How to read specific information from Tweet data

[Attyslogin]

note : raw data line are actually different twitter comments. Need to fetch only those data which has special character (with character) as shown in output and those comment does not contains special character then make it null.

Raw data 

@nice#Greatwork@fake

very nice portal ncs 

Good portal @hatsoff

output needed :

@nice

#Greatwork

@fake 

null

@hatsoff

[Reeza]

Look at the SCAN() function including the third parameter, which allows you specify the delimiter. 

[Attyslogin]

can you show me the exact code ?

[Reeza]

Here's a fully worked example. Specify your delimiters as needed, make sure to include spaces in there if required.

http://blogs.sas.com/content/iml/2016/07/11/break-sentence-into-words-sas.html

I don't think these will give you a NULL value, but if you really need it add in a check if the number of words is 0 then output a null record.

[error_prone · Accepted solution]

A not to complex regular expression and call prxnext can solve the problem:

data work.have;
   length lot $ 100;
   input;

   lot = trim(_infile_);

   datalines;
@nice#Greatwork@fake
very nice portal ncs 
Good portal @hatsoff
;
run;


data work.want;
   set work.have;
   length 
      word $ 100
      rx start p l 8
   ;
   retain rx;

   if _n_ = 1 then do;
      rx= prxparse('/([@#]\w+)/i');
   end;

   start = 1;
   
   call prxnext(rx, start, -1, lot, p, l);
   do while (p > 0);
      word = substr(lot, p, l);
      output;
      call prxnext(rx, start, -1, lot, p, l);
   end;

   if missing(word) then do;
      word = 'null';
      output;
   end;

   keep word;
run;