How to Split a name string with no delimiters into columns/variables

[Rakesh93]

DineshReddy
RajeshRaoReddy
PraveenKumarRaoKumar

ans needed as

    Var1       Var2       Var3      Var4
1. Dinesh   Reddy
2. Rajesh   Rao       Reddy
3. Praveen Kumar    Rao      Kumar

total we need 4 variables

[MargoBlue]

Maybe it helps:

https://communities.sas.com/t5/SAS-Programming/Split-the-string-into-two-parts/td-p/133091

[unison]

Something like this? Your delimiter is an UPPER case letter.

data have;
	input name :$50.;
	datalines;
DineshReddy
RajeshRaoReddy
PraveenKumarRaoKumar
;
run;

data want;
	set have;
	array names {4} $ var1-var4;

	do _i=1 to dim(names);
		_this_upper=anyupper(name, _this_upper+1);
		_next_upper=anyupper(name, _this_upper+1);

		if _next_upper=0 then
			_next_upper=length(name)+1;

		if _this_upper>0 then names[_i]=substr(name, _this_upper, _next_upper-_this_upper);
		else
			leave;
	end;
	drop _:;
run;

[Rakesh93]

Can you please explain how does this code works??

[unison]

Here's a commented version -- tried to describe as best as possible..:

data want;
	set have;
	array names {4} $ var1-var4; *Create an array var1-var4;

	do _i=1 to dim(names); *Work across arrays i=1 to 4 in this case;
		*Use anyupper function to get positions of uppercase characters;
		*First parameter is string to search, second parameter is starting position to search;
		*-- returns 0 if there are no more uppercase characters after specified starting position.;
		_this_upper=anyupper(name, _this_upper+1); *position (index) of THIS upper character;
		_next_upper=anyupper(name, _this_upper+1); *position (index) of NEXT upper character;

		if _next_upper=0 then 
			_next_upper=length(name)+1; *If next upper is 0 (i.e. no more uppercase characters, then change to ending index +1 (helps to get very last chunk out properly);

		if _this_upper>0 then names[_i]=substr(name, _this_upper, _next_upper-_this_upper); *If this upper index is > 0 then capture the substr between this upper and the next one;
		else
			leave; *Else leave and move to next observation (i.e. go from row 1 to row 2);
	end;
	drop _:; *I've named all of my helper variables starting with '_', this drops all of them.;
run;

[Rakesh93]

Thanks

[ed_sas_member]

Hi @Rakesh93 

here is a way to achieve this, using pearl regular expressions.

NB: the proc sql step enables you not to specify arbitrary 4 as the maximum count of words.

data have;
	input name $80.;
	datalines;
DineshReddy
RajeshRaoReddy
PraveenKumarRaoKumar
;
run;

data have1;
	set have;
	_name= prxchange('s/([A-Z])/ $1/',-1,name);
run;

proc sql noprint;
	select max(countw(_name)) into:nb_max from have1;
quit;

data want;
	set have1;
	array var(&nb_max) $20.;
	do i=1 to countw(_name);
		var(i)=scan(_name,i," ");
	end;
	drop i _name;
run;

[novinosrin · Accepted solution]

data have;
	input name :$50.;
	datalines;
DineshReddy
RajeshRaoReddy
PraveenKumarRaoKumar
;
run;

data want;
 set have;
 array names {4} $ var1-var4;
 do _pos=anyupper(name)by 0 while(_pos);
  _p=_pos;
  _pos=anyupper(name,_pos+1);
  _n=sum(_n,1);
  if _pos then  names(_n)=substr(name,_p,_pos-_p);
  else names(_n)=substr(name,_p);
 end;
 drop _:;
run;
proc print noobs;run;

name	var1	var2	var3	var4
DineshReddy	Dinesh	Reddy
RajeshRaoReddy	Rajesh	Rao	Reddy
PraveenKumarRaoKumar	Praveen	Kumar	Rao	Kumar

[novinosrin]

And a variant of the previous with COMPRESS and FINDC

data want;
 set have;
 array names {4} $ var1-var4;
  _t=strip(compress(name,,'kU'));
 do _pos=findc(compress(name),_t)by 0 while(_pos);
  _p=_pos;
  _pos=findc(compress(name),_t,_pos+1);
  _n=sum(_n,1);
  if _pos then  names(_n)=substr(name,_p,_pos-_p);
  else names(_n)=substr(name,_p);
 end;
 drop _:;
run;
proc print noobs;run;

name	var1	var2	var3	var4
DineshReddy	Dinesh	Reddy
RajeshRaoReddy	Rajesh	Rao	Reddy
PraveenKumarRaoKumar	Praveen	Kumar	Rao	Kumar

[mkeintz]

@Rakesh93 

This code loops through the original NAME letter by letter, appending the individual letter to the "current" name component.  Every time the loop encounters an uppercase letter, it just increments which word is considered current:

data have;
  input name :$30.;
datalines;
DineshReddy
RajeshRaoReddy
PraveenKumarRaoKumar
;

data want (drop=L w);
  set have;
  array n {4} $10 name1-name4;
  do L=1 to length(name);
    if char(name,L) = upcase(char(name,L)) then w=sum(w,1);
    n{w}=cats(n{w},char(name,L));
  end;
run;

[Ksharp]

data have;
  input name :$30.;
datalines;
DineshReddy
RajeshRaoReddy
PraveenKumarRaoKumar
;
data temp;
set have;
id+1;
pid=prxparse('/[A-Z][a-z]+/');
s=1;e=length(name);
call prxnext(pid,s,e,name,p,l);
do while(p>0);
 var=substr(name,p,l);
 output;
call prxnext(pid,s,e,name,p,l);
end;
keep id var;
run;
proc transpose data=temp out=want prefix=var_;
by id;
var var;
run;

[Satish_Parida]

This works

data have;
input names:$50.;
cards;
DineshReddy
RajeshRaoReddy
PraveenKumarRaoKumar
;
run;


data temp(drop=i);
set have;
 do i=1 to length(names)+1;
 	if (rank(substr(names,i,1)) gt 64 and rank(substr(names,i,1)) lt 91 and i ne 1) or i=length(names)+1 then do;
 		output;
 		name=substr(names,i,1);
 	end;
 	else name=cats(name,substr(names,i,1));
 end;
run;

proc sort data=temp;
by names;
run;

proc transpose data=temp out=want(drop=_name_);
by names;
var name;
run;

Please let us know if it worked for you.