Find and Remove Repeated Substrings

Question

I've a column in a SQL Server 2008 table where part of the string was accidentally repeated.

Does anyone have a quick and easy way to remove the trailing duplicated substring?

For example,

alpha\bravo\charlie\delta\charlie\delta

should be

alpha\bravo\charlie\delta

Answer 1

If you don't already have a numbers table:

SET NOCOUNT ON;
DECLARE @UpperLimit int = 4000;

;WITH n(rn) AS
(
    SELECT ROW_NUMBER() OVER (ORDER BY [object_id])
    FROM sys.all_columns
)
SELECT [Number] = rn - 1
INTO dbo.Numbers FROM n
WHERE rn <= @UpperLimit + 1;

CREATE UNIQUE CLUSTERED INDEX n ON dbo.Numbers([Number]);

Now a generic split function, that will turn your delimited string into a set:

CREATE FUNCTION dbo.SplitString
(
    @List  nvarchar(max),
    @Delim char(1)
)
RETURNS TABLE
AS
    RETURN ( SELECT 
        rn, 
        vn = ROW_NUMBER() OVER (PARTITION BY [Value] ORDER BY rn), 
        [Value]
      FROM 
      ( 
        SELECT 
          rn = ROW_NUMBER() OVER (ORDER BY CHARINDEX(@Delim, @List + @Delim)),
          [Value] = LTRIM(RTRIM(SUBSTRING(@List, [Number],
          CHARINDEX(@Delim, @List + @Delim, [Number]) - [Number])))
        FROM dbo.Numbers
        WHERE Number <= LEN(@List)
        AND SUBSTRING(@Delim + @List, [Number], 1) = @Delim
      ) AS x
    );
GO

And then a function that puts them back together:

CREATE FUNCTION dbo.DedupeString
(
    @List nvarchar(max)
)
RETURNS nvarchar(max)
AS
BEGIN
    RETURN ( SELECT newval = STUFF((
     SELECT N'\' + x.[Value] FROM dbo.SplitString(@List, '\') AS x
      WHERE (x.vn = 1)
      ORDER BY x.rn
      FOR XML PATH(''), TYPE).value(N'./text()[1]', N'nvarchar(max)'), 1, 1, N'')
    );
END
GO

Sample usage:

SELECT dbo.DedupeString(N'alpha\bravo\bravo\charlie\delta\bravo\charlie\delta');

Results:

alpha\bravo\charlie\delta

You can also say something like:

UPDATE dbo.MessedUpTable
  SET OopsColumn = dbo.DedupeString(OopsColumn);

@MikaelEriksson will probably swoop in with a more efficient way to use XML to eliminate duplicates, but that is what I can offer until then. :-)

Answer 2

create function RemoveDups(@S nvarchar(max)) returns nvarchar(max)
as
begin
  declare @R nvarchar(max)
  declare @W nvarchar(max)
  set @R = ''

  while len(@S) > 1
  begin
    -- Get the first word
    set @W = left(@S, charindex('/', @S+'/')-1)

    -- Add word to result if not already added
    if '/'+@R not like '%/'+@W+'/%'
    begin
      set @R = @R + @W + '/'
    end 

    -- Remove first word
    set @S = stuff(@S, 1, charindex('/', @S+'/'), '')
  end

  return left(@R, len(@R)- 1)
end

As requested by Aaron Bertrand. I will however make no claim on what is the fastest to execute.

-- Table to replace in
declare @T table
(
  ID int identity,
  Value nvarchar(max)
)

-- Add some sample data
insert into @T values ('alpha/beta/alpha/gamma/delta/gamma/delta/alpha')
insert into @T values ('delta/beta/alpha/beta/alpha/gamma/delta/gamma/delta/alpha')

-- Update the column
update T
set Value = NewValue
from (
       select T1.ID,
              Value,
              stuff((select '/' + T4.Value
                     from (
                            select T3.X.value('.', 'nvarchar(max)') as Value,
                                   row_number() over(order by T3.X) as rn
                            from T2.X.nodes('/x') as T3(X)
                          ) as T4
                     group by T4.Value
                     order by min(T4.rn)
                     for xml path(''), type).value('.',  'nvarchar(max)'), 1, 1, '') as NewValue
       from @T as T1
         cross apply (select cast('<x>'+replace(T1.Value, '/', '</x><x>')+'</x>' as xml)) as T2(X)
     ) as T

select *
from @T