oracle regexp_replace to remove duplicate duplicate strings and double separators

Question

Been using the following regexp_replace to create a comma separated string for those values in 'col_name' and remove any duplicates.

REGEXP_REPLACE(RTRIM(XMLAGG(XMLELEMENT(E,col_name,',').EXTRACT('//text()')).GetClobVal(),','),'([^,]+)(,\1)+', '\1')

Works ok but with 1 exception - if 'col_name' has empty string values. What I am ending up with is:

value1,value2,,value3,value4,,value5 etc.

Looking for that elusive regex that can remove any duplicate values and those repeating commas.

Thanks

--- updated ----

col-name ----------------- row1 value 1 row2 value 2 row3 row4 value 2 row5 value 3 row6 value 1 row7

Expected Result: value1,value2,value3

Answer 1

Would the following be acceptable (not a REGEXP_REPLACE, but maybe easier to understand)?

Test table/data:

create table t as
select 'row1' c1 , 'value1' c2 from dual union all
select 'row2' ,  'value2' from dual union all
select 'row3' ,  null from dual union all
select 'row4' ,  'value2' from dual union all
select 'row5' ,  'value3' from dual union all
select 'row6' ,  'value1' from dual union all
select 'row7' ,  null from dual;

select * from t;

C1      C2
row1    value1
row2    value2
row3    null
row4    value2
row5    value3
row6    value1
row7    null

Query:

select 
  listagg(c2 , ',') within group ( order by c2 ) 
from (
  select distinct c2 
  from t
  where c2 is not null
);

result
value1,value2,value3

Dbfiddle here.