Reputation: 15
I need a solution to a problem that has the table structure as listed below. Input
1 1/1/2009 Product1
2 2/2/2009 Product2
3 3/3/2009 Product3
4 4/4/2009 Product4
5 5/5/2009 Product5
Output
1 1/1/2009 2/2009 Product1
2 3/3/2009 4/4/2009 Product3
3 5/5/2009 Product5
I tried using CTE. But was not very sucessful in extracting the second row value. Appreciate any help. Thanks.
Upvotes: 0
Views: 2367
Reputation: 141
You can actually create a dummy column initially and create a cte.
Then, use cte and join them on the dummy key and the row number or a series which simulates sequential numbers.
Then, filter in the dataset to display only odd numbered rows
create table dbo.test
(
id integer,
currdate varchar(20), -- just to keep simple, not to mess with dates, took this as string
productCode varchar(20)
);
insert into test values (1, '1/1/2009', 'product1');
insert into test values (2, '2/2/2009', 'product2');
insert into test values (3, '3/3/2009', 'product3');
insert into test values (4, '4/4/2009', 'product4');
insert into test values (5, '5/5/2009', 'product5');
with ctes as
(
select
't' as joinKey,
id, -- serves as rownum or else create another using row_num() over partition by and order by
currdate,
productCode
from test
)
select
c1.id,
c1.currdate,
c1.productCode,
c2.id,
c2.currdate,
c2.productCode
from
(
select
joinKey,
id,
currdate,
productCode
from
ctes
)c1,
(
select
joinKey,
id,
currdate,
productCode
from
ctes
)c2
where c1.joinKey = c2.joinKey
and c1.id + 1 = c2.id
and c1.id % 2 = 1
The result is as shown below:
id currdate productCode id currdate productCode
1 1/1/2009 product1 2 2/2/2009 product2
3 3/3/2009 product3 4 4/4/2009 product4
Upvotes: 0
Reputation: 22717
I don't know whether Russ' answer helps you at all. Here is a link to an article that explains how to add row numbers to the results of a query. (Search for "row_number" to find the most likely example.)
Once you have a query numbering the rows properly, you should be able to throw that into a CTE, then select from it twice -- once for odd numbers, then again for even numbers. Have each result return the even numbered value for joining (odd - 1 = even). At that point, you can join the results of the queries and get two products on one row.
Upvotes: 1
Reputation: 10865
You are looking for PIVOT: http://msdn.microsoft.com/en-us/library/ms177410.aspx
Here is a best shot with the info you gave, I do something similar in one of my apps. You may need to use a dynamic SQL query if the pivot values change.
SELECT *
FROM (SELECT [Date]
,[Product]
FROM [Values]
PIVOT (Max([Date])
FOR [Product]
IN ('put date ranges here')) pvt
here is what mine looks like, this will allow for a set of different values. This is used in a form builder to retrive the values of user input
--//Get a comma delimited list of field names from the field table for this form
DECLARE @FieldNames varchar(max)
SELECT @FieldNames = COALESCE(@FieldNames + ', ', '') + '[' + CAST([FieldName] AS varchar(max)) + ']'
FROM [Fields]
WHERE [FormID] = @FormID
--//create a dynamic sql pivot table that will bring our
--//fields and values together to look like a real sql table
DECLARE @SQL varchar(max)
SET @SQL =
'SELECT *
FROM (SELECT [Values].[RowID]
,[Fields].[FieldName]
,[Values].[Value]
FROM [Values]
INNER JOIN [Fields] ON [Fields].[FieldID] = [Values].[FieldID]
WHERE [Fields].[FormID] = ''' + @FormID + ''') p
PIVOT (Max([Value])
FOR [FieldName]
IN (' + @FieldNames + ')) pvt'
--//execute our sql to return the data
EXEC(@SQL)
Upvotes: 1