Pivot table to turn rows into columns - sql

Question

I currently run the query
SELECT [PriceAttributeID]
,[PriceID]
,[AttributeID]
,[PriceAttributeComparator]
,[PriceAttributeMin]
,[PriceAttributeMax]
FROM [PriceAttribute]
Which gives the output
1 2 1 1 S NULL
2 3 1 1 M NULL
3 4 1 1 L NULL
4 5 1 1 L NULL
5 5 2 1 Black NULL
I would like to get the output (where _Comp, _Min and _Max relate to PriceAttributeComparator, PriceAttributeMin and PriceAttributeMax)
PriceID 1_Comp 1_Min 1_Max 2_Comp 2_Min 2_Max
2 1 S NULL NULL NULL NULL
3 1 M NULL NULL NULL NULL
4 1 L NULL NULL NULL NULL
5 1 L NULL 1 Black NULL
The same query would also be expected to have 1_ and 2_ prefixes as 4_, 5_, 19_ and 32_ or any other indeterminate number of ID's based on what is in the table at the time.
I have attempted a PIVOT table, but i am new to them and haven't the first clue on how to create what it is i am looking to do.

Answer 1

Part of the problem you are probably having with the PIVOT function is due to the fact you have multiple columns that you want to apply the function to. If you want to use the PIVOT function, then I would suggest first unpivoting the columns PriceAttributeComparator, PriceAttributeMin and PriceAttributeMax. When you unpivot the data you will no longer have multiple columns, you will have multiple rows, then you can apply the pivot to all of the appropriate values.
You did not specify what version of SQL Server you are using but you can use CROSS APPLY with a UNION ALL to unpivot the columns:
select priceid,
col = cast(attributeid as varchar(10))+'_'+ col,
value
from
(
select PriceID,
AttributeID,
comp = cast(PriceAttributeComparator as varchar(10)),
[min] = cast(PriceAttributeMin as varchar(10)),
[max] = cast(PriceAttributeMax as varchar(10))
from PriceAttribute
) d
cross apply
(
select 'comp', comp union all
select 'min', [min] union all
select 'max', [max]
) c (col, value)
See Demo. This process will convert your data into the following format:
| PRICEID | COL | VALUE |
-----------------------------
| 2 | 1_comp | 1 |
| 2 | 1_min | S |
| 2 | 1_max | (null) |
| 3 | 1_comp | 1 |
| 3 | 1_min | M |
| 3 | 1_max | (null) |
Once the data is in multiple rows, then you can apply the PIVOT function to the values in col:
select priceid,
[1_comp], [1_min], [1_max], [2_comp], [2_min], [2_max]
from
(
select priceid,
col = cast(attributeid as varchar(10))+'_'+ col,
value
from
(
select PriceID,
AttributeID,
comp = cast(PriceAttributeComparator as varchar(10)),
[min] = cast(PriceAttributeMin as varchar(10)),
[max] = cast(PriceAttributeMax as varchar(10))
from PriceAttribute
) d
cross apply
(
select 'comp', comp union all
select 'min', [min] union all
select 'max', [max]
) c (col, value)
) src
pivot
(
max(value)
for col in ([1_comp], [1_min], [1_max], [2_comp], [2_min], [2_max])
) piv;
See SQL Fiddle with Demo.
The above versions work great if you have a known number of values but if the values are unknown, then you will need to use dynamic SQL to get the result:
DECLARE #cols AS NVARCHAR(MAX),
#query AS NVARCHAR(MAX)
select #cols = STUFF((SELECT ',' + QUOTENAME(cast(attributeid as varchar(10))+'_'+ col)
from
(
select distinct attributeid
from priceattribute
) d
cross apply
(
select 'comp', 1 union all
select 'min', 2 union all
select 'max', 3
) c (col, so)
group by attributeid, col, so
order by attributeid, so
FOR XML PATH(''), TYPE
).value('.', 'NVARCHAR(MAX)')
,1,1,'')
set #query = 'SELECT priceid, ' + #cols + '
from
(
select priceid,
col = cast(attributeid as varchar(10))+''_''+ col,
value
from
(
select PriceID,
AttributeID,
comp = cast(PriceAttributeComparator as varchar(10)),
[min] = cast(PriceAttributeMin as varchar(10)),
[max] = cast(PriceAttributeMax as varchar(10))
from PriceAttribute
) d
cross apply
(
select ''comp'', comp union all
select ''min'', [min] union all
select ''max'', [max]
) c (col, value)
) x
pivot
(
max(value)
for col in (' + #cols + ')
) p '
execute sp_executesql #query;
See SQL Fiddle with Demo. These solutions will give a result:
| PRICEID | 1_COMP | 1_MIN | 1_MAX | 2_COMP | 2_MIN | 2_MAX |
----------------------------------------------------------------
| 2 | 1 | S | (null) | (null) | (null) | (null) |
| 3 | 1 | M | (null) | (null) | (null) | (null) |
| 4 | 1 | L | (null) | (null) | (null) | (null) |
| 5 | 1 | L | (null) | 1 | Black | (null) |

Answer 2

It might be simplest to do this using conditional aggregation rather than pivot:
SELECT PriceID,
max(case when AttributeID = 1 then PriceAttributeComparator end) as comp_1,
max(case when AttributeID = 1 then PriceAttributeMin end) as min_1,
max(case when AttributeID = 1 then PriceAttributeMax end) as max_1,
max(case when AttributeID = 2 then PriceAttributeComparator end) as comp_2,
max(case when AttributeID = 2 then PriceAttributeMin end) as min_2,
max(case when AttributeID = 2 then PriceAttributeMax end) as max_2
FROM PriceAttribute pa
group by PriceId;