Date:         Thu, 11 Mar 2010 14:19:41 -0500
Reply-To:     D T <sasandstats@LIVE.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         D T <sasandstats@LIVE.COM>
Subject:      Re: efficient way -- prx & catx -- not solved yet
Comments: To: snoopy369@gmail.com
In-Reply-To:  <b7a7fa631003110808t1c62b496r8855e9bc0c7abca3@mail.gmail.com>
Content-Type: text/plain; charset="Windows-1252"

Can anyone help out? catx seems to be a good idea, but I am still not able to get it working. 

See just below for current code (not working correctly) and further down for original problem. I have incorporated two suggestions from Joe, but have not yet solved this.

Thanks for your help!

data end;

     set start;

   

    retain reg1;

    length dairy $30 ;

    start=1;
    stop=length (comment);
   

    if _N_=1 then do;

     reg1=prxparse('/(milk\b|cheese\b|yogurt\b|cheddar\b|cream\b\scheese\b|cream\b)/i');

    end;

call prxnext(reg1, start, stop, comment, position, length);
 do while (position > 0);
  Dairy= catx(' ',dairy,prxposn(reg1,1,comment));
  replace=cats ("s$",dairy,"$$");
  comment=prxchange(replace,1,comment);
 if start > stop then position = 0;
 else
 call prxnext(reg1, start, stop, comment, position, length);
end;

   Here is the problem:
   I am using perl regular expressions in SAS to extract text from a variable with a varying number of comments.
   I would like to be able to write more than one string identified into another variable. Can anyone suggest an
   efficient way to do this?

   In my example a search string such as:
   prxparse('/(milk\b|cheese\b|yoghurt\b|cheddar\b|cream\b\scheese\b|cream\b)/i') would search for all dairy,
   and would have several hits for id 1 and 5. How do I get them most efficiently into variable dairy?
   
   
   Here is the example:
   
   data start;
      length comment $50;
      input id comment &;
      cards;
      1 milk, cheese bacon
      2 oranges/yogurt
      3 cheddar meatballs
      4 bread
      5 yogurt cream cheese, cream
   ;
   
   data end;
     set start;
   
     retain reg1;
   
    if _N_=1 then do;
     reg1=prxparse('/(milk\b|cheese\b|yogurt\b|cheddar\b|cream\b\scheese\b|cream\b)/i');
    end;
   
     if prxmatch(reg1,comment) then do;
     dairy=prxposn(reg1,1,comment);
     end;
   
  run;
   
   data end should look like this:
   
   id    dairy            meats        fruit        grains
    1 milk, cheese     bacon
    2    yoghurt                     oranges/
    3    cheddar       meatballs
    4                                                   bread
    5    yoghurt cream cheese, cream
    
    Thanks!
    
    D.T.                                          

> Date: Thu, 11 Mar 2010 10:08:41 -0600
> From: snoopy369@GMAIL.COM
> Subject: Re: [SAS-L] efficient way to write more than one string              identified into              another variable
> To: SAS-L@LISTSERV.UGA.EDU
> 
> Dairy= catx(' ',dairy,prxposn<stuff>);
> 
> or something to that effect.
> 
> 
> 
> -Joe
> 

                                          
_________________________________________________________________
Hotmail: Trusted email with Microsoft’s powerful SPAM protection.
http://clk.atdmt.com/GBL/go/201469226/direct/01/