Dynamic pivot in SQL Server 2000 without using stored procedure or function

Question

I have these two tables:

Table A

    pj    item     desc        pj_id
------------------------------------
    pjA   iphone   cell phone   1
    pjA   macbook  laptop       2
    pjB   windows  os           3

Table B

     pj   item     pj_id  version   date
-----------------------------------------
     pjA  iphone     1      2       2010
     pjA  iphone     1      3       2011
     pjA  macbook    2      0       2000
     pjB  windows    3      5       2008
     pjB  windows    3      6       2014
     pjB  windows    3      7       2017

And I want

   pj    item       desc       detail1   detail2  detail3
---------------------------------------------------------
   pjA   iphone   cell phone   2010(2)   2011(3)
   pjA   macbook   laptop      2000(0)
   pjB   windows     os        2008(5)   2014(6)  2017(7)

You can see the column "detail" is dynamic, because there are no fix numbers of version of an item in table B

As we know, PIVOT or FOR XML PATH aren't supported on SQL Server 2000.

So I googled some solutions but they usually use stored procedure or function, but I don't have the permissions on this database.

The closest answer for me probably is this , but I have no idea how to use on my table.

Any help would be appreciated.

Answer 1

I have a solution. You can use a temp table to create the 'Details' with a RowNum column and then Pj_id grouped. And by using a left join, you can bring the 'Details' back in. I have only done for 3 'Details'. You can expand as you wish.

 declare @tableA table(pj varchar(20), item varchar(50), descb varchar(100), pj_id int)
insert into @tableA
select 'pjA', 'iphone', 'cell phone', 1 union all
select 'pjA', 'macbook', 'laptop', 2 union all
select 'pjB', 'windows', 'os', 3 

declare @tableB table(pj varchar(20), item varchar(50), pj_id int, version int,   vdate int)
insert into @tableB
select 'pjA', 'iphone', 1, 2, 2010 union all
select 'pjA', 'iphone', 1, 3, 2011 union all
select 'pjA', 'macbook', 2, 0, 2000 union all
select 'pjB', 'windows ', 3, 5, 2008 union all
select 'pjB', 'windows', 3, 6, 2014 union all
select 'pjB', 'windows',3,  7, 2017

declare @Details table(pj_id int, Detail varchar(100), RowNum int)----IDENTITY(1,1)
insert into @Details
SELECT pj_id,
        Detail=cast(vdate as varchar(20))+'('+cast(SUM(version) as varchar(20))+')',
        RowNum = (SELECT COUNT(*) ---Simulating Row_num here
                 FROM @tableB i 
                WHERE i.pj_id  = o.pj_id 
                  AND cast(i.vdate as varchar(20))<=cast(o.vdate as varchar(20))  )
   FROM @tableB  o
group  by
pj_id, cast(o.vdate as varchar(20))
  ORDER BY pj_id, Detail




select 
    A.*, 
    detail1=D1.Detail,
    detail2=D2.Detail,
    detail3=D3.Detail
from
    @tableA A

left join @Details D1 ON
    D1.pj_id=A.pj_id
    AND 
    D1.RowNum=1

left join @Details D2 ON
    D2.pj_id=A.pj_id
    AND 
    D2.RowNum=2

left join @Details D3 ON
    D3.pj_id=A.pj_id
    AND 
    D3.RowNum=3

I have simulated ROW_NUMBER.