Skip over device and character nodes, sockets

Verbose and all_files will now output the source of the dupe or clash.

"Clashes" happen when a symlink is added to the list of scanned files and a
similar file (or the file it points to) registers as a dupe. The symlink will be
removed from the list and the filesystem (assuming -d and -i aren't specified).

2 files changed, 66 insertions, 13 deletions

diff --git a/TODO b/TODO
index cbbeb7c..8b13789 100644
--- a/TODO
+++ b/TODO
@@ -1 +1 @@
-* Find a way to indicate which file a dupe is a copy of.
+
diff --git a/dupekill b/dupekill
index e8521ac..1d63b28 100755
--- a/dupekill
+++ b/dupekill
@@ -6,7 +6,7 @@ import stat
 from optparse import OptionParser
 
 # dupekill - deletes duplicates of existing data
-# Version 1.4 (2012-04-02)
+# Version 1.6 (2012-04-02)
 # Written by zlg <zlg@zlg.space>
 # Original idea and code by NF <radicalmori@gmail.com>
 # License: WTFPL <http://sam.zoy.org/wtfpl>
@@ -58,10 +58,25 @@ def dupekill(dry_run=False, recursive=False, verbose=False, all_files=False, ign
             # Set up a few variables we'll be needing.
             filepath = os.path.join(root, item)
 
+            # We need this to determine the file's type
+            filemode = os.lstat(filepath).st_mode
+
+            # Skip past file types we shouldn't deal with: sockets, device nodes, etc
+            if stat.S_ISSOCK(filemode) or stat.S_ISCHR(filemode) or stat.S_ISBLK(filemode) or stat.S_ISFIFO(filemode):
+                continue
+
+            # Make sure links are handled properly. If they're gonna be ignored,
+            # there's no reason to continue the loop.
+            if stat.S_ISLNK(filemode):
+                if not ignore_links:
+                    if all_files:
+                        print("LINK:", filepath)
+                continue
+
             # Check for information. If we can't fetch any, there's likely
             # a dead symlink or something else wrong.
             try:
-                filesize = os.lstat(filepath).st_size
+                filesize = os.path.getsize(filepath)
             # This occurs when the data points to something that can't be found or
             # resolved.
             except IOError:
@@ -75,11 +90,19 @@ def dupekill(dry_run=False, recursive=False, verbose=False, all_files=False, ign
                     print("DEAD LINK:", filepath)
                 continue
 
-            if ignore_links:
-                if os.lstat(filepath).st_mode == 41471:
-                    continue
+            # No sense in adding an empty file to the list
+            if filesize is 0:
+                continue
+
+            # We need this flag to determine state before adding to the list
+            deleted = False
 
-            deleted = False # We need this flag to determine state before adding to the list
+            # A clash happens when a symlink has been added to the list and is being
+            # checked against a file. In this case, the symlink should be deleted
+            # since the file contains the actual data.
+            clash = False
+
+            # Check to see if we can read or write to the file.
             if not os.access(filepath, os.R_OK):
                 if verbose or all_files:
                     print("NOTICE: Cannot read from", filepath)
@@ -116,17 +139,32 @@ def dupekill(dry_run=False, recursive=False, verbose=False, all_files=False, ign
                     # ---- CHECK FILE PREVIEW ----
                     if len(entry) == 3 and entry[2] == filepreview:
 
-                        # If they match and the files are less than 512 bytes... we don't need to hash!
-                        if entry[1] < 512 and filesize < 512:
+                        # If they match and the files are less than 513 bytes... we don't need to hash!
+                        if entry[1] < 513 and filesize < 513:
                             # We KNOW there's a match now.
                             deleted_files += 1
                             deleted = True
 
+                            # Identify the right file to get rid of
+                            if stat.S_ISLNK(os.lstat(entry[0]).st_mode):
+                                clash = True
+
+                            # It's important to make sure the proper file is being deleted
                             if not dry_run:
-                                os.remove(filepath)
+                                # If the hashlist has a symlink, get rid of it
+                                if not ignore_links and clash:
+                                    os.remove(entry[0])
+                                else:
+                                    os.remove(filepath)
 
                             if verbose or all_files:
-                                print("DUPE:", filepath)
+                                if not ignore_links and clash:
+                                    print("CLASH:", entry[0])
+                                    print("  with", filepath)
+                                    del entry
+                                else:
+                                    print("DUPE:", filepath)
+                                    print("   of", entry[0])
                                 break
                         else:
                             # Open the file in binary mode, to avoid UTF-8 errors.
@@ -145,11 +183,26 @@ def dupekill(dry_run=False, recursive=False, verbose=False, all_files=False, ign
                         deleted_files += 1
                         deleted = True
 
+                        # Identify the right file to get rid of
+                        if stat.S_ISLNK(os.lstat(entry[0]).st_mode):
+                            clash = True
+
+                        # It's important to make sure the proper file is being deleted
                         if not dry_run:
-                            os.remove(filepath)
+                            # If the hashlist has a symlink, get rid of it
+                            if not ignore_links and clash:
+                                os.remove(entry[0])
+                            else:
+                                os.remove(filepath)
 
                         if verbose or all_files:
-                            print("DUPE:", filepath)
+                            if not ignore_links and clash:
+                                print("CLASH:", entry[0])
+                                print("  with", filepath)
+                                del entry
+                            else:
+                                print("DUPE:", filepath)
+                                print("   of", entry[0])
                             break
 
             if need_to_add(filepath, hashList) and not deleted: