Pivot when you dont know amount of columns and coloum names

  • I am trying to figure out how to pivot a temporary table.

    I have a table which starts with a date but the number of columns and columns names will vary but will be type INT (Data, col2,col3,col4………….n)

    So it could look like

    Date , TS-Sales, Budget , Toms sales

    01-Jan-14,100,120,300

    02-Jan-14,80,150,300

    03-Jan-14,100,20,180

    Turned to this

    01-jan-14, 02-jan-14, 03-jan-14

    100,80,100

    120,150,20

    300,300,180

    Or even just the date and a SUM

    What I want is to be able to sum al the columns but without knowing the name and the amount columns to start with this is a manually processes. How could I automate this?

  • Something like this should work

    CREATE TABLE #RawData

    (

    CalDate Date

    ,Value1 INT

    ,Value2 INT

    ,Value3 INT

    )

    INSERT INTO #RawData

    VALUES ('01-Jan-2014',100,120,300),('02-Jan-2014',80,150,300),('03-Jan-2014',100,20,180)

    Select * from #RawData

    DECLARE @SQL NVARCHAR(4000) = 'SELECT '

    DECLARE @SQLFrom NVARCHAR(4000) = 'FROM #RawData'

    SET

    @SQL=@SQL+

    (

    SELECT

    STUFF

    (

    (

    SELECT

    ', SUM(CASE WHEN CalDate = '''

    +CONVERT(Nvarchar(10),CalDate)

    +''' then Value1+Value2+Value3 else 0 end) AS ['

    +CONVERT(Nvarchar(10),CalDate)

    +']'

    FROM #RawData

    FOR XML PATH('')

    )

    ,1,1,'')

    )

    +@SQLFrom

    Print @SQL

    EXEC sp_executesql @SQL

    This will sum up the total, however there are a couple of problems, the length of the string, could breach the max for the sp_executesql.

    Also if you want to split it out you would need to group by something else.

    _________________________________________________________________________
    SSC Guide to Posting and Best Practices

  • Heres another example but with the values split out.

    CREATE TABLE #RawData

    (

    CalDate Date

    ,Value1 INT

    ,Value2 INT

    ,Value3 INT

    )

    INSERT INTO #RawData

    VALUES ('01-Jan-2014',100,120,300),('02-Jan-2014',80,150,300),('03-Jan-2014',100,20,180)

    DECLARE @SQL NVARCHAR(4000) = 'SELECT '

    DECLARE @SQLFrom NVARCHAR(4000) = 'FROM #UnpivtData Group by ValueName'

    /*

    Need to unpivot the raw data set so that VALUE1 etc can be grouped

    */

    SELECT

    Caldate,ValueName,Value

    INTO #UnpivtData

    FROM

    #RawData

    CROSS APPLY

    (

    VALUES

    ('Value1',Value1)

    ,('Value2',Value2)

    ,('Value3',Value3)

    ) x (ValueName,Value)

    /*

    build the Cross apply select

    */

    SET

    @SQL=@SQL+

    (

    SELECT

    STUFF

    (

    (

    SELECT

    ', SUM(CASE WHEN CalDate = '''

    +CONVERT(Nvarchar(10),CalDate)

    +''' then Value else 0 end) AS ['

    +CONVERT(Nvarchar(10),CalDate)

    +']'

    FROM #RawData

    FOR XML PATH('')

    )

    ,1,1,'')

    )

    +@SQLFrom

    Print @SQL

    EXEC sp_executesql @SQL

    drop table #RawData

    Drop table #UnpivtData

    _________________________________________________________________________
    SSC Guide to Posting and Best Practices

Viewing 3 posts - 1 through 2 (of 2 total)

You must be logged in to reply to this topic. Login to reply