Re: Problem with reading lists

[Quentin Spencer]

John W. Eaton wrote:

> On  9-Mar-2005, Quentin Spencer <address@hidden> wrote:
>
> | I think this patch may be all that is necessary. I've never used 
> | textread--does someone want to test it?
>
> Yes, this might be all that is *needed*, but that function would also
> be a lot better if it avoided the use of eval.  I don't see why it is
> necessasry to use eval to create return values with the names a1, a2,
> ..., aN, and then use eval again to assign the temporary variables to
> varargout{1}, varargout{2}, ..., varargout{N}.  It seems it should be
> possbile to do that in one step, without eval.
>  
Good point. I've gone back and removed all references to eval, and 
updated the documentation and changelog in the following patch. I tested 
it on the simple case given in the help text and it seems to work 
properly. If there are no objections, I'll check the updated version 
into octave-forge CVS.
-Quentin


--- textread.m.old      2003-09-12 09:22:42.000000000 -0500
+++ textread.m  2005-03-09 15:03:36.718716296 -0600
@@ -23,7 +23,7 @@
 ## If n is omitted or negative, the entire file will be read.
 ##
 ## %#c formats return a character array of width #
-## %s and %[ formats return a list of strings
+## %s and %[ formats return a cell array of strings
 ## All other formats (see fscanf) return a column vector of doubles
 ##
 ## Example:  say the file 'info.txt' contains the following lines:
@@ -33,8 +33,8 @@
 ## Columns from this file could be read with the line:
 ##   [Lname, Fname, x, a] = textread('info.txt', '%s %s %e age=%d');
 ## then,
-##   nth(Lname,3)  is 'Penguin'
-##   nth(Fname,3)  is 'Tux'
+##   Lname{3}      is 'Penguin'
+##   Fname{3}      is 'Tux'
 ##   x(3)          is  6
 ##   a(3)          is 10
 ##
@@ -72,6 +72,8 @@
 ## 2002-03-15 Paul Kienzle
 ## * %[ produces a string
 ## * additional documentation
+## 2005-03-09 Quentin Spencer
+## * Changed list formatted outputs to cell array
 
 function [varargout] = textread(file, format, n);
 
@@ -122,16 +124,13 @@
   cat = setstr(cat);
   for i=1:length(cat)
     if cat(i) == "s"
-      eval(sprintf("a%d=list();",i));
-    elseif cat(i) == "c"
-      eval(sprintf("a%d="";", i));
+      varargout{i} = {};
     else
-      eval(sprintf("a%d=[];", i));
+      varargout{i} = [];
     endif
   endfor
 
-  
-  
+
   ## Step 3:  Read the contents of the file one term at a time.
   ##          Place the terms into temporary octave matrices
   ##          called 'a1', 'a2', et cetera.  Each pass through
@@ -167,11 +166,11 @@
                    format(idx(i):idx(i+1)-1),row,i);
        endif
         if (cat(i) == "s") # list of strings
-         eval(sprintf("a%d = append(a%d, data);", i, i));
+          varargout{i} = {varargout{i}{:}, data};
        elseif (cat(i) == "c" )  # matrix of characters
-          eval(sprintf("a%d = [ a%d ; data ];", i, i));
+          varargout{i} = [varargout{i}; data];
         else                       # matrix of scalars
-          eval(sprintf("a%d(%d) = data;", i, row));
+          varargout{i}(row) = data;
         endif
       endfor
       ++row;
@@ -187,19 +186,9 @@
     prefer_column_vectors = pcv;
     fclose(fid);
   end_unwind_protect
-  
-  ## Step 4:  populate the return values with the contents of a1, a2, etc.
+
   if nargout == 0
-     ret = list();
-     for i = 1:nFields
-       eval( sprintf("ret(i) = a%d;",i) );
-     endfor
-     vr_val_cnt = 1; varargout{vr_val_cnt++} = ret;
-  else
-    vr_val_cnt = 1;
-    for i = 1:nargout
-      eval( sprintf ("varargout{vr_val_cnt++} = a%d;",i) );
-    endfor
+    varargout = {varargout};
   endif
 
 endfunction