[scrobblers] Use MediaSourceId to avoid multiple Jellyfin scrobbles

vrobbler/apps/scrobbles/models.py

@@ -582,6 +582,7 @@ class Scrobble(TimeStampedModel):
 
     # Metadata
     source = models.CharField(max_length=255, **BNULL)
+    source_id = models.CharField(max_length=255, **BNULL)
     log = models.JSONField(
         **BNULL,
         default=dict,
@@ -1136,6 +1137,23 @@ class Scrobble(TimeStampedModel):
         skip_in_progress_check = kwargs.get("skip_in_progress_check", False)
         read_log_page = kwargs.get("read_log_page", None)
 
+        source_id = scrobble_data.get("source_id")
+
+        # If source_id is provided (e.g., MediaSourceId from Jellyfin), check for existing scrobble first
+        # This prevents duplicates when webhooks arrive faster than scrobble creation
+        if source_id:
+            existing_by_source_id = cls.objects.filter(
+                media_query,
+                user_id=user_id,
+                source_id=source_id,
+            ).first()
+            if existing_by_source_id:
+                logger.info(
+                    "[create_or_update] found existing scrobble by source_id, updating",
+                    extra={"scrobble_id": existing_by_source_id.id, "source_id": source_id},
+                )
+                scrobble_data["playback_status"] = scrobble_data.pop("status", None)
+                return existing_by_source_id.update(scrobble_data)
 
         # Find our last scrobble of this media item (track, video, etc)
         scrobble = (