#!/usr/local/bin/perl5
#
# usage: make_md5 directory >listing.md5
#
# finds all the files under the given directory (avoiding any obvious
# thumbnail directories), and gives their MD5 signatures to stdout.
# progress is shown to stderr.
#
# suggestion: when making a listing, start from a root directory
# that has a unique name, to make later grepping easier. i.e:
#    mv ~/pr0n pr0n2002may
#    make_md5 pr0n2002may >~/Misc/Disc/pr0n2002may.md5
#    (burn CDR from pr0n2002may, remove, label the CDR accordingly)
# then later..
#    grep cat ~/Misc/Disc/everything.md5|grep maid
# might print:
#    "245t98uerguih245mumble   pr0n2002may/Anime/killer_catgirl_maid09.jpg"
#
# --
# Copyright 1995-2002 Chris Baird <cjb@brushtail.apana.org.au>.
# GNU Copyleft V2 copying conditions if you're incapable of writing your
# own from scratch, blah, blah..

$cksum="/usr/bin/md5";  # BSD MD5, maybe?
$N=10;
$|=1;

sub fixname
{
    local($f);
    s/\n//g;
    s/\ /\\ /g;
    s/\'/\\'/g;
    s/\"/\\"/g;
    s/\&/\\&/g;
    s/\`/\\`/g;
    s/\!/\\!/g;
    s/\(/\\(/g;
    s/\)/\\)/g;
    $f = $_;
}

$rootsearch = shift @ARGV;

open(FIND, "find $rootsearch -type f -print|");
while (<FIND>)
{
    $filename = &fixname($_);
    next if /xvpics/;
    next if /\.db\//;
    print STDERR ++$count, "\t $filename  \r";
    push(@allfiles, $filename);   # XXX scary for large filesystems
}
print STDERR "\n\n";

# splice the first N (or whatever's left) filenames
# and feed to $cksum

while ($#allfiles >= 0)
{
    if ($#allfiles < $N) { $N = $#allfiles };
    $a = join(" ", splice (@allfiles, 0, $N +1));
    print STDERR $a, "\n\n";

    open(CKSUM, "$cksum $a |");
    while (<CKSUM>)
    {
        chop;
        # NB: dependent on output format of BSD MD5; GNU might be different
        ($b, $fcksum) = ($_ =~ /^MD5 \((\S+)\) = (\S+)/);
	$fname = $b;
        print "$fcksum\t$fname\n";
    }
}