Question

I have a table eg assume this setup

table MyTable has various columns Id, UserId, col1, col2 col3 including column called Stuff.

I want to output certain Columns from MyTable with a query
but i want to split the ‘Stuff’ column such that 2 new columns are shown in the query

I can define the categories hardcoded, im not sure how this can be represented in sql

Categoy1 = "alpha, bravo, delta, gamma';
Categoy2 = "charlie, echo, hotel';



MyTable

ID |    UserID |      Stuff       | Other Cols....
----------------------------------------------------------
1          1          alpha
2          2          hotel
3          1          charlie
4          1          echo
5          1          gamma
6          2          bravo
7          2          delta

i want the select query to show

UserId  |  Category1           |  Catergory2
----------------------------------------------------------
1            alpha, gamma            charlie, echo
---------------------------------------------------------
2            bravo, delta            hotel
----------------------------------------------------------

i.e produce 2 columns split based on whether the stuff column contains an item from category1 or category2
based on a distinct userId the categories content can be comma separated as hown above

Please can you show how this can be done

Hope this makes sense.

Thanks

Answer 1

You can use the xml extensions to concatenate your strings, then just hard code the categories into each subquery:

CREATE TABLE #T (ID INT, UserID INT, [Stuff] VARCHAR(300))
INSERT #T VALUES
    (1, 1, 'alpha'),
    (2, 2, 'hotel'),
    (3, 1, 'charlie'),
    (4, 1, 'echo'),
    (5, 1, 'gamma'),
    (6, 2, 'bravo'),
    (7, 2, 'delta');

SELECT  UserID,
        [Category1] = STUFF((   SELECT  ', ' + [Stuff]
                                FROM    #T t2
                                WHERE   [Stuff] IN ('alpha', 'bravo', 'delta', 'gamma')
                                AND     t.UserID = t2.UserID
                                FOR XML PATH(''), TYPE
                            ).value('.', 'NVARCHAR(MAX)'), 1, 2, ''),
        [Category2] = STUFF((   SELECT  ', ' + [Stuff]
                                FROM    #T t2
                                WHERE   [Stuff] IN ('charlie', 'echo', 'hotel')
                                AND     t.UserID = t2.UserID
                                FOR XML PATH(''), TYPE
                            ).value('.', 'NVARCHAR(MAX)'), 1, 2, '')
FROM    (   SELECT  DISTINCT UserID
            FROM    #T
        ) t

Example on SQL Fiddle

You could define your categories at the start in a CTE (Categories) for improved readibility:

WITH Categories AS
(   SELECT  Category, Name
    FROM    (VALUES
                (1, 'alpha'), 
                (1, 'bravo'), 
                (1, 'delta'), 
                (1, 'gamma'),
                (2, 'charlie'),
                (2, 'echo'), 
                (2, 'hotel')
            ) t (Category, Name)
), Data AS
(   SELECT  UserID, [Stuff], Category
    FROM    T
            INNER JOIN Categories c
                ON c.Name = T.[Stuff]
)
SELECT  UserID,
        [Category1] = STUFF((   SELECT  ', ' + [Stuff]
                                FROM    Data t2
                                WHERE   Category = 1
                                AND     t.UserID = t2.UserID
                                FOR XML PATH(''), TYPE
                            ).value('.', 'NVARCHAR(MAX)'), 1, 2, ''),
        [Category2] = STUFF((   SELECT  ', ' + [Stuff]
                                FROM    Data t2
                                WHERE   Category = 2
                                AND     t.UserID = t2.UserID
                                FOR XML PATH(''), TYPE
                            ).value('.', 'NVARCHAR(MAX)'), 1, 2, '')
FROM    (   SELECT  DISTINCT UserID
            FROM    T
        ) t

Example on SQL Fiddle