From b9bfa6e1ddc1f2ceaed14ac4c4c7782713b40fda Mon Sep 17 00:00:00 2001
From: Ward Wouts <ward@wouts.nl>
Date: Thu, 13 Aug 2015 13:24:44 +0000
Subject: [PATCH] nieuwe vlaggetjes

---
 list_same/list_same | 87 ++++++++++++++++++++++++++++++++++-----------
 1 file changed, 67 insertions(+), 20 deletions(-)

diff --git a/list_same/list_same b/list_same/list_same
index d7e467d..78a2774 100755
--- a/list_same/list_same
+++ b/list_same/list_same
@@ -1,8 +1,5 @@
 #!/usr/bin/perl -w
 
-# $Id$
-# $Source$
-
 # 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:
@@ -20,26 +17,42 @@ use Getopt::Std;
 
 my %opts = ();
 my @dirs;
+my @files;
 
 &cmdline;
 
 my @filelist = &getdir;
+push @filelist, @files;
+
 if ( $opts{m} ) {
 	&mp3_comp(@filelist);
 } else {
-	my %sizes = &get_sizes(@filelist);
-	&quick_comp(%sizes);
+	if ( $opts{f} ) {
+		&get_md5s(@filelist);
+	} else {
+		my %sizes = &get_sizes(@filelist);
+		&quick_comp(%sizes);
+	}
 }
 
 #################################
 # functions
 
 sub cmdline {
-	getopts('dehlms', \%opts);
+	getopts('adefhlms', \%opts);
 	if ( $opts{h} ) { &help; }
 
-	map { if ( -d $_ ) { s/\/$//; push @dirs, $_; } else { die "$_ not a directory" } } @ARGV;
-	if ( scalar @dirs == 0 ) {
+	map { if ( -d $_ ) {
+		s/\/$//;
+		push @dirs, $_;
+	} else {
+		if ( -f $_ ) {
+			push @files, $_;
+		} else {
+			die "$_ not a directory or file"
+		}
+	} } @ARGV;
+	if ( scalar @dirs == 0 && scalar @files == 0 ) {
 		push @dirs, ".";
 	}
 
@@ -47,7 +60,6 @@ sub cmdline {
 		print "-l doens't combine with -d, -e or -s\n\n";
 		&help;
 	}
-#	map { print "$_\n"; } @dirs;
 }
 
 sub getdir {
@@ -75,6 +87,14 @@ sub get_sizes(@) {
 	return %sizes;
 }
 
+sub get_md5s(@) {
+	my %md5s;
+	foreach (@_) {
+		push @{$md5s{&calc_md5($_)}}, $_;
+	}
+	&output_doubles(%md5s);
+}
+
 sub quick_comp(%) {
 	my %sizes = @_;
 	my ( $size, %md5s );
@@ -93,15 +113,34 @@ sub quick_comp(%) {
 # same md5 calculation i use in mv_wrap
 sub calc_md5($) {
 	my $file = shift;
-	my ( $digest, $md5, $FILE );
+	my ( $digest, $fh, $off, $startoff );
+	my $buf = 4096*1024;
+	my  $md5 = Digest::MD5->new;
 	$md5 = Digest::MD5->new;
-	open $FILE, "<$file" or die "couldn't open file: $!\n";
-	seek($FILE, 0, 0);
-	$md5->reset;
-	$md5->addfile($FILE);
-	$digest = $md5->hexdigest;
-	close($FILE);
-	return $digest;
+	open($fh, "<$file") or die "Couldn't open file: $!\n";
+	binmode($fh);
+	seek $fh, 0, 2;		# go to end of file
+	my $eof = tell $fh;
+
+	seek($fh, 0, 0);
+	if ( $opts{f} ) {
+		readline($fh);
+		$off = tell $fh;
+	} else {
+		$off = 0;
+	}
+
+	while ($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 mp3_comp(@) {
@@ -171,12 +210,18 @@ sub has_v2_tag {
 
 sub output_doubles(@) {
 	my %md5s = @_;
-	my ( $key, @files, $i );
+	my ( $key, @files, $i, $numkeys );
 	my $start = $opts{s} ? 1 : 0;
 
+	if ( $opts{a} ) { $numkeys = 0; } else { $numkeys = 1; }
+
 	foreach $key (keys %md5s) {
-		if ( @{$md5s{$key}} > 1 ) {
-			@files = sort dirsort @{$md5s{$key}};
+		if ( @{$md5s{$key}} > $numkeys ) {
+			if ( scalar @dirs > 0 ) {
+				@files = sort dirsort @{$md5s{$key}};
+			} else {
+				@files = @{$md5s{$key}};
+			}
 			for $i ($start .. $#files) {
 				chomp $files[$i];
 				if ($opts{d}) {
@@ -253,8 +298,10 @@ sub help {
 
 Usage: $name [OPTION] ...
 
+	-a		output all files
 	-d		delete resulting files
 	-e		escape output filenames with backslashes
+	-f		skip first line
 	-h		display this help message
 	-l		hardlink resulting files (no change if on
 			different filesystems)