CODEWITHSUNDEEP
sql-serverstringfunctiont-sql
yassine mhadhbi
yassine mhadhbi

Reputation: 27

Get Distinct characters from a given string in sql server

i need a T-SQL function that remove duplicate charcters in a give String

for exemple Fn_Remove('AVGHAHA') it will returns AVGH

Upvotes: 2

Views: 2418

Answers (4)

yassine mhadhbi
yassine mhadhbi

Reputation: 27

thank you guys for your help but i needed a function and i found a solution 

USE [DATAWARHOUSE]
GO

/****** Object:  UserDefinedFunction [dbo].[EliminateRedoblons]    Script Date: 3/14/2019 1:52:02 PM ******/
SET ANSI_NULLS ON
GO

SET QUOTED_IDENTIFIER ON
GO

 CREATE FUNCTION [dbo].[EliminateRedoblons](@str VARCHAR(500))
  RETURNS varchar(500)
  begin
  DECLARE
          @resultat VARCHAR(500)='',
          @i INT=0,
          @is INT,
          @c NVARCHAR(1)

          while @i<=LEN(@str)
          BEGIN
          SET @i=@i+1
          SET @c=substring(@str,@i,1)
          SET @is =CHARINDEX(@c,@resultat)
          IF @IS=0 
          BEGIN
          SET @resultat=@resultat+@c
          END

          END
          return @resultat


  END
GO

Upvotes: 0

PSK
PSK

Reputation: 17943

Another approach can be using recursive CTE, which I learned yesterday from one of @Gordon Linoff's answer.

;with cte as (
      select v.input, convert(varchar(max), '') as updated, 1 as lev
      from (values ('AVGHAHA')) v(input)
      union all
      select stuff(input, 1, 1, ''),
             (case when charindex(left(input, 1),updated) > 0  then updated  else concat(updated , left(input, 1))  end),               
             lev + 1
      from cte
      where input > ''
     )
select top (1) with ties updated
from cte
order by row_number() over (order by lev desc);

Online Demo

Edit:

As a user defined function.

CREATE FUNCTION dbo.Fn_Remove(@Input varchar(100))  
RETURNS varchar(100)   
AS   
-- Returns the stock level for the product.  
BEGIN  
    DECLARE @ret varchar(100) 
    ;with cte as (
          select v.input, convert(varchar(max), '') as updated, 1 as lev
          from (values (@Input)) v(input)
          union all
          select stuff(input, 1, 1, ''),
                 (case when charindex(left(input, 1),updated) > 0  then updated  else concat(updated , left(input, 1))  end),               
                 lev + 1
          from cte
          where input > ''
         )
    select top (1)  @ret=updated
    from cte
    order by lev desc

    RETURN @ret;  
END;

Upvotes: 1

Martin Smith
Martin Smith

Reputation: 453328

First create a Numbers table with ascending integers from 1 upwards.

Then you can use

SELECT STRING_AGG (Char, '') WITHIN GROUP (ORDER BY Pos ASC)  AS csv
FROM
(
SELECT SUBSTRING(@String, number, 1) AS Char, MIN(number) AS Pos
FROM Nums
WHERE number <= LEN(@String)
GROUP BY SUBSTRING(@String, number, 1)
) T

This uses the semantics of your default collation for equality. Use an explicit COLLATE clause if you want something different.

If you are on a version that does not support STRING_AGG you can use XML PATH for string concatenation.

Upvotes: 1

Thom A
Thom A

Reputation: 95589

Using NGrams8K, you can split the string into individual character "tokens", apply a number to that character set, and then rebuild with only the first of each character:

WITH CTE AS(
    SELECT V.S,
           N.position,
           N.token,
           ROW_NUMBER() OVER (PARTITION BY N.token ORDER BY N.position) AS RN
    FROM (VALUES('AVGHAHA'))V(S)
         CROSS APPLY dbo.NGrams8k(V.S,1) N)
SELECT V.S,
       (SELECT '' + C.token
        FROM CTE C
        WHERE C.S = V.S
          AND C.RN = 1
        ORDER BY C.position
        FOR XML PATH('')) AS Replaced
FROM (VALUES('AVGHAHA'))V(S);

Upvotes: 2

Related Questions