j

SYS/metadata.py

@@ -7,7 +7,7 @@ from SYS.logger import log, debug
 from pathlib import Path
 from typing import Any, Dict, Iterable, List, Optional, Sequence, Set, Tuple
 
-from API.HydrusNetwork import apply_hydrus_tag_mutation, fetch_hydrus_metadata, fetch_hydrus_metadata_by_url
+from SYS.yt_metadata import extract_ytdlp_tags
 
 try:  # Optional; used when available for richer metadata fetches
     import yt_dlp
@@ -918,72 +918,7 @@ def apply_tag_mutation(payload: Dict[str,
             }
 
 
-def extract_ytdlp_tags(entry: Dict[str, Any]) -> List[str]:
-    """ """
-    tags: List[str] = []
-    seen_namespaces: Set[str] = set()
 
-    # Meaningful yt-dlp fields that should become tags
-    # This mapping excludes technical fields: filesize, duration, format_id, vcodec, acodec, ext, etc.
-    field_to_namespace = {
-        "artist": "artist",
-        "album": "album",
-        "creator": "creator",
-        "uploader": "creator",  # Map uploader to creator (deduplicate)
-        "uploader_id": "creator",
-        "channel": "channel",
-        "genre": "genre",
-        "track": "track",
-        "track_number": "track_number",
-        "release_date": "release_date",
-        "upload_date": "upload_date",
-        "title": "title",
-        "license": "license",
-        "location": "location",
-    }
-
-    # Extract simple field mappings
-    for yt_field, namespace in field_to_namespace.items():
-        value = entry.get(yt_field)
-        if value is not None:
-            value_str = value_normalize(str(value))
-            if value_str:
-                # Prevent duplicate creator tags (only use first creator)
-                if namespace == "creator":
-                    if "creator" in seen_namespaces:
-                        continue
-                    seen_namespaces.add("creator")
-
-                _add_tag(tags, namespace, value_str)
-
-    # Handle tags field specially (could be list, dict, or string)
-    # For list/sequence tags, capture as freeform (no namespace prefix)
-    tags_field = entry.get("tags")
-    if tags_field is not None:
-        if isinstance(tags_field, list):
-            # Tags is list: ["tag1", "tag2", ...] → capture as freeform tags (no "tag:" prefix)
-            # These are typically genre/category tags from the source (BandCamp genres, etc.)
-            for tag_value in tags_field:
-                if tag_value:
-                    normalized = value_normalize(str(tag_value))
-                    if normalized and normalized not in tags:
-                        tags.append(normalized)
-        elif isinstance(tags_field, dict):
-            # Tags is dict: {"key": "val"} → tag:key:val
-            for key, val in tags_field.items():
-                if key and val:
-                    key_normalized = value_normalize(str(key))
-                    val_normalized = value_normalize(str(val))
-                    if key_normalized and val_normalized:
-                        _add_tag(tags, f"tag:{key_normalized}", val_normalized)
-        else:
-            # Tags is string or other: add as freeform
-            if tags_field:
-                normalized = value_normalize(str(tags_field))
-                if normalized and normalized not in tags:
-                    tags.append(normalized)
-
-    return tags
 
 
 def dedup_tags_by_namespace(tags: List[str], keep_first: bool = True) -> List[str]: