Find position of each letter a

[hhchenfx]

Hi,

I want to find position for each letter "a" in a series.

Is there any simple way like find(string,'a'2)?

Thank you,

HHCFX

data have;
input string :$50.;
datalines;
098ab-c1za3-8a1-8a-1a01212
;run;



data want;set have;

first_a=find(string,'a');
second_a=find(string,'a',2);
run;

[novinosrin]

data have;
input string :$50.;
datalines;
098ab-c1za3-8a1-8a-1a01212
;run;

data want;
set have;
a=0;
do pos=findc(string,'a') by 0 while (pos);
a+1;
output;
pos= findc(string,'a',pos+1) ;
end;
run;

[KachiM]

novinosrin has provided the idea and I build further on his.

hhchenfx is in looking for a SAS function to provide the  position and order of occurrence of 'a' in a string. I don't find a readymade SAS function for his desire. However, we can build a function using the SAS function compiler (FCMP). Here is a function:

proc fcmp outlib = work.cmput.str;
   function findpatnext(str $, pat $);
      file log;
      static lastpos 1;
      pos = findc(str, pat, 'it', lastpos); /* edit: added semicolon */
      lastpos = pos + 1;
      return(pos);
   endsub;
quit;

In a Data Step we can call the function as:

options cmplib = work.cmput;

data want;
   length str $26;
   str = '098ab-c1za3-8a1-8a-1a01212';
   loc = 1;
   do order = 1 by 1 while(loc);
      loc = findpatnext(str, 'a');
      if loc then output;
   end;
run;

The output Data Set, WANT:

Obs	str	loc	order
1	098ab-c1za3-8a1-8a-1a01212	4	1
2	098ab-c1za3-8a1-8a-1a01212	10	2
3	098ab-c1za3-8a1-8a-1a01212	14	3
4	098ab-c1za3-8a1-8a-1a01212	18	4
5	098ab-c1za3-8a1-8a-1a01212	21	5

We repeat the Data Step using NAME in SASHELP.CLASS as:

data want;
   set sashelp.class(keep = name);
   loc = 1;
   do order = 1 by 1 while(loc);
      loc = findpatnext(name, 'a');
      if loc then output;
   end;
run;

The output Data Set is:

Obs 	Name 	loc 	order
1 	Alfred 	1 	1
2 	Alice 	1 	1
3 	Barbara 	2 	1
4 	Barbara 	5 	2
5 	Barbara 	7 	3
6 	Carol 	2 	1
7 	James 	2 	1
8 	Jane 	2 	1
9 	Janet 	2 	1
10 	Mary 	2 	1
11 	Ronald 	4 	1
12 	Thomas 	5 	1
13 	William 	6 	1

Edited:

 In case, the n-th occurrence of 'a' is needed, the do-loop in the Data Step can be modified. Suppose the 3-rd occurrence is needed then the following Data Step can be used:

data want;
   length str $26;
   str = '098ab-c1za3-8a1-8a-1a01212';
   loc = 1;
   do order = 1 by 1 while(loc);
      loc = findpatnext(str, 'a');
      if order = 3 then do; output; leave; end;
   end;
run;

[PeterClemmensen]

A classic use of the Call PRXNEXT Routine

data have;
input string :$50.;
datalines;
098ab-c1za3-8a1-8a-1a01212
;run;

data want(keep=string pos);
   set have;
   RegExID = prxparse('/a/');
   start=1;
   call prxnext(RegExID, start, length(string), string, pos, length);
      do while (pos > 0);
         output;
         call prxnext(RegExID, start, length(string), string, pos, length);
      end;
run;

[Ksharp · Accepted solution]

data have;
input string :$50.;
datalines;
098ab-c1za3-8a1-8a-1a01212
;run;


data want;
 set have;
 do i=1 to lengthn(string);
  if char(string,i)='a' then do;position=i;output;end;
 end;
run;

[hhchenfx]

Thanks a lot.

Very easy to follow code.

HHC