SQL Server : ROW_NUMBER() OVER on table variable to amend duplicate columns - sql

Question

Lets say I have table variable declared like so...
DECLARE #LocalTable TABLE
(
IdField NVARCHAR(MAX),
NameField NVARCHAR(MAX)
)
And I populate it like so...
INSERT INTO #LocalTable
SELECT
IdColumn,
NameColumn
FROM SourceTable
NameColumn in the source table may have duplicate values, and therefore NameField in the local table will have the same duplicate values.
And let's say I want to insert the local table into a target table like so...
INSERT INTO TargetTable (NewIdColumn, NewNameColumn)
SELECT
IdField,
NameField
FROM
#LocalTable
BUT: NewNameColumn in TargetTable has a UNIQUE constraint, and so duplicates cause an exception.
I want to apply this example,
ROW_NUMBER() OVER(PARTITION BY NameField ORDER BY NameField)
Such that the NameField is appended/suffixed with a number digit indicating its duplication.
I have this working example that can select correctly appended values, but I cannot get this to work in an update statement like this:
UPDATE localtable
SET NameField = AppendedNameField
FROM #LocalTable AS localtable
SELECT
CONCAT(Ref.NameField, ROW_NUMBER() OVER (PARTITION BY Ref.NameField
ORDER BY Source.IdField)), *
FROM
#LocalTable AS Source
INNER JOIN
#LocalTable AS Ref ON Ref.NameField = Source.NameField
AND Ref.IdField != Source.IdField
Thanks in advance.

Answer 1

If I have understood what you are trying to do.
WITH CTE AS
(
SELECT
CONCAT(NameField, ROW_NUMBER()
OVER(PARTITION BY NameField ORDER BY IdField)) AS NewName, *
FROM #LocalTable
)
UPDATE
CTE SET Name = NewName
If you only want to do it to duplicated names you can add a COUNT(*) OVER (PARTITION BY Name) into the CTE and conditional logic using that.

Answer 2

You don't necessarily need to update the table, couldn't you just add the suffix when inserting?
DECLARE #LocalTable TABLE (IdField INT, NameField VARCHAR(50));
INSERT #LocalTable VALUES (1, 'Not Duplicate'), (2, 'Duplicate'), (3, 'Duplicate');
INSERT INTO TargetTable (NewIdColumn, NewNameColumn)
SELECT IdField,
CONCAT(NameField,
CASE WHEN COUNT(*) OVER(PARTITION BY NameField) > 1
THEN ROW_NUMBER() OVER(PARTITION BY NameField ORDER BY IdField)
ELSE ''
END)
FROM #LocalTable
ORDER BY IdField;
Alternatively, you can update by simply wrapping the above select in a subquery, and updating that:
DECLARE #LocalTable TABLE (IdField INT, NameField VARCHAR(50));
INSERT #LocalTable VALUES (1, 'Not Duplicate'), (2, 'Duplicate'), (3, 'Duplicate');
UPDATE t
SET NameField = NewNameField
FROM
(
SELECT IdField, NameField,
NewNameField = CONCAT(NameField,
CASE WHEN COUNT(*) OVER(PARTITION BY NameField) > 1
THEN ROW_NUMBER() OVER(PARTITION BY NameField ORDER BY IdField)
ELSE ''
END)
FROM #LocalTable
) AS t;
SELECT * FROM #LocalTable;

Answer 3

Just add an identity field to the temp table.
DECLARE #LocalTable TABLE
(
ix int identity primary key,
IdField NVARCHAR(MAX),
NameField NVARCHAR(MAX)
)
Insert into #LocalTable(IdColumn, NameColumn)
SELECT
IdColumn,
NameColumn
FROM SourceTable
-- Make sure same names are consecutive in the table
ORDER BY NameColumn
The set NameColumn like so:
update lt set
NameColumn = NameColumn
-- Add a number based on the ix, minus the lowest ix entry for the same name
+ cast(
(select lt.ix - min(lt2.ix) + 1
from #localTable lt2 where lt2.name = lt.name)
as nvarchar(10))
from #LocalTable lt
-- Only do those with duplicated names
where lt.NameColumn in (
select NameColumn from #localtable group by NameColumn having count(1) > 1
)

Answer 4

At first, consider using Temp table instead of temp variable.
Second, try to change NVARCHAR(MAX) by something smaller, like INT
Here is the code to include only unique values in the NameField:
CREATE TABLE #LocalTable
(
IdField NVARCHAR(MAX),
NameField NVARCHAR(MAX)
)
INSERT INTO #LocalTable
VALUES (1,'A'), (2,'B'), (3,'B')
INSERT INTO TargetTable
(
NewIdColumn,
NewNameColumn
)
SELECT IdField, NameField
FROM #LocalTable
WHERE IdField in (
SELECT MIN(IdField) FROM #LocalTable
GROUP BY NameField
);
Be aware that "IdField" of duplicate records will be ignored and not inserted into target table.

Answer 5

Try using below code:
WITH TargetTable AS(
SELECT *,
ROW_NUMBER() OVER(PARTITION BY Ref.NameField ORDER BY Source.IdField) AS UniqueID
FROM #LocalTable AS L
),
UpdatedData AS(
SELECT Source.NameField,
ROW_NUMBER() OVER(PARTITION BY Ref.NameField ORDER BY Source.IdField) AS UniqueID,
CONCAT(Ref.NameField, ROW_NUMBER() OVER(PARTITION BY Ref.NameField ORDER BY Source.IdField)) AS AppendedNameField
FROM #LocalTable AS Source
INNER JOIN #LocalTable AS Ref ON Ref.NameField = Source.NameField AND Ref.IdField != Source.IdField
)
UPDATE T
SET NameField=U.AppendedNameField
FROM TargetTable AS T
JOIN UpdatedData AS U ON T.NameField=U.NameField AND T.UniqueID=U.UniqueID;