From sources-request at octave dot org Thu Apr 21 09:14:39 2005
Subject: Matlab compressed files
From: Philipp =?iso-8859-1?q?G=FChring?= <pg at futureware dot at>
To: sources at octave dot org
Date: Thu, 21 Apr 2005 03:39:02 -0500

--Boundary-00=_rY2ZCDjg0xTH0Hf
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

Hi,

Octave cannot handle Matlab compressed files at the moment.

According the the current fileformat specification:
http://www.mathworks.com/access/helpdesk/help/pdf_doc/matlab/matfile_format=
=2Epdf

compressed Tags have miCOMPRESSED (15), and it is simply zlib compressed.

I have written a decompressor in Perl, that does the job cleanly.
Usage:
perl matlab_decompress.pl <compressed_file.mat>

I tried to write decompression support into Octave itself, but I didn=B4t k=
new=20
how to do the memory handling, since the fileformat does not say, how large=
=20
the decompressed memory is, but the zlib library needs a large enough buffe=
r.

The decompression is really simple, you just need the uncompress function f=
rom=20
zlib:
my $dest =3D uncompress($buffer);

Regards,
=2D-=20
Philipp G=FChring

--Boundary-00=_rY2ZCDjg0xTH0Hf
Content-Type: application/x-perl;
  name="matlab_decompress.pl"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
	filename="matlab_decompress.pl"

#!/usr/bin/perl -w

use Compress::Zlib;

my $filename=$ARGV[0];

$filename=~s/\.mat$/_uncompressed.mat/;

open OUT,">$filename";


sub readmatlab($)
{
  my $d=$_[0];
  my $f=undef;
  if(substr($d,0,4) eq "MATL")
  {
    print "Detected Matlab file\n";
  }
  else { return undef};

  if(substr($d,124,2) eq "\x00\x01")
  {
    print "Version: 1.0\n";
  }
  if(substr($d,126,2) eq "IM")
  {
    print "Big Endian\n";
    $f="V";
  }
  elsif(substr($d,126,2) eq "MI")
  {
    print "Little Endian\n";
    
  }
  else
  {
    print "Dateiformat nicht erkennbar!\n";
  }

  print OUT substr($d,0,128);

  my $pos=128;
  my $todo=1;
  while($pos< length($d))
  { 
    my $DataType=unpack($f,substr($d,$pos,4));
    my $NumberBytes=unpack($f,substr($d,$pos+4,4));
    my $buffer=substr($d,$pos+8,$NumberBytes);

    print "Paket mit Format $DataType mit $NumberBytes Bytes.\n";

    if($DataType == 15)
    {
      print "Compressed Tag detected. Ungzipping ...\n";
      my $dest = uncompress($buffer);
#      my $dest = Compress::Zlib::memGunzip($buffer);
      print "---".$dest."---\n"; 
      print "---".$buffer."---\n"; 
      print OUT $dest;
    }
    else
    {
      print OUT substr($d,$pos,8+$NumberBytes);
    }

    $pos+=$NumberBytes+8;
  }
}

sub readmatlabfile($)
{
  print "Lese $_[0]\n";
  if(open(IN,"<$_[0]"))
  { 
    my $old=$/;
    undef $/;
    my $content=<IN>;
    close IN;
    $/=$old;
    return readmatlab($content);
  }
  return undef;
}

readmatlabfile($ARGV[0]);

close OUT;

--Boundary-00=_rY2ZCDjg0xTH0Hf--