#!ruby

#
# Cut header of SPL in spool folder to raw EMF
#

def spl2emf(fnm)
    File.open(fnm + ".emf", "wb") {|fo|
        File.open(fnm, "rb") {|fi|
            str = fi.read
            pos = str.index(" EMF")
            if pos == nil
                print "No EMF marker found; probably wrong file\n"
                next
            end
            fo.write(str[(pos-40)..-1])
        }
    }
end

if __FILE__ == $0
    ARGV.each {|fnm|
        print fnm, "\n"
        spl2emf(fnm)
    }
end
[download]

This should be close (UNTESTED):

#!/usr/bin/perl

#
# Cut header of SPL in spool folder to raw EMF
#

sub spl2emf {
    my $fnm = shift;
    if ( open my $FO, '>:raw', "$fnm.emf" ) {
        if ( open my $FI, '<:raw', $fnm ) {
            read $FI, my $str, -s $FI;
            my $pos = index $str, ' EMF';
            if ( $pos == -1 ) {
                print "No EMF marker found; probably wrong file\n";
                return;
                }
            print $FO substr $str, $pos - 40;
            }
        }
    }

foreach my $fnm ( @ARGV ) {
    print "$fnm\n";
    spl2emf( $fnm );
    }
[download]

While it probably doesn't matter in this case (since the whole file is being read at once), '<:raw' prevents a buffering layer from being added. You want

open(my $fh, '<:perlio:raw', $qfn);
[download]

or

open(my $fh, '<', $qfn);
binmode($fh);
[download]

Same goes for '>:raw'.

---

The usual idiom is

local $/;
my $str = <$FI>;
[download]

rather than

read $FI, my $str, -s $FI;
[download]

---

A regex (/(.{40} EMF.*)/s) would be a bit simpler than index+pos+substr and more idiomatic. It could be a bit slower, but probably not since it's optimised to look for " EMF" first.

In my defence  ;-) I was just trying to do a direct translation.   If I was going to do it in idiomatic Perl I would have done it differently.

perldoc PerlIO
[ SNIP ]
       :raw
           The ":raw" layer is defined as being identical to calling "
+binmode($fh)" - the stream is
           made suitable for passing binary data i.e. each byte is pas
+sed as-is. The stream will
           still be buffered.
[download]




Also, you should always verify that open worked correctly before trying to use its filehandle.

   :-)

Just curious, but if you've got a perfectly working ruby program already, any particular reason for switching?