Mercurial > hg-public > dreambox_tools / changeset

--- a/dupecheck.py	Sun Nov 22 23:08:17 2020 +0100
+++ b/dupecheck.py	Tue May 06 19:45:12 2025 +0200
@@ -262,6 +262,9 @@
                         del self.filelist[idx]
                         break
             listlen = len(self.filelist)
+            print("%i files left to analyze after revoving duplicates" % (
+                listlen))
+
 
         for idx in range(listlen):
             if not self.filelist[idx]:
@@ -271,8 +274,9 @@
             sys.stdout.flush()
             for idx2 in range(idx + 1, listlen):
                 if self.filelist[idx2]:
+                    if not self.filelist[idx2]:
+                        continue
                     if similarity(self.filelist[idx][0], self.filelist[idx2][0]) > self.ratio:
-                        #print "possible duplicate %d %s" % (idx2, item2[0])
                         key = os.path.join(self.filelist[idx][2], self.filelist[idx][1])
                         if not key in self.duplicates:
                             self.duplicates[key] = []
@@ -336,7 +340,8 @@
     if os.path.isfile("dupecheck-ignore.txt"):
         # read the entire file line by line into buffer
         print("Loading ignore filename prefixes file for dupe checking...")
-        dupe.ignore_fileprefix = [line.rstrip('\n').rstrip('\r') for line in open("dupecheck-ignore.txt", "rb")]
+        dupe.ignore_fileprefix = [line.lower().rstrip('\n').rstrip('\r') for line in open("dupecheck-ignore.txt", "rb")]
+        #print(dupe.ignore_fileprefix)
 
     if args.fixnames:
         for srcstr in args.basedir: