHERE SiteID = '+cast(@siteID as varchar(10))+'
                    -- and VariableID BETWEEN 9 and 10 -- remove the variableID filter
                    and DateTimeUTC >= DATEADD (day, +'cast(@pastDays as varchar(10))+', getdate())
            ) x
            pivot 
            (
                SUM(DataValue)
                for VariableID in (' + @cols + ')
            ) p '

execute(@query)

The key to this is the following code which generates the list of variableIds to become columns:

select @cols = STUFF((SELECT distinct ',' + QUOTENAME(VariableID) 
                    from DataValues
            FOR XML PATH(''), TYPE
            ).value('.', 'NVARCHAR(MAX)') 
        ,1,1,'')

This will create the list of all distinct variable id’s. This list is then added to the SQL query string to return.