X-Git-Url: http://git.hungrycats.org/cgi-bin/gitweb.cgi?p=dupemerge;a=blobdiff_plain;f=faster-dupemerge;h=44d0e91a929c361059b4e5a7df6f0ed6252d4699;hp=7e899ddc8eb8ae4f87770d42a33efe4acca43b95;hb=85b26d64c5fe49e33e7946498d5dd6c756c415ce;hpb=6910de85c5daa8b99a1ea7d6e4614584fbfdfec5

diff --git a/faster-dupemerge b/faster-dupemerge
index 7e899dd..44d0e91 100755
--- a/faster-dupemerge
+++ b/faster-dupemerge
@@ -86,7 +86,9 @@ my $collapse_access = 0;
 my $collapse_timestamp = 0;
 my $collapse_zero = 0;
 my $skip_compares = 0;
+my $skip_compare_preference = 0;
 my $skip_hashes = 0;
+my $skip_hashes_threshold = 0;
 my $progress = 0;
 my $verbose = 0;
 my $debug = 0;
@@ -153,6 +155,7 @@ hard links).
                         S, s - lstat(2) (see source for details)
                         U - unlink(2)
                         . - all inodes with similar attributes done
+                        (123456) - current file size in bytes
 
         --sort          pass next options (up to --) to sort command
 
@@ -160,7 +163,10 @@ hard links).
 
         --skip-compare  skip byte-by-byte file comparisons
 
-        --skip-hash     skip calculation of hash function on files
+        --skip-hash[=N] skip calculation of hash function on files
+                        larger than N bytes (default 1M).
+                        Scalars KMGT specify KiB, MiB, GiB, and TiB.
+                        Scalars kmgt specify KB, MB, GB, and TB.
 
         --trust         old name for --skip-compare
                         (trust the hash function)
@@ -180,9 +186,24 @@ while ($#ARGV >= 0) {
 	} elsif ($arg eq '--zeros') {
 		$collapse_zero = 1;
 	} elsif ($arg eq '--trust' || $arg eq '--skip-compare') {
-		$skip_compares = 1;
-	} elsif ($arg eq '--skip-hash') {
-		$skip_hashes = 1;
+		$skip_compares = $skip_compare_preference = 1;
+	} elsif ($arg =~ /^--skip-hash(?:=(\d+)([KkMmGgTt]?))?$/os) {
+		my ($quantity, $unit) = ($1, $2);
+		$unit ||= '_';
+		$quantity ||= 1048576;
+		my %scale = (
+			_ => 1,
+			k => 1000,
+			K => 1024,
+			m => 1000*1000,
+			M => 1024*1024,
+			g => 1000*1000*1000,
+			G => 1024*1024*1024,
+			t => 1000*1000*1000*1000,
+			T => 1024*1024*1024*1024,
+		);
+		$skip_hashes = 0;
+		$skip_hashes_threshold = $quantity * $scale{$unit};
 	} elsif ($arg eq '--progress') {
 		$progress = 1;
 	} elsif ($arg eq '--verbose') {
@@ -221,10 +242,6 @@ while ($#ARGV >= 0) {
 	}
 }
 
-if ($skip_hashes && $skip_compares) {
-	die "Cannot skip both hashes and compares.\n";
-}
-
 @directories or usage;
 
 if (defined($lock_file) && !$dry_run) {
@@ -339,6 +356,7 @@ hash_file:
 		foreach my $filename (sort keys(%{$inode_to_file_name{$candidate}})) {
 			print STDERR "\t\tDigesting file $filename\n" if $debug;
 			if ((-l $filename) || ! -f _) {
+				print STDERR "\n" if $progress;
 				warn "Bogon file " . tick_quote($filename);
 				$input_bogons++;
 				delete $inode_to_file_name{$candidate}->{$filename};
@@ -524,7 +542,7 @@ candidate_file:
 
 									if ($link_done) {
 										delete $inode_to_file_name{$to_inode}->{$to_file};
-										$inode_to_file_name{$from_inode}->{$to_file} = undef;
+										$inode_to_file_name{$from_inode}->{$to_file} = undef unless ($dry_run);
 										$hash_to_inode{$digest} = [ $from_inode ];
 
 										$hard_links++;
@@ -571,14 +589,34 @@ end_merge:
 	undef %inode_to_file_name;
 }
 
+my $last_time = 0;
+my $last_size = 0;
+
 while (<FIND>) {
-	my ($weak_key, $dev, $ino, $name) = m/^(\d+ \d+ \d+ \d+ -?[\d.]+) (\d+) (\d+) (.+)\0$/so;
+	my ($weak_key, $size, $dev, $ino, $name) = m/^((\d+) \d+ \d+ \d+ -?[\d.]+) (\d+) (\d+) (.+)\0$/so;
 	die "read error: $!\nLast input line was '$_'" unless defined($name);
 
 	my $inode = format_inode($dev, $ino);
 
 	print STDERR "weak_key=$weak_key inode=$inode name=$name\n" if $debug;
 
+	if ($skip_hashes_threshold && $size >= $skip_hashes_threshold) {
+		$skip_hashes = 1;
+		$skip_compares = 0;
+	} else {
+		$skip_hashes = 0;
+		$skip_compares = $skip_compare_preference;
+	}
+
+	if ($progress) {
+		my $this_time = time();
+		if ($this_time != $last_time && $size != $last_size) {
+			$last_time = $this_time;
+			$last_size = $size;
+			print STDERR "($size)";
+		}
+	}
+
 	$input_links++;
 	merge_files if $weak_key ne $current_key;
 	$current_key = $weak_key;