From f6ae3a6bb2cad24d00acb5dfb789a28ad693becf Mon Sep 17 00:00:00 2001
From: cvs <cvs@a5e33b96-951a-0410-ae88-c0fe16d076bb>
Date: Sat, 7 Jan 2006 08:44:02 +0000
Subject: [PATCH] Implement --dry-run and --humane options

git-svn-id: svn+ssh://svn.furryterror.org/r/trunk/mokona/zblaxell@4518 a5e33b96-951a-0410-ae88-c0fe16d076bb
---
 faster-dupemerge | 56 ++++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 49 insertions(+), 7 deletions(-)

diff --git a/faster-dupemerge b/faster-dupemerge
index b7554de..18d88cd 100755
--- a/faster-dupemerge
+++ b/faster-dupemerge
@@ -79,6 +79,8 @@ my $skip_compares = 0;
 my $skip_hashes = 0;
 my $verbose = 0;
 my $debug = 0;
+my $dry_run = 0;
+my $humane = 0;
 my @extra_find_opts = ();
 my @extra_sort_opts = ();
 my $lock_file;
@@ -120,8 +122,12 @@ hard links).
         --debug         show all steps in duplication discovery process
                         (implies --verbose)
 
+	--dry-run	do not lock files or make changes to filesystem
+
         --find          pass next options (up to --) to find command
 
+	--humane	human-readable statistics (e.g. 1 048 576)
+
         --lock FILE     exit immediately (status 10) if unable to obtain a 
                         flock(LOCK_EX|LOCK_NB) on FILE
 
@@ -168,6 +174,10 @@ while ($#ARGV >= 0) {
 		}
 	} elsif ($arg eq '--debug') {
 		$debug = $verbose = 1;
+	} elsif ($arg eq '--dry-run') {
+		$dry_run = 1;
+	} elsif ($arg eq '--humane') {
+		$humane = 1;
 	} elsif ($arg eq '--find') {
 		while ($#ARGV >= 0) {
 			my $extra_arg = shift(@ARGV);
@@ -192,7 +202,9 @@ if ($skip_hashes && $skip_compares) {
 	die "Cannot skip both hashes and compares.\n";
 }
 
-if (defined($lock_file)) {
+@directories or usage;
+
+if (defined($lock_file) && !$dry_run) {
 	sysopen(LOCK_FILE, $lock_file, O_CREAT|O_RDONLY, 0666) or die "open: $lock_file: $!";
 	flock(LOCK_FILE, LOCK_EX|LOCK_NB) or die "flock: $lock_file: LOCK_EX|LOCK_NB: $!";
 	print STDERR "Locked '$lock_file' in LOCK_EX mode.\n" if $verbose;
@@ -200,7 +212,7 @@ if (defined($lock_file)) {
 }
 
 END {
-	if ($lock_obtained) {
+	if ($lock_obtained && !$dry_run) {
 		print STDERR "Removing '$lock_file'.\n" if $verbose;
 		unlink($lock_file) or warn "unlink: $lock_file: $!";
 	}
@@ -244,14 +256,18 @@ my %inode_to_file_name = ();
 # Link files
 sub link_files {
 	my ($from, $to) = (@_);
+
+	my $quoted_from = tick_quote($from);
+	my $quoted_to = tick_quote($to);
+	print STDERR "ln -f $quoted_from $quoted_to\n";
+
+	return if $dry_run;
+
 	my $inode_dir = $to;
 	my $inode_base = $to;
 	$inode_dir =~ s:[^/]*$::o;
 	$inode_base =~ s:^.*/::os;
 	my $tmp_to = File::Temp::tempnam($inode_dir, ".$inode_base.");
-	my $quoted_from = tick_quote($from);
-	my $quoted_to = tick_quote($to);
-	print STDERR "ln -f $quoted_from $quoted_to\n";
 	print STDERR "\tlink: $from -> $tmp_to\n" if $debug;
 	link($from, $tmp_to) or die "link: $from -> $tmp_to: $!";
 	print STDERR "\trename: $tmp_to -> $to\n" if $debug;
@@ -461,9 +477,13 @@ candidate_file:
 									# My random number generator chooses the incumbent's size.
 
 									if ($link_done) {
+										# Since we're in a dry run, the filesystem doesn't change.
+										# Our notion of what the filesystem should look like should not change either.
 										delete $inode_to_file_name{$to_inode}->{$to_file};
-										$inode_to_file_name{$from_inode}->{$to_file} = undef;
-										$hash_to_inode{$digest} = $from_inode;
+										unless ($dry_run) {
+											$inode_to_file_name{$from_inode}->{$to_file} = undef;
+											$hash_to_inode{$digest} = $from_inode;
+										}
 
 										$hard_links++;
 										if ($to_nlink == 1) {
@@ -553,6 +573,28 @@ surprises               $surprises
 trivially_unique        $trivially_unique
 STATS
 
+if ($humane) {
+	my $max_num_len = 0;
+
+	sub measure_numbers {
+		my ($num) = @_;
+		my $len = length($num);
+		$len += int( (length($num) - 1) / 3);
+		$max_num_len = $len if $len > $max_num_len;
+	}
+
+	(my $dummy = $stats_blob) =~ s/\d+/measure_numbers($&)/geos;
+
+	sub space_numbers {
+		my ($num) = @_;
+		1 while $num =~ s/(\d)(\d\d\d)( \d\d\d)*$/$1 $2$3/os;
+		$num = ' ' x ($max_num_len - length($num)) . $num;
+		return $num;
+	}
+
+	$stats_blob =~ s/\d+/space_numbers($&)/geos;
+}
+
 $stats_blob =~ s/([^\n]*\n[^\n]*? )(\s+)( [^\n]*\n)/$1 . ('.' x length($2)) . $3/oemg;
 
 print STDERR $stats_blob;
-- 
2.30.2