Using a delimiter to parse the string into a new variable but don't want delimiter outputted.

[Dregerator]

Hi,

I have this string: 

REFMATS= [Type:ADRG, URL :, Version : , Status : , Date : Notes :,[Type:P21 Report, URL :, Version : , Status : , Date : Notes :,[Type:eSub Source Data Location, URL :, Version : , Status : , Date : Notes :,[Type:eSub Data Package Location, URL :, Version : , Status : , Date : Notes :

I am using "[" as a delimiter to segment this string into clauses

data work._ref_mat_info (label = "Reference Materials") ;
length clause $200 ;
set test;
where REFMATS ne "";
_i = 0 ;
do until(REFMATS = "") ;
_i + 1 ;
if _i >= 100 then stop ;
_pos = index(REFMATS,'[');
if _pos = 0 then do ;
clause = REFMATS ;
REFMATS = "" ;
end ;
else do ;
clause = substr(REFMATS,1,_pos);
REFMATS = substr(REFMATS,sum(_pos,1));
end ;
output;
end ;
run ;  

However, it keeps outputting the first "[" as an individual clause so it is not lining up properly with the REFMATS string on the output dataset. Could you advise how to modify the code to use the delimiter "[" but it doesn't get outputted as an observation so the REFMATS line up properly with the clause. 

[Astounding]

It would be straightforward to use the SCAN function for this job.  That's what it was built for, along these lines:

data work._ref_mat_info (label = "Reference Materials") ;
   length clause $200 ;
   set test;
   where REFMATS ne " ";
   do _n_=1 to 100 until (clause=' ');
      clause = scan(REFMATS, _n_, '[');
      if clause ne ' ' then output;   
   end;
run;

[ballardw]

If you use Countw(Refmat,'[') can tell exactly how many elements there are to parse.

If you don't want the "REFMATS=" as a clause start at the second element.

data example (label = "Reference Materials") ;
   length clause $200 ;
   refmats="REFMATS= [Type:ADRG, URL :, Version : , Status : , Date : Notes :,[Type:P21 Report, URL :, Version : , Status : , Date : Notes :,[Type:eSub Source Data Location, URL :, Version : , Status : , Date : Notes :,[Type:eSub Data Package Location, URL :, Version : , Status : , Date : Notes :";
   do _n_=2 to countw(refmats,'[');
      clause = scan(REFMATS, _n_, '[');
      if clause ne ' ' then output;   
   end;
run;