Anonymous Monk has asked for the wisdom of the Perl Monks concerning the following question:

I'm trying to decode an byte array, which contains both signed and unsigned encoded floats. I've written this code which prints integers. Your help is appreciated. 
my $length = length($contents) / 8;
foreach (my $i = 0; $i < $length; $i ++)
{
   my $offset = $i * 8;
   print ord substr($contents, $offset, 8), "\n";
}

[download]

Replies are listed 'Best First'.
Re: Converting bytes to floats
by ikegami (Patriarch) on Apr 17, 2023 at 18:23 UTC
[reply]
[d/l]
      It's a series of bytes. This powershell works: 
      $len = $bytes.Count/8
for( $i = 0 ; $i -lt $len ; $i++) {
  [System.BitConverter]::ToDouble($bytes, $i * 8 )
}

      [download]
[reply]
[d/l]
[reply]
[d/l]
[select]
        Yes, I think, what you are looking for is $number = unpack('d', $bytes);

        If $bytes hold 8 bytes, then unpack will convert that to a number.

[reply]
Re: Converting bytes to floats
by etj (Priest) on Apr 18, 2023 at 19:04 UTC
    A slightly deranged piece of PDL code I wrote to rapidly decode lots of floats at scale from binary STL files, at https://github.com/PDLPorters/pdl/blob/5b6057ffed3ffc1fe2f8018b28299ecf30e36f35/IO/STL/STL.pm#L152-L170: 
    sub _read_binary {
  my ($fh) = @_;
  barf "bigfloat" unless(length(pack("f", 1)) == 4);
  # TODO try to read part name from header (up to \0)
  seek($fh, 80, 0);
  my $buf; read($fh, $buf, 4) or warn "EOF?"; my $triangles = unpack('
+L<', $buf);
  my $bytes = 50 * $triangles; # norm+3vertices * 3float + short with 
+length of extra
  my $bytespdl = zeroes PDL::byte(), 50, $triangles;
  my $bytesread = read($fh, ${$bytespdl->get_dataref}, $bytes);
  barf "Tried to read $bytes but only got $bytesread" if $bytesread !=
+ $bytes;
  $bytespdl->upd_data;
  my $floatpdl = zeroes PDL::float(), 3, 4, $triangles;
  ${$floatpdl->get_dataref} = ${$bytespdl->slice('0:47')->get_dataref}
+;
  $floatpdl->upd_data;
  $floatpdl->type->bswap->($floatpdl) if isbigendian();
  # TODO check that the unit normal is within a thousandth of a radian
  # (0.001 rad is ~0.06deg)
  _as_ndarray($floatpdl->slice(':,1:3'));
}

    [download]
[reply]
[d/l]
[select]
Re: Converting bytes to floats
by Anonymous Monk on Apr 18, 2023 at 14:16 UTC
    Thanks for the help. I was able to get it working by original reply. 
    my @doubles = unpack "d*", $contents;

    [download]
    Some background. This data was produced by a java app and represents points on a chromatogram.
[reply]
[d/l]

Back to Seekers of Perl Wisdom