How to pivot my data with SQL Server 2005

Question

I have the following data

How do I transform it (with SQL Server 2005) into the following format?

I have a example solution that I came up with but it seems a little clunky. It smells perhaps?

DECLARE @ProductLanguage TABLE
(
    [PRODUCT_ID] int
    , [LANGUAGE] varchar(50)
)

INSERT INTO @ProductLanguage ([PRODUCT_ID],[LANGUAGE]) VALUES (52035,'Czech')
INSERT INTO @ProductLanguage ([PRODUCT_ID],[LANGUAGE]) VALUES (52035,'English')
INSERT INTO @ProductLanguage ([PRODUCT_ID],[LANGUAGE]) VALUES (52035,'German')
INSERT INTO @ProductLanguage ([PRODUCT_ID],[LANGUAGE]) VALUES (54001,'Danish')
INSERT INTO @ProductLanguage ([PRODUCT_ID],[LANGUAGE]) VALUES (54001,'Spanish')
INSERT INTO @ProductLanguage ([PRODUCT_ID],[LANGUAGE]) VALUES (54001,'English')
INSERT INTO @ProductLanguage ([PRODUCT_ID],[LANGUAGE]) VALUES (70501,'Finnish')
INSERT INTO @ProductLanguage ([PRODUCT_ID],[LANGUAGE]) VALUES (70501,'Greek')
INSERT INTO @ProductLanguage ([PRODUCT_ID],[LANGUAGE]) VALUES (70501,'Hungarian')
INSERT INTO @ProductLanguage ([PRODUCT_ID],[LANGUAGE]) VALUES (52044,'Hebrew')

SELECT
    PRODUCT_ID
    ,MAX(CASE WHEN [ROW_ID]=1 THEN LANGUAGE ELSE NULL END) As LANG_1
    ,MAX(CASE WHEN [ROW_ID]=2 THEN LANGUAGE ELSE NULL END) As LANG_2
    ,MAX(CASE WHEN [ROW_ID]=3 THEN LANGUAGE ELSE NULL END) As LANG_3
FROM
    (SELECT
        ROW_NUMBER() OVER (PARTITION BY [PRODUCT_ID] ORDER BY [PRODUCT_ID] ASC) AS [ROW_ID]
        , [PRODUCT_ID]
        , [LANGUAGE]
    FROM
        @ProductLanguage) AS Temp
GROUP BY
    [PRODUCT_ID]

The interesting bit is I do not care about the specific Languages displayed in each LANG_* column. Other questions posted here seem to all refer to knowning the pivoted columns by name. But I do not want to name the columns by the languages found.

NOTE : I know I mention the word "pivot" but the best solution for this problem may not involve the PIVOT clause. I just used that word as my question seemed to suggest pivotting data. Maybe a CTE would help with the solution, I do not know. I just know I am not happy about the example solution above.

Answer 1

You can use the PIVOT() function

SELECT    P.PRODUCT_ID,
          P.Czech,
          P.Other languages
FROM      TABLE AS T
PIVOT     (
              AGGREGATE(LANGUAGE) FOR LANGUAGE IN ([Czech],  ...)
          ) AS P

If you don't name the columns explicitly you are forced into doing tricks as far as I know...

Untested (obviously). See: MSDN

Answer 2

This can be done with a PIVOT , either statically or dynamically.

Static Version the columns are hard-coded (See SQL Fiddle with Demo ):

select *
from 
(
  select product_id, language,
    'Lang_' + cast(row_number() over(partition by PRODUCT_ID order by language) as varchar(10)) rn
  from ProductLanguage
) x
pivot
(
  max(language)
  for rn in ([Lang_1], [Lang_2], [Lang_3])
) p

Dynamic Version will get the columns at run-time (See SQL Fiddle With Demo ):

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

select @cols = STUFF((SELECT distinct ',' + QUOTENAME('Lang_' + cast(row_number() over(partition by PRODUCT_ID order by language) as varchar(10))) 
                    from ProductLanguage
            FOR XML PATH(''), TYPE
            ).value('.', 'NVARCHAR(MAX)') 
        ,1,1,'')


set @query = 'SELECT PRODUCT_ID, ' + @cols + ' from 
             (
              select product_id, language,
                ''Lang_'' + cast(row_number() over(partition by PRODUCT_ID order by language) as varchar(10)) rn
              from ProductLanguage
            ) x
            pivot 
            (
                max(language)
                for rn in (' + @cols + ')
            ) p '

execute(@query)

Answer 3

create table  #ProductLanguage 
(
    PRODUCT_ID int
    , [LANGUAGE] varchar(50)
)

INSERT INTO #ProductLanguage  ([PRODUCT_ID],[LANGUAGE]) VALUES (52035,'Czech')
INSERT INTO #ProductLanguage  ([PRODUCT_ID],[LANGUAGE]) VALUES (52035,'English')
INSERT INTO #ProductLanguage  ([PRODUCT_ID],[LANGUAGE]) VALUES (52035,'German')
INSERT INTO #ProductLanguage  ([PRODUCT_ID],[LANGUAGE]) VALUES (54001,'Danish')
INSERT INTO #ProductLanguage  ([PRODUCT_ID],[LANGUAGE]) VALUES (54001,'Spanish')
INSERT INTO #ProductLanguage  ([PRODUCT_ID],[LANGUAGE]) VALUES (54001,'English')
INSERT INTO #ProductLanguage  ([PRODUCT_ID],[LANGUAGE]) VALUES (70501,'Finnish')
INSERT INTO #ProductLanguage  ([PRODUCT_ID],[LANGUAGE]) VALUES (70501,'Greek')
INSERT INTO #ProductLanguage  ([PRODUCT_ID],[LANGUAGE]) VALUES (70501,'Hungarian')
INSERT INTO #ProductLanguage  ([PRODUCT_ID],[LANGUAGE]) VALUES (52044,'Hebrew')


declare @col as varchar(2000)
declare @query as varchar(4000)

--select language ,@col= replace('[Lang_'+str(row_number() over(order by language))+']',' ','') Lang  from (select distinct language  from #ProductLanguage)p
select  @col = stuff((select distinct '],['+language  from #ProductLanguage for xml path('')),1,2,'')+']' 
print @col

set @query = 'select PRODUCT_ID, '+@col+ ' from #ProductLanguage 
pivot (
max(LANGUAGE) for LANGUAGE in ('+@col+')
) pvt'
execute(@query)

Answer 4

Here's an idea... Use the Pivot Keyword... available in sql2005

Pivot and UnPivot

Then after you have this working, outputting the column names as data values, embed this entire sql statemnt as the subquery inside an outer Select statement, where you Alias the column names as "Language1", "Language2" etc...

Select Z.Arabic as Language1, Z.Botwanese as Language2, etc.
From  (Inner Pivot Query Here ) Z