Remove Duplicates in a String

Question

I have a column that contains a list of users all pipe separated. I am writing a script to rebuild this field and need to find a way to remove any duplicate user entries that might be contained in the string. Can someone please help me with the sql to remove the duplicates from the string and update the field?
Example:
Table: REPORTCONFIG
Column: EMAILNAMES
'A.Anderson||B.Dawkins||D.Cowen||A.Anderson||J.Rowling'
Also I have to include in my script a way to remove pipes that might be at the end of the string. If someone could help me on that front I would appreciate it as well.
Edited by: 943224 on Jun 27, 2012 10:41 AM

Answer 1

It looks like Tom's solution will convert one string. And it looks a little complicated:
with data as
  ( select  token, rownum rn
      from  ( select  distinct
                      trim( substr (txt, instr (txt, ',', 1, level  ) + 1
                          , instr (txt, ',', 1, level+1)
                          - instr (txt, ',', 1, level) -1 )
                          )  as token
                from  ( select  ',' || ltrim(rtrim('two,one,three,one,two,three,two,one,one,four',','),',') || ',' txt
                          from  dual
                      )   connect by level <= length(txt)-length(replace(txt,',',''))-1
select  ltrim( max( sys_connect_by_path(token,',') ), ',' ) || ','  str
  from  data
start with rn = 1
connect by prior rn = rn-1 ;It would be nice to have it work on a whole table of data.
I came up with this, but there's one thing I don't like. The "999" hard-coded in the query. It actually only needs to be bigger than the most entries in a string. I'm not sure how to calculate that value dynamically and still keep the query relatively simple.
with smpl_data  as
  ( select  1 id, 'sam,rob,rob,joe,rob,sam'  as txt               from dual union all
    select  2   , 'two,one,three,one,two,three,two,one,one,four'  from dual union all
    select  3   , 'a,s,g,d,s,d,s,a,d,s,g,d,s,a,g,a,s,d,f'         from dual
select  id
     ,  listagg(sngl_val, ', ') within group (order by sngl_val)  as no_dup_list
  from  ( select  distinct id,  lower(trim(regexp_substr( txt, '[^,]+', 1, seq )))  as sngl_val
            from  ( select  a.id,  b.seq,  a.txt
                      from  ( select id, txt, regexp_count(txt,',')+1 as n from smpl_data )  a
                      join  ( select level as seq from dual connect by level <= 999 )  b  on ( a.n >= b.seq )
group  by  id
order  by  id
;Edited by: SK1 on Jun 27, 2012 1:13 PM