f

SYS/metadata.py

@@ -54,6 +54,18 @@ def _extend_namespaced(
             _append_unique(target, seen, f"{namespace}:{val}")
 
 
+def _add_tag(tags: List[str], namespace: str, value: str) -> None:
+    """Add a namespaced tag if not already present."""
+    if not namespace or not value:
+        return
+    normalized_value = value_normalize(value)
+    if not normalized_value:
+        return
+    candidate = f"{namespace}:{normalized_value}"
+    if candidate not in tags:
+        tags.append(candidate)
+
+
 def _coerce_duration(metadata: Dict[str, Any]) -> Optional[float]:
     for key in ("duration", "duration_seconds", "length", "duration_sec"):
         value = metadata.get(key)
@@ -355,6 +367,17 @@ def normalize_urls(value: Any) -> List[str]:
         if not u:
             return None
 
+        # --- HEURISTIC FILTER ---
+        # Ensure it actually looks like a URL/identifier to avoid tag leakage.
+        # This prevents plain tags ("adam22", "10 books") from entering the URL list.
+        low = u.lower()
+        has_scheme = low.startswith((
+            "http://", "https://", "magnet:", "torrent:", "tidal:",
+            "hydrus:", "ytdl:", "soulseek:", "matrix:", "file:"
+        ))
+        if not (has_scheme or "://" in low):
+            return None
+
         # IMPORTANT: URLs can be case-sensitive in the path/query on some hosts
         # (e.g., https://0x0.st/PzGY.webp). Do not lowercase or otherwise rewrite
         # the URL here; preserve exact casing and percent-encoding.