Count distinct values ​​in a string

[a_ramos]

I would like to know how I can count distinct characters in a string

example:

NAME   PHONE   CHAR_DISTINCT

Ana       21212      2

Pablo    234234    3

Hana    1111111    1

[PeterClemmensen]

data have;
input NAME $ PHONE $;
datalines;
Ana 21212
Pablo 234234
Hana 1111111
;

data want(keep=NAME PHONE CHAR_DISTINCT);
   set have;
   newphone=substr(PHONE, 1, 1);
   do i=2 to length(PHONE);
      found=find(newphone, substr(PHONE, i, 1), 'it');
      if found=0 then newphone=cats(newphone, substr(PHONE, i, 1));
   end;
   CHAR_DISTINCT=length(newphone);
run;

[PeterClemmensen · Accepted solution]

Or using PRXCHANGE Function

data want;
   set have;
   array t(100) $  _temporary_;
   call missing(of t(*));
   do _n_=1 to length(PHONE);
      t(_n_)=char(PHONE,_n_);
   end;
   call sortc(of t(*));
   CHAR_DISTINCT=length(prxchange('s/([0-9])\1+/$1/', -1, cats(of t(*))));
run;

[a_ramos]

Thank you! 🙂

[Kurt_Bremser]

Just for completeness, a solution with SQL.

data have;
input NAME $ PHONE $;
datalines;
Ana 21212
Pablo 234234
Hana 1111111
;
run;

data trans;
set have;
length digit $1;
do i = 1 to length(phone);
  digit = substr(phone,i,1);
  output;
end;
drop i;
run;

proc sql;
create table want as
select
  name,
  phone,
  count(distinct digit) as char_distinct
from trans
group by name, phone;
quit;

proc print data=want noobs;
run;

Result:

                      char_
NAME      PHONE     distinct

Ana      21212          2   
Hana     1111111        1   
Pablo    234234         3   

@PeterClemmensen's first solution is the best, as it solves everything in one sequential pass through have.

[Ksharp]

Just for completeness, a solution with Hash Table.

data have;
input NAME $ PHONE $;
datalines;
Ana 21212
Pablo 234234
Hana 1111111
;
run;
data want;
 if _n_=1 then do;
  length k $ 1;
  declare hash h();
  h.definekey('k');
  h.definedone();
 end;
set have;
 do i = 1 to length(phone);
  k = char(phone,i);
  h.ref(); 
end;
count=h.num_items;
h.clear();
drop k i;
run;

[KachiM]

Here is another version. Look at the first character of PHONE. Count it. Compress the first character from the Phone. Repeat the above step until the phone has left with one or no character.

data have;
input NAME $ PHONE $;
datalines;
Ana 21212
Pablo 234234
Hana 1111111
;
run;

data want;
   set have;
   ph = phone;
   count = 0;
   do until(len <= 1);
      d = char(ph,1);
      count + 1;
      ph = compress(ph,d);
      len = length(ph);
   end;
drop ph len d;
run;

Obs 	NAME 	PHONE 	count
1 	Ana 	21212 	2
2 	Pablo 	234234 	3
3 	Hana 	1111111 	1

[FreelanceReinh]

Note that among the previous solutions only @Kurt_Bremser's produces a count of zero in the case of an empty input string. Of course, you could adapt the other solutions correspondingly if needed. For example, using COMPRESS (like in @KachiM's solution):

data want(drop=_t);
set have;
_t=phone;
do CHAR_DISTINCT=0 by 1 while(_t ne ' ');
  _t=compress(_t,first(_t));
end;
run;

To obtain CHAR_DISTINCT=1 for missing PHONE replace the DO statement with:

do CHAR_DISTINCT=1 by 1 until(_t=' ');