From bug-request at octave dot org Wed Mar 16 14:11:57 2005
Subject: split.m
From: =?ISO-8859-1?Q?S=F8ren_Hauberg?= <soren at hauberg dot org>
To: bugs at octave dot org
Date: Wed, 16 Mar 2005 14:09:40 -0600

This is a multi-part message in MIME format.
--------------090609060104020400060207
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 8bit

Hi
The current version of split.m is very slow on large strings (takes more 
than 3 hours to process a 3MB string on my old laptop). The attached 
version is many times faster (same string takes about 15 seconds to 
process on same machine) because it only allocates memory once.

This is my first patch so please tell me if I've done something wrong

Thanks,
Søren

--------------090609060104020400060207
Content-Type: text/x-patch;
 name="split.m.diff"
Content-Transfer-Encoding: 8bit
Content-Disposition: inline;
 filename="split.m.diff"

*** /usr/share/octave/2.1.64/m/strings/split.m	2004-12-05 05:17:25.000000000 +0100
--- split.m	2005-03-16 21:08:06.980198640 +0100
***************
*** 33,38 ****
--- 33,39 ----
  
  ## Author: Kurt Hornik <Kurt dot Hornik at ci dot tuwien dot ac dot at>
  ## Adapted-By: jwe
+ ## Minor changes by Søren Hauberg
  
  function m = split (s, t)
  
*************** function m = split (s, t)
*** 40,46 ****
      usage ("split (s, t)");
    endif
  
!   if (isstr (s) && isstr (t))
  
    l_s = length (s);
    l_t = length (t);
--- 41,50 ----
      usage ("split (s, t)");
    endif
  
!   if not(ischar (s) && ischar (t))
!     error ("split: both s and t must be strings");
!   endif
! 
  
    l_s = length (s);
    l_t = length (t);
*************** function m = split (s, t)
*** 48,88 ****
    if (l_s == 0)
      m = "";
      return;
    elseif (l_s < l_t)
      error ("split: s must not be shorter than t");
    endif
! 
!   if (l_t == 0)
!     ind = 1 : (l_s + 1);
!   else
!     ind = findstr (s, t, 0);
!     if (length (ind) == 0)
!       m = s;
!       return;
!     endif
!     ind = [1 - l_t, ind, l_s + 1];
    endif
  
!   cmd = "";
! 
!   limit = length (ind) - 1;
! 
!   for k = 1 : limit
! 
!     range = (ind (k) + l_t) : ind (k + 1) - 1;
! 
!     if (k != limit)
!       cmd = sprintf ("%s\"%s\", ", cmd, undo_string_escapes (s (range)));
!     else
!       cmd = sprintf ("%s\"%s\"", cmd, undo_string_escapes (s (range)));
!     endif
! 
!   endfor
! 
!   m = eval (sprintf ("str2mat (%s);", cmd));
! 
!   else
!     error ("split: both s and t must be strings");
    endif
  
  endfunction
--- 52,86 ----
    if (l_s == 0)
      m = "";
      return;
+   elseif (l_t == 0)
+     m = s';
+ 	return;
    elseif (l_s < l_t)
      error ("split: s must not be shorter than t");
    endif
!   
!   if (min(size(s)) ~= 1 | min(size(t)) ~= 1)
!     error("split: multible strings are not supported");
    endif
  
!   ind = findstr (s, t, 0);
!   if (length (ind) == 0)
!     m = s;
!     return;
    endif
+   ind2 = [1, ind+l_t];
+   ind  = [ind, l_s+1];
+ 
+   ind_diff = ind-ind2;
+   % Create a matrix of the correct size that's filled with spaces
+   m_rows = length(ind);
+   m_cols = max(ind_diff);
+   m = char( zeros(m_rows, m_cols) + ' ' );
+ 
+   % Copy the strings to the matrix
+   for i = 1:length(ind)
+     tmp = ind2(i):(ind(i)-1);
+     m(i, 1:length(tmp)) = s(tmp);
+   end
  
  endfunction

--------------090609060104020400060207--



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------