#! /usr/bin/perl -w

use strict;
use Digest::MD5;

my $orig = shift or die "no original given\n";
my $dup  = shift or die "no copy given\n";

open ORIG, $orig || die "Cannot open $orig for input: $!\n";
my $origmd5 = Digest::MD5->new;
my $origdig = $origmd5->addfile( *ORIG )->digest;
close ORIG;

open DUP, $dup || die "Cannot open $dup for input: $!\n";
my $dupmd5 = Digest::MD5->new;
my $dupdig = $dupmd5->addfile( *DUP )->digest;

print( $origdig ne $dupdig ? 'not ' : '', "ok\n" );
[download]

Tried your code and I am seeing the same thing, even on the localhost's own file system. Interrestingly I tried it on two seperate copies, /opt/hello and /tmp/hello, and those two DO compare via Digest::MD5! Neither's hash matches that of the original. This is true on W2K, GNU/Linux, and Solaris, perl 5.6.0.

2001-03-14 Edit by Corion : The content of this node was moved into the root node.

You might wanna add some sysseek's to make sure Digest::MD5 reads the whole file. Moreover, you have the outfile only opened to write (prepend a plus to the opening string of destFile.

Well, that doesn't explain the outcome of the comparison. (have to run....hint: OOP versus function)

Jeroen
"We are not alone"(FZ)

Update: I modified your code, and it works. Tested on linux. BTW, yer original code MD'd the objects!

#!/usr/bin/perl -w
# -*-Perl-*-

use strict;
use FileHandle;
use Digest::MD5;

my $sourceFile = $ARGV[0];
my $destFile   = $ARGV[1];
my $inFile     = new FileHandle;
my $outFile    = new FileHandle;
my $inMD5      = Digest::MD5->new;
my $outMD5     = Digest::MD5->new;
my ( 
    $fileLength,
    $fileBuffer,
    $fileOffset
   );

$inFile->open ( "<$sourceFile" )
  or die "Could not open $sourceFile:$!\n";
$inMD5->addfile ( $inFile );
print $inMD5->b64digest , "\n";

$outFile->open ( "+>$destFile" )
  or die "Could not open $destFile:$!\n";

# borrowed from "Programming Perl"
die "Could not rewind $sourceFile: $!" unless defined sysseek $inFile,
+ 0, 0;


my $blockSize = ( stat $inFile )[11] || 16384;
while ( $fileLength  = sysread $inFile, $fileBuffer, $blockSize ) {
    if ( !defined $fileLength ) {
 next
   if $! =~ /^Interrupted/;
 die "System read error: $!\n";
    }
    my $fileOffset = 0;
    while ( $fileLength ) {
 my $written = syswrite $outFile, $fileBuffer, $fileLength, $fileOffse
+t;
 die "System write error: $!\n"
   unless defined $written;
 $fileLength -= $written;
 $fileOffset += $written;
    };
}
die "Could not rewind $destFile: $!" unless defined sysseek $outFile, 
+0, 0;

$outMD5->addfile ( $outFile );
print $outMD5->b64digest , "\n";

$inFile->close;
$outFile->close;
[download]

Moreover, the documented behaviour of an automagical reset right after the digest, seems not to work, at least on b64digest and my system. Something to watch for.

Works like a charm! A few questions to help illuminate my mind, if I may.

I'm looking at the perlfunc entry for open, and I remain a little hazy regarding the subtle difference between ">" truncating and "+>" clobbering. I'm digging through the holy texts and references now, but if you care to pass more wisdom my way ...

The sysseek is a nice touch. I take it that since $outFile was not at the beginning of the file when I passed the filehandle to Digest::MD5, it somehow changed the hashing algorithm's output? Like only hashing from the current position to the end?

THX!
Dex

Little after-the-fact on my part, but that won't do for the NT/2K hosts, unless anyone knows of a port to Win32. Of course, one ~could~ do a pure-perl version and offer it in humble homage to the PPT project. Hmmmm.

HTH
--
idnopheq
Apply yourself to new problems without preparation, develop confidence in your ability to to meet situations as they arrise.

~ Shunryu Suzuki