major update

- use strict;
- bunch more subroutines
- import mp3md5 stuff
- add help

list_same/list_same

@@ -6,89 +6,172 @@
 # Generates a list of files in cur. dir which are actually the same
 # Compares size and MD5 checksum
 # Handy for cleaning up pictures like this:
-# list_same|cut -f 1 -d " "|xargs rm
+# list_same -es | xargs rm
 
 # wishlist:
 #	- entering a list of file to check on the commandline
-#	- include escapes in output (spaces, brackets, etc) (optional)
 #	- option to ignore 0 length files
 
-
+use strict;
-# Changelog:
-# 2001-08-30:
-#	_much_ saner now. only calculate md5s if sizes of
-#	files are the same
-#	nearly complete rewrite
-# 2001-08-29:
-#	use MD5; instead of a shell call to md5
-#	some cleaning of code
-
 use Digest::MD5;
 use Getopt::Std;
 
-getopts('es');
+my ( $opt_e, $opt_h, $opt_s );
+my %opts = ();
+getopts('ehms', \%opts);
 
-opendir(DIR, ".") or die "can't open . $!";
+if ( $opts{h} ) { &help; }
-while (defined($file = readdir(DIR))) {
-	push @filelist, $file;
-}
-closedir(DIR);
 
-foreach (@filelist) {
+my @filelist = &getdir;
-	if ( -f $_ ) {
+if ( $opts{m} ) {
-		push @{$sizes{&get_size($_)}}, $_;
+	&mp3_comp(@filelist);
-	}
+} else {
+	my %sizes = &get_sizes(@filelist);
+	&quick_comp(%sizes);
 }
 
-$start = $opt_s ? 1 : 0;
-
-foreach $size (keys %sizes) {
-	if (@{$sizes{$size}} > 1) {
-		%md5s = ();
-		foreach (@{$sizes{$size}}) {
-			push @{$md5s{&calc_md5($_)}}, $_;
-		}
-		foreach $key (keys %md5s) {
-			if ( @{$md5s{$key}} > 1 ) {
-				@files = sort @{$md5s{$key}};
-				for $i ($start .. $#files) {
-					chomp $files[$i];
-					&output("$files[$i]\n");
-				}
-				unless($opt_s) { print "\n"; }
-			}
-		}
-	}
-}
 
 #################################
 # functions
 
-sub get_size {
+sub getdir {
-	my $file = shift;
+	my ( $file, @filelist );
-	my @stat;
+	opendir(DIR, ".") or die "can't open . $!";
-	@stat = stat $file;
+	while (defined($file = readdir(DIR))) {
-	return $stat[7];
+		if ( -f $file ) {
+			push @filelist, $file;
+		}
+	}
+	closedir(DIR);
+	return @filelist;
+}
+
+sub get_sizes(@) {
+	my @filelist = @_;
+	my ( @stat, %sizes );
+
+	foreach (@filelist) {
+		@stat = stat $_;
+		push @{$sizes{$stat[7]}}, $_;
+	}
+	return %sizes;
+}
+
+sub quick_comp(%) {
+	my %sizes = @_;
+	my ( $size, %md5s );
+
+	foreach $size (keys %sizes) {
+		if (@{$sizes{$size}} > 1) {
+			%md5s = ();
+			foreach (@{$sizes{$size}}) {
+				push @{$md5s{&calc_md5($_)}}, $_;
+			}
+			&output_doubles(%md5s);
+		}
+	}
 }
 
 # same md5 calculation i use in mv_wrap
 sub calc_md5($) {
-	my ($file, $digest);
+	my ($file, $digest, $md5, $FILE);
 	$file = shift;
 	$md5 = Digest::MD5->new;
-	open FILE, "<$file" or die "couldn't open file: $!\n";
+	open $FILE, "<$file" or die "couldn't open file: $!\n";
-	seek(FILE, 0, 0);
+	seek($FILE, 0, 0);
 	$md5->reset;
-	$md5->addfile(FILE);
+	$md5->addfile($FILE);
 	$digest = $md5->hexdigest;
-	close(FILE);
+	close($FILE);
 	return $digest;
 }
 
+sub mp3_comp(@) {
+	my @filelist = @_;
+	my %md5s = ();
+	foreach ( @filelist ) {
+		push @{$md5s{&calc_mp3md5($_)}}, $_;
+	}
+	&output_doubles(%md5s);
+}
+
+sub calc_mp3md5($) {
+	my $file = shift;
+	my ($fh, $off, $size);
+	my $buf = 4096*1024;
+
+	open($fh, "<$file") or die "Couldn't open file: $!\n";
+	binmode($fh);
+	seek $fh, 0, 2;		# go to end of file
+	my $eof = tell $fh;
+	if ($size = &has_v1_tag($fh)) {
+		$eof -= $size;
+	}
+
+	$off = 0;
+	if ($size = &has_v2_tag($fh)) {
+		$off = $size;
+	}
+	my  $md5 = Digest::MD5->new;
+	while ($off < $eof) {
+		seek $fh, $off, 0;
+		if ($buf > ($eof - $off)) { $buf = $eof - $off; }
+		read $fh, my($bytes), $buf;
+		$md5->add($bytes);
+		$off += $buf;
+	}
+	close $fh;
+	return $md5->hexdigest;
+}
+
+sub has_v1_tag {
+	my $fh = shift;
+	seek $fh, -128, 2;
+	if (<$fh> =~ /^TAG/) {
+		return 128;
+	}
+	return 0;
+}
+
+sub has_v2_tag {
+	my $fh = shift;
+	my ($head, @bytes);
+	my $tagsize = 10;
+	seek $fh, 0, 0;
+	read $fh, $head, 3;
+	if ($head =~ /^ID3/) {
+		read $fh, $head, 3; # skip 3 bytes
+		read $fh, $head, 4;
+		@bytes = reverse unpack 'C4', $head;
+		foreach (0 .. 3) {
+			$tagsize += $bytes[$_] * 128 ** $_;
+		}
+		return $tagsize;
+	}
+	return 0;
+}
+
+sub output_doubles(@) {
+	my %md5s = @_;
+	my ( $key, @files, $i );
+	my $start = $opts{s} ? 1 : 0;
+
+	foreach $key (keys %md5s) {
+		if ( @{$md5s{$key}} > 1 ) {
+			@files = sort @{$md5s{$key}};
+			for $i ($start .. $#files) {
+				chomp $files[$i];
+				&output("$files[$i]\n");
+			}
+			unless($opts{s}) { print "\n"; }
+		}
+	}
+}
+
 # escape output if necessary
 sub output($) {
 	my $string = shift;
-	if ($opt_e) {
+	if ($opts{e}) {
 		$string =~ s/\\/\\\\/g;
 		$string =~ s/ /\\ /g;
 		$string =~ s/'/\\'/g;
@@ -101,5 +184,22 @@ sub output($) {
 	} else {
 		print "$string";
 	}
-	$opt_e = $opt_e;
 }
+
+sub help {
+	my $name = $0;
+	$name =~ s/.*\///;
+	print <<EOF;
+
+Usage: $name [OPTION] ...
+
+	-e		escape output filenames with backslashes
+	-h		display this help message
+	-m		mp3 compare, ignores ID3 tags (slow)
+	-s		skip the first entry for doubles
+
+EOF
+	exit;
+}
+
+