SSIS Combining multiple rows into a single row with multiple columns - sql

Question

I have a table that does not have a primary key, but has an ID field that could be used as a primary key. I would like to transform this table into something useable to track inserts, updates, and deletes. Just adding a Identity column is not a viable solution because the table gets recreated every evening.
The table looks something like:
ID Specialty
-- ---------
01 Fam Med
02 Fam Med
02 Int Med
03 Surgery
03 Thor Sur
03 Card Sur
04 Undersea
I would like to transform it into something like the following and use ID as the primary key:
ID Specialty1 Specialty2 Specialty3
-- ---------- ---------- ----------
01 Fam Med
02 Fam Med Int Med
03 Surgery Thor Sur Card Sur
04 Undersea
I was leaning toward using the SQL pivot operator, but I am not familiar enough with pivots and I am not sure it is appropriate in this case.
Thank you in advance for your input.

Answer 1

The SQL you would need to do the PIVOT is:
SELECT ID,
Specialty1,
Specialty2,
Specialty3
FROM ( SELECT ID,
Specialty,
SpecialtyNum = 'Specialty' +
CAST(ROW_NUMBER() OVER(PARTITION BY ID
ORDER BY Specialty) AS VARCHAR(10))
FROM T
) AS t
PIVOT
( MAX(Specialty)
FOR SpecialtyNum IN ([Specialty1], [Specialty2], [Specialty3])
) AS pvt;
Example on SQL Fiddle
The key is adding a further column that you can pivot on, so the result of the subquery:
SELECT ID,
Specialty,
SpecialtyNum = 'Specialty' +
CAST(ROW_NUMBER() OVER(PARTITION BY ID
ORDER BY Specialty) AS VARCHAR(10))
FROM T;
Gives you:
ID Specialty SpecialtyNum
-- --------- -------------
01 Fam Med Specialty1
02 Fam Med Specialty1
02 Int Med Specialty2
03 Surgery Specialty1
03 Thor Sur Specialty2
03 Card Sur Specialty3
04 Undersea Specialty1
Then you can pivot on the SpecialtyNum column. Since each value of specialtyNum is unique to an ID, there is no data loss through aggregation.
But with an unknown number of specialities you would need to generate this dynamically:
DECLARE #Cols NVARCHAR(MAX) =
STUFF(( SELECT DISTINCT ',[Specialty' + CAST(ROW_NUMBER() OVER(PARTITION BY ID ORDER BY Specialty) AS VARCHAR(10)) + '] '
FROM T
FOR XML PATH(''), TYPE
).value('.', 'NVARCHAR(MAX)'), 1, 1, '');
DECLARE #SQL NVARCHAR(MAX) =
'SELECT ID, ' + #Cols +
'FROM ( SELECT ID,
Specialty,
SpecialtyNum = ''Specialty'' +
CAST(ROW_NUMBER() OVER(PARTITION BY ID
ORDER BY Specialty) AS VARCHAR(10))
FROM T
) AS t
PIVOT
( MAX(Specialty)
FOR SpecialtyNum IN (' + #Cols + ')
) AS pvt;';
EXECUTE sp_executesql #SQL;
Example on SQL Fiddle