From f379ffcc7e85dae8b8c6875fc54dc6d0c4707a9e Mon Sep 17 00:00:00 2001
From: GRayHook <s@marinkevich.ru>
Date: Sat, 9 Jul 2022 18:54:34 +0700
Subject: [PATCH] init

---
 ddp.py | 78 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 78 insertions(+)
 create mode 100755 ddp.py

diff --git a/ddp.py b/ddp.py
new file mode 100755
index 0000000..d1b6b7d
--- /dev/null
+++ b/ddp.py
@@ -0,0 +1,78 @@
+#!/usr/bin/python3
+
+import subprocess
+import itertools
+import os
+import sys
+
+
+GET_ALL_PICS_LIST_STRING="find . -iname '*.jpg' -o -iname '*.png' -o -iname '*.jpeg'"
+COMPARE_PICS_CMD="compare -metric AE -fuzz 0.1% \"{}\" \"{}\" /dev/null"
+
+def main():
+    out = subprocess.Popen(['sh', '-c', GET_ALL_PICS_LIST_STRING],
+                           stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
+    all_pics_string = out.communicate()[0].decode('utf-8')
+    all_pics = all_pics_string.split('\n')
+    print(f"Gathered {len(all_pics)} photos.")
+
+    duplicate_map = {}
+    scanned_files = []
+    for pic in all_pics:
+        if not pic:
+            continue
+        name = os.path.basename('.'.join(pic.split('.')[:-1]))
+        if not name:
+            print("WTF? name is empty")
+            continue
+        if pic in scanned_files:
+            continue
+        for other_pic in all_pics:
+            if pic == other_pic:
+                continue
+            if other_pic in scanned_files:
+                continue
+            if os.path.basename(other_pic).startswith(name):
+                if pic not in duplicate_map:
+                    duplicate_map.update({pic: []})
+                    scanned_files.append(pic)
+                duplicate_map[pic].append(other_pic)
+                scanned_files.append(other_pic)
+
+    origs = len(duplicate_map.keys())
+    clones = sum([len(values) for values in duplicate_map.values()])
+    candidates = origs + clones
+    print(f"Found {origs} + {clones} = {candidates} delete candidate.")
+    sys.exit(0)
+
+    for pic, dups in duplicate_map.items():
+        delete_list = []
+        pics = [pic] + dups
+        for pic_a, pic_b in itertools.combinations(pics, 2):
+            if pic_b in delete_list or pic_a in delete_list:
+                continue
+            if not os.path.exists(pic_a):
+                print(f"pic_a doesnt exist: '{pic_a}'")
+                continue
+            if not os.path.exists(pic_b):
+                print(f"pic_b doesnt exist: '{pic_b}'")
+                continue
+            out = subprocess.Popen(['sh', '-c', COMPARE_PICS_CMD.format(pic_a, pic_b)],
+                                   stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
+            value_string, stderr = out.communicate()
+            if out.returncode >= 2:
+                print(f"HERE!!! stdout: {value_string}\nstderr: {stderr}\nreturn code: {out.returncode}")
+                continue
+            #diff = float(value_string.decode('utf-8'))
+            #if diff == 0.:
+            if out.returncode == 0:
+                delete_list.append(pic_b)
+
+        print(f"In list {pics} i will delete")
+        print(f"this pics: {delete_list}")
+        for delete_elem in delete_list:
+            os.remove(delete_elem)
+
+
+if __name__ == '__main__':
+    main()