Re: appending data in a binary file

[email protected] Thu, 6 Nov 2003 20:39:59 +0000 (UTC)

Newsgroups	gmane.comp.gnu.octave.general,gmane.comp.gnu.octave.sources
Message-ID	<[email protected]>

Fresh from the oven! :-)

Attached, please find my M-file named "apend_save.m" .

function: append_save
arguments: 3+additional
1st: string for file name
2nd: string for options to pass to load & save ( i.e., "-binary" )
3rd: a 1x2 cell with the first cell element being a string for the
variable name, and the 2nd cell being the variable value.
4th: can accept additional cells of the same format as "3rd" above.

Exampe:

octave> A = zeros(2,2);
octave> save -binary "test.txt" A
octave> clear A
octave> B = ones(2,2);
octave> append_save( "test.txt", "-binary", {"B", B } )
octave> clear B
octave> A
error: `A' undefined near line 147 column 1
octave> B
error: `B' undefined near line 147 column 1
octave> load -binary "test.txt"
octave> A
A =
  0  0
  0  0

octave> B
B =
  1  1
  1  1

-- <end example> --

   Please give me feedback on this hack.

Thanks,

~Tomer


On Nov 6, 2003 at 10:27pm, charo wrote:
rbalbo >Hi all,
rbalbo >
rbalbo >is there a way to append data to an Octave binary file?
rbalbo >
rbalbo >Thanks,
rbalbo >
rbalbo >R

append_save.m (text/plain, 2.7 KB)

## Tomer Altman taltman-at-lbl-dot-gov

function [ return_value ] = append_save ( filename,
					  option,
					  var_val_cell,
					  varargin )

  ## Input checking:

  if ( nargin < 3 )
    error("append_save: needs three arguments.");
  elseif ( !isstr(filename) )
    error("append_save: filename must be a string.");
  elseif ( !isstr(option) )
    error("append_save: option must be a string." );
  elseif ( !iscell(var_val_cell) )
    error("append_save: variable-value pairs must be cells.")
  elseif ( nargin > 3 )
    for i=1:(nargin-3)
      current_cell = varargin(i);
      if ( !iscell(current_cell) )
	error("append_save: variable-value pairs must be cells.");
      elseif ( ( columns( current_cell ) != 2 ) 
	        || ( rows( current_cell ) != 1 ) )
	error("append_save: variable-value pairs must be 1x2 cells.")
      elseif ( !isstr(current_cell{1} ) )
	error("append_save: variable in pair must be a string." )
      endif
    endfor
  endif

  ## First step: load into the environment what is already stuffed in
  ## the target file. Then, add their name to the list for "save".

  env1 = who;
  
  eval([ "load -force ", \
	option, " ",    \
	filename ] );

  env2 = who;
  
  num_orig_vars = rows(env1);

				# Not really 'current' env...
  num_current_vars = rows(env2);

  num_new_vars = num_current_vars - num_orig_vars;

  var_str = "";

  ## This double 'for loop' weeds out only the loaded vars for
  ## inclusion.

  if ( num_new_vars )

    for i=1:num_current_vars

      current_var = env2{i,1};

      old_bool = 0;

      for j=1:num_orig_vars

	if ( strcmp( env1{j,1}, env2{i,1} ) 
	     ||
	     strcmp( env2{i,1}, "env1" ) )

	  old_bool = 1;

	endif

      endfor

      if ( old_bool == 0 )

	var_name = env2{i,1};

	var_str = [ var_str, " ", var_name, " " ];

      endif

    endfor

  endif

  
  ## Second step: load into the environment the variable pairs. Then,
  ## add the name to the string for "save".

  var_name = var_val_cell{1};

  var_val = var_val_cell{2};

  temp = var_val;

  eval([ var_name, "=temp;" ]);

  var_str = [ var_str, " ", var_name, " " ];

  ## Third step: do the same as step two, but loop through the possible
  ## variable arguments.

  for i=1:(nargin-3)

    current_cell = varargin(i);

    var_name = current_cell{1};
    
    var_val = current_cell{2};
    
    temp = var_val;
    
    eval([ var_name, "=temp;" ]);
    
    var_str = [ var_str, " ", var_name, " " ];

    var_str

  endfor

  ## Finally, save all of the variables into the target file:

  eval( [ "save ", \
	   option, " ",  \
	   filename, " ", var_str; ] );

endfunction