Frequency of Characters in a string

[OS2Rules]

Hi All (again):

After a little fine tuning, this is what I came up with.....

%macro counter (var_name);                                                                                                             

data work01;                                                                                                                           

attrib char length=$1. format=$char1.;                                                                                                 

set worklib.primebase_last_client (keep=&var_name);                                                                                    

if &var_name ne: '' then do i=1 to length(&var_name);                                                                                                                                                                                                                          

    char = upcase(substr(&var_name,i,1));                                                                                              

    if indexc(char,'ABCDEFGHIJKLMNOPQRSTUVWXYZ') = 0 then output;                                                                      

                                                     else delete;                                                                                                                                                                                                              

    end;                                                                                                                               

drop &var_name i;                                                                                                                      

run;                                                                                                                                   

proc freq data=work01;                                                                                                                 

table char / list missing;                                                                                                             

run;                                                                                                                                   

%mend;

[data_null__]

I don't think you want ELSE DELETE in that context.  I modified your program slightly to use FINDC.  I also added digits, space and '.' to the list of good characters but you may not want that.  You could make the character list a parameter to the macro.

data test;
   input testvar $80.;
   cards4;
Not to worry about EBCDIC or DBCS - the whole thing is in ASCII.
Your solution looks interesting.... I will give it at try as time permits and let you know.
(also - my IBM System/370 Reference Summary GX20-1850-3 circa 1976 spells it:  EBCDIC)
%macro counter(data=,var=);
   data work01(keep=char);
      attrib char length=$1. format=$char1.;
      set &data(keep=&var);
      where not missing(&var);
      do i=1 to length(&var);
         char = char(&var,i);
         if findc(char,' .','AKI') then output;
         end;
      run;
   proc freq data=work01;
      table char / list missing;
      run;
   %mend;
%counter(data=test,var=testvar)
;;;;;
   run;

%macro counter(data=,var=);
   data work01(keep=char);
      attrib char length=$1. format=$char1.;
      set &data(keep=&var);
      where not missing(&var) and findc(&var,' .','ADKI');
      do i=1 to length(&var);
         char = char(&var,i);
         if findc(char,' .','ADKI') then output;
         end;
      run;
   proc freq data=work01;
      table char / list missing;
      run;
   %mend;
%counter(data=test,var=testvar);

[erickbernard]

well hopefully I'm not cluttering this thread by another post. Anyway, I had a similar problem some time ago, and my solution was to summarize data first, then remove all the legal characters. here's my sample code (using the testdata above, unfortunately I can't paste from SAS window, and hence the ugly formatting, sorry about that);

/*first summarize data*/

proc sort data=test by testvar; run;

proc summary data=test; by testvar; output out=test2(drop=_type_);run;;

/*then remove "legal" characters*/

data test3; set test2;

original=testvar;

testvar=upcase(testvar);

testvar=compress(testvar,"ABCDEFGHIJKLMNOPQRSTUVWXYZ");

run;