Comments on: SQL SERVER – Grouping by Multiple Columns to Single Column as A String

By: mis

mis — Thu, 11 Apr 2013 20:35:33 +0000

How to put every value of CSV into another column ?

By: smith

smith — Fri, 29 Mar 2013 14:52:58 +0000

This is great.. but now if the same was to come into one column than..
Col
1
A
B
C
2
A
B
C
D
E

By: Someone

Someone — Thu, 21 Feb 2013 15:31:11 +0000

It should be

FOR XML PATH(”)),1,2,”) AS CSV

and not

FOR XML PATH(”)),1,1,”) AS CSV

as it will leave a comma at the beginning.

By: madhivanan

madhivanan — Fri, 15 Feb 2013 13:04:09 +0000

Are you using MySQL? This site is for SQL Server

By: Bhavik

Bhavik — Mon, 28 Jan 2013 19:24:01 +0000

Dear Pinal,
Solution You have provided is not working for SQL COMPACT EDITION 3.5.

Following is the Error Message:
“There was an error parsing the query. [ Token line number = 2,Token line offset = 8,Token in error = SELECT ]”

Please help on this issue.

Regards,
Bhavik

By: Maheswari

Maheswari — Fri, 18 Jan 2013 08:59:44 +0000

i am also want this type of result for my table. pls help

By: Satish

Satish — Thu, 22 Nov 2012 06:17:43 +0000

Nice Article!
How can I achieve the following. The columns are depend on the values of CSV
EX:
ID CSV
1 A
1 B
2 A
2 B
2 C
2 D

ID CSV1 CSV2 CSV3 CSV4
1 A B C
2 A B C D

Thanks
Sathish

By: Zee

Zee — Tue, 06 Nov 2012 21:00:38 +0000

Hi
I like to GROUP BY a varchar not an int , what changes do I have to do,
Thanks

I am Getting This:
Conversion failed when converting the varchar value ‘H123′ to data type int.
Thanks

By: Daniel

Daniel — Tue, 09 Oct 2012 15:08:18 +0000

NIce! but I have a question.

What is FOR XML PATH(”) for ??

By: Sriram

Sriram — Tue, 18 Sep 2012 15:06:21 +0000

This works great! However with a small modification it can go a wee bit forward:

– Get CSV values
SELECT t.ID, STUFF(
(SELECT ‘,’ + s.Col
FROM TestTable s
WHERE s.ID = t.ID
FOR XML PATH(”), TYPE).value(‘.’, ‘VARCHAR(MAX)’),1,1,”) AS CSV
FROM TestTable AS t
GROUP BY t.ID
GO

Adding TYPE ensures that it is available as XML object and doing a type cast to varchar ensure that all the xml-sensitive characters would be converted back from XML. If one of the value was “Me & You” the output would be the same.

By: Haresh Ambaliya

Haresh Ambaliya — Fri, 14 Sep 2012 16:59:23 +0000

Create function that will return comma separated value by Id like

Create function [dbo].[GetCSV](@Id int)
Returns nvarchar(Max)
as
Begin
Declare @ListStr nvarchar(Max)
set @ListStr = ”
select @ListStr = @ListStr + Col + ‘,’ from TestTable where Id = @Id
Return substring(@ListStr,1,len(@ListStr)-1)
End

Then use that function as

Select Id,dbo.GetCSV(Id) from TestTable group by Id

Its as easy as simple query writing.

By: Rama Mathanmohan

Rama Mathanmohan — Fri, 14 Sep 2012 13:22:20 +0000

Select ID,SUBSTRING(
(Select COALESCE(‘,’+Col,”)
From TestTable
Where ID = Master.ID
For Xml Path(”)),2,500000) As CSV
From TestTable as Master
Group By ID

By: Ranjit

Ranjit — Fri, 14 Sep 2012 10:56:45 +0000

Here is the another solution using Recursive CTEs ..
But FOR XML PATH(”) query giving good performance compare to this..

;WITH Rnumcte1 AS
(
SELECT ROW_NUMBER() OVER (PARTITION BY id ORDER BY id,Col) AS Rnum,id,cast(Col as varchar(max)) Col
FROM TestTable
)
,Rnumcte2 AS
(
SELECT Rnum,id,Col
FROM Rnumcte1
WHERE Rnum = 1

UNION ALL

SELECT r1.Rnum,r1.id,r2.Col + ‘,’ + r1.Col
FROM Rnumcte1 r1
INNER JOIN Rnumcte2 r2 ON r1.id = r2.id AND r1.Rnum = r2.Rnum + 1
) –select * from Rnumcte2
SELECT r2.id,r2.Col
FROM (
SELECT id,MAX(Rnum)maxRnum
FROM Rnumcte1
GROUP BY id
) r1
INNER JOIN Rnumcte2 r2 ON r1.id = r2.id AND r1.maxRnum = r2.Rnum
ORDER BY r2.id

By: Sj

Sj — Fri, 14 Sep 2012 06:56:02 +0000

Here is an interesting read which pretty much talks everything about the various ways to above problem. Author Anith Sen have done quite extensive work to explain various ways:

http://www.simple-talk.com/sql/t-sql-programming/concatenating-row-values-in-transact-sql/

By: harsh

harsh — Fri, 14 Sep 2012 06:48:15 +0000

This could be alternative solution.

;WITH CTE (SRNO,ID,COL)
AS
(
SELECT ROW_NUMBER() OVER (PARTITION BY ID ORDER BY ID,COL) AS SRNO,ID,CAST(COL AS VARCHAR(MAX)) FROM TESTTABLE
)
,CTE1(SRNO,ID,COL)
AS
(
SELECT * FROM CTE WHERE SRNO=1
UNION ALL
SELECT CTE.SRNO AS SRNO, CTE.ID,CAST (CTE1.COL + ‘,’ + CTE.COL AS VARCHAR(MAX)) FROM CTE INNER JOIN CTE1
ON CTE.ID=CTE1.ID AND CTE.SRNO=CTE1.SRNO+1
)
SELECT ID, MAX(COL) AS CSV FROM CTE1 GROUP BY CTE1.ID

By: Chintak

Chintak — Fri, 14 Sep 2012 06:32:49 +0000

Just added info in post
http://beyondrelational.com/modules/2/blogs/115/posts/17683/grouping-by-multiple-columns-to-single-column-as-a-string.aspx

By: visualtalks

visualtalks — Fri, 14 Sep 2012 06:32:00 +0000

1. We’re not allowed CLR into prod, so that solution fails.
2. How about if you have nvarchar() for data types instead of single values? :-). Hint: SPLIT function…
3. Uh-oh, I actually created a different puzzle altogether…

By: Grzegorz Łyp

Grzegorz Łyp — Fri, 14 Sep 2012 05:49:26 +0000

This solution is extreme inefficient if you want to get thousands of records. The best solution for the short concatenations (less then 4000 characters) is to create clr aggregation function with StringBuilder class inside clr code. For the longer concatenations you can build clr function that has query as an argument and results in concatenated value of nvarchar(max) type. This time again you can use StringBuilder class inside clr code. The problem is with resource allocation. Every time you perform concatenation you need to allocate new bigger space. In the StringBuilder solution you take the concatenation elements and create single buffer for all elements. There is no waste of time for memory allocations for single concatenations.

By: Naresh

Naresh — Fri, 14 Sep 2012 05:45:20 +0000

I have been using SQL CLR to do this kind of string aggregation, but not sure if that is more efficient than this one or not.