SQL Query - Row to columns not really a pivot - sql

Question

I'm trying to move certain fields of an ID into columns, but it doesn't appear to match all the pivot examples I am finding. All the examples I can find use some form of a grouping on a field value. I want to use more of a placement regardless of the value in the field. I want to do this in a query without looping via code. Data source example (sorry couldn't figure out how to format a table on the post so I used a code snippet):
+----+--------+--------+
| ID | Field1 | Field2 |
+----+--------+--------+
| 1 | NULL | NULL |
| 2 | Jim | 321 |
| 2 | Jack | 54 |
| 2 | Sue | 985 |
| 2 | Gary | 654 |
| 3 | Herb | 332 |
| 3 | Chevy | 10 |
+----+--------+--------+
Result set I'm trying to generate:
+----+------+------+-------+------+------+------+
| ID | Col1 | Col2 | Col3 | Col4 | Col5 | Col6 |
+----+------+------+-------+------+------+------+
| 1 | NULL | NULL | | | | |
| 2 | Jim | 321 | Jack | 54 | Sue | 985 |
| 3 | Herb | 332 | Chevy | 10 | | |
+----+------+------+-------+------+------+------+

Answer 1

SQL Fiddle: http://sqlfiddle.com/#!3/a225a/1
;with cte as (
select id
, field1
, field2
, ROW_NUMBER() over (partition by id order by field1, field2) r
from #t
)
select c1.id
, c1.field1 col1
, c1.field2 col2
, c2.field1 col3
, c2.field2 col4
, c3.field1 col5
, c3.field2 col6
from cte c1
left outer join cte c2 on c2.id = c1.id and c2.r = c1.r + 1
left outer join cte c3 on c3.id = c1.id and c3.r = c1.r + 2
where (c1.r % 3) = 1
Explanation
ROW_NUMBER() over (partition by id order by field1, field2) r. This line ensures that we have a column counting up from 1 for each id. This allows us to distinguish between the multiple rows.
The CTE is used to save typing the same statement for c1, c2 and c3.
The joins ensure that all items in a row have the same id, and that data for col1, col3 and col5 (likewise for col2, col4 and col6) is taken from consecutive rows. We're using left outer joins because there may not rows in the source table for these columns.
The where statement says to take the first row of each set of 3 for the data in c1 (with c2 and c3 thus being the second and third of each set, thanks to the earlier join).

Answer 2

Here's a solution using dynamic sql that works though I'm sure there's a better way to do it. Caution, it's a bit painful. First it builds the list of columns to pivot and select, builds the dynamic sql and runs it.
DECLARE #PivotColumns as varchar(max), #SelectColumns as varchar(max), #sql as varchar(max)
SELECT #PivotColumns = ISNULL(#PivotColumns + ',', '') + ColNum,
#SelectColumns = ISNULL(#SelectColumns + ',', '') + 'NULLIF(' + ColNum + ', ''NULL'') as ' + ColNum
from (select distinct 'Col' + cast(ROW_NUMBER() OVER (partition by id order by id) as varchar) as ColNum
from (select id,
isnull(field1,'NULL') as field1,
isnull(field2,'NULL') as field2
from weirdpivot) cols
unpivot
(
value
for col in (field1, field2)
) unpivoted) DistinctColumns
set #sql = '
select id, + ' + #SelectColumns + '
from (select
''Col'' + cast(ROW_NUMBER() OVER (partition by id order by id) as varchar) as colnum
,id
,value
from (select id,
isnull(field1,''NULL'') as field1,
isnull(field2,''NULL'') as field2
from weirdpivot) cols
unpivot
(
value
for col in (field1, field2)
) u) unpivoted
pivot
(
max(value)
for colnum in (' + #PivotColumns + ')
) p'
exec (#sql)