added changes

video_transcription/ai_transcriber_v2/utils.py

@@ -253,22 +253,30 @@ def check_srt_duration_match(source_srt_path, target_srt_path, tolerance_seconds
     """
     Compares the duration of two SRT files to ensure they cover roughly the same timeframe.
     Useful for detecting partial translations.
-    
-    Args:
-        source_srt_path (str): Path to the original language SRT.
-        target_srt_path (str): Path to the translated SRT.
-        tolerance_seconds (float): Max allowed difference in seconds.
-        tolerance_percent (float): Max allowed difference as a percentage of source duration.
-        
-    Returns:
-        tuple: (bool, str) -> (passed, message)
     """
     if not os.path.exists(source_srt_path) or not os.path.exists(target_srt_path):
         return False, "One or both SRT files missing."
 
+    def load_subs_robust(path):
+        # Try detected encoding first
+        detected_enc = detect_file_encoding(path)
+        try:
+            return pysubs2.load(path, encoding=detected_enc)
+        except Exception:
+            pass
+        
+        # Fallback encodings
+        encodings = ['utf-8', 'cp1252', 'latin-1', 'utf-16', 'shift_jis']
+        for enc in encodings:
+            try:
+                return pysubs2.load(path, encoding=enc)
+            except Exception:
+                continue
+        raise Exception(f"Could not decode {os.path.basename(path)} with any standard encoding.")
+
     try:
-        source_subs = pysubs2.load(source_srt_path)
-        target_subs = pysubs2.load(target_srt_path)
+        source_subs = load_subs_robust(source_srt_path)
+        target_subs = load_subs_robust(target_srt_path)
     except Exception as e:
         return False, f"Error parsing SRTs: {e}"