Create a table with unique values from another table - sql

Question

I am using MS SQL Server Management Studio. I have table -
+--------+----------+
| Num_ID | Alpha_ID |
+--------+----------+
| 1 | A |
| 1 | B |
| 1 | C |
| 2 | B |
| 2 | C |
| 3 | A |
| 4 | C |
| 5 | A |
| 5 | B |
+--------+----------+
I want to create another table with 2 columns from this table so that column_1 gives Unique values in Num_ID (i.e. 1,2,3,4 and so on) and column_2 gives Unique values in Alpha_ID (A, B, C and so on).
But if an alphabet has already occurred, it should not occur again. So the output will be something like this -
Col_1 Col_2
================
1 - A
----------------
2 - B
----------------
3 - NULL (as A has been chosen by 1, it cannot occur next to 3)
----------------
4 - C
----------------
5 - NULL (both 5 A and 5 B cannot be chosen as A and B were picked up by 1 and 2)
----------------
Hope that makes sense.
I would like to clarify that the IDs in the input table are not numerical as I have shown, but both Num_ID and Alpha_ID are complex strings. I have simplified them to 1,2,3,... and A, B, C .... for the purpose of this question

Answer 1

I don't think this could be done without a cursor.
I added few more rows to your sample data to test how it works with other cases.
The logic is straight-forward. At first get a list of all distinct values of Num_ID. Then loop through them and with each iteration add one row to the destination table. To determine the Alpha_ID value to add I'll use EXCEPT operator that takes all available Alpha_ID values for the current Num_ID from the source table and removes from them all values that have been used before.
It is possible to write that INSERT without using explicit variable #CurrAlphaID, but it looks a bit cleaner with variable.
Here is SQL Fiddle.
DECLARE #TSrc TABLE (Num_ID varchar(10), Alpha_ID varchar(10));
INSERT INTO #TSrc (Num_ID, Alpha_ID) VALUES
('1', 'A'),
('1', 'B'),
('1', 'C'),
('2', 'B'),
('2', 'C'),
('3', 'A'),
('3', 'C'),
('4', 'A'),
('4', 'C'),
('5', 'A'),
('5', 'B'),
('5', 'C'),
('6', 'D'),
('6', 'E');
DECLARE #TDst TABLE (Num_ID varchar(10), Alpha_ID varchar(10));
DECLARE #CurrNumID varchar(10);
DECLARE #CurrAlphaID varchar(10);
DECLARE #iFS int;
DECLARE #VarCursor CURSOR;
SET #VarCursor = CURSOR FAST_FORWARD
FOR
SELECT DISTINCT Num_ID
FROM #TSrc
ORDER BY Num_ID;
OPEN #VarCursor;
FETCH NEXT FROM #VarCursor INTO #CurrNumID;
SET #iFS = ##FETCH_STATUS;
WHILE #iFS = 0
BEGIN
SET #CurrAlphaID =
(
SELECT TOP(1) Diff.Alpha_ID
FROM
(
SELECT Src.Alpha_ID
FROM #TSrc AS Src
WHERE Src.Num_ID = #CurrNumID
EXCEPT
SELECT Dst.Alpha_ID
FROM #TDst AS Dst
) AS Diff
ORDER BY Diff.Alpha_ID
);
INSERT INTO #TDst (Num_ID, Alpha_ID)
VALUES (#CurrNumID, #CurrAlphaID);
FETCH NEXT FROM #VarCursor INTO #CurrNumID;
SET #iFS = ##FETCH_STATUS;
END;
CLOSE #VarCursor;
DEALLOCATE #VarCursor;
SELECT * FROM #TDst;
Result
Num_ID Alpha_ID
1 A
2 B
3 C
4 NULL
5 NULL
6 D
Having index on (Num_ID, Alpha_ID) on the source table would help. Having index on (Alpha_ID) on the destination table would help as well.

Answer 2

I think I've made something not through a recursion (cursor or a while)
First, I created a table with rows.
create table #tmptest
(
Num_ID int
, Alpha_ID varchar(50)
)
insert into #tmptest (Num_ID, Alpha_ID) values
(1,'A'),
(1,'B'),
(1,'C'),
(2,'B'),
(2,'C'),
(3,'A'),
(4,'C'),
(5,'A'),
(5,'B')
// this one, with row column
SELECT
ROW_NUMBER() OVER (PARTITION BY Num_ID ORDER BY Num_ID ASC) as row
, *
INTO #tmp_withrow
FROM #tmptest
and these were the results
Lastly, I made an inner query (could possibly be a left join or better).
SELECT DISTINCT
Num_ID
, (
SELECT
TOP 1
Alpha_ID
FROM #tmp_withrow in1
WHERE
in1.Num_ID = t.Num_ID
AND in1.Alpha_ID NOT IN (
SELECT
Alpha_ID
FROM #tmp_withrow in2
WHERE
in2.Num_ID < in1.Num_ID
AND in2.row = 1
)
ORDER BY in1.Num_ID ASC
) AS [NonRepeatingAlpha]
from #tmptest t
and these were the results
Note : I created a flag (row) which will allow you to query all less than the ID's you're in (in2.Num_ID < in1.Num_ID) then find out what letters where already used (in2.row = 1) and then select / avoid all letters that has already been used from the other Num_ID (
WHERE in1.Num_ID = t.Num_ID
AND in1.Alpha_ID NOT IN (
SELECT
Alpha_ID
FROM #tmp_withrow in2
WHERE
in2.Num_ID < in1.Num_ID
AND in2.row = 1
)
I hope this helps. Thanks!