Force Unique Value During Update, without a good join key - tsql

Question

I have a table "MyTable" with 3 columns:
FilePath
FileName
Status
My Table row example:
FilePath | FileName | Status
c:\temp | Null | 30
c:\temp | Null | 30
c:\temp | Null | 30
c:\temp | Null | 30
c:\temp | a.csv | 40
c:\temp | b.csv | 40
c:\temp | c.csv | 40
c:\temp | d.csv | 40
I want to update rows with the FileName is null, with rows where FileName has value.
Now the following query will update the nulls with the same value for all.
I need to force the update to select a unique value per row during the update
UPDATE t1
SET t1.FileName = t2.FileName
FROM MyTable t1
JOIN MyTable t2 ON t1.FilePath = t2.FilePath
The current update will give the following result:
FilePath | FileName | Status
c:\temp | a.csv | 30
c:\temp | a.csv | 30
c:\temp | a.csv | 30
c:\temp | a.csv | 30
c:\temp | a.csv | 40
c:\temp | b.csv | 40
c:\temp | c.csv | 40
c:\temp | d.csv | 40
And what I need is:
FilePath | FileName | Status
c:\temp | a.csv | 30
c:\temp | b.csv | 30
c:\temp | c.csv | 30
c:\temp | d.csv | 30
c:\temp | a.csv | 40
c:\temp | b.csv | 40
c:\temp | c.csv | 40
c:\temp | d.csv | 40

Answer 1

On the wild assumptions that you are using a reasonably modern version of SQL Server and that you would like the FileName pattern to start anew and repeat as needed for each status that has NULL FileName values:
-- Sample data.
declare #MyTable as Table ( Id Int Identity, FilePath VarChar(16), FileName VarChar(16), Status Int );
insert into #MyTable ( FilePath, FileName, Status ) values
( 'c:\temp', null, 30 ),
( 'c:\temp', null, 30 ),
( 'c:\temp', null, 30 ),
( 'c:\temp', null, 30 ),
( 'c:\temp', 'a.csv', 40 ),
( 'c:\temp', 'b.csv', 40 ),
( 'c:\temp', 'c.csv', 40 ),
( 'c:\temp', 'd.csv', 40 ),
( 'c:\temp', null, 50 ),
( 'c:\temp', null, 60 ),
( 'c:\temp', null, 60 ),
( 'c:\temp', null, 60 ),
( 'c:\temp', null, 60 ),
( 'c:\temp', null, 60 ),
( 'c:\temp', null, 60 ),
( 'c:\temp', null, 60 );
select * from #MyTable;
-- Update the null values.
with
FileNames as (
select FileName, Row_Number() over ( order by FileName ) - 1 as RN
from #MyTable
where FileName is not NULL ),
NullNames as (
select Id, Row_Number() over ( partition by Status order by FileName ) - 1 as RN
from #MyTable
where FileName is NULL )
update MT
set FileName = FN.FileName
from #MyTable as MT inner join
NullNames as NN on NN.Id = MT.Id inner join
FileNames as FN on FN.RN = NN.RN % ( select count(42) from FileNames );
select * from #MyTable;