Merge pull request 'fix: Add filename-based title extraction for TV episodes' (#7) from titlename-fix into dev

Reviewed-on: #7

nfoguard.py

@@ -568,7 +568,7 @@ class TVProcessor:
         for (season, episode), (aired, dateadded, source) in episode_dates.items():
             if (season, episode) in disk_episodes:
                 # Get enhanced episode metadata
-                enhanced_metadata = self._get_episode_metadata(series_metadata, season, episode) if series_metadata else None
+                enhanced_metadata = self._get_episode_metadata(series_metadata, season, episode, season_path) if series_metadata else self._get_episode_metadata(None, season, episode, season_path)
                 
                 # Create NFO
                 if config.manage_nfo:
@@ -659,7 +659,7 @@ class TVProcessor:
         
         # Get enhanced metadata from Sonarr
         series_metadata = self._get_sonarr_series_metadata(imdb_id)
-        enhanced_metadata = self._get_episode_metadata(series_metadata, season_num, episode_num) if series_metadata else None
+        enhanced_metadata = self._get_episode_metadata(series_metadata, season_num, episode_num, season_path) if series_metadata else self._get_episode_metadata(None, season_num, episode_num, season_path)
         
         # Get episode date
         aired, dateadded, source = self._get_single_episode_date(imdb_id, season_num, episode_num, series_metadata)
@@ -739,19 +739,75 @@ class TVProcessor:
             _log("ERROR", f"Error getting Sonarr metadata for {imdb_id}: {e}")
             return None
 
-    def _get_episode_metadata(self, series_metadata: Dict[str, Any], season_num: int, episode_num: int) -> Optional[Dict[str, Any]]:
-        """Extract specific episode metadata from series data"""
-        if not series_metadata or "episodes" not in series_metadata:
-            return None
+    def _get_episode_metadata(self, series_metadata: Dict[str, Any], season_num: int, episode_num: int, season_dir: Optional[Path] = None) -> Optional[Dict[str, Any]]:
+        """Extract specific episode metadata from series data with filename fallback for titles"""
+        metadata = {
+            "title": None,
+            "overview": None,
+            "runtime": None,
+            "ratings": {}
+        }
         
+        # Try to get metadata from Sonarr first
+        if series_metadata and "episodes" in series_metadata:
             for episode in series_metadata["episodes"]:
                 if episode.get("seasonNumber") == season_num and episode.get("episodeNumber") == episode_num:
-                return {
+                    metadata.update({
                         "title": episode.get("title"),
                         "overview": episode.get("overview"),
                         "runtime": episode.get("runtime"),
                         "ratings": episode.get("ratings", {})
-                }
+                    })
+                    break
+        
+        # If no title from Sonarr, try to extract from filename
+        if not metadata["title"] and season_dir:
+            filename_title = self._extract_title_from_filename(season_dir, season_num, episode_num)
+            if filename_title:
+                metadata["title"] = filename_title
+                _log("INFO", f"Using filename-extracted title for S{season_num:02d}E{episode_num:02d}: '{filename_title}'")
+        
+        # Return metadata if we have at least some information
+        if any(metadata.values()):
+            return metadata
+            
+        return None
+
+    def _extract_title_from_filename(self, season_dir: Path, season_num: int, episode_num: int) -> Optional[str]:
+        """Extract episode title from video filename as fallback when Sonarr doesn't provide it"""
+        try:
+            import re
+            # Look for video files matching this episode
+            season_pattern = f"S{season_num:02d}E{episode_num:02d}"
+            
+            for video_file in season_dir.glob("*.mkv"):
+                filename = video_file.name
+                if season_pattern in filename:
+                    # Pattern: SeriesName-S01E01-Episode Title[WEBDL-1080p][AAC2.0][h264].mkv
+                    # Extract the part between season/episode and the first bracket
+                    match = re.search(rf'{season_pattern}-(.*?)\[', filename)
+                    if match:
+                        title = match.group(1).strip()
+                        # Clean up common encoding artifacts and separators
+                        title = title.replace('-', ' ').strip()
+                        if title:
+                            _log("DEBUG", f"Extracted title from filename: '{title}' for {season_pattern}")
+                            return title
+            
+            # Also check .mp4 files
+            for video_file in season_dir.glob("*.mp4"):
+                filename = video_file.name
+                if season_pattern in filename:
+                    match = re.search(rf'{season_pattern}-(.*?)\[', filename)
+                    if match:
+                        title = match.group(1).strip()
+                        title = title.replace('-', ' ').strip()
+                        if title:
+                            _log("DEBUG", f"Extracted title from filename: '{title}' for {season_pattern}")
+                            return title
+                            
+        except Exception as e:
+            _log("DEBUG", f"Error extracting title from filename for S{season_num:02d}E{episode_num:02d}: {e}")
         
         return None
 
@@ -869,7 +925,7 @@ class TVProcessor:
             # Get episode date information - webhook processing prioritizes existing DB entries
             _log("DEBUG", f"Processing webhook episode: IMDb={imdb_id}, S{season_num:02d}E{episode_num:02d}")
             aired, dateadded, source = self._get_webhook_episode_date(imdb_id, season_num, episode_num, series_metadata)
-            enhanced_metadata = self._get_episode_metadata(series_metadata, season_num, episode_num) if series_metadata else None
+            enhanced_metadata = self._get_episode_metadata(series_metadata, season_num, episode_num, season_dir) if series_metadata else self._get_episode_metadata(None, season_num, episode_num, season_dir)
             
             # Create NFO
             if config.manage_nfo: