From: Zygo Blaxell <zblaxell@waya.furryterror.org>
Date: Fri, 8 Jan 2010 14:16:02 +0000 (-0500)
Subject: dupemerge: don't stat during the file collection loop
X-Git-Tag: dm6-0.20100514~21
X-Git-Url: http://git.hungrycats.org/cgi-bin/gitweb.cgi?p=dupemerge;a=commitdiff_plain;h=55ebeb05eb7883f9eb09976ebfe0f8ea7718e933;hp=07bccf09a4f852b9aa5e5665f6a5e1932bc755ce

dupemerge: don't stat during the file collection loop

Remove the lstat from the find output reading loop.  It's a redundant
copy of the same code in merge_files.

Adjust merge_files to filter out possible non-files that will now leak
through from the find output.
---

diff --git a/faster-dupemerge b/faster-dupemerge
index 91138c8..aa8b142 100755
--- a/faster-dupemerge
+++ b/faster-dupemerge
@@ -340,10 +340,11 @@ hash_file:
 			print STDERR "\t\tDigesting file $filename\n" if $debug;
 			if ((-l $filename) || ! -f _) {
 				warn "Bogon file " . tick_quote($filename);
-				$surprises++;
+				$input_bogons++;
+				delete $inode_to_file_name{$candidate}->{$filename};
 				next;
 			}
-			eval { 
+			eval {
 				$digest = digest($filename); 
 			};
 			if ($@) {
@@ -578,12 +579,6 @@ while (<FIND>) {
 
 	print STDERR "weak_key=$weak_key inode=$inode name=$name\n" if $debug;
 
-	unless (! (-l $name) && (-f _)) {
-		warn "Bogon file " . tick_quote($name);
-		$input_bogons++;
-		next;
-	}
-
 	$input_links++;
 	merge_files if $weak_key ne $current_key;
 	$current_key = $weak_key;