main.py

@@ -2,9 +2,6 @@ import glob
 import os
 import hashlib
 import time
-import traceback
-from multiprocessing import Pool
-import queue
 from PIL import Image, UnidentifiedImageError
 
 from imagehash import average_hash
@@ -21,7 +18,7 @@ print('ENTER to continue, ctrl+c to cancel.')
 try:
     input()
 except KeyboardInterrupt:
-    print('Cancelled.')
+    print('\nCancelled.')
     os._exit(0)
 
 print('Sorting file list by smallest size...')
@@ -39,75 +36,51 @@ for filename in filenames:
         hasher.update(buf)
         digest = hasher.hexdigest()
         if digest in hashes:
-            print('Found digest', digest, 'collision for', filename)
             delete.add(filename)
         else:
             hashes.add(digest)
 
 print('Found', len(delete), 'duplicates by md5 hash.')
-
-time.sleep(4)
 print('Deduplicating by media fingerprint...')
 
 def get_image_hash(filename):
-    basename = os.path.basename(os.path.dirname(filename))
     try:
         image = Image.open(filename)
-        return basename + str(average_hash(image)) + '_image'
+        return str(average_hash(image)) + '_image'
     except UnidentifiedImageError:
         return None
 
 def get_video_hash(filename):
-    basename = os.path.basename(os.path.dirname(filename))
     try:
-        v = VideoHash(path=filename)
-        digest = str(v.hash_hex)
-        v.delete_storage_path()
-
-        return basename + digest + '_video'
+        return str(VideoHash(path=filename).hash_hex) + '_video'
     except FFmpegFailedToExtractFrames:
         return None
 
 count = 0
 total = len(filenames)
 
-def hasher(filename):
-    if filename in delete: return None
+for filename in filenames:
+    count += 1
+    print('Hashing file', count, '/', total, end='\r')
 
-    print('Hashing file:', filename)
+    if filename in delete: continue
 
     try:
-        digest = get_image_hash(filename)#  or get_video_hash(filename)
+        digest = get_image_hash(filename) or get_video_hash(filename)
     except KeyboardInterrupt:
-        print('Cancelled.')
+        print('\nCancelled.')
         os._exit(0)
     except BaseException as e:
         print()
         print('Exception', e.__class__.__name__, str(e), 'while hashing:')
         print(filename)
-        print(traceback.format_exc())
-        return None
+        continue
 
-    if not digest: return None
+    time.sleep(5)
 
-    return (filename, digest)
-
-with Pool() as pool:
-    results = pool.map(hasher, filenames)
-
-print('Finished hashing.')
-print()
-print('Checking digests:')
-print()
-
-for result in results:
-    if not result: continue
-    print(result)
-
-    filename, digest = result
+    if not digest: continue
 
     if digest in hashes:
-        print('Found digest', digest, 'collision for', filename)
         delete.add(filename)
     else:
         hashes.add(digest)
@@ -115,6 +88,6 @@ for result in results:
 print()
 print('Found', len(delete), 'total duplicate files.')
 
-for dupe in sorted(list(delete)):
-    print('Deleting:', dupe)
+print('Deleting...')
+for dupe in delete:
     os.remove(dupe)

requirements.txt

@@ -1,13 +0,0 @@
-Brotli==1.1.0
-certifi==2023.7.22
-imagedominantcolor==1.0.1
-ImageHash==4.3.1
-mutagen==1.47.0
-numpy==1.26.1
-Pillow==10.1.0
-pycryptodomex==3.19.0
-PyWavelets==1.4.1
-scipy==1.11.3
-videohash==3.0.1
-websockets==11.0.3
-yt-dlp==2023.10.13