fixes

main.py

@@ -26,6 +26,7 @@ from PIL import Image
 
 from drupal_lookup import DrupalReferenceLookup, node_url, search_url
 from image_optimizer import create_optimized_candidate
+from image_safety import DEFAULT_MAX_IMAGE_PIXELS, safe_image_open
 
 
 def build_parser():
@@ -75,6 +76,8 @@ def build_parser():
     change.add_argument('--jpeg-quality', type=int, default=92, help='Poczatkowa jakosc JPEG')
     change.add_argument('--min-jpeg-quality', type=int, default=85, help='Minimalna dopuszczalna jakosc JPEG')
     change.add_argument('--min-gain-percent', type=float, default=5.0, help='Minimalne zmniejszenie objetosci w procentach')
+    change.add_argument('--max-image-mpixels', type=float, default=DEFAULT_MAX_IMAGE_PIXELS / 1_000_000,
+                        help='Limit megapikseli oryginalnego obrazu; powyzej tego progu plik bedzie pominiety (ochrona RAM)')
 
     report = parser.add_argument_group('Raporty')
     report.add_argument('--report-csv', help='CSV z plikami i linkami (UTF-8 BOM, otwiera sie w Excelu)')
@@ -93,6 +96,8 @@ def validate_args(args, parser):
         parser.error('--min-gain-percent musi byc w przedziale 0..100')
     if args.max_changes < 0 or args.optimize_top < 0:
         parser.error('--max-changes i --optimize-top nie moga byc ujemne')
+    if not (0 < args.max_image_mpixels <= 500):
+        parser.error('--max-image-mpixels musi byc w zakresie (0, 500]')
     if args.apply and not (args.optimize or args.update_db):
         parser.error('--apply wymaga --optimize lub --update-db')
     if args.apply and args.optimize and not args.backup_dir:
@@ -203,9 +208,12 @@ def sha256_file(path):
     return digest.hexdigest()
 
 
-def file_resolution(path):
+def file_resolution(path, *, max_pixels=DEFAULT_MAX_IMAGE_PIXELS):
     try:
-        with Image.open(path) as im:
+        # W metadanych mozemy odczytac wymiary nawet wtedy, gdy uzytkownik
+        # ustawi niski prog optymalizacji. Nie dekodujemy pikseli.
+        with safe_image_open(path, max_pixels=max(max_pixels, DEFAULT_MAX_IMAGE_PIXELS),
+                             check_pixels=False) as im:
             return im.size
     except Exception:
         return (0, 0)
@@ -257,7 +265,7 @@ def collect_images(connection, config, args):
         if stat_result.st_size < args.min_size_mb * 1024 * 1024:
             continue
         if full_path not in groups:
-            width, height = file_resolution(full_path)
+            width, height = file_resolution(full_path, max_pixels=int(args.max_image_mpixels * 1_000_000))
             groups[full_path] = {
                 'path': full_path,
                 'rel_path': rel_path,
@@ -286,6 +294,7 @@ def collect_images(connection, config, args):
 
 def is_candidate(image, args):
     return (
+        not image['width'] or not image['height'] or
         image['width'] > args.max_width or
         image['height'] > args.max_height or
         image['before_size'] > args.max_file_mb * 1024 * 1024
@@ -465,6 +474,7 @@ def process_changes(connection, config, args, images, table_prefix):
             min_jpeg_quality=args.min_jpeg_quality,
             min_gain_percent=args.min_gain_percent,
             temp_dir=os.path.dirname(image['path']) if args.apply else None,
+            max_pixels=int(args.max_image_mpixels * 1_000_000),
         )
         image['optimization_reason'] = result['reason']
         image['optimize_status'] = result['status']
@@ -475,7 +485,9 @@ def process_changes(connection, config, args, images, table_prefix):
             print(f'[ERROR] {image["rel_path"]}: {result["reason"]}')
             continue
         if result['status'] == 'skip':
-            if args.debug:
+            if 'przekracza --max-image-mpixels' in result['reason'] or 'Pillow odrzucil' in result['reason']:
+                print(f'[SKIP-LIMIT] {image["rel_path"]}: {result["reason"]}')
+            elif args.debug:
                 print(f'[SKIP] {image["rel_path"]}: {result["reason"]}')
             # Opcjonalny sync DB dla plikow, ktorych optymalizacja sie nie udala.
             if args.update_db and mismatched_rows: