Dynamic pivot with similar column names - sql

Question

I'm looking for a way to pivot a varying amount of rows to columns in sql server 2008 R2. I created the data column RANK in the query because, ultimately I want the pivoted column names to be labeled the value in the RANK column. Then, if somehow I can STUFF the other 3 field values together into one field I would be able to iterate over the row in my backend lang and split the field results appropriately.
Here is the current data set:
And I would like the end result of the pivot to produce a result like this:
I haven't found anything about being able to pivot in this "dynamic" way. Any help would be much appreciated.

Answer 1

As I mentioned above, you need to distinguish each of your Rank values. You've said that this is a calculated value that you can add a number to the end of each one. Once you've added that number, then you still need to pivot it.
The easiest way to first see this would be to write a hard-coded version of the query first.
Sample Data:
create table yourdata
(
id int,
code varchar(50),
created datetime,
[rank] varchar(50)
);
insert into yourdata
select 285856, 'J7609', '2015-01-19', 'Principle' union all
select 285856, 'J7613', '2015-01-19', 'Other' union all
select 285856, 'J0456', '2015-01-19', 'Other' union all
select 285856, 'J0694', '2015-01-19', 'Other' union all
select 285856, 'J1885', '2015-01-19', 'Other' union all
select 285856, 'J2060', '2015-01-19', 'Other' union all
select 285856, 'J2930', '2015-01-19', 'Other';
Static Query:
select Principle_1, Other_1,
Other_2, Other_3, Other_4,
Other_5, Other_6
from
(
-- using row_number to get unique id for each rank
select
data = cast(id as varchar(10)) +' | '+ code +' | '+ convert(varchar(10), created, 112),
[rank] = [rank] + '_' +cast(row_number() over(partition by id, [rank]
order by id) as varchar(10))
from yourdata
) d
pivot
(
max(data)
for [rank] in (Principle_1, Other_1, Other_2, Other_3, Other_4,
Other_5, Other_6)
) p;
Now to do this dynamic, you will create a sql string with the column names and then execute that string:
DECLARE #cols AS NVARCHAR(MAX),
#query AS NVARCHAR(MAX)
select #cols = STUFF((SELECT ',' + QUOTENAME([rank] + '_' +cast(rn as varchar(10)))
from
(
select [rank],
rn = row_number() over(partition by id, [rank]
order by id)
from yourdata
) d
group by [rank], rn
order by rn, [rank] desc
FOR XML PATH(''), TYPE
).value('.', 'NVARCHAR(MAX)')
,1,1,'')
set #query = 'SELECT ' + #cols + '
from
(
select
data = cast(id as varchar(10)) +'' | ''+ code +'' | ''+ convert(varchar(10), created, 112),
[rank] = [rank] + ''_'' +cast(row_number() over(partition by id, [rank]
order by id) as varchar(10))
from yourdata
) x
pivot
(
max(data)
for [rank] in (' + #cols + ')
) p '
exec sp_executesql #query;
This gets you a result:
+---------------------------+---------------------------+---------------------------+---------------------------+---------------------------+---------------------------+---------------------------+
| Principle_1 | Other_1 | Other_2 | Other_3 | Other_4 | Other_5 | Other_6 |
+---------------------------+---------------------------+---------------------------+---------------------------+---------------------------+---------------------------+---------------------------+
| 285856 | J7609 | 20150119 | 285856 | J7613 | 20150119 | 285856 | J0456 | 20150119 | 285856 | J0694 | 20150119 | 285856 | J1885 | 20150119 | 285856 | J2060 | 20150119 | 285856 | J2930 | 20150119 |
+---------------------------+---------------------------+---------------------------+---------------------------+---------------------------+---------------------------+---------------------------+