From bd7393310542e1de4b59f21259b4bd8d68424058 Mon Sep 17 00:00:00 2001
From: "Peter J. Holzer" <hjp@hjp.at>
Date: Fri, 16 Jan 2015 11:31:56 +0100
Subject: [PATCH] Rename finddup to finddup2 and add new finddup for searching
 one directory

---
 finddup  | 79 ++++++++++++++++++++++++--------------------------------
 finddup2 | 78 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 112 insertions(+), 45 deletions(-)
 mode change 100755 => 100644 finddup
 create mode 100755 finddup2

diff --git a/finddup b/finddup
old mode 100755
new mode 100644
index 73a2b7a..5ed6a26
--- a/finddup
+++ b/finddup
@@ -13,66 +13,55 @@ use File::stat;
 use File::Slurp;
 use Digest::SHA qw(sha256_hex);
 
-my ($needledir, $haystackdir) = @ARGV;
+my ($dir) = @ARGV;
 
-my $haystack;
+my $files;
 
 find(sub {
         if (-f $_) {
             my $st = lstat($_);
             my $size = $st->size;
             my $hash = "";
-            unless ($haystack->{$size}{$hash}) {
-                $haystack->{$size}{$hash} = [];
+            unless ($files->{$size}{$hash}) {
+                $files->{$size}{$hash} = [];
             }
-            push $haystack->{$size}{$hash}, { path => $File::Find::name, 
+            push $files->{$size}{$hash}, { path => $File::Find::name, 
                                               size => $size,
                                               dev  => $st->dev,
                                               ino  => $st->ino,
                                             };
         }
     }, 
-    $haystackdir
+    $filesdir
 );
 
-find(
-    {
-        wanted => sub {
-                    if (-f $_) {
-                        my $st = lstat($_);
-                        my $size = $st->size;
-                        my $found;
-                        if ($haystack->{$size}) {
-                            if ($haystack->{$size}{""}) {
-                                for my $file (@{ $haystack->{$size}{""} }) {
-                                    my $content = read_file($file->{path});
-                                    my $hash = sha256_hex($content);
-                                    unless ($haystack->{$size}{$hash}) {
-                                        $haystack->{$size}{$hash} = [];
-                                    }
-                                    push $haystack->{$size}{$hash}, $file;
-                                }
-                                delete $haystack->{$size}{""};
-                            }
-                            my $content = read_file($_);
-                            my $hash = sha256_hex($content);
-                            if ($haystack->{$size}{$hash}) {
-                                for my $file (@{ $haystack->{$size}{$hash} }) {
-                                    unless ($st->dev == $file->{dev} && $st->ino == $file->{ino}) {
-                                        say "$_ -> $file->{path}";
-                                        $found = 1;
-                                    }
-                                }
-                            }
-                        }
-                        unless ($found) {
-                            say "$_ (NONE)";
-                        }
-                    }
-                },
-        no_chdir => 1,
-    },
-    $needledir
-);
+for my $size (keys $files) {
+    if (@{ $files->{$size}{""} } == 1) {
+        next;
+    }
+    if ($files->{$size}{""}) {
+        for my $file (@{ $files->{$size}{""} }) {
+            my $content = read_file($file->{path});
+            my $hash = sha256_hex($content);
+            unless ($files->{$size}{$hash}) {
+                $files->{$size}{$hash} = [];
+            }
+            push $files->{$size}{$hash}, $file;
+        }
+        delete $files->{$size}{""};
+    }
+    for my $hash (keys $files->{$size}) {
+        if (@{ $files->{$size}{$hash} } > 1) {
+            for my $file (@{ $files->{$size}{$hash} }) {
+                unless ($st->dev == $file->{dev} && $st->ino == $file->{ino}) {
+                    say "$file->{path}";
+                }
+            }
+            say "";
+        }
+    }
+}
+
 
 # vim: tw=132 sw=4 expandtab
+
diff --git a/finddup2 b/finddup2
new file mode 100755
index 0000000..73a2b7a
--- /dev/null
+++ b/finddup2
@@ -0,0 +1,78 @@
+#!/usr/bin/perl
+
+=head1 NAME
+
+finddup - yet another tool for finding duplicates
+
+=cut
+
+use v5.10;
+no autovivification 'fetch';
+use File::Find;
+use File::stat;
+use File::Slurp;
+use Digest::SHA qw(sha256_hex);
+
+my ($needledir, $haystackdir) = @ARGV;
+
+my $haystack;
+
+find(sub {
+        if (-f $_) {
+            my $st = lstat($_);
+            my $size = $st->size;
+            my $hash = "";
+            unless ($haystack->{$size}{$hash}) {
+                $haystack->{$size}{$hash} = [];
+            }
+            push $haystack->{$size}{$hash}, { path => $File::Find::name, 
+                                              size => $size,
+                                              dev  => $st->dev,
+                                              ino  => $st->ino,
+                                            };
+        }
+    }, 
+    $haystackdir
+);
+
+find(
+    {
+        wanted => sub {
+                    if (-f $_) {
+                        my $st = lstat($_);
+                        my $size = $st->size;
+                        my $found;
+                        if ($haystack->{$size}) {
+                            if ($haystack->{$size}{""}) {
+                                for my $file (@{ $haystack->{$size}{""} }) {
+                                    my $content = read_file($file->{path});
+                                    my $hash = sha256_hex($content);
+                                    unless ($haystack->{$size}{$hash}) {
+                                        $haystack->{$size}{$hash} = [];
+                                    }
+                                    push $haystack->{$size}{$hash}, $file;
+                                }
+                                delete $haystack->{$size}{""};
+                            }
+                            my $content = read_file($_);
+                            my $hash = sha256_hex($content);
+                            if ($haystack->{$size}{$hash}) {
+                                for my $file (@{ $haystack->{$size}{$hash} }) {
+                                    unless ($st->dev == $file->{dev} && $st->ino == $file->{ino}) {
+                                        say "$_ -> $file->{path}";
+                                        $found = 1;
+                                    }
+                                }
+                            }
+                        }
+                        unless ($found) {
+                            say "$_ (NONE)";
+                        }
+                    }
+                },
+        no_chdir => 1,
+    },
+    $needledir
+);
+
+# vim: tw=132 sw=4 expandtab