Merge pull request 'fix: multi-format support, configurable fallback ratio (#12, #13)' (#29) from fix/issue-code-fixes-v2 into main

Reviewed-on: https://git.home.ms/jarianc/EpisodeMatcher/pulls/29

episode_matcher.py

@@ -150,6 +150,17 @@ def main():
         action="store_true",
         help="Confirm destructive operations (required with --auto-delete-duplicates to actually delete files)"
     )
+
+    parser.add_argument(
+        "--extensions",
+        help="Comma-separated video extensions to process (e.g., 'mkv,mp4,avi'). Defaults to all common formats."
+    )
+
+    parser.add_argument(
+        "--fallback-ratio",
+        type=float,
+        help="Minutes of video per GB when MediaInfo unavailable (default: 45). Use ~25 for 4K, ~60 for low-bitrate."
+    )
     
     args = parser.parse_args()
     
@@ -212,8 +223,13 @@ def main():
         tvdb_episodes = []
     
     # Initialize file classifier
+    video_extensions = args.extensions.split(',') if args.extensions else None
     try:
-        classifier = FileClassifier(str(folder_path))
+        classifier = FileClassifier(
+            str(folder_path),
+            video_extensions=video_extensions,
+            fallback_duration_minutes_per_gb=args.fallback_ratio,
+        )
     except FileNotFoundError as e:
         print(f"Error: {e}")
         sys.exit(1)

src/Matcher/file_classifier.py

@@ -15,77 +15,125 @@ except ImportError:
     config_manager = None
 
 
+# Supported video extensions (configurable via config or env var)
+DEFAULT_VIDEO_EXTENSIONS = {'.mkv', '.mp4', '.avi', '.webm', '.mov', '.wmv', '.flv', '.m4v'}
+
+
 class FileClassifier:
     """Classifies video files as episodes or extras based on file size and video duration."""
-    
+
-    def __init__(self, folder_path: str):
+    def __init__(self, folder_path: str, video_extensions: List[str] = None,
-        """Initialize with folder path containing video files."""
+                 fallback_duration_minutes_per_gb: float = None):
+        """Initialize with folder path containing video files.
+
+        Args:
+            folder_path: Path to folder with video files.
+            video_extensions: List of extensions to process (e.g., ['.mkv', '.mp4']).
+                Defaults to all common formats.
+            fallback_duration_minutes_per_gb: Minutes of video per GB when MediaInfo
+                is unavailable. Defaults to 45 (configurable for 4K or low-bitrate content).
+        """
         self.folder_path = Path(folder_path)
+        self.video_extensions = self._resolve_extensions(video_extensions)
+        self.fallback_ratio = self._resolve_fallback_ratio(fallback_duration_minutes_per_gb)
         self.video_files = self._get_video_files()
         self.file_info = self._analyze_files()
-    
+
+    def _resolve_extensions(self, extensions: List[str] = None) -> set:
+        """Resolve the set of video extensions to scan."""
+        if extensions:
+            return {e.lower() if e.startswith('.') else f'.{e.lower()}' for e in extensions}
+        env_ext = os.environ.get('VIDEO_EXTENSIONS', '')
+        if env_ext:
+            return {e.strip().lower() for e in env_ext.split(',') if e.strip()}
+        return DEFAULT_VIDEO_EXTENSIONS
+
+    def _resolve_fallback_ratio(self, ratio: float = None) -> float:
+        """Resolve the fallback minutes/GB ratio for duration estimation."""
+        if ratio is not None:
+            return float(ratio)
+        env_ratio = os.environ.get('FALLBACK_DURATION_RATIO', '')
+        if env_ratio:
+            try:
+                return float(env_ratio)
+            except ValueError:
+                pass
+        if config_manager:
+            return config_manager.get_default_episode_duration()
+        return 45.0
+
     def _get_video_files(self) -> List[Path]:
-        """Get all MKV files in the folder, excluding system files."""
+        """Get all supported video files in the folder, excluding system files."""
         if not self.folder_path.exists():
             raise FileNotFoundError(f"Folder not found: {self.folder_path}")
-        
+
-        all_mkv_files = list(self.folder_path.glob("*.mkv"))
-        
-        # Filter out macOS resource fork files and other system files
         video_files = []
-        for file in all_mkv_files:
+        for ext in self.video_extensions:
+            video_files.extend(self.folder_path.glob(f'*{ext}'))
+
+        # Filter out macOS resource fork files and other system files
+        filtered = []
+        for file in video_files:
             if file.name.startswith('._'):
-                continue  # Skip macOS resource fork files
+                continue
             if file.name.startswith('.'):
-                continue  # Skip any hidden files
+                continue
-            video_files.append(file)
+            filtered.append(file)
-        
+
-        if not video_files:
+        # Deduplicate (in case overlapping extensions)
-            print(f"Warning: No valid MKV files found in {self.folder_path}")
+        seen = set()
-        
+        unique = []
-        return video_files
+        for f in filtered:
+            if f not in seen:
+                seen.add(f)
+                unique.append(f)
+
+        if not unique:
+            ext_list = ', '.join(sorted(self.video_extensions))
+            print(f"Warning: No valid video files found in {self.folder_path} (scanned: {ext_list})")
+
+        return unique
     
     def _get_file_size(self, file_path: Path) -> int:
         """Get file size in bytes."""
         return file_path.stat().st_size
     
     def _get_video_duration(self, file_path: Path) -> float:
-        """Get video duration in minutes using MediaInfo."""
+        """Get video duration in minutes using MediaInfo, with configurable fallback."""
         if MediaInfo is None:
-            print("MediaInfo not available, using file size as proxy for duration")
+            print(f"MediaInfo not available, using file size as proxy for duration "
-            # Rough estimate: 1GB ≈ 45 minutes for typical video
+                  f"(ratio: {self.fallback_ratio:.1f} min/GB)")
             size_gb = self._get_file_size(file_path) / (1024**3)
-            return size_gb * 45
+            return size_gb * self.fallback_ratio
-        
+
         try:
             media_info = MediaInfo.parse(str(file_path))
             for track in media_info.tracks:
                 if track.track_type == 'Video':
                     duration_ms = track.duration
                     if duration_ms:
-                        # Handle both string and numeric duration values
                         if isinstance(duration_ms, str):
                             try:
                                 duration_ms = float(duration_ms)
                             except ValueError:
                                 print(f"Warning: Could not parse duration '{duration_ms}' for {file_path}")
                                 continue
-                        
+
                         duration_minutes = float(duration_ms) / (1000 * 60)
-                        
+
-                        # Sanity check: if duration seems unreasonable, fall back to size estimation
+                        # Sanity check: fall back to size estimation for suspicious durations
-                        if duration_minutes > 300:  # More than 5 hours is likely wrong
+                        if duration_minutes > 300:
-                            print(f"Warning: Suspicious duration ({duration_minutes:.1f} min) for {file_path.name}, using size estimation")
+                            print(f"Warning: Suspicious duration ({duration_minutes:.1f} min) for {file_path.name}, "
+                                  f"using size estimation ({self.fallback_ratio:.1f} min/GB)")
                             size_gb = self._get_file_size(file_path) / (1024**3)
-                            return size_gb * 45
+                            return size_gb * self.fallback_ratio
-                        
+
                         return duration_minutes
         except Exception as e:
             print(f"Error getting duration for {file_path}: {e}")
-        
+
-        # Fallback to size-based estimation
         size_gb = self._get_file_size(file_path) / (1024**3)
-        return size_gb * 45
+        return size_gb * self.fallback_ratio
     
     def _analyze_files(self) -> List[Dict]:
         """Analyze all video files to get size and duration info."""