From dddfbb07247801b4bbcead7446362ef37131cf12 Mon Sep 17 00:00:00 2001
From: Tanner Collin <git@tannercollin.com>
Date: Mon, 3 Apr 2023 22:11:34 -0600
Subject: [PATCH] Compare bits on audio hashes

---
 main.py | 23 +++++++++++++++++++----
 1 file changed, 19 insertions(+), 4 deletions(-)

diff --git a/main.py b/main.py
index 6055557..0ec2daa 100644
--- a/main.py
+++ b/main.py
@@ -4,11 +4,13 @@ import hashlib
 from PIL import Image, UnidentifiedImageError
 
 import acoustid
+import chromaprint
 from imagehash import average_hash
 from videohash import VideoHash
 from videohash.exceptions import FFmpegFailedToExtractFrames
 
 hashes = set()
+audio_hashes = []
 delete = set()
 
 filenames = [x for x in glob.glob('**', recursive=True) if os.path.isfile(x)]
@@ -52,7 +54,9 @@ def get_image_hash(filename):
 
 def get_audio_hash(filename):
     try:
-        return str(acoustid.fingerprint_file(filename)[1].decode()) + '_audio'
+        fp = acoustid.fingerprint_file(filename)[1]
+        values, _ = chromaprint.decode_fingerprint(fp)
+        return chromaprint.hash_fingerprint(values)
     except acoustid.FingerprintGenerationError:
         return None
 
@@ -84,13 +88,24 @@ for filename in filenames:
 
     if not digest: continue
 
-    if digest in hashes:
-        delete.add(filename)
+    if type(digest) == int:
+        for h in audio_hashes:
+            if bin(digest ^ h).count('1') <= 5:   # TODO adjust?
+                delete.add(filename)
+                break
+        else:  # for
+            audio_hashes.append(digest)
     else:
-        hashes.add(digest)
+        if digest in hashes:
+            delete.add(filename)
+        else:
+            hashes.add(digest)
 
+print()
+print()
 for dupe in delete:
     print(dupe)
+
 print()
 print('Found', len(delete), 'total duplicate files. Delete them?')
 print('ENTER to continue, ctrl+c to cancel.')