topic Re: Limitation of tranwrd function in SAS Programming

Limitation of tranwrd function

frupaul — Tue, 20 Nov 2018 17:06:24 GMT

Hi Everyone,

I have got the below data step:

data nn;
input email :$20.;
list = catx(';','Juke@hotmail', 'Frida@gmail');
if findc(trim(email),List) then Cleansed_List = tranwrd(List,trim(email),'');
cards;
Juke@hotmail
Frida@gmail
Killian@yahoo.com
;

It creates Cleansed_List with a semi colon as delimiter in the wrong places e.g preceding certain values

What I want to create is Cleansed list, with no semi-colon delimiter at the start and end of the resulting values.

Can anyone help please?

Re: Limitation of tranwrd function

Astounding — Tue, 20 Nov 2018 17:17:26 GMT

Easiest would be to eliminate all the semicolons:

list = catx(' ','Juke@hotmail', 'Frida@gmail');

The rest of the program should still work just fine.

Re: Limitation of tranwrd function

Reeza — Tue, 20 Nov 2018 17:23:09 GMT

Show what your input data looks like and what you expect as output please.

Re: Limitation of tranwrd function

data_null__ — Tue, 20 Nov 2018 18:41:39 GMT

The TRANSTRN function is what you seek.

data nn;
input email :$20.;
list = catx(';','Juke@hotmail', 'Frida@gmail');
if findc(trim(email),List) then Cleansed_List = tranSTRN(List,trim(email),trimn(' '));
x = cleansed_list;
cards;
Juke@hotmail
Frida@gmail
Killian@yahoo.com
;;;;
   run;

Re: Limitation of tranwrd function

FreelanceReinh — Tue, 20 Nov 2018 19:01:39 GMT

Hi @frupaul,

FINDC is most likely not what you want: It searches for characters, not substrings of length >1. FIND would be more appropriate, but with reversed order of arguments. TRANSTRN would be preferable to TRANWRD as has been mentioned already. However, it would still leave double semicolons if an email address in the middle of a longer list was removed or the leading or trailing semicolons you have observed. PRXCHANGE should work, but it's maybe simpler to just divide the list and put it together again after omitting the email address(es) to be deleted:

data nn;
length list Cleansed_List $80;
array _em[99] $20;
input email $20.;
list = catx(';', 'Juke@hotmail', 'Frida@gmail');
do _i=1 to countw(list, ';');
  _emi=scan(list, _i, ';');
  if _emi ne email then _em[_i]=_emi;
end;
Cleansed_List = catx(';', of _em[*]);
drop _:;
cards;
Juke@hotmail
Frida@gmail
Killian@yahoo.com
;

(Make sure to increase lengths and number of array elements if necessary.)

Re: Limitation of tranwrd function

frupaul — Tue, 20 Nov 2018 19:12:40 GMT

Thanks for your solution. I think you might have over complicated it using the do loops and stuff. But thanks for providing a solution

Re: Limitation of tranwrd function

Tom — Tue, 20 Nov 2018 19:19:36 GMT

What are you actually trying to do?

Also please look at the definitions of FINDC() and TRANWRD() as I don't think you are using either of them in an appropriate way for what I think your question might be. FINDC() is for finding first location of any one of the characters in a list. TRANWRD() for change one string to another (as opposed to TRANSLATE() that changes individual characters), but does not really understand the concept of a word despite its name.

Re: Limitation of tranwrd function

PGStats — Tue, 20 Nov 2018 20:25:59 GMT

@FreelanceReinh's solution might look too complicated, but it is the only one that really works. PRXCHANGE would also work well, but compiling a new pattern for every observation would be very expensive.