Date:         Thu, 12 Jul 2001 14:29:21 -0400
Reply-To:     "Droogendyk, Harry" <Harry.Droogendyk@CIBC.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Droogendyk, Harry" <Harry.Droogendyk@CIBC.COM>
Subject:      Re: Is there a function to return how many characters exactly pre
              fix match between two strings?
Content-Type: text/plain; charset="iso-8859-1"

Richard:

Interesting, the SUBSTR takes less time than the PEEKC for a million
iterations ( PIII 700, 128M, NT 4.0, v8.2 )

1093
1094  data _null_;
1095
1096    a= 'I am happy';
1097    b= 'I am happier';
1098
1099    do j = 1 to 1000000;
1100      do i = 1 to max(length(a),length(b));
1101         if substr(a,i,1) ne substr(b,i,1) then
1102               leave;
1103      end;
1104    end;
1105  run;

NOTE: DATA statement used:
      real time           7.88 seconds
      cpu time            7.72 seconds


1106
1107  data _null_;
1108
1109    a= 'I am happy';
1110    b= 'I am happier';
1111
1112    do j = 1 to 1000000;
1113    do i = 0 to max(length(a),length(b))-1;
1114       if peekc(addr(a)+i,1) ne peekc(addr(b)+i,1) then
1115           leave;
1116    end;
1117    end;
1118  run;

NOTE: DATA statement used:
      real time           12.07 seconds
      cpu time            11.87 seconds


                -----Original Message-----
                From:   Gregg P. Snell
[mailto:gsnell@datasavantconsulting.com]
                Sent:   Thursday, July 12, 2001 1:38 PM
                To:     SAS-L@LISTSERV.UGA.EDU
                Subject:        Re: Is there a function to return how many
characters exactly prefix match between two strings?

                Greetings Richard!

                This is a tough one.  I've played with it and don't think it
can be done
                without at least one loop and one substring. Though I would
love to see any
                other responses as I can see the value is such a function.
Here is the best
                solution I could come up with.  It searches from the end of
the shortest
                string towards the front.

                data _null_;
                  s1="I am Happy";
                  s2="I am Happier";
                  retain order 0;
                  if length(s1) < length(s2) then order=1;
                  retain x;
                  x=min(length(s1),length(s2));
                  do until(done);
                     if order then done=index(s2,substr(s1,1,x));
                     else             done=index(s1,substr(s2,1,x));
                     x=x-1+done;
                     if x=0 then done=1;
                  end;
                  put x=;
                run;

                Anyone else have a suggestion?

                Gregg P. Snell
                Data Savant Consulting
                (913) 638-4640
                (208) 977-1943 fax
                http://www.datasavantconsulting.com

                Co-chair of MWSUG 2001
                http://www.mwsug.org/mw_2001/index.htm


                Richard A. DeVenezia <radevenz@IX.NETCOM.COM> wrote in
message
                news:e320eda1.0107120737.2b80c12c@posting.google.com...
                > I am looking for a function something like this:
                >
                > prefixMatchCount = prefixMatch (s1, s2);
                >
                > s1 = 'abcdefgh';
                > s2 = 'a bcdefg';
                > prefixMatch (s1, s2) would return 1, indicating
substr(s1,1,1) =
                > substr(s2,1,1)
                >
                > s1 = 'I am happy';
                > s2 = 'I am happier';
                > prefixMatch (s1, s2) would return 9, indicating
substr(s1,1,9) =
                > substr(s2,1,9)
                >
                > I would prefer _not_ to have to use loops comparing
consecutively
                > longer substr() until non match.
                >
                >