topic Re: How to split a string when AND or OR is found? in SAS Programming

How to split a string when AND or OR is found?

inavjots — Fri, 03 Jul 2020 18:51:38 GMT

Sample string-

Hobbies= Reading and singing.

How to store Reading and singing in two separate variables such as Hobby 1 and Hobby 2?

Hobby1= Reading

Hobby2= Singing

Thanks

Re: How to split a string when AND or OR is found?

Shmuel — Fri, 03 Jul 2020 18:38:41 GMT

Are you looking for verbs?

What would you subset into verbs having a text like:

a = "I like to read books and to draw pictures" - ?

You should post more examples and try to define the rules of words to look for.

Re: How to split a string when AND or OR is found?

inavjots — Fri, 03 Jul 2020 18:52:45 GMT

The question has been updated. Please take a look now. I hope it is understandable now.

Re: How to split a string when AND or OR is found?

Reeza — Fri, 03 Jul 2020 19:05:27 GMT

You can use SCAN() and FINDW()

Use FINDW() to see if you have an AND/OR and then if you do, use SCAN() to portion them out. Or use SCAN to loop through the words and drop any and/or's.

Re: How to split a string when AND or OR is found?

inavjots — Fri, 03 Jul 2020 19:24:28 GMT

Thank you for your reply, Could you please explain it with an example? I am not familiar with FINDW().

Re: How to split a string when AND or OR is found?

Reeza — Fri, 03 Jul 2020 19:40:21 GMT

This is the closest example I have. You can add conditions to it.
https://gist.github.com/statgeek/bed5ea2c12903b38fdcf19f3f1f1aae9

If you need further assistance please show what you've tried and what issues you're having.

Re: How to split a string when AND or OR is found?

Shmuel — Fri, 03 Jul 2020 19:47:03 GMT

You gave only one example and did not relate to the example I have posted.

For your specific example you can use next code, though it's not a complete, because

you will get a bad result for the 2nd example. Check output.

data test;
   infile cards truncover;
   input hobbies $char50. ;
cards;
Reading and Singing
I like to read books and to draw pictures
;
run;
data want;
 set test;
    do i=1 to countw(hobbies);
       if scan(lowcase(hobbies),i) in ('and', 'or') then do;
          hobby1 = scan(hobbies, i-1);
          hobby2 = scan(hobbies, i+1);
          output;
       end;
    end;
run;

To get a better code you need to post more examples with the desired output.

Assuming hobbies are usually verbs, it may be more difficult to decide by a program which word is a verb.

Re: How to split a string when AND or OR is found?

inavjots — Fri, 03 Jul 2020 20:07:09 GMT

This is perfect. Much obliged.