hz

ddp.py

@@ -4,11 +4,14 @@ import subprocess
 import itertools
 import os
 import sys
+import threading
 
 
 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"
 
+PICS_MAP_SQUSHED = {}
+
 def main():
     out = subprocess.Popen(['sh', '-c', GET_ALL_PICS_LIST_STRING],
                            stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
@@ -16,38 +19,49 @@ def main():
     all_pics = all_pics_string.split('\n')
     print(f"Gathered {len(all_pics)} photos.")
 
-    duplicate_map = {}
-    scanned_files = []
+    # I. basenames
+    pics_map = {}
     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")
+        key = os.path.basename('.'.join(pic.split('.')[:-1]))
+        if key not in pics_map:
+            pics_map.update({key: [pic]})
             continue
-        if pic in scanned_files:
+        pics_map[key].append(pic)
+    print(f"Hashtable for {len(pics_map)} keys.")
+
+    # II. sort
+    pics_map_keys = sorted(pics_map.keys(), key=len)
+
+    # III. squash
+    pics_map_squashed = {}
+    squash_marker = ['-', ' ', '(']
+    while pics_map_keys:
+        key = pics_map_keys.pop(0)
+        if len(key) < 7:
             continue
-        for other_pic in all_pics:
-            if pic == other_pic:
+        if not key:
             continue
-            if other_pic in scanned_files:
+        squashed_keys = []
+        for other_key in pics_map_keys:
+            if other_key.startswith(key) and other_key[len(key)] in squash_marker:
+                squashed_keys.append(other_key)
+        if not squashed_keys:
             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)
+        pics_map_squashed.update({key: pics_map[key]})
+        for squashed_key in squashed_keys:
+            pics_map_squashed[key] += pics_map[squashed_key]
+            pics_map_keys.remove(squashed_key)
 
-    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)
+    print(f"Squashed hashtable to {len(pics_map_squashed)} keys.")
 
-    for pic, dups in duplicate_map.items():
+    estimated = len(pics_map_squashed)
+    for pics in pics_map_squashed.values():
+        if estimated % 10 == 0:
+            print(f"Estimated: {estimated}")
+        print(f"Do {len(pics)} photos: '{pics}'")
+        estimated -= 1
         delete_list = []
-        pics = [pic] + dups
+        compars_count = 0
         for pic_a, pic_b in itertools.combinations(pics, 2):
             if pic_b in delete_list or pic_a in delete_list:
                 continue
@@ -57,6 +71,7 @@ def main():
             if not os.path.exists(pic_b):
                 print(f"pic_b doesnt exist: '{pic_b}'")
                 continue
+            compars_count += 1
             out = subprocess.Popen(['sh', '-c', COMPARE_PICS_CMD.format(pic_a, pic_b)],
                                    stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
             value_string, stderr = out.communicate()
@@ -68,8 +83,10 @@ def main():
             if out.returncode == 0:
                 delete_list.append(pic_b)
 
-        print(f"In list {pics} i will delete")
-        print(f"this pics: {delete_list}")
+        print(f"Cmps: {compars_count}")
+        if not delete_list:
+            continue
+        print(f"Delete: {delete_list}")
         for delete_elem in delete_list:
             os.remove(delete_elem)